blob: 7ab6e9e5e77c0beda01c3d4bcfb6d95551127b61 [file] [log] [blame]
Mark Fashehccd979b2005-12-15 14:31:24 -08001/* -*- 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"
Tao Ma5a7bc8e2008-08-18 17:38:46 +080067#define OCFS2_XATTR_BLOCK_SIGNATURE "XATTR01"
Mark Fasheh87d35a72008-12-10 17:36:25 -080068#define OCFS2_DIR_TRAILER_SIGNATURE "DIRTRL1"
Mark Fasheh9b7895e2008-11-12 16:27:44 -080069#define OCFS2_DX_ROOT_SIGNATURE "DXDIR01"
70#define OCFS2_DX_LEAF_SIGNATURE "DXLEAF1"
Mark Fashehccd979b2005-12-15 14:31:24 -080071
72/* Compatibility flags */
73#define OCFS2_HAS_COMPAT_FEATURE(sb,mask) \
74 ( OCFS2_SB(sb)->s_feature_compat & (mask) )
75#define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask) \
76 ( OCFS2_SB(sb)->s_feature_ro_compat & (mask) )
77#define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask) \
78 ( OCFS2_SB(sb)->s_feature_incompat & (mask) )
79#define OCFS2_SET_COMPAT_FEATURE(sb,mask) \
80 OCFS2_SB(sb)->s_feature_compat |= (mask)
81#define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask) \
82 OCFS2_SB(sb)->s_feature_ro_compat |= (mask)
83#define OCFS2_SET_INCOMPAT_FEATURE(sb,mask) \
84 OCFS2_SB(sb)->s_feature_incompat |= (mask)
85#define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask) \
86 OCFS2_SB(sb)->s_feature_compat &= ~(mask)
87#define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
88 OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask)
89#define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask) \
90 OCFS2_SB(sb)->s_feature_incompat &= ~(mask)
91
Joel Beckera9772182008-12-16 18:10:18 -080092#define OCFS2_FEATURE_COMPAT_SUPP (OCFS2_FEATURE_COMPAT_BACKUP_SB \
93 | OCFS2_FEATURE_COMPAT_JBD2_SB)
Mark Fashehdcd05382007-01-16 11:32:23 -080094#define OCFS2_FEATURE_INCOMPAT_SUPP (OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \
Mark Fasheh15b1e362007-09-07 13:58:15 -070095 | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \
Joel Becker386a2ef2008-02-01 12:06:54 -080096 | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \
Joel Beckerb61817e2008-02-01 15:08:23 -080097 | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \
Tiger Yang8154da32008-08-18 17:11:46 +080098 | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \
Joel Becker9d28cfb2008-10-16 17:53:29 -070099 | OCFS2_FEATURE_INCOMPAT_XATTR \
Mark Fasheh3a8df2b2008-11-24 17:14:09 -0800100 | OCFS2_FEATURE_INCOMPAT_META_ECC \
101 | OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS)
Jan Kara19ece542008-08-21 20:13:17 +0200102#define OCFS2_FEATURE_RO_COMPAT_SUPP (OCFS2_FEATURE_RO_COMPAT_UNWRITTEN \
103 | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \
104 | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)
Mark Fashehccd979b2005-12-15 14:31:24 -0800105
106/*
107 * Heartbeat-only devices are missing journals and other files. The
108 * filesystem driver can't load them, but the library can. Never put
109 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
110 */
111#define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002
112
Mark Fasheh89039012006-12-07 18:05:37 -0800113/*
114 * tunefs sets this incompat flag before starting the resize and clears it
115 * at the end. This flag protects users from inadvertently mounting the fs
116 * after an aborted run without fsck-ing.
117 */
118#define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG 0x0004
Mark Fashehccd979b2005-12-15 14:31:24 -0800119
Sunil Mushranc271c5c2006-12-05 17:56:35 -0800120/* Used to denote a non-clustered volume */
121#define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT 0x0008
122
Mark Fasheh89039012006-12-07 18:05:37 -0800123/* Support for sparse allocation in b-trees */
124#define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC 0x0010
125
Mark Fashehccd979b2005-12-15 14:31:24 -0800126/*
Mark Fasheh92e91ce2007-09-20 11:19:20 -0700127 * Tunefs sets this incompat flag before starting an operation which
128 * would require cleanup on abort. This is done to protect users from
129 * inadvertently mounting the fs after an aborted run without
130 * fsck-ing.
131 *
132 * s_tunefs_flags on the super block describes precisely which
133 * operations were in progress.
134 */
135#define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG 0x0020
136
Mark Fasheh15b1e362007-09-07 13:58:15 -0700137/* Support for data packed into inode blocks */
138#define OCFS2_FEATURE_INCOMPAT_INLINE_DATA 0x0040
139
Mark Fasheh92e91ce2007-09-20 11:19:20 -0700140/*
Joel Beckerb61817e2008-02-01 15:08:23 -0800141 * Support for alternate, userspace cluster stacks. If set, the superblock
142 * field s_cluster_info contains a tag for the alternate stack in use as
143 * well as the name of the cluster being joined.
144 * mount.ocfs2 must pass in a matching stack name.
145 *
146 * If not set, the classic stack will be used. This is compatbile with
147 * all older versions.
148 */
149#define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK 0x0080
150
Tiger Yang8154da32008-08-18 17:11:46 +0800151/* Support for the extended slot map */
152#define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
153
154/* Support for extended attributes */
155#define OCFS2_FEATURE_INCOMPAT_XATTR 0x0200
156
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800157/* Support for indexed directores */
158#define OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS 0x0400
159
Joel Beckerab552d52008-10-16 17:50:30 -0700160/* Metadata checksum and error correction */
161#define OCFS2_FEATURE_INCOMPAT_META_ECC 0x0800
162
Joel Beckerb61817e2008-02-01 15:08:23 -0800163/*
Mark Fasheh50af94b2007-01-21 14:44:59 -0800164 * backup superblock flag is used to indicate that this volume
165 * has backup superblocks.
166 */
167#define OCFS2_FEATURE_COMPAT_BACKUP_SB 0x0001
168
Mark Fasheh328d5752007-06-18 10:48:04 -0700169/*
Joel Beckera9772182008-12-16 18:10:18 -0800170 * The filesystem will correctly handle journal feature bits.
171 */
172#define OCFS2_FEATURE_COMPAT_JBD2_SB 0x0002
173
174/*
Mark Fasheh328d5752007-06-18 10:48:04 -0700175 * Unwritten extents support.
176 */
177#define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN 0x0001
178
Jan Kara1a224ad2008-08-20 15:43:36 +0200179/*
180 * Maintain quota information for this filesystem
181 */
182#define OCFS2_FEATURE_RO_COMPAT_USRQUOTA 0x0002
183#define OCFS2_FEATURE_RO_COMPAT_GRPQUOTA 0x0004
184
Mark Fasheh50af94b2007-01-21 14:44:59 -0800185/* The byte offset of the first backup block will be 1G.
186 * The following will be 4G, 16G, 64G, 256G and 1T.
187 */
188#define OCFS2_BACKUP_SB_START 1 << 30
189
190/* the max backup superblock nums */
191#define OCFS2_MAX_BACKUP_SUPERBLOCKS 6
192
193/*
Mark Fasheh92e91ce2007-09-20 11:19:20 -0700194 * Flags on ocfs2_super_block.s_tunefs_flags
195 */
196#define OCFS2_TUNEFS_INPROG_REMOVE_SLOT 0x0001 /* Removing slots */
197
198/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800199 * Flags on ocfs2_dinode.i_flags
200 */
201#define OCFS2_VALID_FL (0x00000001) /* Inode is valid */
202#define OCFS2_UNUSED2_FL (0x00000002)
203#define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */
204#define OCFS2_UNUSED3_FL (0x00000008)
205/* System inode flags */
206#define OCFS2_SYSTEM_FL (0x00000010) /* System inode */
207#define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */
208#define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */
209#define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */
210#define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */
211#define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */
212#define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */
213#define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */
Jan Kara1a224ad2008-08-20 15:43:36 +0200214#define OCFS2_QUOTA_FL (0x00001000) /* Quota file */
Mark Fashehccd979b2005-12-15 14:31:24 -0800215
Mark Fasheh15b1e362007-09-07 13:58:15 -0700216/*
217 * Flags on ocfs2_dinode.i_dyn_features
218 *
219 * These can change much more often than i_flags. When adding flags,
220 * keep in mind that i_dyn_features is only 16 bits wide.
221 */
222#define OCFS2_INLINE_DATA_FL (0x0001) /* Data stored in inode block */
223#define OCFS2_HAS_XATTR_FL (0x0002)
224#define OCFS2_INLINE_XATTR_FL (0x0004)
225#define OCFS2_INDEXED_DIR_FL (0x0008)
226
Herbert Poetzlca4d1472006-07-03 17:27:12 -0700227/* Inode attributes, keep in sync with EXT2 */
228#define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
229#define OCFS2_UNRM_FL (0x00000002) /* Undelete */
230#define OCFS2_COMPR_FL (0x00000004) /* Compress file */
231#define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
232#define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
233#define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
234#define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
235#define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
236#define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
237
238#define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
239#define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
240
241/*
Mark Fashehe48edee2007-03-07 16:46:57 -0800242 * Extent record flags (e_node.leaf.flags)
243 */
244#define OCFS2_EXT_UNWRITTEN (0x01) /* Extent is allocated but
245 * unwritten */
246
247/*
Herbert Poetzlca4d1472006-07-03 17:27:12 -0700248 * ioctl commands
249 */
250#define OCFS2_IOC_GETFLAGS _IOR('f', 1, long)
251#define OCFS2_IOC_SETFLAGS _IOW('f', 2, long)
Mark Fasheh586d2322007-03-09 15:56:28 -0800252#define OCFS2_IOC32_GETFLAGS _IOR('f', 1, int)
253#define OCFS2_IOC32_SETFLAGS _IOW('f', 2, int)
Herbert Poetzlca4d1472006-07-03 17:27:12 -0700254
Mark Fashehccd979b2005-12-15 14:31:24 -0800255/*
Mark Fashehb2580102007-03-09 16:53:21 -0800256 * Space reservation / allocation / free ioctls and argument structure
257 * are designed to be compatible with XFS.
258 *
259 * ALLOCSP* and FREESP* are not and will never be supported, but are
260 * included here for completeness.
261 */
262struct ocfs2_space_resv {
263 __s16 l_type;
264 __s16 l_whence;
265 __s64 l_start;
266 __s64 l_len; /* len == 0 means until end of file */
267 __s32 l_sysid;
268 __u32 l_pid;
269 __s32 l_pad[4]; /* reserve area */
270};
271
272#define OCFS2_IOC_ALLOCSP _IOW ('X', 10, struct ocfs2_space_resv)
273#define OCFS2_IOC_FREESP _IOW ('X', 11, struct ocfs2_space_resv)
274#define OCFS2_IOC_RESVSP _IOW ('X', 40, struct ocfs2_space_resv)
275#define OCFS2_IOC_UNRESVSP _IOW ('X', 41, struct ocfs2_space_resv)
276#define OCFS2_IOC_ALLOCSP64 _IOW ('X', 36, struct ocfs2_space_resv)
277#define OCFS2_IOC_FREESP64 _IOW ('X', 37, struct ocfs2_space_resv)
278#define OCFS2_IOC_RESVSP64 _IOW ('X', 42, struct ocfs2_space_resv)
279#define OCFS2_IOC_UNRESVSP64 _IOW ('X', 43, struct ocfs2_space_resv)
280
Tao Ma7909f2b2007-12-18 15:47:25 +0800281/* Used to pass group descriptor data when online resize is done */
282struct ocfs2_new_group_input {
283 __u64 group; /* Group descriptor's blkno. */
284 __u32 clusters; /* Total number of clusters in this group */
285 __u32 frees; /* Total free clusters in this group */
286 __u16 chain; /* Chain for this group */
287 __u16 reserved1;
288 __u32 reserved2;
289};
290
Tao Mad6590722007-12-18 15:47:03 +0800291#define OCFS2_IOC_GROUP_EXTEND _IOW('o', 1, int)
Tao Ma7909f2b2007-12-18 15:47:25 +0800292#define OCFS2_IOC_GROUP_ADD _IOW('o', 2,struct ocfs2_new_group_input)
293#define OCFS2_IOC_GROUP_ADD64 _IOW('o', 3,struct ocfs2_new_group_input)
Tao Mad6590722007-12-18 15:47:03 +0800294
Mark Fashehb2580102007-03-09 16:53:21 -0800295/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800296 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
297 */
298#define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */
299
300/*
301 * superblock s_state flags
302 */
303#define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */
304
305/* Limit of space in ocfs2_dir_entry */
306#define OCFS2_MAX_FILENAME_LEN 255
307
308/* Maximum slots on an ocfs2 file system */
309#define OCFS2_MAX_SLOTS 255
310
311/* Slot map indicator for an empty slot */
312#define OCFS2_INVALID_SLOT -1
313
314#define OCFS2_VOL_UUID_LEN 16
315#define OCFS2_MAX_VOL_LABEL_LEN 64
316
Joel Beckerb61817e2008-02-01 15:08:23 -0800317/* The alternate, userspace stack fields */
318#define OCFS2_STACK_LABEL_LEN 4
319#define OCFS2_CLUSTER_NAME_LEN 16
320
Mark Fashehccd979b2005-12-15 14:31:24 -0800321/* Journal limits (in bytes) */
322#define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
Mark Fashehccd979b2005-12-15 14:31:24 -0800323
Sunil Mushran2fbe8d12007-12-20 14:58:11 -0800324/*
325 * Default local alloc size (in megabytes)
326 *
327 * The value chosen should be such that most allocations, including new
328 * block groups, use local alloc.
329 */
330#define OCFS2_DEFAULT_LOCAL_ALLOC_SIZE 8
331
Tiger Yangfdd77702008-08-18 17:08:55 +0800332/*
333 * Inline extended attribute size (in bytes)
334 * The value chosen should be aligned to 16 byte boundaries.
335 */
336#define OCFS2_MIN_XATTR_INLINE_SIZE 256
337
Mark Fashehccd979b2005-12-15 14:31:24 -0800338struct ocfs2_system_inode_info {
339 char *si_name;
340 int si_iflags;
341 int si_mode;
342};
343
344/* System file index */
345enum {
346 BAD_BLOCK_SYSTEM_INODE = 0,
347 GLOBAL_INODE_ALLOC_SYSTEM_INODE,
348 SLOT_MAP_SYSTEM_INODE,
349#define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
350 HEARTBEAT_SYSTEM_INODE,
351 GLOBAL_BITMAP_SYSTEM_INODE,
Jan Kara1a224ad2008-08-20 15:43:36 +0200352 USER_QUOTA_SYSTEM_INODE,
353 GROUP_QUOTA_SYSTEM_INODE,
354#define OCFS2_LAST_GLOBAL_SYSTEM_INODE GROUP_QUOTA_SYSTEM_INODE
Mark Fashehccd979b2005-12-15 14:31:24 -0800355 ORPHAN_DIR_SYSTEM_INODE,
356 EXTENT_ALLOC_SYSTEM_INODE,
357 INODE_ALLOC_SYSTEM_INODE,
358 JOURNAL_SYSTEM_INODE,
359 LOCAL_ALLOC_SYSTEM_INODE,
360 TRUNCATE_LOG_SYSTEM_INODE,
Jan Kara1a224ad2008-08-20 15:43:36 +0200361 LOCAL_USER_QUOTA_SYSTEM_INODE,
362 LOCAL_GROUP_QUOTA_SYSTEM_INODE,
Mark Fashehccd979b2005-12-15 14:31:24 -0800363 NUM_SYSTEM_INODES
364};
365
366static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
367 /* Global system inodes (single copy) */
368 /* The first two are only used from userspace mfks/tunefs */
369 [BAD_BLOCK_SYSTEM_INODE] = { "bad_blocks", 0, S_IFREG | 0644 },
370 [GLOBAL_INODE_ALLOC_SYSTEM_INODE] = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
371
372 /* These are used by the running filesystem */
373 [SLOT_MAP_SYSTEM_INODE] = { "slot_map", 0, S_IFREG | 0644 },
374 [HEARTBEAT_SYSTEM_INODE] = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
375 [GLOBAL_BITMAP_SYSTEM_INODE] = { "global_bitmap", 0, S_IFREG | 0644 },
Jan Kara1a224ad2008-08-20 15:43:36 +0200376 [USER_QUOTA_SYSTEM_INODE] = { "aquota.user", OCFS2_QUOTA_FL, S_IFREG | 0644 },
377 [GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group", OCFS2_QUOTA_FL, S_IFREG | 0644 },
Mark Fashehccd979b2005-12-15 14:31:24 -0800378
379 /* Slot-specific system inodes (one copy per slot) */
380 [ORPHAN_DIR_SYSTEM_INODE] = { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
381 [EXTENT_ALLOC_SYSTEM_INODE] = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
382 [INODE_ALLOC_SYSTEM_INODE] = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
383 [JOURNAL_SYSTEM_INODE] = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
384 [LOCAL_ALLOC_SYSTEM_INODE] = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
Jan Kara1a224ad2008-08-20 15:43:36 +0200385 [TRUNCATE_LOG_SYSTEM_INODE] = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 },
386 [LOCAL_USER_QUOTA_SYSTEM_INODE] = { "aquota.user:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
387 [LOCAL_GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
Mark Fashehccd979b2005-12-15 14:31:24 -0800388};
389
390/* Parameter passed from mount.ocfs2 to module */
391#define OCFS2_HB_NONE "heartbeat=none"
392#define OCFS2_HB_LOCAL "heartbeat=local"
393
394/*
395 * OCFS2 directory file types. Only the low 3 bits are used. The
396 * other bits are reserved for now.
397 */
398#define OCFS2_FT_UNKNOWN 0
399#define OCFS2_FT_REG_FILE 1
400#define OCFS2_FT_DIR 2
401#define OCFS2_FT_CHRDEV 3
402#define OCFS2_FT_BLKDEV 4
403#define OCFS2_FT_FIFO 5
404#define OCFS2_FT_SOCK 6
405#define OCFS2_FT_SYMLINK 7
406
407#define OCFS2_FT_MAX 8
408
409/*
410 * OCFS2_DIR_PAD defines the directory entries boundaries
411 *
412 * NOTE: It must be a multiple of 4
413 */
414#define OCFS2_DIR_PAD 4
415#define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
416#define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
417#define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
418 OCFS2_DIR_ROUND) & \
419 ~OCFS2_DIR_ROUND)
Mark Fashehe7c17e42009-01-29 18:17:46 -0800420#define OCFS2_DIR_MIN_REC_LEN OCFS2_DIR_REC_LEN(1)
Mark Fashehccd979b2005-12-15 14:31:24 -0800421
422#define OCFS2_LINK_MAX 32000
Mark Fasheh198a1ca2008-11-20 17:54:57 -0800423#define OCFS2_DX_LINK_MAX ((1U << 31) - 1U)
424#define OCFS2_LINKS_HI_SHIFT 16
Mark Fashehe3a93c22009-02-17 15:29:35 -0800425#define OCFS2_DX_ENTRIES_MAX (0xffffffffU)
Mark Fashehccd979b2005-12-15 14:31:24 -0800426
427#define S_SHIFT 12
428static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = {
429 [S_IFREG >> S_SHIFT] = OCFS2_FT_REG_FILE,
430 [S_IFDIR >> S_SHIFT] = OCFS2_FT_DIR,
431 [S_IFCHR >> S_SHIFT] = OCFS2_FT_CHRDEV,
432 [S_IFBLK >> S_SHIFT] = OCFS2_FT_BLKDEV,
433 [S_IFIFO >> S_SHIFT] = OCFS2_FT_FIFO,
434 [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK,
435 [S_IFLNK >> S_SHIFT] = OCFS2_FT_SYMLINK,
436};
437
438
439/*
440 * Convenience casts
441 */
442#define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
443
444/*
Joel Beckerab552d52008-10-16 17:50:30 -0700445 * Block checking structure. This is used in metadata to validate the
446 * contents. If OCFS2_FEATURE_INCOMPAT_META_ECC is not set, it is all
447 * zeros.
448 */
449struct ocfs2_block_check {
450/*00*/ __le32 bc_crc32e; /* 802.3 Ethernet II CRC32 */
451 __le16 bc_ecc; /* Single-error-correction parity vector.
452 This is a simple Hamming code dependant
453 on the blocksize. OCFS2's maximum
454 blocksize, 4K, requires 16 parity bits,
455 so we fit in __le16. */
456 __le16 bc_reserved1;
457/*08*/
458};
459
460/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800461 * On disk extent record for OCFS2
462 * It describes a range of clusters on disk.
Mark Fashehe48edee2007-03-07 16:46:57 -0800463 *
464 * Length fields are divided into interior and leaf node versions.
465 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
Mark Fashehccd979b2005-12-15 14:31:24 -0800466 */
467struct ocfs2_extent_rec {
468/*00*/ __le32 e_cpos; /* Offset into the file, in clusters */
Mark Fashehe48edee2007-03-07 16:46:57 -0800469 union {
470 __le32 e_int_clusters; /* Clusters covered by all children */
471 struct {
472 __le16 e_leaf_clusters; /* Clusters covered by this
473 extent */
474 __u8 e_reserved1;
475 __u8 e_flags; /* Extent flags */
476 };
477 };
Mark Fashehccd979b2005-12-15 14:31:24 -0800478 __le64 e_blkno; /* Physical disk offset, in blocks */
479/*10*/
480};
481
482struct ocfs2_chain_rec {
483 __le32 c_free; /* Number of free bits in this chain. */
484 __le32 c_total; /* Number of total bits in this chain */
485 __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */
486};
487
488struct ocfs2_truncate_rec {
489 __le32 t_start; /* 1st cluster in this log */
490 __le32 t_clusters; /* Number of total clusters covered */
491};
492
493/*
494 * On disk extent list for OCFS2 (node in the tree). Note that this
495 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
496 * offsets are relative to ocfs2_dinode.id2.i_list or
497 * ocfs2_extent_block.h_list, respectively.
498 */
499struct ocfs2_extent_list {
500/*00*/ __le16 l_tree_depth; /* Extent tree depth from this
501 point. 0 means data extents
502 hang directly off this
Mark Fashehdcd05382007-01-16 11:32:23 -0800503 header (a leaf)
504 NOTE: The high 8 bits cannot be
505 used - tree_depth is never that big.
506 */
Mark Fashehccd979b2005-12-15 14:31:24 -0800507 __le16 l_count; /* Number of extent records */
508 __le16 l_next_free_rec; /* Next unused extent slot */
509 __le16 l_reserved1;
510 __le64 l_reserved2; /* Pad to
511 sizeof(ocfs2_extent_rec) */
512/*10*/ struct ocfs2_extent_rec l_recs[0]; /* Extent records */
513};
514
515/*
516 * On disk allocation chain list for OCFS2. Note that this is
517 * contained inside ocfs2_dinode, so the offsets are relative to
518 * ocfs2_dinode.id2.i_chain.
519 */
520struct ocfs2_chain_list {
521/*00*/ __le16 cl_cpg; /* Clusters per Block Group */
522 __le16 cl_bpc; /* Bits per cluster */
523 __le16 cl_count; /* Total chains in this list */
524 __le16 cl_next_free_rec; /* Next unused chain slot */
525 __le64 cl_reserved1;
526/*10*/ struct ocfs2_chain_rec cl_recs[0]; /* Chain records */
527};
528
529/*
530 * On disk deallocation log for OCFS2. Note that this is
531 * contained inside ocfs2_dinode, so the offsets are relative to
532 * ocfs2_dinode.id2.i_dealloc.
533 */
534struct ocfs2_truncate_log {
535/*00*/ __le16 tl_count; /* Total records in this log */
536 __le16 tl_used; /* Number of records in use */
537 __le32 tl_reserved1;
538/*08*/ struct ocfs2_truncate_rec tl_recs[0]; /* Truncate records */
539};
540
541/*
542 * On disk extent block (indirect block) for OCFS2
543 */
544struct ocfs2_extent_block
545{
546/*00*/ __u8 h_signature[8]; /* Signature for verification */
Joel Beckerab552d52008-10-16 17:50:30 -0700547 struct ocfs2_block_check h_check; /* Error checking */
Mark Fashehccd979b2005-12-15 14:31:24 -0800548/*10*/ __le16 h_suballoc_slot; /* Slot suballocator this
549 extent_header belongs to */
550 __le16 h_suballoc_bit; /* Bit offset in suballocator
551 block group */
552 __le32 h_fs_generation; /* Must match super block */
553 __le64 h_blkno; /* Offset on disk, in blocks */
554/*20*/ __le64 h_reserved3;
555 __le64 h_next_leaf_blk; /* Offset on disk, in blocks,
556 of next leaf header pointing
557 to data */
558/*30*/ struct ocfs2_extent_list h_list; /* Extent record list */
559/* Actual on-disk size is one block */
560};
561
562/*
Joel Beckerfb86b1f2008-02-01 11:59:05 -0800563 * On disk slot map for OCFS2. This defines the contents of the "slot_map"
Joel Becker386a2ef2008-02-01 12:06:54 -0800564 * system file. A slot is valid if it contains a node number >= 0. The
565 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty.
Joel Beckerfb86b1f2008-02-01 11:59:05 -0800566 */
567struct ocfs2_slot_map {
568/*00*/ __le16 sm_slots[0];
569/*
570 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255,
571 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
572 */
573};
574
Joel Becker386a2ef2008-02-01 12:06:54 -0800575struct ocfs2_extended_slot {
576/*00*/ __u8 es_valid;
577 __u8 es_reserved1[3];
578 __le32 es_node_num;
579/*10*/
580};
581
582/*
583 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
584 * is set. It separates out the valid marker from the node number, and
585 * has room to grow. Unlike the old slot map, this format is defined by
586 * i_size.
587 */
588struct ocfs2_slot_map_extended {
589/*00*/ struct ocfs2_extended_slot se_slots[0];
590/*
591 * Actual size is i_size of the slot_map system file. It should
592 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
593 */
594};
595
Joel Beckerb61817e2008-02-01 15:08:23 -0800596struct ocfs2_cluster_info {
597/*00*/ __u8 ci_stack[OCFS2_STACK_LABEL_LEN];
598 __le32 ci_reserved;
599/*08*/ __u8 ci_cluster[OCFS2_CLUSTER_NAME_LEN];
600/*18*/
601};
602
Joel Beckerfb86b1f2008-02-01 11:59:05 -0800603/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800604 * On disk superblock for OCFS2
605 * Note that it is contained inside an ocfs2_dinode, so all offsets
606 * are relative to the start of ocfs2_dinode.id2.
607 */
608struct ocfs2_super_block {
609/*00*/ __le16 s_major_rev_level;
610 __le16 s_minor_rev_level;
611 __le16 s_mnt_count;
612 __le16 s_max_mnt_count;
613 __le16 s_state; /* File system state */
614 __le16 s_errors; /* Behaviour when detecting errors */
615 __le32 s_checkinterval; /* Max time between checks */
616/*10*/ __le64 s_lastcheck; /* Time of last check */
617 __le32 s_creator_os; /* OS */
618 __le32 s_feature_compat; /* Compatible feature set */
619/*20*/ __le32 s_feature_incompat; /* Incompatible feature set */
620 __le32 s_feature_ro_compat; /* Readonly-compatible feature set */
621 __le64 s_root_blkno; /* Offset, in blocks, of root directory
622 dinode */
623/*30*/ __le64 s_system_dir_blkno; /* Offset, in blocks, of system
624 directory dinode */
625 __le32 s_blocksize_bits; /* Blocksize for this fs */
626 __le32 s_clustersize_bits; /* Clustersize for this fs */
627/*40*/ __le16 s_max_slots; /* Max number of simultaneous mounts
628 before tunefs required */
Mark Fasheh92e91ce2007-09-20 11:19:20 -0700629 __le16 s_tunefs_flag;
Tiger Yangcf1d6c72008-08-18 17:11:00 +0800630 __le32 s_uuid_hash; /* hash value of uuid */
Mark Fashehccd979b2005-12-15 14:31:24 -0800631 __le64 s_first_cluster_group; /* Block offset of 1st cluster
632 * group header */
633/*50*/ __u8 s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */
634/*90*/ __u8 s_uuid[OCFS2_VOL_UUID_LEN]; /* 128-bit uuid */
Joel Beckerb61817e2008-02-01 15:08:23 -0800635/*A0*/ struct ocfs2_cluster_info s_cluster_info; /* Selected userspace
636 stack. Only valid
637 with INCOMPAT flag. */
Tiger Yang8154da32008-08-18 17:11:46 +0800638/*B8*/ __le16 s_xattr_inline_size; /* extended attribute inline size
639 for this fs*/
640 __le16 s_reserved0;
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800641 __le32 s_dx_seed[3]; /* seed[0-2] for dx dir hash.
642 * s_uuid_hash serves as seed[3]. */
643/*C0*/ __le64 s_reserved2[15]; /* Fill out superblock */
Joel Beckerb61817e2008-02-01 15:08:23 -0800644/*140*/
645
646 /*
647 * NOTE: As stated above, all offsets are relative to
648 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
649 * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within
650 * our smallest blocksize, which is 512 bytes. To ensure this,
651 * we reserve the space in s_reserved2. Anything past s_reserved2
652 * will not be available on the smallest blocksize.
653 */
Mark Fashehccd979b2005-12-15 14:31:24 -0800654};
655
656/*
657 * Local allocation bitmap for OCFS2 slots
658 * Note that it exists inside an ocfs2_dinode, so all offsets are
659 * relative to the start of ocfs2_dinode.id2.
660 */
661struct ocfs2_local_alloc
662{
663/*00*/ __le32 la_bm_off; /* Starting bit offset in main bitmap */
664 __le16 la_size; /* Size of included bitmap, in bytes */
665 __le16 la_reserved1;
666 __le64 la_reserved2;
667/*10*/ __u8 la_bitmap[0];
668};
669
670/*
Mark Fasheh15b1e362007-09-07 13:58:15 -0700671 * Data-in-inode header. This is only used if i_dyn_features has
672 * OCFS2_INLINE_DATA_FL set.
673 */
674struct ocfs2_inline_data
675{
676/*00*/ __le16 id_count; /* Number of bytes that can be used
677 * for data, starting at id_data */
678 __le16 id_reserved0;
679 __le32 id_reserved1;
680 __u8 id_data[0]; /* Start of user data */
681};
682
683/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800684 * On disk inode for OCFS2
685 */
686struct ocfs2_dinode {
687/*00*/ __u8 i_signature[8]; /* Signature for validation */
688 __le32 i_generation; /* Generation number */
689 __le16 i_suballoc_slot; /* Slot suballocator this inode
690 belongs to */
691 __le16 i_suballoc_bit; /* Bit offset in suballocator
692 block group */
Mark Fasheh198a1ca2008-11-20 17:54:57 -0800693/*10*/ __le16 i_links_count_hi; /* High 16 bits of links count */
Tiger Yangfdd77702008-08-18 17:08:55 +0800694 __le16 i_xattr_inline_size;
Mark Fashehccd979b2005-12-15 14:31:24 -0800695 __le32 i_clusters; /* Cluster count */
696 __le32 i_uid; /* Owner UID */
697 __le32 i_gid; /* Owning GID */
698/*20*/ __le64 i_size; /* Size in bytes */
699 __le16 i_mode; /* File mode */
700 __le16 i_links_count; /* Links count */
701 __le32 i_flags; /* File flags */
702/*30*/ __le64 i_atime; /* Access time */
703 __le64 i_ctime; /* Creation time */
704/*40*/ __le64 i_mtime; /* Modification time */
705 __le64 i_dtime; /* Deletion time */
706/*50*/ __le64 i_blkno; /* Offset on disk, in blocks */
707 __le64 i_last_eb_blk; /* Pointer to last extent
708 block */
709/*60*/ __le32 i_fs_generation; /* Generation per fs-instance */
710 __le32 i_atime_nsec;
711 __le32 i_ctime_nsec;
712 __le32 i_mtime_nsec;
Tiger Yangfdd77702008-08-18 17:08:55 +0800713/*70*/ __le32 i_attr;
Tiger Yang50008632007-03-20 16:01:38 -0700714 __le16 i_orphaned_slot; /* Only valid when OCFS2_ORPHANED_FL
715 was set in i_flags */
Mark Fasheh15b1e362007-09-07 13:58:15 -0700716 __le16 i_dyn_features;
Tiger Yangfdd77702008-08-18 17:08:55 +0800717 __le64 i_xattr_loc;
Joel Beckerab552d52008-10-16 17:50:30 -0700718/*80*/ struct ocfs2_block_check i_check; /* Error checking */
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800719/*88*/ __le64 i_dx_root; /* Pointer to dir index root block */
720 __le64 i_reserved2[5];
Mark Fashehccd979b2005-12-15 14:31:24 -0800721/*B8*/ union {
722 __le64 i_pad1; /* Generic way to refer to this
723 64bit union */
724 struct {
725 __le64 i_rdev; /* Device number */
726 } dev1;
727 struct { /* Info for bitmap system
728 inodes */
729 __le32 i_used; /* Bits (ie, clusters) used */
730 __le32 i_total; /* Total bits (clusters)
731 available */
732 } bitmap1;
733 struct { /* Info for journal system
734 inodes */
735 __le32 ij_flags; /* Mounted, version, etc. */
Sunil Mushranc69991a2008-07-14 17:31:09 -0700736 __le32 ij_recovery_generation; /* Incremented when the
737 journal is recovered
738 after an unclean
739 shutdown */
Mark Fashehccd979b2005-12-15 14:31:24 -0800740 } journal1;
741 } id1; /* Inode type dependant 1 */
742/*C0*/ union {
743 struct ocfs2_super_block i_super;
744 struct ocfs2_local_alloc i_lab;
745 struct ocfs2_chain_list i_chain;
746 struct ocfs2_extent_list i_list;
747 struct ocfs2_truncate_log i_dealloc;
Mark Fasheh15b1e362007-09-07 13:58:15 -0700748 struct ocfs2_inline_data i_data;
Mark Fashehccd979b2005-12-15 14:31:24 -0800749 __u8 i_symlink[0];
750 } id2;
751/* Actual on-disk size is one block */
752};
753
754/*
755 * On-disk directory entry structure for OCFS2
756 *
757 * Packed as this structure could be accessed unaligned on 64-bit platforms
758 */
759struct ocfs2_dir_entry {
760/*00*/ __le64 inode; /* Inode number */
761 __le16 rec_len; /* Directory entry length */
762 __u8 name_len; /* Name length */
763 __u8 file_type;
764/*0C*/ char name[OCFS2_MAX_FILENAME_LEN]; /* File name */
765/* Actual on-disk length specified by rec_len */
766} __attribute__ ((packed));
767
768/*
Mark Fasheh87d35a72008-12-10 17:36:25 -0800769 * Per-block record for the unindexed directory btree. This is carefully
770 * crafted so that the rec_len and name_len records of an ocfs2_dir_entry are
771 * mirrored. That way, the directory manipulation code needs a minimal amount
772 * of update.
773 *
774 * NOTE: Keep this structure aligned to a multiple of 4 bytes.
775 */
776struct ocfs2_dir_block_trailer {
777/*00*/ __le64 db_compat_inode; /* Always zero. Was inode */
778
779 __le16 db_compat_rec_len; /* Backwards compatible with
780 * ocfs2_dir_entry. */
781 __u8 db_compat_name_len; /* Always zero. Was name_len */
782 __u8 db_reserved0;
783 __le16 db_reserved1;
784 __le16 db_free_rec_len; /* Size of largest empty hole
785 * in this block. (unused) */
786/*10*/ __u8 db_signature[8]; /* Signature for verification */
787 __le64 db_reserved2;
788 __le64 db_free_next; /* Next block in list (unused) */
789/*20*/ __le64 db_blkno; /* Offset on disk, in blocks */
790 __le64 db_parent_dinode; /* dinode which owns me, in
791 blocks */
Joel Beckerc175a512008-12-10 17:58:22 -0800792/*30*/ struct ocfs2_block_check db_check; /* Error checking */
Mark Fasheh87d35a72008-12-10 17:36:25 -0800793/*40*/
794};
795
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800796 /*
797 * A directory entry in the indexed tree. We don't store the full name here,
798 * but instead provide a pointer to the full dirent in the unindexed tree.
799 *
800 * We also store name_len here so as to reduce the number of leaf blocks we
801 * need to search in case of collisions.
802 */
803struct ocfs2_dx_entry {
804 __le32 dx_major_hash; /* Used to find logical
805 * cluster in index */
806 __le32 dx_minor_hash; /* Lower bits used to find
807 * block in cluster */
808 __le64 dx_dirent_blk; /* Physical block in unindexed
809 * tree holding this dirent. */
810};
811
812struct ocfs2_dx_entry_list {
813 __le32 de_reserved;
814 __le16 de_count; /* Maximum number of entries
815 * possible in de_entries */
816 __le16 de_num_used; /* Current number of
817 * de_entries entries */
818 struct ocfs2_dx_entry de_entries[0]; /* Indexed dir entries
819 * in a packed array of
820 * length de_num_used */
821};
822
Mark Fasheh4ed8a6b2008-11-24 17:02:08 -0800823#define OCFS2_DX_FLAG_INLINE 0x01
824
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800825/*
826 * A directory indexing block. Each indexed directory has one of these,
827 * pointed to by ocfs2_dinode.
828 *
829 * This block stores an indexed btree root, and a set of free space
830 * start-of-list pointers.
831 */
832struct ocfs2_dx_root_block {
833 __u8 dr_signature[8]; /* Signature for verification */
834 struct ocfs2_block_check dr_check; /* Error checking */
835 __le16 dr_suballoc_slot; /* Slot suballocator this
836 * block belongs to. */
837 __le16 dr_suballoc_bit; /* Bit offset in suballocator
838 * block group */
839 __le32 dr_fs_generation; /* Must match super block */
840 __le64 dr_blkno; /* Offset on disk, in blocks */
841 __le64 dr_last_eb_blk; /* Pointer to last
842 * extent block */
843 __le32 dr_clusters; /* Clusters allocated
844 * to the indexed tree. */
Mark Fasheh4ed8a6b2008-11-24 17:02:08 -0800845 __u8 dr_flags; /* OCFS2_DX_FLAG_* flags */
846 __u8 dr_reserved0;
847 __le16 dr_reserved1;
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800848 __le64 dr_dir_blkno; /* Pointer to parent inode */
Mark Fashehe3a93c22009-02-17 15:29:35 -0800849 __le32 dr_num_entries; /* Total number of
850 * names stored in
851 * this directory.*/
852 __le32 dr_reserved2;
Mark Fashehe7c17e42009-01-29 18:17:46 -0800853 __le64 dr_free_blk; /* Pointer to head of free
854 * unindexed block list. */
855 __le64 dr_reserved3[15];
Mark Fasheh4ed8a6b2008-11-24 17:02:08 -0800856 union {
857 struct ocfs2_extent_list dr_list; /* Keep this aligned to 128
858 * bits for maximum space
859 * efficiency. */
860 struct ocfs2_dx_entry_list dr_entries; /* In-root-block list of
861 * entries. We grow out
862 * to extents if this
863 * gets too big. */
864 };
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800865};
866
867/*
868 * The header of a leaf block in the indexed tree.
869 */
870struct ocfs2_dx_leaf {
871 __u8 dl_signature[8];/* Signature for verification */
872 struct ocfs2_block_check dl_check; /* Error checking */
873 __le64 dl_blkno; /* Offset on disk, in blocks */
874 __le32 dl_fs_generation;/* Must match super block */
875 __le32 dl_reserved0;
876 __le64 dl_reserved1;
877 struct ocfs2_dx_entry_list dl_list;
878};
879
Mark Fasheh87d35a72008-12-10 17:36:25 -0800880/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800881 * On disk allocator group structure for OCFS2
882 */
883struct ocfs2_group_desc
884{
885/*00*/ __u8 bg_signature[8]; /* Signature for validation */
886 __le16 bg_size; /* Size of included bitmap in
887 bytes. */
888 __le16 bg_bits; /* Bits represented by this
889 group. */
890 __le16 bg_free_bits_count; /* Free bits count */
891 __le16 bg_chain; /* What chain I am in. */
892/*10*/ __le32 bg_generation;
893 __le32 bg_reserved1;
894 __le64 bg_next_group; /* Next group in my list, in
895 blocks */
896/*20*/ __le64 bg_parent_dinode; /* dinode which owns me, in
897 blocks */
898 __le64 bg_blkno; /* Offset on disk, in blocks */
Joel Beckerab552d52008-10-16 17:50:30 -0700899/*30*/ struct ocfs2_block_check bg_check; /* Error checking */
900 __le64 bg_reserved2;
Mark Fashehccd979b2005-12-15 14:31:24 -0800901/*40*/ __u8 bg_bitmap[0];
902};
903
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800904/*
905 * On disk extended attribute structure for OCFS2.
906 */
907
908/*
909 * ocfs2_xattr_entry indicates one extend attribute.
910 *
911 * Note that it can be stored in inode, one block or one xattr bucket.
912 */
913struct ocfs2_xattr_entry {
914 __le32 xe_name_hash; /* hash value of xattr prefix+suffix. */
Tao Ma8573f792008-10-24 22:24:17 +0800915 __le16 xe_name_offset; /* byte offset from the 1st entry in the
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800916 local xattr storage(inode, xattr block or
917 xattr bucket). */
918 __u8 xe_name_len; /* xattr name len, does't include prefix. */
Tao Ma8573f792008-10-24 22:24:17 +0800919 __u8 xe_type; /* the low 7 bits indicate the name prefix
920 * type and the highest bit indicates whether
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800921 * the EA is stored in the local storage. */
922 __le64 xe_value_size; /* real xattr value length. */
923};
924
925/*
926 * On disk structure for xattr header.
927 *
928 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
929 * the local xattr storage.
930 */
931struct ocfs2_xattr_header {
932 __le16 xh_count; /* contains the count of how
933 many records are in the
934 local xattr storage. */
Tao Ma0c044f02008-08-18 17:38:50 +0800935 __le16 xh_free_start; /* current offset for storing
936 xattr. */
937 __le16 xh_name_value_len; /* total length of name/value
938 length in this bucket. */
Tao Ma8573f792008-10-24 22:24:17 +0800939 __le16 xh_num_buckets; /* Number of xattr buckets
940 in this extent record,
941 only valid in the first
942 bucket. */
Joel Beckerab552d52008-10-16 17:50:30 -0700943 struct ocfs2_block_check xh_check; /* Error checking
944 (Note, this is only
945 used for xattr
946 buckets. A block uses
947 xb_check and sets
948 this field to zero.) */
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800949 struct ocfs2_xattr_entry xh_entries[0]; /* xattr entry list. */
950};
951
952/*
953 * On disk structure for xattr value root.
954 *
Tao Ma8573f792008-10-24 22:24:17 +0800955 * When an xattr's value is large enough, it is stored in an external
956 * b-tree like file data. The xattr value root points to this structure.
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800957 */
958struct ocfs2_xattr_value_root {
959/*00*/ __le32 xr_clusters; /* clusters covered by xattr value. */
960 __le32 xr_reserved0;
961 __le64 xr_last_eb_blk; /* Pointer to last extent block */
962/*10*/ struct ocfs2_extent_list xr_list; /* Extent record list */
963};
964
965/*
966 * On disk structure for xattr tree root.
967 *
968 * It is used when there are too many extended attributes for one file. These
969 * attributes will be organized and stored in an indexed-btree.
970 */
971struct ocfs2_xattr_tree_root {
972/*00*/ __le32 xt_clusters; /* clusters covered by xattr. */
973 __le32 xt_reserved0;
974 __le64 xt_last_eb_blk; /* Pointer to last extent block */
975/*10*/ struct ocfs2_extent_list xt_list; /* Extent record list */
976};
977
Tiger Yangcf1d6c72008-08-18 17:11:00 +0800978#define OCFS2_XATTR_INDEXED 0x1
979#define OCFS2_HASH_SHIFT 5
980#define OCFS2_XATTR_ROUND 3
981#define OCFS2_XATTR_SIZE(size) (((size) + OCFS2_XATTR_ROUND) & \
982 ~(OCFS2_XATTR_ROUND))
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800983
Tao Ma0c044f02008-08-18 17:38:50 +0800984#define OCFS2_XATTR_BUCKET_SIZE 4096
985#define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET (OCFS2_XATTR_BUCKET_SIZE \
986 / OCFS2_MIN_BLOCKSIZE)
987
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800988/*
989 * On disk structure for xattr block.
990 */
991struct ocfs2_xattr_block {
992/*00*/ __u8 xb_signature[8]; /* Signature for verification */
993 __le16 xb_suballoc_slot; /* Slot suballocator this
994 block belongs to. */
995 __le16 xb_suballoc_bit; /* Bit offset in suballocator
996 block group */
997 __le32 xb_fs_generation; /* Must match super block */
998/*10*/ __le64 xb_blkno; /* Offset on disk, in blocks */
Joel Beckerab552d52008-10-16 17:50:30 -0700999 struct ocfs2_block_check xb_check; /* Error checking */
Tao Ma5a7bc8e2008-08-18 17:38:46 +08001000/*20*/ __le16 xb_flags; /* Indicates whether this block contains
1001 real xattr or a xattr tree. */
1002 __le16 xb_reserved0;
1003 __le32 xb_reserved1;
1004 __le64 xb_reserved2;
1005/*30*/ union {
1006 struct ocfs2_xattr_header xb_header; /* xattr header if this
1007 block contains xattr */
1008 struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
1009 block cotains xattr
1010 tree. */
1011 } xb_attrs;
1012};
1013
1014#define OCFS2_XATTR_ENTRY_LOCAL 0x80
1015#define OCFS2_XATTR_TYPE_MASK 0x7F
1016static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
1017 int local)
1018{
1019 if (local)
1020 xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
1021 else
1022 xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
1023}
1024
1025static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
1026{
1027 return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
1028}
1029
1030static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
1031{
1032 xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
1033}
1034
1035static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
1036{
1037 return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
1038}
1039
Jan Kara9e33d692008-08-25 19:56:50 +02001040/*
1041 * On disk structures for global quota file
1042 */
1043
1044/* Magic numbers and known versions for global quota files */
1045#define OCFS2_GLOBAL_QMAGICS {\
1046 0x0cf52470, /* USRQUOTA */ \
1047 0x0cf52471 /* GRPQUOTA */ \
1048}
1049
1050#define OCFS2_GLOBAL_QVERSIONS {\
1051 0, \
1052 0, \
1053}
1054
1055
1056/* Each block of each quota file has a certain fixed number of bytes reserved
1057 * for OCFS2 internal use at its end. OCFS2 can use it for things like
1058 * checksums, etc. */
1059#define OCFS2_QBLK_RESERVED_SPACE 8
1060
1061/* Generic header of all quota files */
1062struct ocfs2_disk_dqheader {
1063 __le32 dqh_magic; /* Magic number identifying file */
1064 __le32 dqh_version; /* Quota format version */
1065};
1066
1067#define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1068
1069/* Information header of global quota file (immediately follows the generic
1070 * header) */
1071struct ocfs2_global_disk_dqinfo {
1072/*00*/ __le32 dqi_bgrace; /* Grace time for space softlimit excess */
1073 __le32 dqi_igrace; /* Grace time for inode softlimit excess */
1074 __le32 dqi_syncms; /* Time after which we sync local changes to
1075 * global quota file */
1076 __le32 dqi_blocks; /* Number of blocks in quota file */
1077/*10*/ __le32 dqi_free_blk; /* First free block in quota file */
1078 __le32 dqi_free_entry; /* First block with free dquot entry in quota
1079 * file */
1080};
1081
1082/* Structure with global user / group information. We reserve some space
1083 * for future use. */
1084struct ocfs2_global_disk_dqblk {
1085/*00*/ __le32 dqb_id; /* ID the structure belongs to */
1086 __le32 dqb_use_count; /* Number of nodes having reference to this structure */
1087 __le64 dqb_ihardlimit; /* absolute limit on allocated inodes */
1088/*10*/ __le64 dqb_isoftlimit; /* preferred inode limit */
1089 __le64 dqb_curinodes; /* current # allocated inodes */
1090/*20*/ __le64 dqb_bhardlimit; /* absolute limit on disk space */
1091 __le64 dqb_bsoftlimit; /* preferred limit on disk space */
1092/*30*/ __le64 dqb_curspace; /* current space occupied */
1093 __le64 dqb_btime; /* time limit for excessive disk use */
1094/*40*/ __le64 dqb_itime; /* time limit for excessive inode use */
1095 __le64 dqb_pad1;
1096/*50*/ __le64 dqb_pad2;
1097};
1098
1099/*
1100 * On-disk structures for local quota file
1101 */
1102
1103/* Magic numbers and known versions for local quota files */
1104#define OCFS2_LOCAL_QMAGICS {\
1105 0x0cf524c0, /* USRQUOTA */ \
1106 0x0cf524c1 /* GRPQUOTA */ \
1107}
1108
1109#define OCFS2_LOCAL_QVERSIONS {\
1110 0, \
1111 0, \
1112}
1113
1114/* Quota flags in dqinfo header */
1115#define OLQF_CLEAN 0x0001 /* Quota file is empty (this should be after\
1116 * quota has been cleanly turned off) */
1117
1118#define OCFS2_LOCAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1119
1120/* Information header of local quota file (immediately follows the generic
1121 * header) */
1122struct ocfs2_local_disk_dqinfo {
1123 __le32 dqi_flags; /* Flags for quota file */
1124 __le32 dqi_chunks; /* Number of chunks of quota structures
1125 * with a bitmap */
1126 __le32 dqi_blocks; /* Number of blocks allocated for quota file */
1127};
1128
1129/* Header of one chunk of a quota file */
1130struct ocfs2_local_disk_chunk {
1131 __le32 dqc_free; /* Number of free entries in the bitmap */
1132 u8 dqc_bitmap[0]; /* Bitmap of entries in the corresponding
1133 * chunk of quota file */
1134};
1135
1136/* One entry in local quota file */
1137struct ocfs2_local_disk_dqblk {
1138/*00*/ __le64 dqb_id; /* id this quota applies to */
1139 __le64 dqb_spacemod; /* Change in the amount of used space */
1140/*10*/ __le64 dqb_inodemod; /* Change in the amount of used inodes */
1141};
1142
Joel Beckerab552d52008-10-16 17:50:30 -07001143
1144/*
1145 * The quota trailer lives at the end of each quota block.
1146 */
1147
1148struct ocfs2_disk_dqtrailer {
1149/*00*/ struct ocfs2_block_check dq_check; /* Error checking */
1150/*08*/ /* Cannot be larger than OCFS2_QBLK_RESERVED_SPACE */
1151};
1152
1153static inline struct ocfs2_disk_dqtrailer *ocfs2_block_dqtrailer(int blocksize,
1154 void *buf)
1155{
1156 char *ptr = buf;
1157 ptr += blocksize - OCFS2_QBLK_RESERVED_SPACE;
1158
1159 return (struct ocfs2_disk_dqtrailer *)ptr;
1160}
1161
Mark Fashehccd979b2005-12-15 14:31:24 -08001162#ifdef __KERNEL__
1163static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
1164{
1165 return sb->s_blocksize -
1166 offsetof(struct ocfs2_dinode, id2.i_symlink);
1167}
1168
Tiger Yangfdd77702008-08-18 17:08:55 +08001169static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
1170 struct ocfs2_dinode *di)
1171{
1172 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1173
1174 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1175 return sb->s_blocksize -
1176 offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1177 xattrsize;
1178 else
1179 return sb->s_blocksize -
1180 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1181}
1182
Mark Fashehccd979b2005-12-15 14:31:24 -08001183static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
1184{
1185 int size;
1186
1187 size = sb->s_blocksize -
1188 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1189
1190 return size / sizeof(struct ocfs2_extent_rec);
1191}
1192
Tiger Yangfdd77702008-08-18 17:08:55 +08001193static inline int ocfs2_extent_recs_per_inode_with_xattr(
1194 struct super_block *sb,
1195 struct ocfs2_dinode *di)
1196{
1197 int size;
1198 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1199
1200 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1201 size = sb->s_blocksize -
1202 offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
1203 xattrsize;
1204 else
1205 size = sb->s_blocksize -
1206 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1207
1208 return size / sizeof(struct ocfs2_extent_rec);
1209}
1210
Mark Fasheh9b7895e2008-11-12 16:27:44 -08001211static inline int ocfs2_extent_recs_per_dx_root(struct super_block *sb)
1212{
1213 int size;
1214
1215 size = sb->s_blocksize -
1216 offsetof(struct ocfs2_dx_root_block, dr_list.l_recs);
1217
1218 return size / sizeof(struct ocfs2_extent_rec);
1219}
1220
Mark Fashehccd979b2005-12-15 14:31:24 -08001221static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
1222{
1223 int size;
1224
1225 size = sb->s_blocksize -
1226 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1227
1228 return size / sizeof(struct ocfs2_chain_rec);
1229}
1230
1231static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
1232{
1233 int size;
1234
1235 size = sb->s_blocksize -
1236 offsetof(struct ocfs2_extent_block, h_list.l_recs);
1237
1238 return size / sizeof(struct ocfs2_extent_rec);
1239}
1240
Mark Fasheh9b7895e2008-11-12 16:27:44 -08001241static inline int ocfs2_dx_entries_per_leaf(struct super_block *sb)
1242{
1243 int size;
1244
1245 size = sb->s_blocksize -
1246 offsetof(struct ocfs2_dx_leaf, dl_list.de_entries);
1247
1248 return size / sizeof(struct ocfs2_dx_entry);
1249}
1250
Mark Fasheh4ed8a6b2008-11-24 17:02:08 -08001251static inline int ocfs2_dx_entries_per_root(struct super_block *sb)
1252{
1253 int size;
1254
1255 size = sb->s_blocksize -
1256 offsetof(struct ocfs2_dx_root_block, dr_entries.de_entries);
1257
1258 return size / sizeof(struct ocfs2_dx_entry);
1259}
1260
Mark Fashehccd979b2005-12-15 14:31:24 -08001261static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
1262{
1263 u16 size;
1264
1265 size = sb->s_blocksize -
1266 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1267
1268 return size;
1269}
1270
1271static inline int ocfs2_group_bitmap_size(struct super_block *sb)
1272{
1273 int size;
1274
1275 size = sb->s_blocksize -
1276 offsetof(struct ocfs2_group_desc, bg_bitmap);
1277
1278 return size;
1279}
1280
1281static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
1282{
1283 int size;
1284
1285 size = sb->s_blocksize -
1286 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1287
1288 return size / sizeof(struct ocfs2_truncate_rec);
1289}
Mark Fasheh50af94b2007-01-21 14:44:59 -08001290
1291static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
1292{
1293 u64 offset = OCFS2_BACKUP_SB_START;
1294
1295 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1296 offset <<= (2 * index);
Mark Fasheha8a75a22007-01-26 10:46:59 -08001297 offset >>= sb->s_blocksize_bits;
Mark Fasheh50af94b2007-01-21 14:44:59 -08001298 return offset;
1299 }
1300
1301 return 0;
1302
1303}
Tao Ma0c044f02008-08-18 17:38:50 +08001304
1305static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
1306{
1307 int size;
1308
1309 size = sb->s_blocksize -
1310 offsetof(struct ocfs2_xattr_block,
1311 xb_attrs.xb_root.xt_list.l_recs);
1312
1313 return size / sizeof(struct ocfs2_extent_rec);
1314}
Mark Fashehccd979b2005-12-15 14:31:24 -08001315#else
1316static inline int ocfs2_fast_symlink_chars(int blocksize)
1317{
1318 return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
1319}
1320
Mark Fasheh15b1e362007-09-07 13:58:15 -07001321static inline int ocfs2_max_inline_data(int blocksize)
1322{
1323 return blocksize - offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1324}
1325
Mark Fashehccd979b2005-12-15 14:31:24 -08001326static inline int ocfs2_extent_recs_per_inode(int blocksize)
1327{
1328 int size;
1329
1330 size = blocksize -
1331 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1332
1333 return size / sizeof(struct ocfs2_extent_rec);
1334}
1335
1336static inline int ocfs2_chain_recs_per_inode(int blocksize)
1337{
1338 int size;
1339
1340 size = blocksize -
1341 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1342
1343 return size / sizeof(struct ocfs2_chain_rec);
1344}
1345
1346static inline int ocfs2_extent_recs_per_eb(int blocksize)
1347{
1348 int size;
1349
1350 size = blocksize -
1351 offsetof(struct ocfs2_extent_block, h_list.l_recs);
1352
1353 return size / sizeof(struct ocfs2_extent_rec);
1354}
1355
1356static inline int ocfs2_local_alloc_size(int blocksize)
1357{
1358 int size;
1359
1360 size = blocksize -
1361 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1362
1363 return size;
1364}
1365
1366static inline int ocfs2_group_bitmap_size(int blocksize)
1367{
1368 int size;
1369
1370 size = blocksize -
1371 offsetof(struct ocfs2_group_desc, bg_bitmap);
1372
1373 return size;
1374}
1375
1376static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1377{
1378 int size;
1379
1380 size = blocksize -
1381 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1382
1383 return size / sizeof(struct ocfs2_truncate_rec);
1384}
Mark Fasheh50af94b2007-01-21 14:44:59 -08001385
1386static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1387{
1388 uint64_t offset = OCFS2_BACKUP_SB_START;
1389
1390 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1391 offset <<= (2 * index);
1392 offset /= blocksize;
1393 return offset;
1394 }
1395
1396 return 0;
1397}
Tao Ma0c044f02008-08-18 17:38:50 +08001398
1399static inline int ocfs2_xattr_recs_per_xb(int blocksize)
1400{
1401 int size;
1402
1403 size = blocksize -
1404 offsetof(struct ocfs2_xattr_block,
1405 xb_attrs.xb_root.xt_list.l_recs);
1406
1407 return size / sizeof(struct ocfs2_extent_rec);
1408}
Mark Fashehccd979b2005-12-15 14:31:24 -08001409#endif /* __KERNEL__ */
1410
1411
1412static inline int ocfs2_system_inode_is_global(int type)
1413{
1414 return ((type >= 0) &&
1415 (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1416}
1417
1418static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1419 int type, int slot)
1420{
1421 int chars;
1422
1423 /*
1424 * Global system inodes can only have one copy. Everything
1425 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1426 * list has a copy per slot.
1427 */
1428 if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
Joel Beckerfe9f3872008-06-12 22:39:18 -07001429 chars = snprintf(buf, len, "%s",
Mark Fashehccd979b2005-12-15 14:31:24 -08001430 ocfs2_system_inodes[type].si_name);
1431 else
1432 chars = snprintf(buf, len,
1433 ocfs2_system_inodes[type].si_name,
1434 slot);
1435
1436 return chars;
1437}
1438
1439static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1440 umode_t mode)
1441{
1442 de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
1443}
1444
1445#endif /* _OCFS2_FS_H */
1446