blob: f7a5201410c83e802338ce950f5ee8149b916620 [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 \
100 | OCFS2_FEATURE_INCOMPAT_META_ECC)
Jan Kara19ece542008-08-21 20:13:17 +0200101#define OCFS2_FEATURE_RO_COMPAT_SUPP (OCFS2_FEATURE_RO_COMPAT_UNWRITTEN \
102 | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \
103 | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)
Mark Fashehccd979b2005-12-15 14:31:24 -0800104
105/*
106 * Heartbeat-only devices are missing journals and other files. The
107 * filesystem driver can't load them, but the library can. Never put
108 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
109 */
110#define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV 0x0002
111
Mark Fasheh89039012006-12-07 18:05:37 -0800112/*
113 * tunefs sets this incompat flag before starting the resize and clears it
114 * at the end. This flag protects users from inadvertently mounting the fs
115 * after an aborted run without fsck-ing.
116 */
117#define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG 0x0004
Mark Fashehccd979b2005-12-15 14:31:24 -0800118
Sunil Mushranc271c5c2006-12-05 17:56:35 -0800119/* Used to denote a non-clustered volume */
120#define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT 0x0008
121
Mark Fasheh89039012006-12-07 18:05:37 -0800122/* Support for sparse allocation in b-trees */
123#define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC 0x0010
124
Mark Fashehccd979b2005-12-15 14:31:24 -0800125/*
Mark Fasheh92e91ce2007-09-20 11:19:20 -0700126 * Tunefs sets this incompat flag before starting an operation which
127 * would require cleanup on abort. This is done to protect users from
128 * inadvertently mounting the fs after an aborted run without
129 * fsck-ing.
130 *
131 * s_tunefs_flags on the super block describes precisely which
132 * operations were in progress.
133 */
134#define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG 0x0020
135
Mark Fasheh15b1e362007-09-07 13:58:15 -0700136/* Support for data packed into inode blocks */
137#define OCFS2_FEATURE_INCOMPAT_INLINE_DATA 0x0040
138
Mark Fasheh92e91ce2007-09-20 11:19:20 -0700139/*
Joel Beckerb61817e2008-02-01 15:08:23 -0800140 * Support for alternate, userspace cluster stacks. If set, the superblock
141 * field s_cluster_info contains a tag for the alternate stack in use as
142 * well as the name of the cluster being joined.
143 * mount.ocfs2 must pass in a matching stack name.
144 *
145 * If not set, the classic stack will be used. This is compatbile with
146 * all older versions.
147 */
148#define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK 0x0080
149
Tiger Yang8154da32008-08-18 17:11:46 +0800150/* Support for the extended slot map */
151#define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
152
153/* Support for extended attributes */
154#define OCFS2_FEATURE_INCOMPAT_XATTR 0x0200
155
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800156/* Support for indexed directores */
157#define OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS 0x0400
158
Joel Beckerab552d52008-10-16 17:50:30 -0700159/* Metadata checksum and error correction */
160#define OCFS2_FEATURE_INCOMPAT_META_ECC 0x0800
161
Joel Beckerb61817e2008-02-01 15:08:23 -0800162/*
Mark Fasheh50af94b2007-01-21 14:44:59 -0800163 * backup superblock flag is used to indicate that this volume
164 * has backup superblocks.
165 */
166#define OCFS2_FEATURE_COMPAT_BACKUP_SB 0x0001
167
Mark Fasheh328d5752007-06-18 10:48:04 -0700168/*
Joel Beckera9772182008-12-16 18:10:18 -0800169 * The filesystem will correctly handle journal feature bits.
170 */
171#define OCFS2_FEATURE_COMPAT_JBD2_SB 0x0002
172
173/*
Mark Fasheh328d5752007-06-18 10:48:04 -0700174 * Unwritten extents support.
175 */
176#define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN 0x0001
177
Jan Kara1a224ad2008-08-20 15:43:36 +0200178/*
179 * Maintain quota information for this filesystem
180 */
181#define OCFS2_FEATURE_RO_COMPAT_USRQUOTA 0x0002
182#define OCFS2_FEATURE_RO_COMPAT_GRPQUOTA 0x0004
183
Mark Fasheh50af94b2007-01-21 14:44:59 -0800184/* The byte offset of the first backup block will be 1G.
185 * The following will be 4G, 16G, 64G, 256G and 1T.
186 */
187#define OCFS2_BACKUP_SB_START 1 << 30
188
189/* the max backup superblock nums */
190#define OCFS2_MAX_BACKUP_SUPERBLOCKS 6
191
192/*
Mark Fasheh92e91ce2007-09-20 11:19:20 -0700193 * Flags on ocfs2_super_block.s_tunefs_flags
194 */
195#define OCFS2_TUNEFS_INPROG_REMOVE_SLOT 0x0001 /* Removing slots */
196
197/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800198 * Flags on ocfs2_dinode.i_flags
199 */
200#define OCFS2_VALID_FL (0x00000001) /* Inode is valid */
201#define OCFS2_UNUSED2_FL (0x00000002)
202#define OCFS2_ORPHANED_FL (0x00000004) /* On the orphan list */
203#define OCFS2_UNUSED3_FL (0x00000008)
204/* System inode flags */
205#define OCFS2_SYSTEM_FL (0x00000010) /* System inode */
206#define OCFS2_SUPER_BLOCK_FL (0x00000020) /* Super block */
207#define OCFS2_LOCAL_ALLOC_FL (0x00000040) /* Slot local alloc bitmap */
208#define OCFS2_BITMAP_FL (0x00000080) /* Allocation bitmap */
209#define OCFS2_JOURNAL_FL (0x00000100) /* Slot local journal */
210#define OCFS2_HEARTBEAT_FL (0x00000200) /* Heartbeat area */
211#define OCFS2_CHAIN_FL (0x00000400) /* Chain allocator */
212#define OCFS2_DEALLOC_FL (0x00000800) /* Truncate log */
Jan Kara1a224ad2008-08-20 15:43:36 +0200213#define OCFS2_QUOTA_FL (0x00001000) /* Quota file */
Mark Fashehccd979b2005-12-15 14:31:24 -0800214
Mark Fasheh15b1e362007-09-07 13:58:15 -0700215/*
216 * Flags on ocfs2_dinode.i_dyn_features
217 *
218 * These can change much more often than i_flags. When adding flags,
219 * keep in mind that i_dyn_features is only 16 bits wide.
220 */
221#define OCFS2_INLINE_DATA_FL (0x0001) /* Data stored in inode block */
222#define OCFS2_HAS_XATTR_FL (0x0002)
223#define OCFS2_INLINE_XATTR_FL (0x0004)
224#define OCFS2_INDEXED_DIR_FL (0x0008)
225
Herbert Poetzlca4d1472006-07-03 17:27:12 -0700226/* Inode attributes, keep in sync with EXT2 */
227#define OCFS2_SECRM_FL (0x00000001) /* Secure deletion */
228#define OCFS2_UNRM_FL (0x00000002) /* Undelete */
229#define OCFS2_COMPR_FL (0x00000004) /* Compress file */
230#define OCFS2_SYNC_FL (0x00000008) /* Synchronous updates */
231#define OCFS2_IMMUTABLE_FL (0x00000010) /* Immutable file */
232#define OCFS2_APPEND_FL (0x00000020) /* writes to file may only append */
233#define OCFS2_NODUMP_FL (0x00000040) /* do not dump file */
234#define OCFS2_NOATIME_FL (0x00000080) /* do not update atime */
235#define OCFS2_DIRSYNC_FL (0x00010000) /* dirsync behaviour (directories only) */
236
237#define OCFS2_FL_VISIBLE (0x000100FF) /* User visible flags */
238#define OCFS2_FL_MODIFIABLE (0x000100FF) /* User modifiable flags */
239
240/*
Mark Fashehe48edee2007-03-07 16:46:57 -0800241 * Extent record flags (e_node.leaf.flags)
242 */
243#define OCFS2_EXT_UNWRITTEN (0x01) /* Extent is allocated but
244 * unwritten */
245
246/*
Herbert Poetzlca4d1472006-07-03 17:27:12 -0700247 * ioctl commands
248 */
249#define OCFS2_IOC_GETFLAGS _IOR('f', 1, long)
250#define OCFS2_IOC_SETFLAGS _IOW('f', 2, long)
Mark Fasheh586d2322007-03-09 15:56:28 -0800251#define OCFS2_IOC32_GETFLAGS _IOR('f', 1, int)
252#define OCFS2_IOC32_SETFLAGS _IOW('f', 2, int)
Herbert Poetzlca4d1472006-07-03 17:27:12 -0700253
Mark Fashehccd979b2005-12-15 14:31:24 -0800254/*
Mark Fashehb2580102007-03-09 16:53:21 -0800255 * Space reservation / allocation / free ioctls and argument structure
256 * are designed to be compatible with XFS.
257 *
258 * ALLOCSP* and FREESP* are not and will never be supported, but are
259 * included here for completeness.
260 */
261struct ocfs2_space_resv {
262 __s16 l_type;
263 __s16 l_whence;
264 __s64 l_start;
265 __s64 l_len; /* len == 0 means until end of file */
266 __s32 l_sysid;
267 __u32 l_pid;
268 __s32 l_pad[4]; /* reserve area */
269};
270
271#define OCFS2_IOC_ALLOCSP _IOW ('X', 10, struct ocfs2_space_resv)
272#define OCFS2_IOC_FREESP _IOW ('X', 11, struct ocfs2_space_resv)
273#define OCFS2_IOC_RESVSP _IOW ('X', 40, struct ocfs2_space_resv)
274#define OCFS2_IOC_UNRESVSP _IOW ('X', 41, struct ocfs2_space_resv)
275#define OCFS2_IOC_ALLOCSP64 _IOW ('X', 36, struct ocfs2_space_resv)
276#define OCFS2_IOC_FREESP64 _IOW ('X', 37, struct ocfs2_space_resv)
277#define OCFS2_IOC_RESVSP64 _IOW ('X', 42, struct ocfs2_space_resv)
278#define OCFS2_IOC_UNRESVSP64 _IOW ('X', 43, struct ocfs2_space_resv)
279
Tao Ma7909f2b2007-12-18 15:47:25 +0800280/* Used to pass group descriptor data when online resize is done */
281struct ocfs2_new_group_input {
282 __u64 group; /* Group descriptor's blkno. */
283 __u32 clusters; /* Total number of clusters in this group */
284 __u32 frees; /* Total free clusters in this group */
285 __u16 chain; /* Chain for this group */
286 __u16 reserved1;
287 __u32 reserved2;
288};
289
Tao Mad6590722007-12-18 15:47:03 +0800290#define OCFS2_IOC_GROUP_EXTEND _IOW('o', 1, int)
Tao Ma7909f2b2007-12-18 15:47:25 +0800291#define OCFS2_IOC_GROUP_ADD _IOW('o', 2,struct ocfs2_new_group_input)
292#define OCFS2_IOC_GROUP_ADD64 _IOW('o', 3,struct ocfs2_new_group_input)
Tao Mad6590722007-12-18 15:47:03 +0800293
Mark Fashehb2580102007-03-09 16:53:21 -0800294/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800295 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
296 */
297#define OCFS2_JOURNAL_DIRTY_FL (0x00000001) /* Journal needs recovery */
298
299/*
300 * superblock s_state flags
301 */
302#define OCFS2_ERROR_FS (0x00000001) /* FS saw errors */
303
304/* Limit of space in ocfs2_dir_entry */
305#define OCFS2_MAX_FILENAME_LEN 255
306
307/* Maximum slots on an ocfs2 file system */
308#define OCFS2_MAX_SLOTS 255
309
310/* Slot map indicator for an empty slot */
311#define OCFS2_INVALID_SLOT -1
312
313#define OCFS2_VOL_UUID_LEN 16
314#define OCFS2_MAX_VOL_LABEL_LEN 64
315
Joel Beckerb61817e2008-02-01 15:08:23 -0800316/* The alternate, userspace stack fields */
317#define OCFS2_STACK_LABEL_LEN 4
318#define OCFS2_CLUSTER_NAME_LEN 16
319
Mark Fashehccd979b2005-12-15 14:31:24 -0800320/* Journal limits (in bytes) */
321#define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
Mark Fashehccd979b2005-12-15 14:31:24 -0800322
Sunil Mushran2fbe8d12007-12-20 14:58:11 -0800323/*
324 * Default local alloc size (in megabytes)
325 *
326 * The value chosen should be such that most allocations, including new
327 * block groups, use local alloc.
328 */
329#define OCFS2_DEFAULT_LOCAL_ALLOC_SIZE 8
330
Tiger Yangfdd77702008-08-18 17:08:55 +0800331/*
332 * Inline extended attribute size (in bytes)
333 * The value chosen should be aligned to 16 byte boundaries.
334 */
335#define OCFS2_MIN_XATTR_INLINE_SIZE 256
336
Mark Fashehccd979b2005-12-15 14:31:24 -0800337struct ocfs2_system_inode_info {
338 char *si_name;
339 int si_iflags;
340 int si_mode;
341};
342
343/* System file index */
344enum {
345 BAD_BLOCK_SYSTEM_INODE = 0,
346 GLOBAL_INODE_ALLOC_SYSTEM_INODE,
347 SLOT_MAP_SYSTEM_INODE,
348#define OCFS2_FIRST_ONLINE_SYSTEM_INODE SLOT_MAP_SYSTEM_INODE
349 HEARTBEAT_SYSTEM_INODE,
350 GLOBAL_BITMAP_SYSTEM_INODE,
Jan Kara1a224ad2008-08-20 15:43:36 +0200351 USER_QUOTA_SYSTEM_INODE,
352 GROUP_QUOTA_SYSTEM_INODE,
353#define OCFS2_LAST_GLOBAL_SYSTEM_INODE GROUP_QUOTA_SYSTEM_INODE
Mark Fashehccd979b2005-12-15 14:31:24 -0800354 ORPHAN_DIR_SYSTEM_INODE,
355 EXTENT_ALLOC_SYSTEM_INODE,
356 INODE_ALLOC_SYSTEM_INODE,
357 JOURNAL_SYSTEM_INODE,
358 LOCAL_ALLOC_SYSTEM_INODE,
359 TRUNCATE_LOG_SYSTEM_INODE,
Jan Kara1a224ad2008-08-20 15:43:36 +0200360 LOCAL_USER_QUOTA_SYSTEM_INODE,
361 LOCAL_GROUP_QUOTA_SYSTEM_INODE,
Mark Fashehccd979b2005-12-15 14:31:24 -0800362 NUM_SYSTEM_INODES
363};
364
365static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
366 /* Global system inodes (single copy) */
367 /* The first two are only used from userspace mfks/tunefs */
368 [BAD_BLOCK_SYSTEM_INODE] = { "bad_blocks", 0, S_IFREG | 0644 },
369 [GLOBAL_INODE_ALLOC_SYSTEM_INODE] = { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
370
371 /* These are used by the running filesystem */
372 [SLOT_MAP_SYSTEM_INODE] = { "slot_map", 0, S_IFREG | 0644 },
373 [HEARTBEAT_SYSTEM_INODE] = { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
374 [GLOBAL_BITMAP_SYSTEM_INODE] = { "global_bitmap", 0, S_IFREG | 0644 },
Jan Kara1a224ad2008-08-20 15:43:36 +0200375 [USER_QUOTA_SYSTEM_INODE] = { "aquota.user", OCFS2_QUOTA_FL, S_IFREG | 0644 },
376 [GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group", OCFS2_QUOTA_FL, S_IFREG | 0644 },
Mark Fashehccd979b2005-12-15 14:31:24 -0800377
378 /* Slot-specific system inodes (one copy per slot) */
379 [ORPHAN_DIR_SYSTEM_INODE] = { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
380 [EXTENT_ALLOC_SYSTEM_INODE] = { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
381 [INODE_ALLOC_SYSTEM_INODE] = { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
382 [JOURNAL_SYSTEM_INODE] = { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
383 [LOCAL_ALLOC_SYSTEM_INODE] = { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
Jan Kara1a224ad2008-08-20 15:43:36 +0200384 [TRUNCATE_LOG_SYSTEM_INODE] = { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 },
385 [LOCAL_USER_QUOTA_SYSTEM_INODE] = { "aquota.user:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
386 [LOCAL_GROUP_QUOTA_SYSTEM_INODE] = { "aquota.group:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
Mark Fashehccd979b2005-12-15 14:31:24 -0800387};
388
389/* Parameter passed from mount.ocfs2 to module */
390#define OCFS2_HB_NONE "heartbeat=none"
391#define OCFS2_HB_LOCAL "heartbeat=local"
392
393/*
394 * OCFS2 directory file types. Only the low 3 bits are used. The
395 * other bits are reserved for now.
396 */
397#define OCFS2_FT_UNKNOWN 0
398#define OCFS2_FT_REG_FILE 1
399#define OCFS2_FT_DIR 2
400#define OCFS2_FT_CHRDEV 3
401#define OCFS2_FT_BLKDEV 4
402#define OCFS2_FT_FIFO 5
403#define OCFS2_FT_SOCK 6
404#define OCFS2_FT_SYMLINK 7
405
406#define OCFS2_FT_MAX 8
407
408/*
409 * OCFS2_DIR_PAD defines the directory entries boundaries
410 *
411 * NOTE: It must be a multiple of 4
412 */
413#define OCFS2_DIR_PAD 4
414#define OCFS2_DIR_ROUND (OCFS2_DIR_PAD - 1)
415#define OCFS2_DIR_MEMBER_LEN offsetof(struct ocfs2_dir_entry, name)
416#define OCFS2_DIR_REC_LEN(name_len) (((name_len) + OCFS2_DIR_MEMBER_LEN + \
417 OCFS2_DIR_ROUND) & \
418 ~OCFS2_DIR_ROUND)
Mark Fashehe7c17e42009-01-29 18:17:46 -0800419#define OCFS2_DIR_MIN_REC_LEN OCFS2_DIR_REC_LEN(1)
Mark Fashehccd979b2005-12-15 14:31:24 -0800420
421#define OCFS2_LINK_MAX 32000
Mark Fasheh198a1ca2008-11-20 17:54:57 -0800422#define OCFS2_DX_LINK_MAX ((1U << 31) - 1U)
423#define OCFS2_LINKS_HI_SHIFT 16
Mark Fashehe3a93c22009-02-17 15:29:35 -0800424#define OCFS2_DX_ENTRIES_MAX (0xffffffffU)
Mark Fashehccd979b2005-12-15 14:31:24 -0800425
426#define S_SHIFT 12
427static unsigned char ocfs2_type_by_mode[S_IFMT >> S_SHIFT] = {
428 [S_IFREG >> S_SHIFT] = OCFS2_FT_REG_FILE,
429 [S_IFDIR >> S_SHIFT] = OCFS2_FT_DIR,
430 [S_IFCHR >> S_SHIFT] = OCFS2_FT_CHRDEV,
431 [S_IFBLK >> S_SHIFT] = OCFS2_FT_BLKDEV,
432 [S_IFIFO >> S_SHIFT] = OCFS2_FT_FIFO,
433 [S_IFSOCK >> S_SHIFT] = OCFS2_FT_SOCK,
434 [S_IFLNK >> S_SHIFT] = OCFS2_FT_SYMLINK,
435};
436
437
438/*
439 * Convenience casts
440 */
441#define OCFS2_RAW_SB(dinode) (&((dinode)->id2.i_super))
442
443/*
Joel Beckerab552d52008-10-16 17:50:30 -0700444 * Block checking structure. This is used in metadata to validate the
445 * contents. If OCFS2_FEATURE_INCOMPAT_META_ECC is not set, it is all
446 * zeros.
447 */
448struct ocfs2_block_check {
449/*00*/ __le32 bc_crc32e; /* 802.3 Ethernet II CRC32 */
450 __le16 bc_ecc; /* Single-error-correction parity vector.
451 This is a simple Hamming code dependant
452 on the blocksize. OCFS2's maximum
453 blocksize, 4K, requires 16 parity bits,
454 so we fit in __le16. */
455 __le16 bc_reserved1;
456/*08*/
457};
458
459/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800460 * On disk extent record for OCFS2
461 * It describes a range of clusters on disk.
Mark Fashehe48edee2007-03-07 16:46:57 -0800462 *
463 * Length fields are divided into interior and leaf node versions.
464 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
Mark Fashehccd979b2005-12-15 14:31:24 -0800465 */
466struct ocfs2_extent_rec {
467/*00*/ __le32 e_cpos; /* Offset into the file, in clusters */
Mark Fashehe48edee2007-03-07 16:46:57 -0800468 union {
469 __le32 e_int_clusters; /* Clusters covered by all children */
470 struct {
471 __le16 e_leaf_clusters; /* Clusters covered by this
472 extent */
473 __u8 e_reserved1;
474 __u8 e_flags; /* Extent flags */
475 };
476 };
Mark Fashehccd979b2005-12-15 14:31:24 -0800477 __le64 e_blkno; /* Physical disk offset, in blocks */
478/*10*/
479};
480
481struct ocfs2_chain_rec {
482 __le32 c_free; /* Number of free bits in this chain. */
483 __le32 c_total; /* Number of total bits in this chain */
484 __le64 c_blkno; /* Physical disk offset (blocks) of 1st group */
485};
486
487struct ocfs2_truncate_rec {
488 __le32 t_start; /* 1st cluster in this log */
489 __le32 t_clusters; /* Number of total clusters covered */
490};
491
492/*
493 * On disk extent list for OCFS2 (node in the tree). Note that this
494 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
495 * offsets are relative to ocfs2_dinode.id2.i_list or
496 * ocfs2_extent_block.h_list, respectively.
497 */
498struct ocfs2_extent_list {
499/*00*/ __le16 l_tree_depth; /* Extent tree depth from this
500 point. 0 means data extents
501 hang directly off this
Mark Fashehdcd05382007-01-16 11:32:23 -0800502 header (a leaf)
503 NOTE: The high 8 bits cannot be
504 used - tree_depth is never that big.
505 */
Mark Fashehccd979b2005-12-15 14:31:24 -0800506 __le16 l_count; /* Number of extent records */
507 __le16 l_next_free_rec; /* Next unused extent slot */
508 __le16 l_reserved1;
509 __le64 l_reserved2; /* Pad to
510 sizeof(ocfs2_extent_rec) */
511/*10*/ struct ocfs2_extent_rec l_recs[0]; /* Extent records */
512};
513
514/*
515 * On disk allocation chain list for OCFS2. Note that this is
516 * contained inside ocfs2_dinode, so the offsets are relative to
517 * ocfs2_dinode.id2.i_chain.
518 */
519struct ocfs2_chain_list {
520/*00*/ __le16 cl_cpg; /* Clusters per Block Group */
521 __le16 cl_bpc; /* Bits per cluster */
522 __le16 cl_count; /* Total chains in this list */
523 __le16 cl_next_free_rec; /* Next unused chain slot */
524 __le64 cl_reserved1;
525/*10*/ struct ocfs2_chain_rec cl_recs[0]; /* Chain records */
526};
527
528/*
529 * On disk deallocation log for OCFS2. Note that this is
530 * contained inside ocfs2_dinode, so the offsets are relative to
531 * ocfs2_dinode.id2.i_dealloc.
532 */
533struct ocfs2_truncate_log {
534/*00*/ __le16 tl_count; /* Total records in this log */
535 __le16 tl_used; /* Number of records in use */
536 __le32 tl_reserved1;
537/*08*/ struct ocfs2_truncate_rec tl_recs[0]; /* Truncate records */
538};
539
540/*
541 * On disk extent block (indirect block) for OCFS2
542 */
543struct ocfs2_extent_block
544{
545/*00*/ __u8 h_signature[8]; /* Signature for verification */
Joel Beckerab552d52008-10-16 17:50:30 -0700546 struct ocfs2_block_check h_check; /* Error checking */
Mark Fashehccd979b2005-12-15 14:31:24 -0800547/*10*/ __le16 h_suballoc_slot; /* Slot suballocator this
548 extent_header belongs to */
549 __le16 h_suballoc_bit; /* Bit offset in suballocator
550 block group */
551 __le32 h_fs_generation; /* Must match super block */
552 __le64 h_blkno; /* Offset on disk, in blocks */
553/*20*/ __le64 h_reserved3;
554 __le64 h_next_leaf_blk; /* Offset on disk, in blocks,
555 of next leaf header pointing
556 to data */
557/*30*/ struct ocfs2_extent_list h_list; /* Extent record list */
558/* Actual on-disk size is one block */
559};
560
561/*
Joel Beckerfb86b1f2008-02-01 11:59:05 -0800562 * On disk slot map for OCFS2. This defines the contents of the "slot_map"
Joel Becker386a2ef2008-02-01 12:06:54 -0800563 * system file. A slot is valid if it contains a node number >= 0. The
564 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT. This marks a slot empty.
Joel Beckerfb86b1f2008-02-01 11:59:05 -0800565 */
566struct ocfs2_slot_map {
567/*00*/ __le16 sm_slots[0];
568/*
569 * Actual on-disk size is one block. OCFS2_MAX_SLOTS is 255,
570 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
571 */
572};
573
Joel Becker386a2ef2008-02-01 12:06:54 -0800574struct ocfs2_extended_slot {
575/*00*/ __u8 es_valid;
576 __u8 es_reserved1[3];
577 __le32 es_node_num;
578/*10*/
579};
580
581/*
582 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
583 * is set. It separates out the valid marker from the node number, and
584 * has room to grow. Unlike the old slot map, this format is defined by
585 * i_size.
586 */
587struct ocfs2_slot_map_extended {
588/*00*/ struct ocfs2_extended_slot se_slots[0];
589/*
590 * Actual size is i_size of the slot_map system file. It should
591 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
592 */
593};
594
Joel Beckerb61817e2008-02-01 15:08:23 -0800595struct ocfs2_cluster_info {
596/*00*/ __u8 ci_stack[OCFS2_STACK_LABEL_LEN];
597 __le32 ci_reserved;
598/*08*/ __u8 ci_cluster[OCFS2_CLUSTER_NAME_LEN];
599/*18*/
600};
601
Joel Beckerfb86b1f2008-02-01 11:59:05 -0800602/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800603 * On disk superblock for OCFS2
604 * Note that it is contained inside an ocfs2_dinode, so all offsets
605 * are relative to the start of ocfs2_dinode.id2.
606 */
607struct ocfs2_super_block {
608/*00*/ __le16 s_major_rev_level;
609 __le16 s_minor_rev_level;
610 __le16 s_mnt_count;
611 __le16 s_max_mnt_count;
612 __le16 s_state; /* File system state */
613 __le16 s_errors; /* Behaviour when detecting errors */
614 __le32 s_checkinterval; /* Max time between checks */
615/*10*/ __le64 s_lastcheck; /* Time of last check */
616 __le32 s_creator_os; /* OS */
617 __le32 s_feature_compat; /* Compatible feature set */
618/*20*/ __le32 s_feature_incompat; /* Incompatible feature set */
619 __le32 s_feature_ro_compat; /* Readonly-compatible feature set */
620 __le64 s_root_blkno; /* Offset, in blocks, of root directory
621 dinode */
622/*30*/ __le64 s_system_dir_blkno; /* Offset, in blocks, of system
623 directory dinode */
624 __le32 s_blocksize_bits; /* Blocksize for this fs */
625 __le32 s_clustersize_bits; /* Clustersize for this fs */
626/*40*/ __le16 s_max_slots; /* Max number of simultaneous mounts
627 before tunefs required */
Mark Fasheh92e91ce2007-09-20 11:19:20 -0700628 __le16 s_tunefs_flag;
Tiger Yangcf1d6c72008-08-18 17:11:00 +0800629 __le32 s_uuid_hash; /* hash value of uuid */
Mark Fashehccd979b2005-12-15 14:31:24 -0800630 __le64 s_first_cluster_group; /* Block offset of 1st cluster
631 * group header */
632/*50*/ __u8 s_label[OCFS2_MAX_VOL_LABEL_LEN]; /* Label for mounting, etc. */
633/*90*/ __u8 s_uuid[OCFS2_VOL_UUID_LEN]; /* 128-bit uuid */
Joel Beckerb61817e2008-02-01 15:08:23 -0800634/*A0*/ struct ocfs2_cluster_info s_cluster_info; /* Selected userspace
635 stack. Only valid
636 with INCOMPAT flag. */
Tiger Yang8154da32008-08-18 17:11:46 +0800637/*B8*/ __le16 s_xattr_inline_size; /* extended attribute inline size
638 for this fs*/
639 __le16 s_reserved0;
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800640 __le32 s_dx_seed[3]; /* seed[0-2] for dx dir hash.
641 * s_uuid_hash serves as seed[3]. */
642/*C0*/ __le64 s_reserved2[15]; /* Fill out superblock */
Joel Beckerb61817e2008-02-01 15:08:23 -0800643/*140*/
644
645 /*
646 * NOTE: As stated above, all offsets are relative to
647 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
648 * 0xC0 + 0x140 = 0x200 or 512 bytes. A superblock must fit within
649 * our smallest blocksize, which is 512 bytes. To ensure this,
650 * we reserve the space in s_reserved2. Anything past s_reserved2
651 * will not be available on the smallest blocksize.
652 */
Mark Fashehccd979b2005-12-15 14:31:24 -0800653};
654
655/*
656 * Local allocation bitmap for OCFS2 slots
657 * Note that it exists inside an ocfs2_dinode, so all offsets are
658 * relative to the start of ocfs2_dinode.id2.
659 */
660struct ocfs2_local_alloc
661{
662/*00*/ __le32 la_bm_off; /* Starting bit offset in main bitmap */
663 __le16 la_size; /* Size of included bitmap, in bytes */
664 __le16 la_reserved1;
665 __le64 la_reserved2;
666/*10*/ __u8 la_bitmap[0];
667};
668
669/*
Mark Fasheh15b1e362007-09-07 13:58:15 -0700670 * Data-in-inode header. This is only used if i_dyn_features has
671 * OCFS2_INLINE_DATA_FL set.
672 */
673struct ocfs2_inline_data
674{
675/*00*/ __le16 id_count; /* Number of bytes that can be used
676 * for data, starting at id_data */
677 __le16 id_reserved0;
678 __le32 id_reserved1;
679 __u8 id_data[0]; /* Start of user data */
680};
681
682/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800683 * On disk inode for OCFS2
684 */
685struct ocfs2_dinode {
686/*00*/ __u8 i_signature[8]; /* Signature for validation */
687 __le32 i_generation; /* Generation number */
688 __le16 i_suballoc_slot; /* Slot suballocator this inode
689 belongs to */
690 __le16 i_suballoc_bit; /* Bit offset in suballocator
691 block group */
Mark Fasheh198a1ca2008-11-20 17:54:57 -0800692/*10*/ __le16 i_links_count_hi; /* High 16 bits of links count */
Tiger Yangfdd77702008-08-18 17:08:55 +0800693 __le16 i_xattr_inline_size;
Mark Fashehccd979b2005-12-15 14:31:24 -0800694 __le32 i_clusters; /* Cluster count */
695 __le32 i_uid; /* Owner UID */
696 __le32 i_gid; /* Owning GID */
697/*20*/ __le64 i_size; /* Size in bytes */
698 __le16 i_mode; /* File mode */
699 __le16 i_links_count; /* Links count */
700 __le32 i_flags; /* File flags */
701/*30*/ __le64 i_atime; /* Access time */
702 __le64 i_ctime; /* Creation time */
703/*40*/ __le64 i_mtime; /* Modification time */
704 __le64 i_dtime; /* Deletion time */
705/*50*/ __le64 i_blkno; /* Offset on disk, in blocks */
706 __le64 i_last_eb_blk; /* Pointer to last extent
707 block */
708/*60*/ __le32 i_fs_generation; /* Generation per fs-instance */
709 __le32 i_atime_nsec;
710 __le32 i_ctime_nsec;
711 __le32 i_mtime_nsec;
Tiger Yangfdd77702008-08-18 17:08:55 +0800712/*70*/ __le32 i_attr;
Tiger Yang50008632007-03-20 16:01:38 -0700713 __le16 i_orphaned_slot; /* Only valid when OCFS2_ORPHANED_FL
714 was set in i_flags */
Mark Fasheh15b1e362007-09-07 13:58:15 -0700715 __le16 i_dyn_features;
Tiger Yangfdd77702008-08-18 17:08:55 +0800716 __le64 i_xattr_loc;
Joel Beckerab552d52008-10-16 17:50:30 -0700717/*80*/ struct ocfs2_block_check i_check; /* Error checking */
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800718/*88*/ __le64 i_dx_root; /* Pointer to dir index root block */
719 __le64 i_reserved2[5];
Mark Fashehccd979b2005-12-15 14:31:24 -0800720/*B8*/ union {
721 __le64 i_pad1; /* Generic way to refer to this
722 64bit union */
723 struct {
724 __le64 i_rdev; /* Device number */
725 } dev1;
726 struct { /* Info for bitmap system
727 inodes */
728 __le32 i_used; /* Bits (ie, clusters) used */
729 __le32 i_total; /* Total bits (clusters)
730 available */
731 } bitmap1;
732 struct { /* Info for journal system
733 inodes */
734 __le32 ij_flags; /* Mounted, version, etc. */
Sunil Mushranc69991a2008-07-14 17:31:09 -0700735 __le32 ij_recovery_generation; /* Incremented when the
736 journal is recovered
737 after an unclean
738 shutdown */
Mark Fashehccd979b2005-12-15 14:31:24 -0800739 } journal1;
740 } id1; /* Inode type dependant 1 */
741/*C0*/ union {
742 struct ocfs2_super_block i_super;
743 struct ocfs2_local_alloc i_lab;
744 struct ocfs2_chain_list i_chain;
745 struct ocfs2_extent_list i_list;
746 struct ocfs2_truncate_log i_dealloc;
Mark Fasheh15b1e362007-09-07 13:58:15 -0700747 struct ocfs2_inline_data i_data;
Mark Fashehccd979b2005-12-15 14:31:24 -0800748 __u8 i_symlink[0];
749 } id2;
750/* Actual on-disk size is one block */
751};
752
753/*
754 * On-disk directory entry structure for OCFS2
755 *
756 * Packed as this structure could be accessed unaligned on 64-bit platforms
757 */
758struct ocfs2_dir_entry {
759/*00*/ __le64 inode; /* Inode number */
760 __le16 rec_len; /* Directory entry length */
761 __u8 name_len; /* Name length */
762 __u8 file_type;
763/*0C*/ char name[OCFS2_MAX_FILENAME_LEN]; /* File name */
764/* Actual on-disk length specified by rec_len */
765} __attribute__ ((packed));
766
767/*
Mark Fasheh87d35a72008-12-10 17:36:25 -0800768 * Per-block record for the unindexed directory btree. This is carefully
769 * crafted so that the rec_len and name_len records of an ocfs2_dir_entry are
770 * mirrored. That way, the directory manipulation code needs a minimal amount
771 * of update.
772 *
773 * NOTE: Keep this structure aligned to a multiple of 4 bytes.
774 */
775struct ocfs2_dir_block_trailer {
776/*00*/ __le64 db_compat_inode; /* Always zero. Was inode */
777
778 __le16 db_compat_rec_len; /* Backwards compatible with
779 * ocfs2_dir_entry. */
780 __u8 db_compat_name_len; /* Always zero. Was name_len */
781 __u8 db_reserved0;
782 __le16 db_reserved1;
783 __le16 db_free_rec_len; /* Size of largest empty hole
784 * in this block. (unused) */
785/*10*/ __u8 db_signature[8]; /* Signature for verification */
786 __le64 db_reserved2;
787 __le64 db_free_next; /* Next block in list (unused) */
788/*20*/ __le64 db_blkno; /* Offset on disk, in blocks */
789 __le64 db_parent_dinode; /* dinode which owns me, in
790 blocks */
Joel Beckerc175a512008-12-10 17:58:22 -0800791/*30*/ struct ocfs2_block_check db_check; /* Error checking */
Mark Fasheh87d35a72008-12-10 17:36:25 -0800792/*40*/
793};
794
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800795 /*
796 * A directory entry in the indexed tree. We don't store the full name here,
797 * but instead provide a pointer to the full dirent in the unindexed tree.
798 *
799 * We also store name_len here so as to reduce the number of leaf blocks we
800 * need to search in case of collisions.
801 */
802struct ocfs2_dx_entry {
803 __le32 dx_major_hash; /* Used to find logical
804 * cluster in index */
805 __le32 dx_minor_hash; /* Lower bits used to find
806 * block in cluster */
807 __le64 dx_dirent_blk; /* Physical block in unindexed
808 * tree holding this dirent. */
809};
810
811struct ocfs2_dx_entry_list {
812 __le32 de_reserved;
813 __le16 de_count; /* Maximum number of entries
814 * possible in de_entries */
815 __le16 de_num_used; /* Current number of
816 * de_entries entries */
817 struct ocfs2_dx_entry de_entries[0]; /* Indexed dir entries
818 * in a packed array of
819 * length de_num_used */
820};
821
Mark Fasheh4ed8a6b2008-11-24 17:02:08 -0800822#define OCFS2_DX_FLAG_INLINE 0x01
823
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800824/*
825 * A directory indexing block. Each indexed directory has one of these,
826 * pointed to by ocfs2_dinode.
827 *
828 * This block stores an indexed btree root, and a set of free space
829 * start-of-list pointers.
830 */
831struct ocfs2_dx_root_block {
832 __u8 dr_signature[8]; /* Signature for verification */
833 struct ocfs2_block_check dr_check; /* Error checking */
834 __le16 dr_suballoc_slot; /* Slot suballocator this
835 * block belongs to. */
836 __le16 dr_suballoc_bit; /* Bit offset in suballocator
837 * block group */
838 __le32 dr_fs_generation; /* Must match super block */
839 __le64 dr_blkno; /* Offset on disk, in blocks */
840 __le64 dr_last_eb_blk; /* Pointer to last
841 * extent block */
842 __le32 dr_clusters; /* Clusters allocated
843 * to the indexed tree. */
Mark Fasheh4ed8a6b2008-11-24 17:02:08 -0800844 __u8 dr_flags; /* OCFS2_DX_FLAG_* flags */
845 __u8 dr_reserved0;
846 __le16 dr_reserved1;
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800847 __le64 dr_dir_blkno; /* Pointer to parent inode */
Mark Fashehe3a93c22009-02-17 15:29:35 -0800848 __le32 dr_num_entries; /* Total number of
849 * names stored in
850 * this directory.*/
851 __le32 dr_reserved2;
Mark Fashehe7c17e42009-01-29 18:17:46 -0800852 __le64 dr_free_blk; /* Pointer to head of free
853 * unindexed block list. */
854 __le64 dr_reserved3[15];
Mark Fasheh4ed8a6b2008-11-24 17:02:08 -0800855 union {
856 struct ocfs2_extent_list dr_list; /* Keep this aligned to 128
857 * bits for maximum space
858 * efficiency. */
859 struct ocfs2_dx_entry_list dr_entries; /* In-root-block list of
860 * entries. We grow out
861 * to extents if this
862 * gets too big. */
863 };
Mark Fasheh9b7895e2008-11-12 16:27:44 -0800864};
865
866/*
867 * The header of a leaf block in the indexed tree.
868 */
869struct ocfs2_dx_leaf {
870 __u8 dl_signature[8];/* Signature for verification */
871 struct ocfs2_block_check dl_check; /* Error checking */
872 __le64 dl_blkno; /* Offset on disk, in blocks */
873 __le32 dl_fs_generation;/* Must match super block */
874 __le32 dl_reserved0;
875 __le64 dl_reserved1;
876 struct ocfs2_dx_entry_list dl_list;
877};
878
Mark Fasheh87d35a72008-12-10 17:36:25 -0800879/*
Mark Fashehccd979b2005-12-15 14:31:24 -0800880 * On disk allocator group structure for OCFS2
881 */
882struct ocfs2_group_desc
883{
884/*00*/ __u8 bg_signature[8]; /* Signature for validation */
885 __le16 bg_size; /* Size of included bitmap in
886 bytes. */
887 __le16 bg_bits; /* Bits represented by this
888 group. */
889 __le16 bg_free_bits_count; /* Free bits count */
890 __le16 bg_chain; /* What chain I am in. */
891/*10*/ __le32 bg_generation;
892 __le32 bg_reserved1;
893 __le64 bg_next_group; /* Next group in my list, in
894 blocks */
895/*20*/ __le64 bg_parent_dinode; /* dinode which owns me, in
896 blocks */
897 __le64 bg_blkno; /* Offset on disk, in blocks */
Joel Beckerab552d52008-10-16 17:50:30 -0700898/*30*/ struct ocfs2_block_check bg_check; /* Error checking */
899 __le64 bg_reserved2;
Mark Fashehccd979b2005-12-15 14:31:24 -0800900/*40*/ __u8 bg_bitmap[0];
901};
902
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800903/*
904 * On disk extended attribute structure for OCFS2.
905 */
906
907/*
908 * ocfs2_xattr_entry indicates one extend attribute.
909 *
910 * Note that it can be stored in inode, one block or one xattr bucket.
911 */
912struct ocfs2_xattr_entry {
913 __le32 xe_name_hash; /* hash value of xattr prefix+suffix. */
Tao Ma8573f792008-10-24 22:24:17 +0800914 __le16 xe_name_offset; /* byte offset from the 1st entry in the
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800915 local xattr storage(inode, xattr block or
916 xattr bucket). */
917 __u8 xe_name_len; /* xattr name len, does't include prefix. */
Tao Ma8573f792008-10-24 22:24:17 +0800918 __u8 xe_type; /* the low 7 bits indicate the name prefix
919 * type and the highest bit indicates whether
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800920 * the EA is stored in the local storage. */
921 __le64 xe_value_size; /* real xattr value length. */
922};
923
924/*
925 * On disk structure for xattr header.
926 *
927 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
928 * the local xattr storage.
929 */
930struct ocfs2_xattr_header {
931 __le16 xh_count; /* contains the count of how
932 many records are in the
933 local xattr storage. */
Tao Ma0c044f02008-08-18 17:38:50 +0800934 __le16 xh_free_start; /* current offset for storing
935 xattr. */
936 __le16 xh_name_value_len; /* total length of name/value
937 length in this bucket. */
Tao Ma8573f792008-10-24 22:24:17 +0800938 __le16 xh_num_buckets; /* Number of xattr buckets
939 in this extent record,
940 only valid in the first
941 bucket. */
Joel Beckerab552d52008-10-16 17:50:30 -0700942 struct ocfs2_block_check xh_check; /* Error checking
943 (Note, this is only
944 used for xattr
945 buckets. A block uses
946 xb_check and sets
947 this field to zero.) */
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800948 struct ocfs2_xattr_entry xh_entries[0]; /* xattr entry list. */
949};
950
951/*
952 * On disk structure for xattr value root.
953 *
Tao Ma8573f792008-10-24 22:24:17 +0800954 * When an xattr's value is large enough, it is stored in an external
955 * b-tree like file data. The xattr value root points to this structure.
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800956 */
957struct ocfs2_xattr_value_root {
958/*00*/ __le32 xr_clusters; /* clusters covered by xattr value. */
959 __le32 xr_reserved0;
960 __le64 xr_last_eb_blk; /* Pointer to last extent block */
961/*10*/ struct ocfs2_extent_list xr_list; /* Extent record list */
962};
963
964/*
965 * On disk structure for xattr tree root.
966 *
967 * It is used when there are too many extended attributes for one file. These
968 * attributes will be organized and stored in an indexed-btree.
969 */
970struct ocfs2_xattr_tree_root {
971/*00*/ __le32 xt_clusters; /* clusters covered by xattr. */
972 __le32 xt_reserved0;
973 __le64 xt_last_eb_blk; /* Pointer to last extent block */
974/*10*/ struct ocfs2_extent_list xt_list; /* Extent record list */
975};
976
Tiger Yangcf1d6c72008-08-18 17:11:00 +0800977#define OCFS2_XATTR_INDEXED 0x1
978#define OCFS2_HASH_SHIFT 5
979#define OCFS2_XATTR_ROUND 3
980#define OCFS2_XATTR_SIZE(size) (((size) + OCFS2_XATTR_ROUND) & \
981 ~(OCFS2_XATTR_ROUND))
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800982
Tao Ma0c044f02008-08-18 17:38:50 +0800983#define OCFS2_XATTR_BUCKET_SIZE 4096
984#define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET (OCFS2_XATTR_BUCKET_SIZE \
985 / OCFS2_MIN_BLOCKSIZE)
986
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800987/*
988 * On disk structure for xattr block.
989 */
990struct ocfs2_xattr_block {
991/*00*/ __u8 xb_signature[8]; /* Signature for verification */
992 __le16 xb_suballoc_slot; /* Slot suballocator this
993 block belongs to. */
994 __le16 xb_suballoc_bit; /* Bit offset in suballocator
995 block group */
996 __le32 xb_fs_generation; /* Must match super block */
997/*10*/ __le64 xb_blkno; /* Offset on disk, in blocks */
Joel Beckerab552d52008-10-16 17:50:30 -0700998 struct ocfs2_block_check xb_check; /* Error checking */
Tao Ma5a7bc8e2008-08-18 17:38:46 +0800999/*20*/ __le16 xb_flags; /* Indicates whether this block contains
1000 real xattr or a xattr tree. */
1001 __le16 xb_reserved0;
1002 __le32 xb_reserved1;
1003 __le64 xb_reserved2;
1004/*30*/ union {
1005 struct ocfs2_xattr_header xb_header; /* xattr header if this
1006 block contains xattr */
1007 struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
1008 block cotains xattr
1009 tree. */
1010 } xb_attrs;
1011};
1012
1013#define OCFS2_XATTR_ENTRY_LOCAL 0x80
1014#define OCFS2_XATTR_TYPE_MASK 0x7F
1015static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
1016 int local)
1017{
1018 if (local)
1019 xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
1020 else
1021 xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
1022}
1023
1024static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
1025{
1026 return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
1027}
1028
1029static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
1030{
1031 xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
1032}
1033
1034static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
1035{
1036 return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
1037}
1038
Jan Kara9e33d692008-08-25 19:56:50 +02001039/*
1040 * On disk structures for global quota file
1041 */
1042
1043/* Magic numbers and known versions for global quota files */
1044#define OCFS2_GLOBAL_QMAGICS {\
1045 0x0cf52470, /* USRQUOTA */ \
1046 0x0cf52471 /* GRPQUOTA */ \
1047}
1048
1049#define OCFS2_GLOBAL_QVERSIONS {\
1050 0, \
1051 0, \
1052}
1053
1054
1055/* Each block of each quota file has a certain fixed number of bytes reserved
1056 * for OCFS2 internal use at its end. OCFS2 can use it for things like
1057 * checksums, etc. */
1058#define OCFS2_QBLK_RESERVED_SPACE 8
1059
1060/* Generic header of all quota files */
1061struct ocfs2_disk_dqheader {
1062 __le32 dqh_magic; /* Magic number identifying file */
1063 __le32 dqh_version; /* Quota format version */
1064};
1065
1066#define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1067
1068/* Information header of global quota file (immediately follows the generic
1069 * header) */
1070struct ocfs2_global_disk_dqinfo {
1071/*00*/ __le32 dqi_bgrace; /* Grace time for space softlimit excess */
1072 __le32 dqi_igrace; /* Grace time for inode softlimit excess */
1073 __le32 dqi_syncms; /* Time after which we sync local changes to
1074 * global quota file */
1075 __le32 dqi_blocks; /* Number of blocks in quota file */
1076/*10*/ __le32 dqi_free_blk; /* First free block in quota file */
1077 __le32 dqi_free_entry; /* First block with free dquot entry in quota
1078 * file */
1079};
1080
1081/* Structure with global user / group information. We reserve some space
1082 * for future use. */
1083struct ocfs2_global_disk_dqblk {
1084/*00*/ __le32 dqb_id; /* ID the structure belongs to */
1085 __le32 dqb_use_count; /* Number of nodes having reference to this structure */
1086 __le64 dqb_ihardlimit; /* absolute limit on allocated inodes */
1087/*10*/ __le64 dqb_isoftlimit; /* preferred inode limit */
1088 __le64 dqb_curinodes; /* current # allocated inodes */
1089/*20*/ __le64 dqb_bhardlimit; /* absolute limit on disk space */
1090 __le64 dqb_bsoftlimit; /* preferred limit on disk space */
1091/*30*/ __le64 dqb_curspace; /* current space occupied */
1092 __le64 dqb_btime; /* time limit for excessive disk use */
1093/*40*/ __le64 dqb_itime; /* time limit for excessive inode use */
1094 __le64 dqb_pad1;
1095/*50*/ __le64 dqb_pad2;
1096};
1097
1098/*
1099 * On-disk structures for local quota file
1100 */
1101
1102/* Magic numbers and known versions for local quota files */
1103#define OCFS2_LOCAL_QMAGICS {\
1104 0x0cf524c0, /* USRQUOTA */ \
1105 0x0cf524c1 /* GRPQUOTA */ \
1106}
1107
1108#define OCFS2_LOCAL_QVERSIONS {\
1109 0, \
1110 0, \
1111}
1112
1113/* Quota flags in dqinfo header */
1114#define OLQF_CLEAN 0x0001 /* Quota file is empty (this should be after\
1115 * quota has been cleanly turned off) */
1116
1117#define OCFS2_LOCAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1118
1119/* Information header of local quota file (immediately follows the generic
1120 * header) */
1121struct ocfs2_local_disk_dqinfo {
1122 __le32 dqi_flags; /* Flags for quota file */
1123 __le32 dqi_chunks; /* Number of chunks of quota structures
1124 * with a bitmap */
1125 __le32 dqi_blocks; /* Number of blocks allocated for quota file */
1126};
1127
1128/* Header of one chunk of a quota file */
1129struct ocfs2_local_disk_chunk {
1130 __le32 dqc_free; /* Number of free entries in the bitmap */
1131 u8 dqc_bitmap[0]; /* Bitmap of entries in the corresponding
1132 * chunk of quota file */
1133};
1134
1135/* One entry in local quota file */
1136struct ocfs2_local_disk_dqblk {
1137/*00*/ __le64 dqb_id; /* id this quota applies to */
1138 __le64 dqb_spacemod; /* Change in the amount of used space */
1139/*10*/ __le64 dqb_inodemod; /* Change in the amount of used inodes */
1140};
1141
Joel Beckerab552d52008-10-16 17:50:30 -07001142
1143/*
1144 * The quota trailer lives at the end of each quota block.
1145 */
1146
1147struct ocfs2_disk_dqtrailer {
1148/*00*/ struct ocfs2_block_check dq_check; /* Error checking */
1149/*08*/ /* Cannot be larger than OCFS2_QBLK_RESERVED_SPACE */
1150};
1151
1152static inline struct ocfs2_disk_dqtrailer *ocfs2_block_dqtrailer(int blocksize,
1153 void *buf)
1154{
1155 char *ptr = buf;
1156 ptr += blocksize - OCFS2_QBLK_RESERVED_SPACE;
1157
1158 return (struct ocfs2_disk_dqtrailer *)ptr;
1159}
1160
Mark Fashehccd979b2005-12-15 14:31:24 -08001161#ifdef __KERNEL__
1162static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
1163{
1164 return sb->s_blocksize -
1165 offsetof(struct ocfs2_dinode, id2.i_symlink);
1166}
1167
Tiger Yangfdd77702008-08-18 17:08:55 +08001168static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
1169 struct ocfs2_dinode *di)
1170{
1171 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1172
1173 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1174 return sb->s_blocksize -
1175 offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1176 xattrsize;
1177 else
1178 return sb->s_blocksize -
1179 offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1180}
1181
Mark Fashehccd979b2005-12-15 14:31:24 -08001182static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
1183{
1184 int size;
1185
1186 size = sb->s_blocksize -
1187 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1188
1189 return size / sizeof(struct ocfs2_extent_rec);
1190}
1191
Tiger Yangfdd77702008-08-18 17:08:55 +08001192static inline int ocfs2_extent_recs_per_inode_with_xattr(
1193 struct super_block *sb,
1194 struct ocfs2_dinode *di)
1195{
1196 int size;
1197 unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1198
1199 if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1200 size = sb->s_blocksize -
1201 offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
1202 xattrsize;
1203 else
1204 size = sb->s_blocksize -
1205 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1206
1207 return size / sizeof(struct ocfs2_extent_rec);
1208}
1209
Mark Fasheh9b7895e2008-11-12 16:27:44 -08001210static inline int ocfs2_extent_recs_per_dx_root(struct super_block *sb)
1211{
1212 int size;
1213
1214 size = sb->s_blocksize -
1215 offsetof(struct ocfs2_dx_root_block, dr_list.l_recs);
1216
1217 return size / sizeof(struct ocfs2_extent_rec);
1218}
1219
Mark Fashehccd979b2005-12-15 14:31:24 -08001220static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
1221{
1222 int size;
1223
1224 size = sb->s_blocksize -
1225 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1226
1227 return size / sizeof(struct ocfs2_chain_rec);
1228}
1229
1230static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
1231{
1232 int size;
1233
1234 size = sb->s_blocksize -
1235 offsetof(struct ocfs2_extent_block, h_list.l_recs);
1236
1237 return size / sizeof(struct ocfs2_extent_rec);
1238}
1239
Mark Fasheh9b7895e2008-11-12 16:27:44 -08001240static inline int ocfs2_dx_entries_per_leaf(struct super_block *sb)
1241{
1242 int size;
1243
1244 size = sb->s_blocksize -
1245 offsetof(struct ocfs2_dx_leaf, dl_list.de_entries);
1246
1247 return size / sizeof(struct ocfs2_dx_entry);
1248}
1249
Mark Fasheh4ed8a6b2008-11-24 17:02:08 -08001250static inline int ocfs2_dx_entries_per_root(struct super_block *sb)
1251{
1252 int size;
1253
1254 size = sb->s_blocksize -
1255 offsetof(struct ocfs2_dx_root_block, dr_entries.de_entries);
1256
1257 return size / sizeof(struct ocfs2_dx_entry);
1258}
1259
Mark Fashehccd979b2005-12-15 14:31:24 -08001260static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
1261{
1262 u16 size;
1263
1264 size = sb->s_blocksize -
1265 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1266
1267 return size;
1268}
1269
1270static inline int ocfs2_group_bitmap_size(struct super_block *sb)
1271{
1272 int size;
1273
1274 size = sb->s_blocksize -
1275 offsetof(struct ocfs2_group_desc, bg_bitmap);
1276
1277 return size;
1278}
1279
1280static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
1281{
1282 int size;
1283
1284 size = sb->s_blocksize -
1285 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1286
1287 return size / sizeof(struct ocfs2_truncate_rec);
1288}
Mark Fasheh50af94b2007-01-21 14:44:59 -08001289
1290static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
1291{
1292 u64 offset = OCFS2_BACKUP_SB_START;
1293
1294 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1295 offset <<= (2 * index);
Mark Fasheha8a75a22007-01-26 10:46:59 -08001296 offset >>= sb->s_blocksize_bits;
Mark Fasheh50af94b2007-01-21 14:44:59 -08001297 return offset;
1298 }
1299
1300 return 0;
1301
1302}
Tao Ma0c044f02008-08-18 17:38:50 +08001303
1304static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
1305{
1306 int size;
1307
1308 size = sb->s_blocksize -
1309 offsetof(struct ocfs2_xattr_block,
1310 xb_attrs.xb_root.xt_list.l_recs);
1311
1312 return size / sizeof(struct ocfs2_extent_rec);
1313}
Mark Fashehccd979b2005-12-15 14:31:24 -08001314#else
1315static inline int ocfs2_fast_symlink_chars(int blocksize)
1316{
1317 return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
1318}
1319
Mark Fasheh15b1e362007-09-07 13:58:15 -07001320static inline int ocfs2_max_inline_data(int blocksize)
1321{
1322 return blocksize - offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1323}
1324
Mark Fashehccd979b2005-12-15 14:31:24 -08001325static inline int ocfs2_extent_recs_per_inode(int blocksize)
1326{
1327 int size;
1328
1329 size = blocksize -
1330 offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1331
1332 return size / sizeof(struct ocfs2_extent_rec);
1333}
1334
1335static inline int ocfs2_chain_recs_per_inode(int blocksize)
1336{
1337 int size;
1338
1339 size = blocksize -
1340 offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1341
1342 return size / sizeof(struct ocfs2_chain_rec);
1343}
1344
1345static inline int ocfs2_extent_recs_per_eb(int blocksize)
1346{
1347 int size;
1348
1349 size = blocksize -
1350 offsetof(struct ocfs2_extent_block, h_list.l_recs);
1351
1352 return size / sizeof(struct ocfs2_extent_rec);
1353}
1354
1355static inline int ocfs2_local_alloc_size(int blocksize)
1356{
1357 int size;
1358
1359 size = blocksize -
1360 offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1361
1362 return size;
1363}
1364
1365static inline int ocfs2_group_bitmap_size(int blocksize)
1366{
1367 int size;
1368
1369 size = blocksize -
1370 offsetof(struct ocfs2_group_desc, bg_bitmap);
1371
1372 return size;
1373}
1374
1375static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1376{
1377 int size;
1378
1379 size = blocksize -
1380 offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1381
1382 return size / sizeof(struct ocfs2_truncate_rec);
1383}
Mark Fasheh50af94b2007-01-21 14:44:59 -08001384
1385static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1386{
1387 uint64_t offset = OCFS2_BACKUP_SB_START;
1388
1389 if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1390 offset <<= (2 * index);
1391 offset /= blocksize;
1392 return offset;
1393 }
1394
1395 return 0;
1396}
Tao Ma0c044f02008-08-18 17:38:50 +08001397
1398static inline int ocfs2_xattr_recs_per_xb(int blocksize)
1399{
1400 int size;
1401
1402 size = blocksize -
1403 offsetof(struct ocfs2_xattr_block,
1404 xb_attrs.xb_root.xt_list.l_recs);
1405
1406 return size / sizeof(struct ocfs2_extent_rec);
1407}
Mark Fashehccd979b2005-12-15 14:31:24 -08001408#endif /* __KERNEL__ */
1409
1410
1411static inline int ocfs2_system_inode_is_global(int type)
1412{
1413 return ((type >= 0) &&
1414 (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1415}
1416
1417static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1418 int type, int slot)
1419{
1420 int chars;
1421
1422 /*
1423 * Global system inodes can only have one copy. Everything
1424 * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1425 * list has a copy per slot.
1426 */
1427 if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
Joel Beckerfe9f3872008-06-12 22:39:18 -07001428 chars = snprintf(buf, len, "%s",
Mark Fashehccd979b2005-12-15 14:31:24 -08001429 ocfs2_system_inodes[type].si_name);
1430 else
1431 chars = snprintf(buf, len,
1432 ocfs2_system_inodes[type].si_name,
1433 slot);
1434
1435 return chars;
1436}
1437
1438static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1439 umode_t mode)
1440{
1441 de->file_type = ocfs2_type_by_mode[(mode & S_IFMT)>>S_SHIFT];
1442}
1443
1444#endif /* _OCFS2_FS_H */
1445