1/* SPDX-License-Identifier: GPL-2.0-only */
2/*
3 * ocfs2_fs.h
4 *
5 * On-disk structures for OCFS2.
6 *
7 * Copyright (C) 2002, 2004 Oracle.  All rights reserved.
8 */
9
10#ifndef _OCFS2_FS_H
11#define _OCFS2_FS_H
12
13#include <linux/magic.h>
14
15/* Version */
16#define OCFS2_MAJOR_REV_LEVEL		0
17#define OCFS2_MINOR_REV_LEVEL          	90
18
19/*
20 * An OCFS2 volume starts this way:
21 * Sector 0: Valid ocfs1_vol_disk_hdr that cleanly fails to mount OCFS.
22 * Sector 1: Valid ocfs1_vol_label that cleanly fails to mount OCFS.
23 * Block OCFS2_SUPER_BLOCK_BLKNO: OCFS2 superblock.
24 *
25 * All other structures are found from the superblock information.
26 *
27 * OCFS2_SUPER_BLOCK_BLKNO is in blocks, not sectors.  eg, for a
28 * blocksize of 2K, it is 4096 bytes into disk.
29 */
30#define OCFS2_SUPER_BLOCK_BLKNO		2
31
32/*
33 * Cluster size limits. The maximum is kept arbitrarily at 1 MB, and could
34 * grow if needed.
35 */
36#define OCFS2_MIN_CLUSTERSIZE		4096
37#define OCFS2_MAX_CLUSTERSIZE		1048576
38
39/*
40 * Blocks cannot be bigger than clusters, so the maximum blocksize is the
41 * minimum cluster size.
42 */
43#define OCFS2_MIN_BLOCKSIZE		512
44#define OCFS2_MAX_BLOCKSIZE		OCFS2_MIN_CLUSTERSIZE
45
46/* Object signatures */
47#define OCFS2_SUPER_BLOCK_SIGNATURE	"OCFSV2"
48#define OCFS2_INODE_SIGNATURE		"INODE01"
49#define OCFS2_EXTENT_BLOCK_SIGNATURE	"EXBLK01"
50#define OCFS2_GROUP_DESC_SIGNATURE      "GROUP01"
51#define OCFS2_XATTR_BLOCK_SIGNATURE	"XATTR01"
52#define OCFS2_DIR_TRAILER_SIGNATURE	"DIRTRL1"
53#define OCFS2_DX_ROOT_SIGNATURE		"DXDIR01"
54#define OCFS2_DX_LEAF_SIGNATURE		"DXLEAF1"
55#define OCFS2_REFCOUNT_BLOCK_SIGNATURE	"REFCNT1"
56
57/* Compatibility flags */
58#define OCFS2_HAS_COMPAT_FEATURE(sb,mask)			\
59	( OCFS2_SB(sb)->s_feature_compat & (mask) )
60#define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask)			\
61	( OCFS2_SB(sb)->s_feature_ro_compat & (mask) )
62#define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask)			\
63	( OCFS2_SB(sb)->s_feature_incompat & (mask) )
64#define OCFS2_SET_COMPAT_FEATURE(sb,mask)			\
65	OCFS2_SB(sb)->s_feature_compat |= (mask)
66#define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask)			\
67	OCFS2_SB(sb)->s_feature_ro_compat |= (mask)
68#define OCFS2_SET_INCOMPAT_FEATURE(sb,mask)			\
69	OCFS2_SB(sb)->s_feature_incompat |= (mask)
70#define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask)			\
71	OCFS2_SB(sb)->s_feature_compat &= ~(mask)
72#define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
73	OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask)
74#define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
75	OCFS2_SB(sb)->s_feature_incompat &= ~(mask)
76
77#define OCFS2_FEATURE_COMPAT_SUPP	(OCFS2_FEATURE_COMPAT_BACKUP_SB	\
78					 | OCFS2_FEATURE_COMPAT_JBD2_SB)
79#define OCFS2_FEATURE_INCOMPAT_SUPP	(OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \
80					 | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \
81					 | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \
82					 | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \
83					 | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \
84					 | OCFS2_FEATURE_INCOMPAT_XATTR \
85					 | OCFS2_FEATURE_INCOMPAT_META_ECC \
86					 | OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS \
87					 | OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE \
88					 | OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG	\
89					 | OCFS2_FEATURE_INCOMPAT_CLUSTERINFO \
90					 | OCFS2_FEATURE_INCOMPAT_APPEND_DIO)
91#define OCFS2_FEATURE_RO_COMPAT_SUPP	(OCFS2_FEATURE_RO_COMPAT_UNWRITTEN \
92					 | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \
93					 | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)
94
95/*
96 * Heartbeat-only devices are missing journals and other files.  The
97 * filesystem driver can't load them, but the library can.  Never put
98 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
99 */
100#define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV	0x0002
101
102/*
103 * tunefs sets this incompat flag before starting the resize and clears it
104 * at the end. This flag protects users from inadvertently mounting the fs
105 * after an aborted run without fsck-ing.
106 */
107#define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG    0x0004
108
109/* Used to denote a non-clustered volume */
110#define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT	0x0008
111
112/* Support for sparse allocation in b-trees */
113#define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC	0x0010
114
115/*
116 * Tunefs sets this incompat flag before starting an operation which
117 * would require cleanup on abort. This is done to protect users from
118 * inadvertently mounting the fs after an aborted run without
119 * fsck-ing.
120 *
121 * s_tunefs_flags on the super block describes precisely which
122 * operations were in progress.
123 */
124#define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG	0x0020
125
126/* Support for data packed into inode blocks */
127#define OCFS2_FEATURE_INCOMPAT_INLINE_DATA	0x0040
128
129/*
130 * Support for alternate, userspace cluster stacks.  If set, the superblock
131 * field s_cluster_info contains a tag for the alternate stack in use as
132 * well as the name of the cluster being joined.
133 * mount.ocfs2 must pass in a matching stack name.
134 *
135 * If not set, the classic stack will be used.  This is compatbile with
136 * all older versions.
137 */
138#define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK	0x0080
139
140/* Support for the extended slot map */
141#define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
142
143/* Support for extended attributes */
144#define OCFS2_FEATURE_INCOMPAT_XATTR		0x0200
145
146/* Support for indexed directores */
147#define OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS	0x0400
148
149/* Metadata checksum and error correction */
150#define OCFS2_FEATURE_INCOMPAT_META_ECC		0x0800
151
152/* Refcount tree support */
153#define OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE	0x1000
154
155/* Discontiguous block groups */
156#define OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG	0x2000
157
158/*
159 * Incompat bit to indicate useable clusterinfo with stackflags for all
160 * cluster stacks (userspace adnd o2cb). If this bit is set,
161 * INCOMPAT_USERSPACE_STACK becomes superfluous and thus should not be set.
162 */
163#define OCFS2_FEATURE_INCOMPAT_CLUSTERINFO	0x4000
164
165/*
166 * Append Direct IO support
167 */
168#define OCFS2_FEATURE_INCOMPAT_APPEND_DIO	0x8000
169
170/*
171 * backup superblock flag is used to indicate that this volume
172 * has backup superblocks.
173 */
174#define OCFS2_FEATURE_COMPAT_BACKUP_SB		0x0001
175
176/*
177 * The filesystem will correctly handle journal feature bits.
178 */
179#define OCFS2_FEATURE_COMPAT_JBD2_SB		0x0002
180
181/*
182 * Unwritten extents support.
183 */
184#define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN	0x0001
185
186/*
187 * Maintain quota information for this filesystem
188 */
189#define OCFS2_FEATURE_RO_COMPAT_USRQUOTA	0x0002
190#define OCFS2_FEATURE_RO_COMPAT_GRPQUOTA	0x0004
191
192
193/* The byte offset of the first backup block will be 1G.
194 * The following will be 4G, 16G, 64G, 256G and 1T.
195 */
196#define OCFS2_BACKUP_SB_START			1 << 30
197
198/* the max backup superblock nums */
199#define OCFS2_MAX_BACKUP_SUPERBLOCKS	6
200
201/*
202 * Flags on ocfs2_super_block.s_tunefs_flags
203 */
204#define OCFS2_TUNEFS_INPROG_REMOVE_SLOT		0x0001	/* Removing slots */
205
206/*
207 * Flags on ocfs2_dinode.i_flags
208 */
209#define OCFS2_VALID_FL		(0x00000001)	/* Inode is valid */
210#define OCFS2_UNUSED2_FL	(0x00000002)
211#define OCFS2_ORPHANED_FL	(0x00000004)	/* On the orphan list */
212#define OCFS2_UNUSED3_FL	(0x00000008)
213/* System inode flags */
214#define OCFS2_SYSTEM_FL		(0x00000010)	/* System inode */
215#define OCFS2_SUPER_BLOCK_FL	(0x00000020)	/* Super block */
216#define OCFS2_LOCAL_ALLOC_FL	(0x00000040)	/* Slot local alloc bitmap */
217#define OCFS2_BITMAP_FL		(0x00000080)	/* Allocation bitmap */
218#define OCFS2_JOURNAL_FL	(0x00000100)	/* Slot local journal */
219#define OCFS2_HEARTBEAT_FL	(0x00000200)	/* Heartbeat area */
220#define OCFS2_CHAIN_FL		(0x00000400)	/* Chain allocator */
221#define OCFS2_DEALLOC_FL	(0x00000800)	/* Truncate log */
222#define OCFS2_QUOTA_FL		(0x00001000)	/* Quota file */
223#define OCFS2_DIO_ORPHANED_FL	(0X00002000)	/* On the orphan list especially
224						 * for dio */
225
226/*
227 * Flags on ocfs2_dinode.i_dyn_features
228 *
229 * These can change much more often than i_flags. When adding flags,
230 * keep in mind that i_dyn_features is only 16 bits wide.
231 */
232#define OCFS2_INLINE_DATA_FL	(0x0001)	/* Data stored in inode block */
233#define OCFS2_HAS_XATTR_FL	(0x0002)
234#define OCFS2_INLINE_XATTR_FL	(0x0004)
235#define OCFS2_INDEXED_DIR_FL	(0x0008)
236#define OCFS2_HAS_REFCOUNT_FL   (0x0010)
237
238/* Inode attributes, keep in sync with EXT2 */
239#define OCFS2_SECRM_FL			FS_SECRM_FL	/* Secure deletion */
240#define OCFS2_UNRM_FL			FS_UNRM_FL	/* Undelete */
241#define OCFS2_COMPR_FL			FS_COMPR_FL	/* Compress file */
242#define OCFS2_SYNC_FL			FS_SYNC_FL	/* Synchronous updates */
243#define OCFS2_IMMUTABLE_FL		FS_IMMUTABLE_FL	/* Immutable file */
244#define OCFS2_APPEND_FL			FS_APPEND_FL	/* writes to file may only append */
245#define OCFS2_NODUMP_FL			FS_NODUMP_FL	/* do not dump file */
246#define OCFS2_NOATIME_FL		FS_NOATIME_FL	/* do not update atime */
247/* Reserved for compression usage... */
248#define OCFS2_DIRTY_FL			FS_DIRTY_FL
249#define OCFS2_COMPRBLK_FL		FS_COMPRBLK_FL	/* One or more compressed clusters */
250#define OCFS2_NOCOMP_FL			FS_NOCOMP_FL	/* Don't compress */
251#define OCFS2_ECOMPR_FL			FS_ECOMPR_FL	/* Compression error */
252/* End compression flags --- maybe not all used */
253#define OCFS2_BTREE_FL			FS_BTREE_FL	/* btree format dir */
254#define OCFS2_INDEX_FL			FS_INDEX_FL	/* hash-indexed directory */
255#define OCFS2_IMAGIC_FL			FS_IMAGIC_FL	/* AFS directory */
256#define OCFS2_JOURNAL_DATA_FL		FS_JOURNAL_DATA_FL /* Reserved for ext3 */
257#define OCFS2_NOTAIL_FL			FS_NOTAIL_FL	/* file tail should not be merged */
258#define OCFS2_DIRSYNC_FL		FS_DIRSYNC_FL	/* dirsync behaviour (directories only) */
259#define OCFS2_TOPDIR_FL			FS_TOPDIR_FL	/* Top of directory hierarchies*/
260#define OCFS2_RESERVED_FL		FS_RESERVED_FL	/* reserved for ext2 lib */
261
262#define OCFS2_FL_VISIBLE		FS_FL_USER_VISIBLE	/* User visible flags */
263#define OCFS2_FL_MODIFIABLE		FS_FL_USER_MODIFIABLE	/* User modifiable flags */
264
265/*
266 * Extent record flags (e_node.leaf.flags)
267 */
268#define OCFS2_EXT_UNWRITTEN		(0x01)	/* Extent is allocated but
269						 * unwritten */
270#define OCFS2_EXT_REFCOUNTED		(0x02)  /* Extent is reference
271						 * counted in an associated
272						 * refcount tree */
273
274/*
275 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
276 */
277#define OCFS2_JOURNAL_DIRTY_FL	(0x00000001)	/* Journal needs recovery */
278
279/*
280 * superblock s_state flags
281 */
282#define OCFS2_ERROR_FS		(0x00000001)	/* FS saw errors */
283
284/* Limit of space in ocfs2_dir_entry */
285#define OCFS2_MAX_FILENAME_LEN		255
286
287/* Maximum slots on an ocfs2 file system */
288#define OCFS2_MAX_SLOTS			255
289
290/* Slot map indicator for an empty slot */
291#define OCFS2_INVALID_SLOT		((u16)-1)
292
293#define OCFS2_VOL_UUID_LEN		16
294#define OCFS2_MAX_VOL_LABEL_LEN		64
295
296/* The cluster stack fields */
297#define OCFS2_STACK_LABEL_LEN		4
298#define OCFS2_CLUSTER_NAME_LEN		16
299
300/* Classic (historically speaking) cluster stack */
301#define OCFS2_CLASSIC_CLUSTER_STACK	"o2cb"
302
303/* Journal limits (in bytes) */
304#define OCFS2_MIN_JOURNAL_SIZE		(4 * 1024 * 1024)
305
306/*
307 * Inline extended attribute size (in bytes)
308 * The value chosen should be aligned to 16 byte boundaries.
309 */
310#define OCFS2_MIN_XATTR_INLINE_SIZE     256
311
312/*
313 * Cluster info flags (ocfs2_cluster_info.ci_stackflags)
314 */
315#define OCFS2_CLUSTER_O2CB_GLOBAL_HEARTBEAT	(0x01)
316
317struct ocfs2_system_inode_info {
318	char	*si_name;
319	int	si_iflags;
320	int	si_mode;
321};
322
323/* System file index */
324enum {
325	BAD_BLOCK_SYSTEM_INODE = 0,
326	GLOBAL_INODE_ALLOC_SYSTEM_INODE,
327#define OCFS2_FIRST_ONLINE_SYSTEM_INODE GLOBAL_INODE_ALLOC_SYSTEM_INODE
328	SLOT_MAP_SYSTEM_INODE,
329	HEARTBEAT_SYSTEM_INODE,
330	GLOBAL_BITMAP_SYSTEM_INODE,
331	USER_QUOTA_SYSTEM_INODE,
332	GROUP_QUOTA_SYSTEM_INODE,
333#define OCFS2_LAST_GLOBAL_SYSTEM_INODE GROUP_QUOTA_SYSTEM_INODE
334#define OCFS2_FIRST_LOCAL_SYSTEM_INODE ORPHAN_DIR_SYSTEM_INODE
335	ORPHAN_DIR_SYSTEM_INODE,
336	EXTENT_ALLOC_SYSTEM_INODE,
337	INODE_ALLOC_SYSTEM_INODE,
338	JOURNAL_SYSTEM_INODE,
339	LOCAL_ALLOC_SYSTEM_INODE,
340	TRUNCATE_LOG_SYSTEM_INODE,
341	LOCAL_USER_QUOTA_SYSTEM_INODE,
342	LOCAL_GROUP_QUOTA_SYSTEM_INODE,
343#define OCFS2_LAST_LOCAL_SYSTEM_INODE LOCAL_GROUP_QUOTA_SYSTEM_INODE
344	NUM_SYSTEM_INODES
345};
346#define NUM_GLOBAL_SYSTEM_INODES OCFS2_FIRST_LOCAL_SYSTEM_INODE
347#define NUM_LOCAL_SYSTEM_INODES	\
348		(NUM_SYSTEM_INODES - OCFS2_FIRST_LOCAL_SYSTEM_INODE)
349
350static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
351	/* Global system inodes (single copy) */
352	/* The first two are only used from userspace mfks/tunefs */
353	[BAD_BLOCK_SYSTEM_INODE]		= { "bad_blocks", 0, S_IFREG | 0644 },
354	[GLOBAL_INODE_ALLOC_SYSTEM_INODE] 	= { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
355
356	/* These are used by the running filesystem */
357	[SLOT_MAP_SYSTEM_INODE]			= { "slot_map", 0, S_IFREG | 0644 },
358	[HEARTBEAT_SYSTEM_INODE]		= { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
359	[GLOBAL_BITMAP_SYSTEM_INODE]		= { "global_bitmap", 0, S_IFREG | 0644 },
360	[USER_QUOTA_SYSTEM_INODE]		= { "aquota.user", OCFS2_QUOTA_FL, S_IFREG | 0644 },
361	[GROUP_QUOTA_SYSTEM_INODE]		= { "aquota.group", OCFS2_QUOTA_FL, S_IFREG | 0644 },
362
363	/* Slot-specific system inodes (one copy per slot) */
364	[ORPHAN_DIR_SYSTEM_INODE]		= { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
365	[EXTENT_ALLOC_SYSTEM_INODE]		= { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
366	[INODE_ALLOC_SYSTEM_INODE]		= { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
367	[JOURNAL_SYSTEM_INODE]			= { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
368	[LOCAL_ALLOC_SYSTEM_INODE]		= { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
369	[TRUNCATE_LOG_SYSTEM_INODE]		= { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 },
370	[LOCAL_USER_QUOTA_SYSTEM_INODE]		= { "aquota.user:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
371	[LOCAL_GROUP_QUOTA_SYSTEM_INODE]	= { "aquota.group:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
372};
373
374/* Parameter passed from mount.ocfs2 to module */
375#define OCFS2_HB_NONE			"heartbeat=none"
376#define OCFS2_HB_LOCAL			"heartbeat=local"
377#define OCFS2_HB_GLOBAL			"heartbeat=global"
378
379/*
380 * OCFS2_DIR_PAD defines the directory entries boundaries
381 *
382 * NOTE: It must be a multiple of 4
383 */
384#define OCFS2_DIR_PAD			4
385#define OCFS2_DIR_ROUND			(OCFS2_DIR_PAD - 1)
386#define OCFS2_DIR_MEMBER_LEN 		offsetof(struct ocfs2_dir_entry, name)
387#define OCFS2_DIR_REC_LEN(name_len)	(((name_len) + OCFS2_DIR_MEMBER_LEN + \
388                                          OCFS2_DIR_ROUND) & \
389					 ~OCFS2_DIR_ROUND)
390#define OCFS2_DIR_MIN_REC_LEN	OCFS2_DIR_REC_LEN(1)
391
392#define OCFS2_LINK_MAX		32000
393#define	OCFS2_DX_LINK_MAX	((1U << 31) - 1U)
394#define	OCFS2_LINKS_HI_SHIFT	16
395#define	OCFS2_DX_ENTRIES_MAX	(0xffffffffU)
396
397
398/*
399 * Convenience casts
400 */
401#define OCFS2_RAW_SB(dinode)		(&((dinode)->id2.i_super))
402
403/*
404 * Block checking structure.  This is used in metadata to validate the
405 * contents.  If OCFS2_FEATURE_INCOMPAT_META_ECC is not set, it is all
406 * zeros.
407 */
408struct ocfs2_block_check {
409/*00*/	__le32 bc_crc32e;	/* 802.3 Ethernet II CRC32 */
410	__le16 bc_ecc;		/* Single-error-correction parity vector.
411				   This is a simple Hamming code dependent
412				   on the blocksize.  OCFS2's maximum
413				   blocksize, 4K, requires 16 parity bits,
414				   so we fit in __le16. */
415	__le16 bc_reserved1;
416/*08*/
417};
418
419/*
420 * On disk extent record for OCFS2
421 * It describes a range of clusters on disk.
422 *
423 * Length fields are divided into interior and leaf node versions.
424 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
425 */
426struct ocfs2_extent_rec {
427/*00*/	__le32 e_cpos;		/* Offset into the file, in clusters */
428	union {
429		__le32 e_int_clusters; /* Clusters covered by all children */
430		struct {
431			__le16 e_leaf_clusters; /* Clusters covered by this
432						   extent */
433			__u8 e_reserved1;
434			__u8 e_flags; /* Extent flags */
435		};
436	};
437	__le64 e_blkno;		/* Physical disk offset, in blocks */
438/*10*/
439};
440
441struct ocfs2_chain_rec {
442	__le32 c_free;	/* Number of free bits in this chain. */
443	__le32 c_total;	/* Number of total bits in this chain */
444	__le64 c_blkno;	/* Physical disk offset (blocks) of 1st group */
445};
446
447struct ocfs2_truncate_rec {
448	__le32 t_start;		/* 1st cluster in this log */
449	__le32 t_clusters;	/* Number of total clusters covered */
450};
451
452/*
453 * On disk extent list for OCFS2 (node in the tree).  Note that this
454 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
455 * offsets are relative to ocfs2_dinode.id2.i_list or
456 * ocfs2_extent_block.h_list, respectively.
457 */
458struct ocfs2_extent_list {
459/*00*/	__le16 l_tree_depth;		/* Extent tree depth from this
460					   point.  0 means data extents
461					   hang directly off this
462					   header (a leaf)
463					   NOTE: The high 8 bits cannot be
464					   used - tree_depth is never that big.
465					*/
466	__le16 l_count;			/* Number of extent records */
467	__le16 l_next_free_rec;		/* Next unused extent slot */
468	__le16 l_reserved1;
469	__le64 l_reserved2;		/* Pad to
470					   sizeof(ocfs2_extent_rec) */
471/*10*/	struct ocfs2_extent_rec l_recs[];	/* Extent records */
472};
473
474/*
475 * On disk allocation chain list for OCFS2.  Note that this is
476 * contained inside ocfs2_dinode, so the offsets are relative to
477 * ocfs2_dinode.id2.i_chain.
478 */
479struct ocfs2_chain_list {
480/*00*/	__le16 cl_cpg;			/* Clusters per Block Group */
481	__le16 cl_bpc;			/* Bits per cluster */
482	__le16 cl_count;		/* Total chains in this list */
483	__le16 cl_next_free_rec;	/* Next unused chain slot */
484	__le64 cl_reserved1;
485/*10*/	struct ocfs2_chain_rec cl_recs[];	/* Chain records */
486};
487
488/*
489 * On disk deallocation log for OCFS2.  Note that this is
490 * contained inside ocfs2_dinode, so the offsets are relative to
491 * ocfs2_dinode.id2.i_dealloc.
492 */
493struct ocfs2_truncate_log {
494/*00*/	__le16 tl_count;		/* Total records in this log */
495	__le16 tl_used;			/* Number of records in use */
496	__le32 tl_reserved1;
497/*08*/	struct ocfs2_truncate_rec tl_recs[];	/* Truncate records */
498};
499
500/*
501 * On disk extent block (indirect block) for OCFS2
502 */
503struct ocfs2_extent_block
504{
505/*00*/	__u8 h_signature[8];		/* Signature for verification */
506	struct ocfs2_block_check h_check;	/* Error checking */
507/*10*/	__le16 h_suballoc_slot;		/* Slot suballocator this
508					   extent_header belongs to */
509	__le16 h_suballoc_bit;		/* Bit offset in suballocator
510					   block group */
511	__le32 h_fs_generation;		/* Must match super block */
512	__le64 h_blkno;			/* Offset on disk, in blocks */
513/*20*/	__le64 h_suballoc_loc;		/* Suballocator block group this
514					   eb belongs to.  Only valid
515					   if allocated from a
516					   discontiguous block group */
517	__le64 h_next_leaf_blk;		/* Offset on disk, in blocks,
518					   of next leaf header pointing
519					   to data */
520/*30*/	struct ocfs2_extent_list h_list;	/* Extent record list */
521/* Actual on-disk size is one block */
522};
523
524/*
525 * On disk slot map for OCFS2.  This defines the contents of the "slot_map"
526 * system file.  A slot is valid if it contains a node number >= 0.  The
527 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT.  This marks a slot empty.
528 */
529struct ocfs2_slot_map {
530/*00*/	DECLARE_FLEX_ARRAY(__le16, sm_slots);
531/*
532 * Actual on-disk size is one block.  OCFS2_MAX_SLOTS is 255,
533 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
534 */
535};
536
537struct ocfs2_extended_slot {
538/*00*/	__u8	es_valid;
539	__u8	es_reserved1[3];
540	__le32	es_node_num;
541/*08*/
542};
543
544/*
545 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
546 * is set.  It separates out the valid marker from the node number, and
547 * has room to grow.  Unlike the old slot map, this format is defined by
548 * i_size.
549 */
550struct ocfs2_slot_map_extended {
551/*00*/	DECLARE_FLEX_ARRAY(struct ocfs2_extended_slot, se_slots);
552/*
553 * Actual size is i_size of the slot_map system file.  It should
554 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
555 */
556};
557
558/*
559 * ci_stackflags is only valid if the incompat bit
560 * OCFS2_FEATURE_INCOMPAT_CLUSTERINFO is set.
561 */
562struct ocfs2_cluster_info {
563/*00*/	__u8   ci_stack[OCFS2_STACK_LABEL_LEN];
564	union {
565		__le32 ci_reserved;
566		struct {
567			__u8 ci_stackflags;
568			__u8 ci_reserved1;
569			__u8 ci_reserved2;
570			__u8 ci_reserved3;
571		};
572	};
573/*08*/	__u8   ci_cluster[OCFS2_CLUSTER_NAME_LEN];
574/*18*/
575};
576
577/*
578 * On disk superblock for OCFS2
579 * Note that it is contained inside an ocfs2_dinode, so all offsets
580 * are relative to the start of ocfs2_dinode.id2.
581 */
582struct ocfs2_super_block {
583/*00*/	__le16 s_major_rev_level;
584	__le16 s_minor_rev_level;
585	__le16 s_mnt_count;
586	__le16 s_max_mnt_count;
587	__le16 s_state;			/* File system state */
588	__le16 s_errors;			/* Behaviour when detecting errors */
589	__le32 s_checkinterval;		/* Max time between checks */
590/*10*/	__le64 s_lastcheck;		/* Time of last check */
591	__le32 s_creator_os;		/* OS */
592	__le32 s_feature_compat;		/* Compatible feature set */
593/*20*/	__le32 s_feature_incompat;	/* Incompatible feature set */
594	__le32 s_feature_ro_compat;	/* Readonly-compatible feature set */
595	__le64 s_root_blkno;		/* Offset, in blocks, of root directory
596					   dinode */
597/*30*/	__le64 s_system_dir_blkno;	/* Offset, in blocks, of system
598					   directory dinode */
599	__le32 s_blocksize_bits;		/* Blocksize for this fs */
600	__le32 s_clustersize_bits;	/* Clustersize for this fs */
601/*40*/	__le16 s_max_slots;		/* Max number of simultaneous mounts
602					   before tunefs required */
603	__le16 s_tunefs_flag;
604	__le32 s_uuid_hash;		/* hash value of uuid */
605	__le64 s_first_cluster_group;	/* Block offset of 1st cluster
606					 * group header */
607/*50*/	__u8  s_label[OCFS2_MAX_VOL_LABEL_LEN];	/* Label for mounting, etc. */
608/*90*/	__u8  s_uuid[OCFS2_VOL_UUID_LEN];	/* 128-bit uuid */
609/*A0*/  struct ocfs2_cluster_info s_cluster_info; /* Only valid if either
610						     userspace or clusterinfo
611						     INCOMPAT flag set. */
612/*B8*/	__le16 s_xattr_inline_size;	/* extended attribute inline size
613					   for this fs*/
614	__le16 s_reserved0;
615	__le32 s_dx_seed[3];		/* seed[0-2] for dx dir hash.
616					 * s_uuid_hash serves as seed[3]. */
617/*C0*/  __le64 s_reserved2[15];		/* Fill out superblock */
618/*140*/
619
620	/*
621	 * NOTE: As stated above, all offsets are relative to
622	 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
623	 * 0xC0 + 0x140 = 0x200 or 512 bytes.  A superblock must fit within
624	 * our smallest blocksize, which is 512 bytes.  To ensure this,
625	 * we reserve the space in s_reserved2.  Anything past s_reserved2
626	 * will not be available on the smallest blocksize.
627	 */
628};
629
630/*
631 * Local allocation bitmap for OCFS2 slots
632 * Note that it exists inside an ocfs2_dinode, so all offsets are
633 * relative to the start of ocfs2_dinode.id2.
634 */
635struct ocfs2_local_alloc
636{
637/*00*/	__le32 la_bm_off;	/* Starting bit offset in main bitmap */
638	__le16 la_size;		/* Size of included bitmap, in bytes */
639	__le16 la_reserved1;
640	__le64 la_reserved2;
641/*10*/	__u8   la_bitmap[];
642};
643
644/*
645 * Data-in-inode header. This is only used if i_dyn_features has
646 * OCFS2_INLINE_DATA_FL set.
647 */
648struct ocfs2_inline_data
649{
650/*00*/	__le16	id_count;	/* Number of bytes that can be used
651				 * for data, starting at id_data */
652	__le16	id_reserved0;
653	__le32	id_reserved1;
654	__u8	id_data[];	/* Start of user data */
655};
656
657/*
658 * On disk inode for OCFS2
659 */
660struct ocfs2_dinode {
661/*00*/	__u8 i_signature[8];		/* Signature for validation */
662	__le32 i_generation;		/* Generation number */
663	__le16 i_suballoc_slot;		/* Slot suballocator this inode
664					   belongs to */
665	__le16 i_suballoc_bit;		/* Bit offset in suballocator
666					   block group */
667/*10*/	__le16 i_links_count_hi;	/* High 16 bits of links count */
668	__le16 i_xattr_inline_size;
669	__le32 i_clusters;		/* Cluster count */
670	__le32 i_uid;			/* Owner UID */
671	__le32 i_gid;			/* Owning GID */
672/*20*/	__le64 i_size;			/* Size in bytes */
673	__le16 i_mode;			/* File mode */
674	__le16 i_links_count;		/* Links count */
675	__le32 i_flags;			/* File flags */
676/*30*/	__le64 i_atime;			/* Access time */
677	__le64 i_ctime;			/* Creation time */
678/*40*/	__le64 i_mtime;			/* Modification time */
679	__le64 i_dtime;			/* Deletion time */
680/*50*/	__le64 i_blkno;			/* Offset on disk, in blocks */
681	__le64 i_last_eb_blk;		/* Pointer to last extent
682					   block */
683/*60*/	__le32 i_fs_generation;		/* Generation per fs-instance */
684	__le32 i_atime_nsec;
685	__le32 i_ctime_nsec;
686	__le32 i_mtime_nsec;
687/*70*/	__le32 i_attr;
688	__le16 i_orphaned_slot;		/* Only valid when OCFS2_ORPHANED_FL
689					   was set in i_flags */
690	__le16 i_dyn_features;
691	__le64 i_xattr_loc;
692/*80*/	struct ocfs2_block_check i_check;	/* Error checking */
693/*88*/	__le64 i_dx_root;		/* Pointer to dir index root block */
694/*90*/	__le64 i_refcount_loc;
695	__le64 i_suballoc_loc;		/* Suballocator block group this
696					   inode belongs to.  Only valid
697					   if allocated from a
698					   discontiguous block group */
699/*A0*/	__le16 i_dio_orphaned_slot;	/* only used for append dio write */
700	__le16 i_reserved1[3];
701	__le64 i_reserved2[2];
702/*B8*/	union {
703		__le64 i_pad1;		/* Generic way to refer to this
704					   64bit union */
705		struct {
706			__le64 i_rdev;	/* Device number */
707		} dev1;
708		struct {		/* Info for bitmap system
709					   inodes */
710			__le32 i_used;	/* Bits (ie, clusters) used  */
711			__le32 i_total;	/* Total bits (clusters)
712					   available */
713		} bitmap1;
714		struct {		/* Info for journal system
715					   inodes */
716			__le32 ij_flags;	/* Mounted, version, etc. */
717			__le32 ij_recovery_generation; /* Incremented when the
718							  journal is recovered
719							  after an unclean
720							  shutdown */
721		} journal1;
722	} id1;				/* Inode type dependent 1 */
723/*C0*/	union {
724		struct ocfs2_super_block	i_super;
725		struct ocfs2_local_alloc	i_lab;
726		struct ocfs2_chain_list		i_chain;
727		struct ocfs2_extent_list	i_list;
728		struct ocfs2_truncate_log	i_dealloc;
729		struct ocfs2_inline_data	i_data;
730		DECLARE_FLEX_ARRAY(__u8,	i_symlink);
731	} id2;
732/* Actual on-disk size is one block */
733};
734
735/*
736 * On-disk directory entry structure for OCFS2
737 *
738 * Packed as this structure could be accessed unaligned on 64-bit platforms
739 */
740struct ocfs2_dir_entry {
741/*00*/	__le64   inode;                  /* Inode number */
742	__le16   rec_len;                /* Directory entry length */
743	__u8    name_len;               /* Name length */
744	__u8    file_type;
745/*0C*/	char    name[OCFS2_MAX_FILENAME_LEN];   /* File name */
746/* Actual on-disk length specified by rec_len */
747} __attribute__ ((packed));
748
749/*
750 * Per-block record for the unindexed directory btree. This is carefully
751 * crafted so that the rec_len and name_len records of an ocfs2_dir_entry are
752 * mirrored. That way, the directory manipulation code needs a minimal amount
753 * of update.
754 *
755 * NOTE: Keep this structure aligned to a multiple of 4 bytes.
756 */
757struct ocfs2_dir_block_trailer {
758/*00*/	__le64		db_compat_inode;	/* Always zero. Was inode */
759
760	__le16		db_compat_rec_len;	/* Backwards compatible with
761						 * ocfs2_dir_entry. */
762	__u8		db_compat_name_len;	/* Always zero. Was name_len */
763	__u8		db_reserved0;
764	__le16		db_reserved1;
765	__le16		db_free_rec_len;	/* Size of largest empty hole
766						 * in this block. (unused) */
767/*10*/	__u8		db_signature[8];	/* Signature for verification */
768	__le64		db_reserved2;
769/*20*/	__le64		db_free_next;		/* Next block in list (unused) */
770	__le64		db_blkno;		/* Offset on disk, in blocks */
771/*30*/	__le64		db_parent_dinode;	/* dinode which owns me, in
772						   blocks */
773	struct ocfs2_block_check db_check;	/* Error checking */
774/*40*/
775};
776
777 /*
778 * A directory entry in the indexed tree. We don't store the full name here,
779 * but instead provide a pointer to the full dirent in the unindexed tree.
780 *
781 * We also store name_len here so as to reduce the number of leaf blocks we
782 * need to search in case of collisions.
783 */
784struct ocfs2_dx_entry {
785	__le32		dx_major_hash;	/* Used to find logical
786					 * cluster in index */
787	__le32		dx_minor_hash;	/* Lower bits used to find
788					 * block in cluster */
789	__le64		dx_dirent_blk;	/* Physical block in unindexed
790					 * tree holding this dirent. */
791};
792
793struct ocfs2_dx_entry_list {
794	__le32		de_reserved;
795	__le16		de_count;	/* Maximum number of entries
796					 * possible in de_entries */
797	__le16		de_num_used;	/* Current number of
798					 * de_entries entries */
799	struct	ocfs2_dx_entry		de_entries[];	/* Indexed dir entries
800							 * in a packed array of
801							 * length de_num_used */
802};
803
804#define OCFS2_DX_FLAG_INLINE	0x01
805
806/*
807 * A directory indexing block. Each indexed directory has one of these,
808 * pointed to by ocfs2_dinode.
809 *
810 * This block stores an indexed btree root, and a set of free space
811 * start-of-list pointers.
812 */
813struct ocfs2_dx_root_block {
814	__u8		dr_signature[8];	/* Signature for verification */
815	struct ocfs2_block_check dr_check;	/* Error checking */
816	__le16		dr_suballoc_slot;	/* Slot suballocator this
817						 * block belongs to. */
818	__le16		dr_suballoc_bit;	/* Bit offset in suballocator
819						 * block group */
820	__le32		dr_fs_generation;	/* Must match super block */
821	__le64		dr_blkno;		/* Offset on disk, in blocks */
822	__le64		dr_last_eb_blk;		/* Pointer to last
823						 * extent block */
824	__le32		dr_clusters;		/* Clusters allocated
825						 * to the indexed tree. */
826	__u8		dr_flags;		/* OCFS2_DX_FLAG_* flags */
827	__u8		dr_reserved0;
828	__le16		dr_reserved1;
829	__le64		dr_dir_blkno;		/* Pointer to parent inode */
830	__le32		dr_num_entries;		/* Total number of
831						 * names stored in
832						 * this directory.*/
833	__le32		dr_reserved2;
834	__le64		dr_free_blk;		/* Pointer to head of free
835						 * unindexed block list. */
836	__le64		dr_suballoc_loc;	/* Suballocator block group
837						   this root belongs to.
838						   Only valid if allocated
839						   from a discontiguous
840						   block group */
841	__le64		dr_reserved3[14];
842	union {
843		struct ocfs2_extent_list dr_list; /* Keep this aligned to 128
844						   * bits for maximum space
845						   * efficiency. */
846		struct ocfs2_dx_entry_list dr_entries; /* In-root-block list of
847							* entries. We grow out
848							* to extents if this
849							* gets too big. */
850	};
851};
852
853/*
854 * The header of a leaf block in the indexed tree.
855 */
856struct ocfs2_dx_leaf {
857	__u8		dl_signature[8];/* Signature for verification */
858	struct ocfs2_block_check dl_check;	/* Error checking */
859	__le64		dl_blkno;	/* Offset on disk, in blocks */
860	__le32		dl_fs_generation;/* Must match super block */
861	__le32		dl_reserved0;
862	__le64		dl_reserved1;
863	struct ocfs2_dx_entry_list	dl_list;
864};
865
866/*
867 * Largest bitmap for a block (suballocator) group in bytes.  This limit
868 * does not affect cluster groups (global allocator).  Cluster group
869 * bitmaps run to the end of the block.
870 */
871#define OCFS2_MAX_BG_BITMAP_SIZE	256
872
873/*
874 * On disk allocator group structure for OCFS2
875 */
876struct ocfs2_group_desc
877{
878/*00*/	__u8    bg_signature[8];        /* Signature for validation */
879	__le16   bg_size;                /* Size of included bitmap in
880					   bytes. */
881	__le16   bg_bits;                /* Bits represented by this
882					   group. */
883	__le16	bg_free_bits_count;     /* Free bits count */
884	__le16   bg_chain;               /* What chain I am in. */
885/*10*/	__le32   bg_generation;
886	__le16   bg_contig_free_bits;   /* max contig free bits length */
887	__le16   bg_reserved1;
888	__le64   bg_next_group;          /* Next group in my list, in
889					   blocks */
890/*20*/	__le64   bg_parent_dinode;       /* dinode which owns me, in
891					   blocks */
892	__le64   bg_blkno;               /* Offset on disk, in blocks */
893/*30*/	struct ocfs2_block_check bg_check;	/* Error checking */
894	__le64   bg_reserved2;
895/*40*/	union {
896		DECLARE_FLEX_ARRAY(__u8, bg_bitmap);
897		struct {
898			/*
899			 * Block groups may be discontiguous when
900			 * OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG is set.
901			 * The extents of a discontiguous block group are
902			 * stored in bg_list.  It is a flat list.
903			 * l_tree_depth must always be zero.  A
904			 * discontiguous group is signified by a non-zero
905			 * bg_list->l_next_free_rec.  Only block groups
906			 * can be discontiguous; Cluster groups cannot.
907			 * We've never made a block group with more than
908			 * 2048 blocks (256 bytes of bg_bitmap).  This
909			 * codifies that limit so that we can fit bg_list.
910			 * bg_size of a discontiguous block group will
911			 * be 256 to match bg_bitmap_filler.
912			 */
913			__u8 bg_bitmap_filler[OCFS2_MAX_BG_BITMAP_SIZE];
914/*140*/			struct ocfs2_extent_list bg_list;
915		};
916	};
917/* Actual on-disk size is one block */
918};
919
920struct ocfs2_refcount_rec {
921/*00*/	__le64 r_cpos;		/* Physical offset, in clusters */
922	__le32 r_clusters;	/* Clusters covered by this extent */
923	__le32 r_refcount;	/* Reference count of this extent */
924/*10*/
925};
926#define OCFS2_32BIT_POS_MASK		(0xffffffffULL)
927
928#define OCFS2_REFCOUNT_LEAF_FL          (0x00000001)
929#define OCFS2_REFCOUNT_TREE_FL          (0x00000002)
930
931struct ocfs2_refcount_list {
932/*00*/	__le16 rl_count;	/* Maximum number of entries possible
933				   in rl_records */
934	__le16 rl_used;		/* Current number of used records */
935	__le32 rl_reserved2;
936	__le64 rl_reserved1;	/* Pad to sizeof(ocfs2_refcount_record) */
937/*10*/	struct ocfs2_refcount_rec rl_recs[];	/* Refcount records */
938};
939
940
941struct ocfs2_refcount_block {
942/*00*/	__u8 rf_signature[8];		/* Signature for verification */
943	__le16 rf_suballoc_slot;	/* Slot suballocator this block
944					   belongs to */
945	__le16 rf_suballoc_bit;		/* Bit offset in suballocator
946					   block group */
947	__le32 rf_fs_generation;	/* Must match superblock */
948/*10*/	__le64 rf_blkno;		/* Offset on disk, in blocks */
949	__le64 rf_parent;		/* Parent block, only valid if
950					   OCFS2_REFCOUNT_LEAF_FL is set in
951					   rf_flags */
952/*20*/	struct ocfs2_block_check rf_check;	/* Error checking */
953	__le64 rf_last_eb_blk;		/* Pointer to last extent block */
954/*30*/	__le32 rf_count;		/* Number of inodes sharing this
955					   refcount tree */
956	__le32 rf_flags;		/* See the flags above */
957	__le32 rf_clusters;		/* clusters covered by refcount tree. */
958	__le32 rf_cpos;			/* cluster offset in refcount tree.*/
959/*40*/	__le32 rf_generation;		/* generation number. all be the same
960					 * for the same refcount tree. */
961	__le32 rf_reserved0;
962	__le64 rf_suballoc_loc;		/* Suballocator block group this
963					   refcount block belongs to. Only
964					   valid if allocated from a
965					   discontiguous block group */
966/*50*/	__le64 rf_reserved1[6];
967/*80*/	union {
968		struct ocfs2_refcount_list rf_records;  /* List of refcount
969							  records */
970		struct ocfs2_extent_list rf_list;	/* Extent record list,
971							only valid if
972							OCFS2_REFCOUNT_TREE_FL
973							is set in rf_flags */
974	};
975/* Actual on-disk size is one block */
976};
977
978/*
979 * On disk extended attribute structure for OCFS2.
980 */
981
982/*
983 * ocfs2_xattr_entry indicates one extend attribute.
984 *
985 * Note that it can be stored in inode, one block or one xattr bucket.
986 */
987struct ocfs2_xattr_entry {
988	__le32	xe_name_hash;    /* hash value of xattr prefix+suffix. */
989	__le16	xe_name_offset;  /* byte offset from the 1st entry in the
990				    local xattr storage(inode, xattr block or
991				    xattr bucket). */
992	__u8	xe_name_len;	 /* xattr name len, doesn't include prefix. */
993	__u8	xe_type;         /* the low 7 bits indicate the name prefix
994				  * type and the highest bit indicates whether
995				  * the EA is stored in the local storage. */
996	__le64	xe_value_size;	 /* real xattr value length. */
997};
998
999/*
1000 * On disk structure for xattr header.
1001 *
1002 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
1003 * the local xattr storage.
1004 */
1005struct ocfs2_xattr_header {
1006	__le16	xh_count;                       /* contains the count of how
1007						   many records are in the
1008						   local xattr storage. */
1009	__le16	xh_free_start;                  /* current offset for storing
1010						   xattr. */
1011	__le16	xh_name_value_len;              /* total length of name/value
1012						   length in this bucket. */
1013	__le16	xh_num_buckets;                 /* Number of xattr buckets
1014						   in this extent record,
1015						   only valid in the first
1016						   bucket. */
1017	struct ocfs2_block_check xh_check;	/* Error checking
1018						   (Note, this is only
1019						    used for xattr
1020						    buckets.  A block uses
1021						    xb_check and sets
1022						    this field to zero.) */
1023	struct ocfs2_xattr_entry xh_entries[]; /* xattr entry list. */
1024};
1025
1026/*
1027 * On disk structure for xattr value root.
1028 *
1029 * When an xattr's value is large enough, it is stored in an external
1030 * b-tree like file data.  The xattr value root points to this structure.
1031 */
1032struct ocfs2_xattr_value_root {
1033/*00*/	__le32	xr_clusters;              /* clusters covered by xattr value. */
1034	__le32	xr_reserved0;
1035	__le64	xr_last_eb_blk;           /* Pointer to last extent block */
1036/*10*/	struct ocfs2_extent_list xr_list; /* Extent record list */
1037};
1038
1039/*
1040 * On disk structure for xattr tree root.
1041 *
1042 * It is used when there are too many extended attributes for one file. These
1043 * attributes will be organized and stored in an indexed-btree.
1044 */
1045struct ocfs2_xattr_tree_root {
1046/*00*/	__le32	xt_clusters;              /* clusters covered by xattr. */
1047	__le32	xt_reserved0;
1048	__le64	xt_last_eb_blk;           /* Pointer to last extent block */
1049/*10*/	struct ocfs2_extent_list xt_list; /* Extent record list */
1050};
1051
1052#define OCFS2_XATTR_INDEXED	0x1
1053#define OCFS2_HASH_SHIFT	5
1054#define OCFS2_XATTR_ROUND	3
1055#define OCFS2_XATTR_SIZE(size)	(((size) + OCFS2_XATTR_ROUND) & \
1056				~(OCFS2_XATTR_ROUND))
1057
1058#define OCFS2_XATTR_BUCKET_SIZE			4096
1059#define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET 	(OCFS2_XATTR_BUCKET_SIZE \
1060						 / OCFS2_MIN_BLOCKSIZE)
1061
1062/*
1063 * On disk structure for xattr block.
1064 */
1065struct ocfs2_xattr_block {
1066/*00*/	__u8	xb_signature[8];     /* Signature for verification */
1067	__le16	xb_suballoc_slot;    /* Slot suballocator this
1068					block belongs to. */
1069	__le16	xb_suballoc_bit;     /* Bit offset in suballocator
1070					block group */
1071	__le32	xb_fs_generation;    /* Must match super block */
1072/*10*/	__le64	xb_blkno;            /* Offset on disk, in blocks */
1073	struct ocfs2_block_check xb_check;	/* Error checking */
1074/*20*/	__le16	xb_flags;            /* Indicates whether this block contains
1075					real xattr or a xattr tree. */
1076	__le16	xb_reserved0;
1077	__le32  xb_reserved1;
1078	__le64	xb_suballoc_loc;	/* Suballocator block group this
1079					   xattr block belongs to. Only
1080					   valid if allocated from a
1081					   discontiguous block group */
1082/*30*/	union {
1083		struct ocfs2_xattr_header xb_header; /* xattr header if this
1084							block contains xattr */
1085		struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
1086							block cotains xattr
1087							tree. */
1088	} xb_attrs;
1089};
1090
1091#define OCFS2_XATTR_ENTRY_LOCAL		0x80
1092#define OCFS2_XATTR_TYPE_MASK		0x7F
1093static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
1094					 int local)
1095{
1096	if (local)
1097		xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
1098	else
1099		xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
1100}
1101
1102static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
1103{
1104	return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
1105}
1106
1107static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
1108{
1109	xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
1110}
1111
1112static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
1113{
1114	return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
1115}
1116
1117/*
1118 *  On disk structures for global quota file
1119 */
1120
1121/* Magic numbers and known versions for global quota files */
1122#define OCFS2_GLOBAL_QMAGICS {\
1123	0x0cf52470, /* USRQUOTA */ \
1124	0x0cf52471  /* GRPQUOTA */ \
1125}
1126
1127#define OCFS2_GLOBAL_QVERSIONS {\
1128	0, \
1129	0, \
1130}
1131
1132
1133/* Each block of each quota file has a certain fixed number of bytes reserved
1134 * for OCFS2 internal use at its end. OCFS2 can use it for things like
1135 * checksums, etc. */
1136#define OCFS2_QBLK_RESERVED_SPACE 8
1137
1138/* Generic header of all quota files */
1139struct ocfs2_disk_dqheader {
1140	__le32 dqh_magic;	/* Magic number identifying file */
1141	__le32 dqh_version;	/* Quota format version */
1142};
1143
1144#define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1145
1146/* Information header of global quota file (immediately follows the generic
1147 * header) */
1148struct ocfs2_global_disk_dqinfo {
1149/*00*/	__le32 dqi_bgrace;	/* Grace time for space softlimit excess */
1150	__le32 dqi_igrace;	/* Grace time for inode softlimit excess */
1151	__le32 dqi_syncms;	/* Time after which we sync local changes to
1152				 * global quota file */
1153	__le32 dqi_blocks;	/* Number of blocks in quota file */
1154/*10*/	__le32 dqi_free_blk;	/* First free block in quota file */
1155	__le32 dqi_free_entry;	/* First block with free dquot entry in quota
1156				 * file */
1157};
1158
1159/* Structure with global user / group information. We reserve some space
1160 * for future use. */
1161struct ocfs2_global_disk_dqblk {
1162/*00*/	__le32 dqb_id;          /* ID the structure belongs to */
1163	__le32 dqb_use_count;   /* Number of nodes having reference to this structure */
1164	__le64 dqb_ihardlimit;  /* absolute limit on allocated inodes */
1165/*10*/	__le64 dqb_isoftlimit;  /* preferred inode limit */
1166	__le64 dqb_curinodes;   /* current # allocated inodes */
1167/*20*/	__le64 dqb_bhardlimit;  /* absolute limit on disk space */
1168	__le64 dqb_bsoftlimit;  /* preferred limit on disk space */
1169/*30*/	__le64 dqb_curspace;    /* current space occupied */
1170	__le64 dqb_btime;       /* time limit for excessive disk use */
1171/*40*/	__le64 dqb_itime;       /* time limit for excessive inode use */
1172	__le64 dqb_pad1;
1173/*50*/	__le64 dqb_pad2;
1174};
1175
1176/*
1177 *  On-disk structures for local quota file
1178 */
1179
1180/* Magic numbers and known versions for local quota files */
1181#define OCFS2_LOCAL_QMAGICS {\
1182	0x0cf524c0, /* USRQUOTA */ \
1183	0x0cf524c1  /* GRPQUOTA */ \
1184}
1185
1186#define OCFS2_LOCAL_QVERSIONS {\
1187	0, \
1188	0, \
1189}
1190
1191/* Quota flags in dqinfo header */
1192#define OLQF_CLEAN	0x0001	/* Quota file is empty (this should be after\
1193				 * quota has been cleanly turned off) */
1194
1195#define OCFS2_LOCAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1196
1197/* Information header of local quota file (immediately follows the generic
1198 * header) */
1199struct ocfs2_local_disk_dqinfo {
1200	__le32 dqi_flags;	/* Flags for quota file */
1201	__le32 dqi_chunks;	/* Number of chunks of quota structures
1202				 * with a bitmap */
1203	__le32 dqi_blocks;	/* Number of blocks allocated for quota file */
1204};
1205
1206/* Header of one chunk of a quota file */
1207struct ocfs2_local_disk_chunk {
1208	__le32 dqc_free;	/* Number of free entries in the bitmap */
1209	__u8 dqc_bitmap[];	/* Bitmap of entries in the corresponding
1210				 * chunk of quota file */
1211};
1212
1213/* One entry in local quota file */
1214struct ocfs2_local_disk_dqblk {
1215/*00*/	__le64 dqb_id;		/* id this quota applies to */
1216	__le64 dqb_spacemod;	/* Change in the amount of used space */
1217/*10*/	__le64 dqb_inodemod;	/* Change in the amount of used inodes */
1218};
1219
1220
1221/*
1222 * The quota trailer lives at the end of each quota block.
1223 */
1224
1225struct ocfs2_disk_dqtrailer {
1226/*00*/	struct ocfs2_block_check dq_check;	/* Error checking */
1227/*08*/	/* Cannot be larger than OCFS2_QBLK_RESERVED_SPACE */
1228};
1229
1230static inline struct ocfs2_disk_dqtrailer *ocfs2_block_dqtrailer(int blocksize,
1231								 void *buf)
1232{
1233	char *ptr = buf;
1234	ptr += blocksize - OCFS2_QBLK_RESERVED_SPACE;
1235
1236	return (struct ocfs2_disk_dqtrailer *)ptr;
1237}
1238
1239#ifdef __KERNEL__
1240static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
1241{
1242	return  sb->s_blocksize -
1243		 offsetof(struct ocfs2_dinode, id2.i_symlink);
1244}
1245
1246static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
1247						   struct ocfs2_dinode *di)
1248{
1249	unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1250
1251	if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1252		return sb->s_blocksize -
1253			offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1254			xattrsize;
1255	else
1256		return sb->s_blocksize -
1257			offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1258}
1259
1260static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
1261{
1262	int size;
1263
1264	size = sb->s_blocksize -
1265		offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1266
1267	return size / sizeof(struct ocfs2_extent_rec);
1268}
1269
1270static inline int ocfs2_extent_recs_per_inode_with_xattr(
1271						struct super_block *sb,
1272						struct ocfs2_dinode *di)
1273{
1274	int size;
1275	unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1276
1277	if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1278		size = sb->s_blocksize -
1279			offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
1280			xattrsize;
1281	else
1282		size = sb->s_blocksize -
1283			offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1284
1285	return size / sizeof(struct ocfs2_extent_rec);
1286}
1287
1288static inline int ocfs2_extent_recs_per_dx_root(struct super_block *sb)
1289{
1290	int size;
1291
1292	size = sb->s_blocksize -
1293		offsetof(struct ocfs2_dx_root_block, dr_list.l_recs);
1294
1295	return size / sizeof(struct ocfs2_extent_rec);
1296}
1297
1298static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
1299{
1300	int size;
1301
1302	size = sb->s_blocksize -
1303		offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1304
1305	return size / sizeof(struct ocfs2_chain_rec);
1306}
1307
1308static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
1309{
1310	int size;
1311
1312	size = sb->s_blocksize -
1313		offsetof(struct ocfs2_extent_block, h_list.l_recs);
1314
1315	return size / sizeof(struct ocfs2_extent_rec);
1316}
1317
1318static inline u16 ocfs2_extent_recs_per_gd(struct super_block *sb)
1319{
1320	int size;
1321
1322	size = sb->s_blocksize -
1323		offsetof(struct ocfs2_group_desc, bg_list.l_recs);
1324
1325	return size / sizeof(struct ocfs2_extent_rec);
1326}
1327
1328static inline int ocfs2_dx_entries_per_leaf(struct super_block *sb)
1329{
1330	int size;
1331
1332	size = sb->s_blocksize -
1333		offsetof(struct ocfs2_dx_leaf, dl_list.de_entries);
1334
1335	return size / sizeof(struct ocfs2_dx_entry);
1336}
1337
1338static inline int ocfs2_dx_entries_per_root(struct super_block *sb)
1339{
1340	int size;
1341
1342	size = sb->s_blocksize -
1343		offsetof(struct ocfs2_dx_root_block, dr_entries.de_entries);
1344
1345	return size / sizeof(struct ocfs2_dx_entry);
1346}
1347
1348static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
1349{
1350	u16 size;
1351
1352	size = sb->s_blocksize -
1353		offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1354
1355	return size;
1356}
1357
1358static inline int ocfs2_group_bitmap_size(struct super_block *sb,
1359					  int suballocator,
1360					  u32 feature_incompat)
1361{
1362	int size = sb->s_blocksize -
1363		offsetof(struct ocfs2_group_desc, bg_bitmap);
1364
1365	/*
1366	 * The cluster allocator uses the entire block.  Suballocators have
1367	 * never used more than OCFS2_MAX_BG_BITMAP_SIZE.  Unfortunately, older
1368	 * code expects bg_size set to the maximum.  Thus we must keep
1369	 * bg_size as-is unless discontig_bg is enabled.
1370	 */
1371	if (suballocator &&
1372	    (feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG))
1373		size = OCFS2_MAX_BG_BITMAP_SIZE;
1374
1375	return size;
1376}
1377
1378static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
1379{
1380	int size;
1381
1382	size = sb->s_blocksize -
1383		offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1384
1385	return size / sizeof(struct ocfs2_truncate_rec);
1386}
1387
1388static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
1389{
1390	u64 offset = OCFS2_BACKUP_SB_START;
1391
1392	if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1393		offset <<= (2 * index);
1394		offset >>= sb->s_blocksize_bits;
1395		return offset;
1396	}
1397
1398	return 0;
1399
1400}
1401
1402static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
1403{
1404	int size;
1405
1406	size = sb->s_blocksize -
1407		offsetof(struct ocfs2_xattr_block,
1408			 xb_attrs.xb_root.xt_list.l_recs);
1409
1410	return size / sizeof(struct ocfs2_extent_rec);
1411}
1412
1413static inline u16 ocfs2_extent_recs_per_rb(struct super_block *sb)
1414{
1415	int size;
1416
1417	size = sb->s_blocksize -
1418		offsetof(struct ocfs2_refcount_block, rf_list.l_recs);
1419
1420	return size / sizeof(struct ocfs2_extent_rec);
1421}
1422
1423static inline u16 ocfs2_refcount_recs_per_rb(struct super_block *sb)
1424{
1425	int size;
1426
1427	size = sb->s_blocksize -
1428		offsetof(struct ocfs2_refcount_block, rf_records.rl_recs);
1429
1430	return size / sizeof(struct ocfs2_refcount_rec);
1431}
1432
1433static inline u32
1434ocfs2_get_ref_rec_low_cpos(const struct ocfs2_refcount_rec *rec)
1435{
1436	return le64_to_cpu(rec->r_cpos) & OCFS2_32BIT_POS_MASK;
1437}
1438#else
1439static inline int ocfs2_fast_symlink_chars(int blocksize)
1440{
1441	return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
1442}
1443
1444static inline int ocfs2_max_inline_data_with_xattr(int blocksize,
1445						   struct ocfs2_dinode *di)
1446{
1447	if (di && (di->i_dyn_features & OCFS2_INLINE_XATTR_FL))
1448		return blocksize -
1449			offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1450			di->i_xattr_inline_size;
1451	else
1452		return blocksize -
1453			offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1454}
1455
1456static inline int ocfs2_extent_recs_per_inode(int blocksize)
1457{
1458	int size;
1459
1460	size = blocksize -
1461		offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1462
1463	return size / sizeof(struct ocfs2_extent_rec);
1464}
1465
1466static inline int ocfs2_chain_recs_per_inode(int blocksize)
1467{
1468	int size;
1469
1470	size = blocksize -
1471		offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1472
1473	return size / sizeof(struct ocfs2_chain_rec);
1474}
1475
1476static inline int ocfs2_extent_recs_per_eb(int blocksize)
1477{
1478	int size;
1479
1480	size = blocksize -
1481		offsetof(struct ocfs2_extent_block, h_list.l_recs);
1482
1483	return size / sizeof(struct ocfs2_extent_rec);
1484}
1485
1486static inline int ocfs2_extent_recs_per_gd(int blocksize)
1487{
1488	int size;
1489
1490	size = blocksize -
1491		offsetof(struct ocfs2_group_desc, bg_list.l_recs);
1492
1493	return size / sizeof(struct ocfs2_extent_rec);
1494}
1495
1496static inline int ocfs2_local_alloc_size(int blocksize)
1497{
1498	int size;
1499
1500	size = blocksize -
1501		offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1502
1503	return size;
1504}
1505
1506static inline int ocfs2_group_bitmap_size(int blocksize,
1507					  int suballocator,
1508					  uint32_t feature_incompat)
1509{
1510	int size = sb->s_blocksize -
1511		offsetof(struct ocfs2_group_desc, bg_bitmap);
1512
1513	/*
1514	 * The cluster allocator uses the entire block.  Suballocators have
1515	 * never used more than OCFS2_MAX_BG_BITMAP_SIZE.  Unfortunately, older
1516	 * code expects bg_size set to the maximum.  Thus we must keep
1517	 * bg_size as-is unless discontig_bg is enabled.
1518	 */
1519	if (suballocator &&
1520	    (feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG))
1521		size = OCFS2_MAX_BG_BITMAP_SIZE;
1522
1523	return size;
1524}
1525
1526static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1527{
1528	int size;
1529
1530	size = blocksize -
1531		offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1532
1533	return size / sizeof(struct ocfs2_truncate_rec);
1534}
1535
1536static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1537{
1538	uint64_t offset = OCFS2_BACKUP_SB_START;
1539
1540	if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1541		offset <<= (2 * index);
1542		offset /= blocksize;
1543		return offset;
1544	}
1545
1546	return 0;
1547}
1548
1549static inline int ocfs2_xattr_recs_per_xb(int blocksize)
1550{
1551	int size;
1552
1553	size = blocksize -
1554		offsetof(struct ocfs2_xattr_block,
1555			 xb_attrs.xb_root.xt_list.l_recs);
1556
1557	return size / sizeof(struct ocfs2_extent_rec);
1558}
1559#endif  /* __KERNEL__ */
1560
1561
1562static inline int ocfs2_system_inode_is_global(int type)
1563{
1564	return ((type >= 0) &&
1565		(type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1566}
1567
1568static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1569						  int type, int slot)
1570{
1571	int chars;
1572
1573        /*
1574         * Global system inodes can only have one copy.  Everything
1575         * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1576         * list has a copy per slot.
1577         */
1578	if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
1579		chars = snprintf(buf, len, "%s",
1580				 ocfs2_system_inodes[type].si_name);
1581	else
1582		chars = snprintf(buf, len,
1583				 ocfs2_system_inodes[type].si_name,
1584				 slot);
1585
1586	return chars;
1587}
1588
1589static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1590				    umode_t mode)
1591{
1592	de->file_type = fs_umode_to_ftype(mode);
1593}
1594
1595static inline int ocfs2_gd_is_discontig(struct ocfs2_group_desc *gd)
1596{
1597	if ((offsetof(struct ocfs2_group_desc, bg_bitmap) +
1598	     le16_to_cpu(gd->bg_size)) !=
1599	    offsetof(struct ocfs2_group_desc, bg_list))
1600		return 0;
1601	/*
1602	 * Only valid to check l_next_free_rec if
1603	 * bg_bitmap + bg_size == bg_list.
1604	 */
1605	if (!gd->bg_list.l_next_free_rec)
1606		return 0;
1607	return 1;
1608}
1609#endif  /* _OCFS2_FS_H */
1610
1611