xref: /kernel/linux/linux-5.10/fs/ocfs2/ocfs2_fs.h (revision 8c2ecf20)
1/* SPDX-License-Identifier: GPL-2.0-only */
2/* -*- mode: c; c-basic-offset: 8; -*-
3 * vim: noexpandtab sw=8 ts=8 sts=0:
4 *
5 * ocfs2_fs.h
6 *
7 * On-disk structures for OCFS2.
8 *
9 * Copyright (C) 2002, 2004 Oracle.  All rights reserved.
10 */
11
12#ifndef _OCFS2_FS_H
13#define _OCFS2_FS_H
14
15#include <linux/magic.h>
16
17/* Version */
18#define OCFS2_MAJOR_REV_LEVEL		0
19#define OCFS2_MINOR_REV_LEVEL          	90
20
21/*
22 * An OCFS2 volume starts this way:
23 * Sector 0: Valid ocfs1_vol_disk_hdr that cleanly fails to mount OCFS.
24 * Sector 1: Valid ocfs1_vol_label that cleanly fails to mount OCFS.
25 * Block OCFS2_SUPER_BLOCK_BLKNO: OCFS2 superblock.
26 *
27 * All other structures are found from the superblock information.
28 *
29 * OCFS2_SUPER_BLOCK_BLKNO is in blocks, not sectors.  eg, for a
30 * blocksize of 2K, it is 4096 bytes into disk.
31 */
32#define OCFS2_SUPER_BLOCK_BLKNO		2
33
34/*
35 * Cluster size limits. The maximum is kept arbitrarily at 1 MB, and could
36 * grow if needed.
37 */
38#define OCFS2_MIN_CLUSTERSIZE		4096
39#define OCFS2_MAX_CLUSTERSIZE		1048576
40
41/*
42 * Blocks cannot be bigger than clusters, so the maximum blocksize is the
43 * minimum cluster size.
44 */
45#define OCFS2_MIN_BLOCKSIZE		512
46#define OCFS2_MAX_BLOCKSIZE		OCFS2_MIN_CLUSTERSIZE
47
48/* Object signatures */
49#define OCFS2_SUPER_BLOCK_SIGNATURE	"OCFSV2"
50#define OCFS2_INODE_SIGNATURE		"INODE01"
51#define OCFS2_EXTENT_BLOCK_SIGNATURE	"EXBLK01"
52#define OCFS2_GROUP_DESC_SIGNATURE      "GROUP01"
53#define OCFS2_XATTR_BLOCK_SIGNATURE	"XATTR01"
54#define OCFS2_DIR_TRAILER_SIGNATURE	"DIRTRL1"
55#define OCFS2_DX_ROOT_SIGNATURE		"DXDIR01"
56#define OCFS2_DX_LEAF_SIGNATURE		"DXLEAF1"
57#define OCFS2_REFCOUNT_BLOCK_SIGNATURE	"REFCNT1"
58
59/* Compatibility flags */
60#define OCFS2_HAS_COMPAT_FEATURE(sb,mask)			\
61	( OCFS2_SB(sb)->s_feature_compat & (mask) )
62#define OCFS2_HAS_RO_COMPAT_FEATURE(sb,mask)			\
63	( OCFS2_SB(sb)->s_feature_ro_compat & (mask) )
64#define OCFS2_HAS_INCOMPAT_FEATURE(sb,mask)			\
65	( OCFS2_SB(sb)->s_feature_incompat & (mask) )
66#define OCFS2_SET_COMPAT_FEATURE(sb,mask)			\
67	OCFS2_SB(sb)->s_feature_compat |= (mask)
68#define OCFS2_SET_RO_COMPAT_FEATURE(sb,mask)			\
69	OCFS2_SB(sb)->s_feature_ro_compat |= (mask)
70#define OCFS2_SET_INCOMPAT_FEATURE(sb,mask)			\
71	OCFS2_SB(sb)->s_feature_incompat |= (mask)
72#define OCFS2_CLEAR_COMPAT_FEATURE(sb,mask)			\
73	OCFS2_SB(sb)->s_feature_compat &= ~(mask)
74#define OCFS2_CLEAR_RO_COMPAT_FEATURE(sb,mask)			\
75	OCFS2_SB(sb)->s_feature_ro_compat &= ~(mask)
76#define OCFS2_CLEAR_INCOMPAT_FEATURE(sb,mask)			\
77	OCFS2_SB(sb)->s_feature_incompat &= ~(mask)
78
79#define OCFS2_FEATURE_COMPAT_SUPP	(OCFS2_FEATURE_COMPAT_BACKUP_SB	\
80					 | OCFS2_FEATURE_COMPAT_JBD2_SB)
81#define OCFS2_FEATURE_INCOMPAT_SUPP	(OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT \
82					 | OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC \
83					 | OCFS2_FEATURE_INCOMPAT_INLINE_DATA \
84					 | OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP \
85					 | OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK \
86					 | OCFS2_FEATURE_INCOMPAT_XATTR \
87					 | OCFS2_FEATURE_INCOMPAT_META_ECC \
88					 | OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS \
89					 | OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE \
90					 | OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG	\
91					 | OCFS2_FEATURE_INCOMPAT_CLUSTERINFO \
92					 | OCFS2_FEATURE_INCOMPAT_APPEND_DIO)
93#define OCFS2_FEATURE_RO_COMPAT_SUPP	(OCFS2_FEATURE_RO_COMPAT_UNWRITTEN \
94					 | OCFS2_FEATURE_RO_COMPAT_USRQUOTA \
95					 | OCFS2_FEATURE_RO_COMPAT_GRPQUOTA)
96
97/*
98 * Heartbeat-only devices are missing journals and other files.  The
99 * filesystem driver can't load them, but the library can.  Never put
100 * this in OCFS2_FEATURE_INCOMPAT_SUPP, *ever*.
101 */
102#define OCFS2_FEATURE_INCOMPAT_HEARTBEAT_DEV	0x0002
103
104/*
105 * tunefs sets this incompat flag before starting the resize and clears it
106 * at the end. This flag protects users from inadvertently mounting the fs
107 * after an aborted run without fsck-ing.
108 */
109#define OCFS2_FEATURE_INCOMPAT_RESIZE_INPROG    0x0004
110
111/* Used to denote a non-clustered volume */
112#define OCFS2_FEATURE_INCOMPAT_LOCAL_MOUNT	0x0008
113
114/* Support for sparse allocation in b-trees */
115#define OCFS2_FEATURE_INCOMPAT_SPARSE_ALLOC	0x0010
116
117/*
118 * Tunefs sets this incompat flag before starting an operation which
119 * would require cleanup on abort. This is done to protect users from
120 * inadvertently mounting the fs after an aborted run without
121 * fsck-ing.
122 *
123 * s_tunefs_flags on the super block describes precisely which
124 * operations were in progress.
125 */
126#define OCFS2_FEATURE_INCOMPAT_TUNEFS_INPROG	0x0020
127
128/* Support for data packed into inode blocks */
129#define OCFS2_FEATURE_INCOMPAT_INLINE_DATA	0x0040
130
131/*
132 * Support for alternate, userspace cluster stacks.  If set, the superblock
133 * field s_cluster_info contains a tag for the alternate stack in use as
134 * well as the name of the cluster being joined.
135 * mount.ocfs2 must pass in a matching stack name.
136 *
137 * If not set, the classic stack will be used.  This is compatbile with
138 * all older versions.
139 */
140#define OCFS2_FEATURE_INCOMPAT_USERSPACE_STACK	0x0080
141
142/* Support for the extended slot map */
143#define OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP 0x100
144
145/* Support for extended attributes */
146#define OCFS2_FEATURE_INCOMPAT_XATTR		0x0200
147
148/* Support for indexed directores */
149#define OCFS2_FEATURE_INCOMPAT_INDEXED_DIRS	0x0400
150
151/* Metadata checksum and error correction */
152#define OCFS2_FEATURE_INCOMPAT_META_ECC		0x0800
153
154/* Refcount tree support */
155#define OCFS2_FEATURE_INCOMPAT_REFCOUNT_TREE	0x1000
156
157/* Discontiguous block groups */
158#define OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG	0x2000
159
160/*
161 * Incompat bit to indicate useable clusterinfo with stackflags for all
162 * cluster stacks (userspace adnd o2cb). If this bit is set,
163 * INCOMPAT_USERSPACE_STACK becomes superfluous and thus should not be set.
164 */
165#define OCFS2_FEATURE_INCOMPAT_CLUSTERINFO	0x4000
166
167/*
168 * Append Direct IO support
169 */
170#define OCFS2_FEATURE_INCOMPAT_APPEND_DIO	0x8000
171
172/*
173 * backup superblock flag is used to indicate that this volume
174 * has backup superblocks.
175 */
176#define OCFS2_FEATURE_COMPAT_BACKUP_SB		0x0001
177
178/*
179 * The filesystem will correctly handle journal feature bits.
180 */
181#define OCFS2_FEATURE_COMPAT_JBD2_SB		0x0002
182
183/*
184 * Unwritten extents support.
185 */
186#define OCFS2_FEATURE_RO_COMPAT_UNWRITTEN	0x0001
187
188/*
189 * Maintain quota information for this filesystem
190 */
191#define OCFS2_FEATURE_RO_COMPAT_USRQUOTA	0x0002
192#define OCFS2_FEATURE_RO_COMPAT_GRPQUOTA	0x0004
193
194
195/* The byte offset of the first backup block will be 1G.
196 * The following will be 4G, 16G, 64G, 256G and 1T.
197 */
198#define OCFS2_BACKUP_SB_START			1 << 30
199
200/* the max backup superblock nums */
201#define OCFS2_MAX_BACKUP_SUPERBLOCKS	6
202
203/*
204 * Flags on ocfs2_super_block.s_tunefs_flags
205 */
206#define OCFS2_TUNEFS_INPROG_REMOVE_SLOT		0x0001	/* Removing slots */
207
208/*
209 * Flags on ocfs2_dinode.i_flags
210 */
211#define OCFS2_VALID_FL		(0x00000001)	/* Inode is valid */
212#define OCFS2_UNUSED2_FL	(0x00000002)
213#define OCFS2_ORPHANED_FL	(0x00000004)	/* On the orphan list */
214#define OCFS2_UNUSED3_FL	(0x00000008)
215/* System inode flags */
216#define OCFS2_SYSTEM_FL		(0x00000010)	/* System inode */
217#define OCFS2_SUPER_BLOCK_FL	(0x00000020)	/* Super block */
218#define OCFS2_LOCAL_ALLOC_FL	(0x00000040)	/* Slot local alloc bitmap */
219#define OCFS2_BITMAP_FL		(0x00000080)	/* Allocation bitmap */
220#define OCFS2_JOURNAL_FL	(0x00000100)	/* Slot local journal */
221#define OCFS2_HEARTBEAT_FL	(0x00000200)	/* Heartbeat area */
222#define OCFS2_CHAIN_FL		(0x00000400)	/* Chain allocator */
223#define OCFS2_DEALLOC_FL	(0x00000800)	/* Truncate log */
224#define OCFS2_QUOTA_FL		(0x00001000)	/* Quota file */
225#define OCFS2_DIO_ORPHANED_FL	(0X00002000)	/* On the orphan list especially
226						 * for dio */
227
228/*
229 * Flags on ocfs2_dinode.i_dyn_features
230 *
231 * These can change much more often than i_flags. When adding flags,
232 * keep in mind that i_dyn_features is only 16 bits wide.
233 */
234#define OCFS2_INLINE_DATA_FL	(0x0001)	/* Data stored in inode block */
235#define OCFS2_HAS_XATTR_FL	(0x0002)
236#define OCFS2_INLINE_XATTR_FL	(0x0004)
237#define OCFS2_INDEXED_DIR_FL	(0x0008)
238#define OCFS2_HAS_REFCOUNT_FL   (0x0010)
239
240/* Inode attributes, keep in sync with EXT2 */
241#define OCFS2_SECRM_FL			FS_SECRM_FL	/* Secure deletion */
242#define OCFS2_UNRM_FL			FS_UNRM_FL	/* Undelete */
243#define OCFS2_COMPR_FL			FS_COMPR_FL	/* Compress file */
244#define OCFS2_SYNC_FL			FS_SYNC_FL	/* Synchronous updates */
245#define OCFS2_IMMUTABLE_FL		FS_IMMUTABLE_FL	/* Immutable file */
246#define OCFS2_APPEND_FL			FS_APPEND_FL	/* writes to file may only append */
247#define OCFS2_NODUMP_FL			FS_NODUMP_FL	/* do not dump file */
248#define OCFS2_NOATIME_FL		FS_NOATIME_FL	/* do not update atime */
249/* Reserved for compression usage... */
250#define OCFS2_DIRTY_FL			FS_DIRTY_FL
251#define OCFS2_COMPRBLK_FL		FS_COMPRBLK_FL	/* One or more compressed clusters */
252#define OCFS2_NOCOMP_FL			FS_NOCOMP_FL	/* Don't compress */
253#define OCFS2_ECOMPR_FL			FS_ECOMPR_FL	/* Compression error */
254/* End compression flags --- maybe not all used */
255#define OCFS2_BTREE_FL			FS_BTREE_FL	/* btree format dir */
256#define OCFS2_INDEX_FL			FS_INDEX_FL	/* hash-indexed directory */
257#define OCFS2_IMAGIC_FL			FS_IMAGIC_FL	/* AFS directory */
258#define OCFS2_JOURNAL_DATA_FL		FS_JOURNAL_DATA_FL /* Reserved for ext3 */
259#define OCFS2_NOTAIL_FL			FS_NOTAIL_FL	/* file tail should not be merged */
260#define OCFS2_DIRSYNC_FL		FS_DIRSYNC_FL	/* dirsync behaviour (directories only) */
261#define OCFS2_TOPDIR_FL			FS_TOPDIR_FL	/* Top of directory hierarchies*/
262#define OCFS2_RESERVED_FL		FS_RESERVED_FL	/* reserved for ext2 lib */
263
264#define OCFS2_FL_VISIBLE		FS_FL_USER_VISIBLE	/* User visible flags */
265#define OCFS2_FL_MODIFIABLE		FS_FL_USER_MODIFIABLE	/* User modifiable flags */
266
267/*
268 * Extent record flags (e_node.leaf.flags)
269 */
270#define OCFS2_EXT_UNWRITTEN		(0x01)	/* Extent is allocated but
271						 * unwritten */
272#define OCFS2_EXT_REFCOUNTED		(0x02)  /* Extent is reference
273						 * counted in an associated
274						 * refcount tree */
275
276/*
277 * Journal Flags (ocfs2_dinode.id1.journal1.i_flags)
278 */
279#define OCFS2_JOURNAL_DIRTY_FL	(0x00000001)	/* Journal needs recovery */
280
281/*
282 * superblock s_state flags
283 */
284#define OCFS2_ERROR_FS		(0x00000001)	/* FS saw errors */
285
286/* Limit of space in ocfs2_dir_entry */
287#define OCFS2_MAX_FILENAME_LEN		255
288
289/* Maximum slots on an ocfs2 file system */
290#define OCFS2_MAX_SLOTS			255
291
292/* Slot map indicator for an empty slot */
293#define OCFS2_INVALID_SLOT		((u16)-1)
294
295#define OCFS2_VOL_UUID_LEN		16
296#define OCFS2_MAX_VOL_LABEL_LEN		64
297
298/* The cluster stack fields */
299#define OCFS2_STACK_LABEL_LEN		4
300#define OCFS2_CLUSTER_NAME_LEN		16
301
302/* Classic (historically speaking) cluster stack */
303#define OCFS2_CLASSIC_CLUSTER_STACK	"o2cb"
304
305/* Journal limits (in bytes) */
306#define OCFS2_MIN_JOURNAL_SIZE		(4 * 1024 * 1024)
307
308/*
309 * Inline extended attribute size (in bytes)
310 * The value chosen should be aligned to 16 byte boundaries.
311 */
312#define OCFS2_MIN_XATTR_INLINE_SIZE     256
313
314/*
315 * Cluster info flags (ocfs2_cluster_info.ci_stackflags)
316 */
317#define OCFS2_CLUSTER_O2CB_GLOBAL_HEARTBEAT	(0x01)
318
319struct ocfs2_system_inode_info {
320	char	*si_name;
321	int	si_iflags;
322	int	si_mode;
323};
324
325/* System file index */
326enum {
327	BAD_BLOCK_SYSTEM_INODE = 0,
328	GLOBAL_INODE_ALLOC_SYSTEM_INODE,
329#define OCFS2_FIRST_ONLINE_SYSTEM_INODE GLOBAL_INODE_ALLOC_SYSTEM_INODE
330	SLOT_MAP_SYSTEM_INODE,
331	HEARTBEAT_SYSTEM_INODE,
332	GLOBAL_BITMAP_SYSTEM_INODE,
333	USER_QUOTA_SYSTEM_INODE,
334	GROUP_QUOTA_SYSTEM_INODE,
335#define OCFS2_LAST_GLOBAL_SYSTEM_INODE GROUP_QUOTA_SYSTEM_INODE
336#define OCFS2_FIRST_LOCAL_SYSTEM_INODE ORPHAN_DIR_SYSTEM_INODE
337	ORPHAN_DIR_SYSTEM_INODE,
338	EXTENT_ALLOC_SYSTEM_INODE,
339	INODE_ALLOC_SYSTEM_INODE,
340	JOURNAL_SYSTEM_INODE,
341	LOCAL_ALLOC_SYSTEM_INODE,
342	TRUNCATE_LOG_SYSTEM_INODE,
343	LOCAL_USER_QUOTA_SYSTEM_INODE,
344	LOCAL_GROUP_QUOTA_SYSTEM_INODE,
345#define OCFS2_LAST_LOCAL_SYSTEM_INODE LOCAL_GROUP_QUOTA_SYSTEM_INODE
346	NUM_SYSTEM_INODES
347};
348#define NUM_GLOBAL_SYSTEM_INODES OCFS2_FIRST_LOCAL_SYSTEM_INODE
349#define NUM_LOCAL_SYSTEM_INODES	\
350		(NUM_SYSTEM_INODES - OCFS2_FIRST_LOCAL_SYSTEM_INODE)
351
352static struct ocfs2_system_inode_info ocfs2_system_inodes[NUM_SYSTEM_INODES] = {
353	/* Global system inodes (single copy) */
354	/* The first two are only used from userspace mfks/tunefs */
355	[BAD_BLOCK_SYSTEM_INODE]		= { "bad_blocks", 0, S_IFREG | 0644 },
356	[GLOBAL_INODE_ALLOC_SYSTEM_INODE] 	= { "global_inode_alloc", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
357
358	/* These are used by the running filesystem */
359	[SLOT_MAP_SYSTEM_INODE]			= { "slot_map", 0, S_IFREG | 0644 },
360	[HEARTBEAT_SYSTEM_INODE]		= { "heartbeat", OCFS2_HEARTBEAT_FL, S_IFREG | 0644 },
361	[GLOBAL_BITMAP_SYSTEM_INODE]		= { "global_bitmap", 0, S_IFREG | 0644 },
362	[USER_QUOTA_SYSTEM_INODE]		= { "aquota.user", OCFS2_QUOTA_FL, S_IFREG | 0644 },
363	[GROUP_QUOTA_SYSTEM_INODE]		= { "aquota.group", OCFS2_QUOTA_FL, S_IFREG | 0644 },
364
365	/* Slot-specific system inodes (one copy per slot) */
366	[ORPHAN_DIR_SYSTEM_INODE]		= { "orphan_dir:%04d", 0, S_IFDIR | 0755 },
367	[EXTENT_ALLOC_SYSTEM_INODE]		= { "extent_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
368	[INODE_ALLOC_SYSTEM_INODE]		= { "inode_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_CHAIN_FL, S_IFREG | 0644 },
369	[JOURNAL_SYSTEM_INODE]			= { "journal:%04d", OCFS2_JOURNAL_FL, S_IFREG | 0644 },
370	[LOCAL_ALLOC_SYSTEM_INODE]		= { "local_alloc:%04d", OCFS2_BITMAP_FL | OCFS2_LOCAL_ALLOC_FL, S_IFREG | 0644 },
371	[TRUNCATE_LOG_SYSTEM_INODE]		= { "truncate_log:%04d", OCFS2_DEALLOC_FL, S_IFREG | 0644 },
372	[LOCAL_USER_QUOTA_SYSTEM_INODE]		= { "aquota.user:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
373	[LOCAL_GROUP_QUOTA_SYSTEM_INODE]	= { "aquota.group:%04d", OCFS2_QUOTA_FL, S_IFREG | 0644 },
374};
375
376/* Parameter passed from mount.ocfs2 to module */
377#define OCFS2_HB_NONE			"heartbeat=none"
378#define OCFS2_HB_LOCAL			"heartbeat=local"
379#define OCFS2_HB_GLOBAL			"heartbeat=global"
380
381/*
382 * OCFS2_DIR_PAD defines the directory entries boundaries
383 *
384 * NOTE: It must be a multiple of 4
385 */
386#define OCFS2_DIR_PAD			4
387#define OCFS2_DIR_ROUND			(OCFS2_DIR_PAD - 1)
388#define OCFS2_DIR_MEMBER_LEN 		offsetof(struct ocfs2_dir_entry, name)
389#define OCFS2_DIR_REC_LEN(name_len)	(((name_len) + OCFS2_DIR_MEMBER_LEN + \
390                                          OCFS2_DIR_ROUND) & \
391					 ~OCFS2_DIR_ROUND)
392#define OCFS2_DIR_MIN_REC_LEN	OCFS2_DIR_REC_LEN(1)
393
394#define OCFS2_LINK_MAX		32000
395#define	OCFS2_DX_LINK_MAX	((1U << 31) - 1U)
396#define	OCFS2_LINKS_HI_SHIFT	16
397#define	OCFS2_DX_ENTRIES_MAX	(0xffffffffU)
398
399
400/*
401 * Convenience casts
402 */
403#define OCFS2_RAW_SB(dinode)		(&((dinode)->id2.i_super))
404
405/*
406 * Block checking structure.  This is used in metadata to validate the
407 * contents.  If OCFS2_FEATURE_INCOMPAT_META_ECC is not set, it is all
408 * zeros.
409 */
410struct ocfs2_block_check {
411/*00*/	__le32 bc_crc32e;	/* 802.3 Ethernet II CRC32 */
412	__le16 bc_ecc;		/* Single-error-correction parity vector.
413				   This is a simple Hamming code dependent
414				   on the blocksize.  OCFS2's maximum
415				   blocksize, 4K, requires 16 parity bits,
416				   so we fit in __le16. */
417	__le16 bc_reserved1;
418/*08*/
419};
420
421/*
422 * On disk extent record for OCFS2
423 * It describes a range of clusters on disk.
424 *
425 * Length fields are divided into interior and leaf node versions.
426 * This leaves room for a flags field (OCFS2_EXT_*) in the leaf nodes.
427 */
428struct ocfs2_extent_rec {
429/*00*/	__le32 e_cpos;		/* Offset into the file, in clusters */
430	union {
431		__le32 e_int_clusters; /* Clusters covered by all children */
432		struct {
433			__le16 e_leaf_clusters; /* Clusters covered by this
434						   extent */
435			__u8 e_reserved1;
436			__u8 e_flags; /* Extent flags */
437		};
438	};
439	__le64 e_blkno;		/* Physical disk offset, in blocks */
440/*10*/
441};
442
443struct ocfs2_chain_rec {
444	__le32 c_free;	/* Number of free bits in this chain. */
445	__le32 c_total;	/* Number of total bits in this chain */
446	__le64 c_blkno;	/* Physical disk offset (blocks) of 1st group */
447};
448
449struct ocfs2_truncate_rec {
450	__le32 t_start;		/* 1st cluster in this log */
451	__le32 t_clusters;	/* Number of total clusters covered */
452};
453
454/*
455 * On disk extent list for OCFS2 (node in the tree).  Note that this
456 * is contained inside ocfs2_dinode or ocfs2_extent_block, so the
457 * offsets are relative to ocfs2_dinode.id2.i_list or
458 * ocfs2_extent_block.h_list, respectively.
459 */
460struct ocfs2_extent_list {
461/*00*/	__le16 l_tree_depth;		/* Extent tree depth from this
462					   point.  0 means data extents
463					   hang directly off this
464					   header (a leaf)
465					   NOTE: The high 8 bits cannot be
466					   used - tree_depth is never that big.
467					*/
468	__le16 l_count;			/* Number of extent records */
469	__le16 l_next_free_rec;		/* Next unused extent slot */
470	__le16 l_reserved1;
471	__le64 l_reserved2;		/* Pad to
472					   sizeof(ocfs2_extent_rec) */
473/*10*/	struct ocfs2_extent_rec l_recs[];	/* Extent records */
474};
475
476/*
477 * On disk allocation chain list for OCFS2.  Note that this is
478 * contained inside ocfs2_dinode, so the offsets are relative to
479 * ocfs2_dinode.id2.i_chain.
480 */
481struct ocfs2_chain_list {
482/*00*/	__le16 cl_cpg;			/* Clusters per Block Group */
483	__le16 cl_bpc;			/* Bits per cluster */
484	__le16 cl_count;		/* Total chains in this list */
485	__le16 cl_next_free_rec;	/* Next unused chain slot */
486	__le64 cl_reserved1;
487/*10*/	struct ocfs2_chain_rec cl_recs[];	/* Chain records */
488};
489
490/*
491 * On disk deallocation log for OCFS2.  Note that this is
492 * contained inside ocfs2_dinode, so the offsets are relative to
493 * ocfs2_dinode.id2.i_dealloc.
494 */
495struct ocfs2_truncate_log {
496/*00*/	__le16 tl_count;		/* Total records in this log */
497	__le16 tl_used;			/* Number of records in use */
498	__le32 tl_reserved1;
499/*08*/	struct ocfs2_truncate_rec tl_recs[];	/* Truncate records */
500};
501
502/*
503 * On disk extent block (indirect block) for OCFS2
504 */
505struct ocfs2_extent_block
506{
507/*00*/	__u8 h_signature[8];		/* Signature for verification */
508	struct ocfs2_block_check h_check;	/* Error checking */
509/*10*/	__le16 h_suballoc_slot;		/* Slot suballocator this
510					   extent_header belongs to */
511	__le16 h_suballoc_bit;		/* Bit offset in suballocator
512					   block group */
513	__le32 h_fs_generation;		/* Must match super block */
514	__le64 h_blkno;			/* Offset on disk, in blocks */
515/*20*/	__le64 h_suballoc_loc;		/* Suballocator block group this
516					   eb belongs to.  Only valid
517					   if allocated from a
518					   discontiguous block group */
519	__le64 h_next_leaf_blk;		/* Offset on disk, in blocks,
520					   of next leaf header pointing
521					   to data */
522/*30*/	struct ocfs2_extent_list h_list;	/* Extent record list */
523/* Actual on-disk size is one block */
524};
525
526/*
527 * On disk slot map for OCFS2.  This defines the contents of the "slot_map"
528 * system file.  A slot is valid if it contains a node number >= 0.  The
529 * value -1 (0xFFFF) is OCFS2_INVALID_SLOT.  This marks a slot empty.
530 */
531struct ocfs2_slot_map {
532/*00*/	__le16 sm_slots[0];
533/*
534 * Actual on-disk size is one block.  OCFS2_MAX_SLOTS is 255,
535 * 255 * sizeof(__le16) == 512B, within the 512B block minimum blocksize.
536 */
537};
538
539struct ocfs2_extended_slot {
540/*00*/	__u8	es_valid;
541	__u8	es_reserved1[3];
542	__le32	es_node_num;
543/*08*/
544};
545
546/*
547 * The extended slot map, used when OCFS2_FEATURE_INCOMPAT_EXTENDED_SLOT_MAP
548 * is set.  It separates out the valid marker from the node number, and
549 * has room to grow.  Unlike the old slot map, this format is defined by
550 * i_size.
551 */
552struct ocfs2_slot_map_extended {
553/*00*/	struct ocfs2_extended_slot se_slots[0];
554/*
555 * Actual size is i_size of the slot_map system file.  It should
556 * match s_max_slots * sizeof(struct ocfs2_extended_slot)
557 */
558};
559
560/*
561 * ci_stackflags is only valid if the incompat bit
562 * OCFS2_FEATURE_INCOMPAT_CLUSTERINFO is set.
563 */
564struct ocfs2_cluster_info {
565/*00*/	__u8   ci_stack[OCFS2_STACK_LABEL_LEN];
566	union {
567		__le32 ci_reserved;
568		struct {
569			__u8 ci_stackflags;
570			__u8 ci_reserved1;
571			__u8 ci_reserved2;
572			__u8 ci_reserved3;
573		};
574	};
575/*08*/	__u8   ci_cluster[OCFS2_CLUSTER_NAME_LEN];
576/*18*/
577};
578
579/*
580 * On disk superblock for OCFS2
581 * Note that it is contained inside an ocfs2_dinode, so all offsets
582 * are relative to the start of ocfs2_dinode.id2.
583 */
584struct ocfs2_super_block {
585/*00*/	__le16 s_major_rev_level;
586	__le16 s_minor_rev_level;
587	__le16 s_mnt_count;
588	__le16 s_max_mnt_count;
589	__le16 s_state;			/* File system state */
590	__le16 s_errors;			/* Behaviour when detecting errors */
591	__le32 s_checkinterval;		/* Max time between checks */
592/*10*/	__le64 s_lastcheck;		/* Time of last check */
593	__le32 s_creator_os;		/* OS */
594	__le32 s_feature_compat;		/* Compatible feature set */
595/*20*/	__le32 s_feature_incompat;	/* Incompatible feature set */
596	__le32 s_feature_ro_compat;	/* Readonly-compatible feature set */
597	__le64 s_root_blkno;		/* Offset, in blocks, of root directory
598					   dinode */
599/*30*/	__le64 s_system_dir_blkno;	/* Offset, in blocks, of system
600					   directory dinode */
601	__le32 s_blocksize_bits;		/* Blocksize for this fs */
602	__le32 s_clustersize_bits;	/* Clustersize for this fs */
603/*40*/	__le16 s_max_slots;		/* Max number of simultaneous mounts
604					   before tunefs required */
605	__le16 s_tunefs_flag;
606	__le32 s_uuid_hash;		/* hash value of uuid */
607	__le64 s_first_cluster_group;	/* Block offset of 1st cluster
608					 * group header */
609/*50*/	__u8  s_label[OCFS2_MAX_VOL_LABEL_LEN];	/* Label for mounting, etc. */
610/*90*/	__u8  s_uuid[OCFS2_VOL_UUID_LEN];	/* 128-bit uuid */
611/*A0*/  struct ocfs2_cluster_info s_cluster_info; /* Only valid if either
612						     userspace or clusterinfo
613						     INCOMPAT flag set. */
614/*B8*/	__le16 s_xattr_inline_size;	/* extended attribute inline size
615					   for this fs*/
616	__le16 s_reserved0;
617	__le32 s_dx_seed[3];		/* seed[0-2] for dx dir hash.
618					 * s_uuid_hash serves as seed[3]. */
619/*C0*/  __le64 s_reserved2[15];		/* Fill out superblock */
620/*140*/
621
622	/*
623	 * NOTE: As stated above, all offsets are relative to
624	 * ocfs2_dinode.id2, which is at 0xC0 in the inode.
625	 * 0xC0 + 0x140 = 0x200 or 512 bytes.  A superblock must fit within
626	 * our smallest blocksize, which is 512 bytes.  To ensure this,
627	 * we reserve the space in s_reserved2.  Anything past s_reserved2
628	 * will not be available on the smallest blocksize.
629	 */
630};
631
632/*
633 * Local allocation bitmap for OCFS2 slots
634 * Note that it exists inside an ocfs2_dinode, so all offsets are
635 * relative to the start of ocfs2_dinode.id2.
636 */
637struct ocfs2_local_alloc
638{
639/*00*/	__le32 la_bm_off;	/* Starting bit offset in main bitmap */
640	__le16 la_size;		/* Size of included bitmap, in bytes */
641	__le16 la_reserved1;
642	__le64 la_reserved2;
643/*10*/	__u8   la_bitmap[];
644};
645
646/*
647 * Data-in-inode header. This is only used if i_dyn_features has
648 * OCFS2_INLINE_DATA_FL set.
649 */
650struct ocfs2_inline_data
651{
652/*00*/	__le16	id_count;	/* Number of bytes that can be used
653				 * for data, starting at id_data */
654	__le16	id_reserved0;
655	__le32	id_reserved1;
656	__u8	id_data[];	/* Start of user data */
657};
658
659/*
660 * On disk inode for OCFS2
661 */
662struct ocfs2_dinode {
663/*00*/	__u8 i_signature[8];		/* Signature for validation */
664	__le32 i_generation;		/* Generation number */
665	__le16 i_suballoc_slot;		/* Slot suballocator this inode
666					   belongs to */
667	__le16 i_suballoc_bit;		/* Bit offset in suballocator
668					   block group */
669/*10*/	__le16 i_links_count_hi;	/* High 16 bits of links count */
670	__le16 i_xattr_inline_size;
671	__le32 i_clusters;		/* Cluster count */
672	__le32 i_uid;			/* Owner UID */
673	__le32 i_gid;			/* Owning GID */
674/*20*/	__le64 i_size;			/* Size in bytes */
675	__le16 i_mode;			/* File mode */
676	__le16 i_links_count;		/* Links count */
677	__le32 i_flags;			/* File flags */
678/*30*/	__le64 i_atime;			/* Access time */
679	__le64 i_ctime;			/* Creation time */
680/*40*/	__le64 i_mtime;			/* Modification time */
681	__le64 i_dtime;			/* Deletion time */
682/*50*/	__le64 i_blkno;			/* Offset on disk, in blocks */
683	__le64 i_last_eb_blk;		/* Pointer to last extent
684					   block */
685/*60*/	__le32 i_fs_generation;		/* Generation per fs-instance */
686	__le32 i_atime_nsec;
687	__le32 i_ctime_nsec;
688	__le32 i_mtime_nsec;
689/*70*/	__le32 i_attr;
690	__le16 i_orphaned_slot;		/* Only valid when OCFS2_ORPHANED_FL
691					   was set in i_flags */
692	__le16 i_dyn_features;
693	__le64 i_xattr_loc;
694/*80*/	struct ocfs2_block_check i_check;	/* Error checking */
695/*88*/	__le64 i_dx_root;		/* Pointer to dir index root block */
696/*90*/	__le64 i_refcount_loc;
697	__le64 i_suballoc_loc;		/* Suballocator block group this
698					   inode belongs to.  Only valid
699					   if allocated from a
700					   discontiguous block group */
701/*A0*/	__le16 i_dio_orphaned_slot;	/* only used for append dio write */
702	__le16 i_reserved1[3];
703	__le64 i_reserved2[2];
704/*B8*/	union {
705		__le64 i_pad1;		/* Generic way to refer to this
706					   64bit union */
707		struct {
708			__le64 i_rdev;	/* Device number */
709		} dev1;
710		struct {		/* Info for bitmap system
711					   inodes */
712			__le32 i_used;	/* Bits (ie, clusters) used  */
713			__le32 i_total;	/* Total bits (clusters)
714					   available */
715		} bitmap1;
716		struct {		/* Info for journal system
717					   inodes */
718			__le32 ij_flags;	/* Mounted, version, etc. */
719			__le32 ij_recovery_generation; /* Incremented when the
720							  journal is recovered
721							  after an unclean
722							  shutdown */
723		} journal1;
724	} id1;				/* Inode type dependent 1 */
725/*C0*/	union {
726		struct ocfs2_super_block	i_super;
727		struct ocfs2_local_alloc	i_lab;
728		struct ocfs2_chain_list		i_chain;
729		struct ocfs2_extent_list	i_list;
730		struct ocfs2_truncate_log	i_dealloc;
731		struct ocfs2_inline_data	i_data;
732		__u8               		i_symlink[0];
733	} id2;
734/* Actual on-disk size is one block */
735};
736
737/*
738 * On-disk directory entry structure for OCFS2
739 *
740 * Packed as this structure could be accessed unaligned on 64-bit platforms
741 */
742struct ocfs2_dir_entry {
743/*00*/	__le64   inode;                  /* Inode number */
744	__le16   rec_len;                /* Directory entry length */
745	__u8    name_len;               /* Name length */
746	__u8    file_type;
747/*0C*/	char    name[OCFS2_MAX_FILENAME_LEN];   /* File name */
748/* Actual on-disk length specified by rec_len */
749} __attribute__ ((packed));
750
751/*
752 * Per-block record for the unindexed directory btree. This is carefully
753 * crafted so that the rec_len and name_len records of an ocfs2_dir_entry are
754 * mirrored. That way, the directory manipulation code needs a minimal amount
755 * of update.
756 *
757 * NOTE: Keep this structure aligned to a multiple of 4 bytes.
758 */
759struct ocfs2_dir_block_trailer {
760/*00*/	__le64		db_compat_inode;	/* Always zero. Was inode */
761
762	__le16		db_compat_rec_len;	/* Backwards compatible with
763						 * ocfs2_dir_entry. */
764	__u8		db_compat_name_len;	/* Always zero. Was name_len */
765	__u8		db_reserved0;
766	__le16		db_reserved1;
767	__le16		db_free_rec_len;	/* Size of largest empty hole
768						 * in this block. (unused) */
769/*10*/	__u8		db_signature[8];	/* Signature for verification */
770	__le64		db_reserved2;
771/*20*/	__le64		db_free_next;		/* Next block in list (unused) */
772	__le64		db_blkno;		/* Offset on disk, in blocks */
773/*30*/	__le64		db_parent_dinode;	/* dinode which owns me, in
774						   blocks */
775	struct ocfs2_block_check db_check;	/* Error checking */
776/*40*/
777};
778
779 /*
780 * A directory entry in the indexed tree. We don't store the full name here,
781 * but instead provide a pointer to the full dirent in the unindexed tree.
782 *
783 * We also store name_len here so as to reduce the number of leaf blocks we
784 * need to search in case of collisions.
785 */
786struct ocfs2_dx_entry {
787	__le32		dx_major_hash;	/* Used to find logical
788					 * cluster in index */
789	__le32		dx_minor_hash;	/* Lower bits used to find
790					 * block in cluster */
791	__le64		dx_dirent_blk;	/* Physical block in unindexed
792					 * tree holding this dirent. */
793};
794
795struct ocfs2_dx_entry_list {
796	__le32		de_reserved;
797	__le16		de_count;	/* Maximum number of entries
798					 * possible in de_entries */
799	__le16		de_num_used;	/* Current number of
800					 * de_entries entries */
801	struct	ocfs2_dx_entry		de_entries[];	/* Indexed dir entries
802							 * in a packed array of
803							 * length de_num_used */
804};
805
806#define OCFS2_DX_FLAG_INLINE	0x01
807
808/*
809 * A directory indexing block. Each indexed directory has one of these,
810 * pointed to by ocfs2_dinode.
811 *
812 * This block stores an indexed btree root, and a set of free space
813 * start-of-list pointers.
814 */
815struct ocfs2_dx_root_block {
816	__u8		dr_signature[8];	/* Signature for verification */
817	struct ocfs2_block_check dr_check;	/* Error checking */
818	__le16		dr_suballoc_slot;	/* Slot suballocator this
819						 * block belongs to. */
820	__le16		dr_suballoc_bit;	/* Bit offset in suballocator
821						 * block group */
822	__le32		dr_fs_generation;	/* Must match super block */
823	__le64		dr_blkno;		/* Offset on disk, in blocks */
824	__le64		dr_last_eb_blk;		/* Pointer to last
825						 * extent block */
826	__le32		dr_clusters;		/* Clusters allocated
827						 * to the indexed tree. */
828	__u8		dr_flags;		/* OCFS2_DX_FLAG_* flags */
829	__u8		dr_reserved0;
830	__le16		dr_reserved1;
831	__le64		dr_dir_blkno;		/* Pointer to parent inode */
832	__le32		dr_num_entries;		/* Total number of
833						 * names stored in
834						 * this directory.*/
835	__le32		dr_reserved2;
836	__le64		dr_free_blk;		/* Pointer to head of free
837						 * unindexed block list. */
838	__le64		dr_suballoc_loc;	/* Suballocator block group
839						   this root belongs to.
840						   Only valid if allocated
841						   from a discontiguous
842						   block group */
843	__le64		dr_reserved3[14];
844	union {
845		struct ocfs2_extent_list dr_list; /* Keep this aligned to 128
846						   * bits for maximum space
847						   * efficiency. */
848		struct ocfs2_dx_entry_list dr_entries; /* In-root-block list of
849							* entries. We grow out
850							* to extents if this
851							* gets too big. */
852	};
853};
854
855/*
856 * The header of a leaf block in the indexed tree.
857 */
858struct ocfs2_dx_leaf {
859	__u8		dl_signature[8];/* Signature for verification */
860	struct ocfs2_block_check dl_check;	/* Error checking */
861	__le64		dl_blkno;	/* Offset on disk, in blocks */
862	__le32		dl_fs_generation;/* Must match super block */
863	__le32		dl_reserved0;
864	__le64		dl_reserved1;
865	struct ocfs2_dx_entry_list	dl_list;
866};
867
868/*
869 * Largest bitmap for a block (suballocator) group in bytes.  This limit
870 * does not affect cluster groups (global allocator).  Cluster group
871 * bitmaps run to the end of the block.
872 */
873#define OCFS2_MAX_BG_BITMAP_SIZE	256
874
875/*
876 * On disk allocator group structure for OCFS2
877 */
878struct ocfs2_group_desc
879{
880/*00*/	__u8    bg_signature[8];        /* Signature for validation */
881	__le16   bg_size;                /* Size of included bitmap in
882					   bytes. */
883	__le16   bg_bits;                /* Bits represented by this
884					   group. */
885	__le16	bg_free_bits_count;     /* Free bits count */
886	__le16   bg_chain;               /* What chain I am in. */
887/*10*/	__le32   bg_generation;
888	__le32	bg_reserved1;
889	__le64   bg_next_group;          /* Next group in my list, in
890					   blocks */
891/*20*/	__le64   bg_parent_dinode;       /* dinode which owns me, in
892					   blocks */
893	__le64   bg_blkno;               /* Offset on disk, in blocks */
894/*30*/	struct ocfs2_block_check bg_check;	/* Error checking */
895	__le64   bg_reserved2;
896/*40*/	union {
897		__u8    bg_bitmap[0];
898		struct {
899			/*
900			 * Block groups may be discontiguous when
901			 * OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG is set.
902			 * The extents of a discontiguous block group are
903			 * stored in bg_list.  It is a flat list.
904			 * l_tree_depth must always be zero.  A
905			 * discontiguous group is signified by a non-zero
906			 * bg_list->l_next_free_rec.  Only block groups
907			 * can be discontiguous; Cluster groups cannot.
908			 * We've never made a block group with more than
909			 * 2048 blocks (256 bytes of bg_bitmap).  This
910			 * codifies that limit so that we can fit bg_list.
911			 * bg_size of a discontiguous block group will
912			 * be 256 to match bg_bitmap_filler.
913			 */
914			__u8 bg_bitmap_filler[OCFS2_MAX_BG_BITMAP_SIZE];
915/*140*/			struct ocfs2_extent_list bg_list;
916		};
917	};
918/* Actual on-disk size is one block */
919};
920
921struct ocfs2_refcount_rec {
922/*00*/	__le64 r_cpos;		/* Physical offset, in clusters */
923	__le32 r_clusters;	/* Clusters covered by this extent */
924	__le32 r_refcount;	/* Reference count of this extent */
925/*10*/
926};
927#define OCFS2_32BIT_POS_MASK		(0xffffffffULL)
928
929#define OCFS2_REFCOUNT_LEAF_FL          (0x00000001)
930#define OCFS2_REFCOUNT_TREE_FL          (0x00000002)
931
932struct ocfs2_refcount_list {
933/*00*/	__le16 rl_count;	/* Maximum number of entries possible
934				   in rl_records */
935	__le16 rl_used;		/* Current number of used records */
936	__le32 rl_reserved2;
937	__le64 rl_reserved1;	/* Pad to sizeof(ocfs2_refcount_record) */
938/*10*/	struct ocfs2_refcount_rec rl_recs[];	/* Refcount records */
939};
940
941
942struct ocfs2_refcount_block {
943/*00*/	__u8 rf_signature[8];		/* Signature for verification */
944	__le16 rf_suballoc_slot;	/* Slot suballocator this block
945					   belongs to */
946	__le16 rf_suballoc_bit;		/* Bit offset in suballocator
947					   block group */
948	__le32 rf_fs_generation;	/* Must match superblock */
949/*10*/	__le64 rf_blkno;		/* Offset on disk, in blocks */
950	__le64 rf_parent;		/* Parent block, only valid if
951					   OCFS2_REFCOUNT_LEAF_FL is set in
952					   rf_flags */
953/*20*/	struct ocfs2_block_check rf_check;	/* Error checking */
954	__le64 rf_last_eb_blk;		/* Pointer to last extent block */
955/*30*/	__le32 rf_count;		/* Number of inodes sharing this
956					   refcount tree */
957	__le32 rf_flags;		/* See the flags above */
958	__le32 rf_clusters;		/* clusters covered by refcount tree. */
959	__le32 rf_cpos;			/* cluster offset in refcount tree.*/
960/*40*/	__le32 rf_generation;		/* generation number. all be the same
961					 * for the same refcount tree. */
962	__le32 rf_reserved0;
963	__le64 rf_suballoc_loc;		/* Suballocator block group this
964					   refcount block belongs to. Only
965					   valid if allocated from a
966					   discontiguous block group */
967/*50*/	__le64 rf_reserved1[6];
968/*80*/	union {
969		struct ocfs2_refcount_list rf_records;  /* List of refcount
970							  records */
971		struct ocfs2_extent_list rf_list;	/* Extent record list,
972							only valid if
973							OCFS2_REFCOUNT_TREE_FL
974							is set in rf_flags */
975	};
976/* Actual on-disk size is one block */
977};
978
979/*
980 * On disk extended attribute structure for OCFS2.
981 */
982
983/*
984 * ocfs2_xattr_entry indicates one extend attribute.
985 *
986 * Note that it can be stored in inode, one block or one xattr bucket.
987 */
988struct ocfs2_xattr_entry {
989	__le32	xe_name_hash;    /* hash value of xattr prefix+suffix. */
990	__le16	xe_name_offset;  /* byte offset from the 1st entry in the
991				    local xattr storage(inode, xattr block or
992				    xattr bucket). */
993	__u8	xe_name_len;	 /* xattr name len, doesn't include prefix. */
994	__u8	xe_type;         /* the low 7 bits indicate the name prefix
995				  * type and the highest bit indicates whether
996				  * the EA is stored in the local storage. */
997	__le64	xe_value_size;	 /* real xattr value length. */
998};
999
1000/*
1001 * On disk structure for xattr header.
1002 *
1003 * One ocfs2_xattr_header describes how many ocfs2_xattr_entry records in
1004 * the local xattr storage.
1005 */
1006struct ocfs2_xattr_header {
1007	__le16	xh_count;                       /* contains the count of how
1008						   many records are in the
1009						   local xattr storage. */
1010	__le16	xh_free_start;                  /* current offset for storing
1011						   xattr. */
1012	__le16	xh_name_value_len;              /* total length of name/value
1013						   length in this bucket. */
1014	__le16	xh_num_buckets;                 /* Number of xattr buckets
1015						   in this extent record,
1016						   only valid in the first
1017						   bucket. */
1018	struct ocfs2_block_check xh_check;	/* Error checking
1019						   (Note, this is only
1020						    used for xattr
1021						    buckets.  A block uses
1022						    xb_check and sets
1023						    this field to zero.) */
1024	struct ocfs2_xattr_entry xh_entries[]; /* xattr entry list. */
1025};
1026
1027/*
1028 * On disk structure for xattr value root.
1029 *
1030 * When an xattr's value is large enough, it is stored in an external
1031 * b-tree like file data.  The xattr value root points to this structure.
1032 */
1033struct ocfs2_xattr_value_root {
1034/*00*/	__le32	xr_clusters;              /* clusters covered by xattr value. */
1035	__le32	xr_reserved0;
1036	__le64	xr_last_eb_blk;           /* Pointer to last extent block */
1037/*10*/	struct ocfs2_extent_list xr_list; /* Extent record list */
1038};
1039
1040/*
1041 * On disk structure for xattr tree root.
1042 *
1043 * It is used when there are too many extended attributes for one file. These
1044 * attributes will be organized and stored in an indexed-btree.
1045 */
1046struct ocfs2_xattr_tree_root {
1047/*00*/	__le32	xt_clusters;              /* clusters covered by xattr. */
1048	__le32	xt_reserved0;
1049	__le64	xt_last_eb_blk;           /* Pointer to last extent block */
1050/*10*/	struct ocfs2_extent_list xt_list; /* Extent record list */
1051};
1052
1053#define OCFS2_XATTR_INDEXED	0x1
1054#define OCFS2_HASH_SHIFT	5
1055#define OCFS2_XATTR_ROUND	3
1056#define OCFS2_XATTR_SIZE(size)	(((size) + OCFS2_XATTR_ROUND) & \
1057				~(OCFS2_XATTR_ROUND))
1058
1059#define OCFS2_XATTR_BUCKET_SIZE			4096
1060#define OCFS2_XATTR_MAX_BLOCKS_PER_BUCKET 	(OCFS2_XATTR_BUCKET_SIZE \
1061						 / OCFS2_MIN_BLOCKSIZE)
1062
1063/*
1064 * On disk structure for xattr block.
1065 */
1066struct ocfs2_xattr_block {
1067/*00*/	__u8	xb_signature[8];     /* Signature for verification */
1068	__le16	xb_suballoc_slot;    /* Slot suballocator this
1069					block belongs to. */
1070	__le16	xb_suballoc_bit;     /* Bit offset in suballocator
1071					block group */
1072	__le32	xb_fs_generation;    /* Must match super block */
1073/*10*/	__le64	xb_blkno;            /* Offset on disk, in blocks */
1074	struct ocfs2_block_check xb_check;	/* Error checking */
1075/*20*/	__le16	xb_flags;            /* Indicates whether this block contains
1076					real xattr or a xattr tree. */
1077	__le16	xb_reserved0;
1078	__le32  xb_reserved1;
1079	__le64	xb_suballoc_loc;	/* Suballocator block group this
1080					   xattr block belongs to. Only
1081					   valid if allocated from a
1082					   discontiguous block group */
1083/*30*/	union {
1084		struct ocfs2_xattr_header xb_header; /* xattr header if this
1085							block contains xattr */
1086		struct ocfs2_xattr_tree_root xb_root;/* xattr tree root if this
1087							block cotains xattr
1088							tree. */
1089	} xb_attrs;
1090};
1091
1092#define OCFS2_XATTR_ENTRY_LOCAL		0x80
1093#define OCFS2_XATTR_TYPE_MASK		0x7F
1094static inline void ocfs2_xattr_set_local(struct ocfs2_xattr_entry *xe,
1095					 int local)
1096{
1097	if (local)
1098		xe->xe_type |= OCFS2_XATTR_ENTRY_LOCAL;
1099	else
1100		xe->xe_type &= ~OCFS2_XATTR_ENTRY_LOCAL;
1101}
1102
1103static inline int ocfs2_xattr_is_local(struct ocfs2_xattr_entry *xe)
1104{
1105	return xe->xe_type & OCFS2_XATTR_ENTRY_LOCAL;
1106}
1107
1108static inline void ocfs2_xattr_set_type(struct ocfs2_xattr_entry *xe, int type)
1109{
1110	xe->xe_type |= type & OCFS2_XATTR_TYPE_MASK;
1111}
1112
1113static inline int ocfs2_xattr_get_type(struct ocfs2_xattr_entry *xe)
1114{
1115	return xe->xe_type & OCFS2_XATTR_TYPE_MASK;
1116}
1117
1118/*
1119 *  On disk structures for global quota file
1120 */
1121
1122/* Magic numbers and known versions for global quota files */
1123#define OCFS2_GLOBAL_QMAGICS {\
1124	0x0cf52470, /* USRQUOTA */ \
1125	0x0cf52471  /* GRPQUOTA */ \
1126}
1127
1128#define OCFS2_GLOBAL_QVERSIONS {\
1129	0, \
1130	0, \
1131}
1132
1133
1134/* Each block of each quota file has a certain fixed number of bytes reserved
1135 * for OCFS2 internal use at its end. OCFS2 can use it for things like
1136 * checksums, etc. */
1137#define OCFS2_QBLK_RESERVED_SPACE 8
1138
1139/* Generic header of all quota files */
1140struct ocfs2_disk_dqheader {
1141	__le32 dqh_magic;	/* Magic number identifying file */
1142	__le32 dqh_version;	/* Quota format version */
1143};
1144
1145#define OCFS2_GLOBAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1146
1147/* Information header of global quota file (immediately follows the generic
1148 * header) */
1149struct ocfs2_global_disk_dqinfo {
1150/*00*/	__le32 dqi_bgrace;	/* Grace time for space softlimit excess */
1151	__le32 dqi_igrace;	/* Grace time for inode softlimit excess */
1152	__le32 dqi_syncms;	/* Time after which we sync local changes to
1153				 * global quota file */
1154	__le32 dqi_blocks;	/* Number of blocks in quota file */
1155/*10*/	__le32 dqi_free_blk;	/* First free block in quota file */
1156	__le32 dqi_free_entry;	/* First block with free dquot entry in quota
1157				 * file */
1158};
1159
1160/* Structure with global user / group information. We reserve some space
1161 * for future use. */
1162struct ocfs2_global_disk_dqblk {
1163/*00*/	__le32 dqb_id;          /* ID the structure belongs to */
1164	__le32 dqb_use_count;   /* Number of nodes having reference to this structure */
1165	__le64 dqb_ihardlimit;  /* absolute limit on allocated inodes */
1166/*10*/	__le64 dqb_isoftlimit;  /* preferred inode limit */
1167	__le64 dqb_curinodes;   /* current # allocated inodes */
1168/*20*/	__le64 dqb_bhardlimit;  /* absolute limit on disk space */
1169	__le64 dqb_bsoftlimit;  /* preferred limit on disk space */
1170/*30*/	__le64 dqb_curspace;    /* current space occupied */
1171	__le64 dqb_btime;       /* time limit for excessive disk use */
1172/*40*/	__le64 dqb_itime;       /* time limit for excessive inode use */
1173	__le64 dqb_pad1;
1174/*50*/	__le64 dqb_pad2;
1175};
1176
1177/*
1178 *  On-disk structures for local quota file
1179 */
1180
1181/* Magic numbers and known versions for local quota files */
1182#define OCFS2_LOCAL_QMAGICS {\
1183	0x0cf524c0, /* USRQUOTA */ \
1184	0x0cf524c1  /* GRPQUOTA */ \
1185}
1186
1187#define OCFS2_LOCAL_QVERSIONS {\
1188	0, \
1189	0, \
1190}
1191
1192/* Quota flags in dqinfo header */
1193#define OLQF_CLEAN	0x0001	/* Quota file is empty (this should be after\
1194				 * quota has been cleanly turned off) */
1195
1196#define OCFS2_LOCAL_INFO_OFF (sizeof(struct ocfs2_disk_dqheader))
1197
1198/* Information header of local quota file (immediately follows the generic
1199 * header) */
1200struct ocfs2_local_disk_dqinfo {
1201	__le32 dqi_flags;	/* Flags for quota file */
1202	__le32 dqi_chunks;	/* Number of chunks of quota structures
1203				 * with a bitmap */
1204	__le32 dqi_blocks;	/* Number of blocks allocated for quota file */
1205};
1206
1207/* Header of one chunk of a quota file */
1208struct ocfs2_local_disk_chunk {
1209	__le32 dqc_free;	/* Number of free entries in the bitmap */
1210	__u8 dqc_bitmap[];	/* Bitmap of entries in the corresponding
1211				 * chunk of quota file */
1212};
1213
1214/* One entry in local quota file */
1215struct ocfs2_local_disk_dqblk {
1216/*00*/	__le64 dqb_id;		/* id this quota applies to */
1217	__le64 dqb_spacemod;	/* Change in the amount of used space */
1218/*10*/	__le64 dqb_inodemod;	/* Change in the amount of used inodes */
1219};
1220
1221
1222/*
1223 * The quota trailer lives at the end of each quota block.
1224 */
1225
1226struct ocfs2_disk_dqtrailer {
1227/*00*/	struct ocfs2_block_check dq_check;	/* Error checking */
1228/*08*/	/* Cannot be larger than OCFS2_QBLK_RESERVED_SPACE */
1229};
1230
1231static inline struct ocfs2_disk_dqtrailer *ocfs2_block_dqtrailer(int blocksize,
1232								 void *buf)
1233{
1234	char *ptr = buf;
1235	ptr += blocksize - OCFS2_QBLK_RESERVED_SPACE;
1236
1237	return (struct ocfs2_disk_dqtrailer *)ptr;
1238}
1239
1240#ifdef __KERNEL__
1241static inline int ocfs2_fast_symlink_chars(struct super_block *sb)
1242{
1243	return  sb->s_blocksize -
1244		 offsetof(struct ocfs2_dinode, id2.i_symlink);
1245}
1246
1247static inline int ocfs2_max_inline_data_with_xattr(struct super_block *sb,
1248						   struct ocfs2_dinode *di)
1249{
1250	unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1251
1252	if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1253		return sb->s_blocksize -
1254			offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1255			xattrsize;
1256	else
1257		return sb->s_blocksize -
1258			offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1259}
1260
1261static inline int ocfs2_extent_recs_per_inode(struct super_block *sb)
1262{
1263	int size;
1264
1265	size = sb->s_blocksize -
1266		offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1267
1268	return size / sizeof(struct ocfs2_extent_rec);
1269}
1270
1271static inline int ocfs2_extent_recs_per_inode_with_xattr(
1272						struct super_block *sb,
1273						struct ocfs2_dinode *di)
1274{
1275	int size;
1276	unsigned int xattrsize = le16_to_cpu(di->i_xattr_inline_size);
1277
1278	if (le16_to_cpu(di->i_dyn_features) & OCFS2_INLINE_XATTR_FL)
1279		size = sb->s_blocksize -
1280			offsetof(struct ocfs2_dinode, id2.i_list.l_recs) -
1281			xattrsize;
1282	else
1283		size = sb->s_blocksize -
1284			offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1285
1286	return size / sizeof(struct ocfs2_extent_rec);
1287}
1288
1289static inline int ocfs2_extent_recs_per_dx_root(struct super_block *sb)
1290{
1291	int size;
1292
1293	size = sb->s_blocksize -
1294		offsetof(struct ocfs2_dx_root_block, dr_list.l_recs);
1295
1296	return size / sizeof(struct ocfs2_extent_rec);
1297}
1298
1299static inline int ocfs2_chain_recs_per_inode(struct super_block *sb)
1300{
1301	int size;
1302
1303	size = sb->s_blocksize -
1304		offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1305
1306	return size / sizeof(struct ocfs2_chain_rec);
1307}
1308
1309static inline u16 ocfs2_extent_recs_per_eb(struct super_block *sb)
1310{
1311	int size;
1312
1313	size = sb->s_blocksize -
1314		offsetof(struct ocfs2_extent_block, h_list.l_recs);
1315
1316	return size / sizeof(struct ocfs2_extent_rec);
1317}
1318
1319static inline u16 ocfs2_extent_recs_per_gd(struct super_block *sb)
1320{
1321	int size;
1322
1323	size = sb->s_blocksize -
1324		offsetof(struct ocfs2_group_desc, bg_list.l_recs);
1325
1326	return size / sizeof(struct ocfs2_extent_rec);
1327}
1328
1329static inline int ocfs2_dx_entries_per_leaf(struct super_block *sb)
1330{
1331	int size;
1332
1333	size = sb->s_blocksize -
1334		offsetof(struct ocfs2_dx_leaf, dl_list.de_entries);
1335
1336	return size / sizeof(struct ocfs2_dx_entry);
1337}
1338
1339static inline int ocfs2_dx_entries_per_root(struct super_block *sb)
1340{
1341	int size;
1342
1343	size = sb->s_blocksize -
1344		offsetof(struct ocfs2_dx_root_block, dr_entries.de_entries);
1345
1346	return size / sizeof(struct ocfs2_dx_entry);
1347}
1348
1349static inline u16 ocfs2_local_alloc_size(struct super_block *sb)
1350{
1351	u16 size;
1352
1353	size = sb->s_blocksize -
1354		offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1355
1356	return size;
1357}
1358
1359static inline int ocfs2_group_bitmap_size(struct super_block *sb,
1360					  int suballocator,
1361					  u32 feature_incompat)
1362{
1363	int size = sb->s_blocksize -
1364		offsetof(struct ocfs2_group_desc, bg_bitmap);
1365
1366	/*
1367	 * The cluster allocator uses the entire block.  Suballocators have
1368	 * never used more than OCFS2_MAX_BG_BITMAP_SIZE.  Unfortunately, older
1369	 * code expects bg_size set to the maximum.  Thus we must keep
1370	 * bg_size as-is unless discontig_bg is enabled.
1371	 */
1372	if (suballocator &&
1373	    (feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG))
1374		size = OCFS2_MAX_BG_BITMAP_SIZE;
1375
1376	return size;
1377}
1378
1379static inline int ocfs2_truncate_recs_per_inode(struct super_block *sb)
1380{
1381	int size;
1382
1383	size = sb->s_blocksize -
1384		offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1385
1386	return size / sizeof(struct ocfs2_truncate_rec);
1387}
1388
1389static inline u64 ocfs2_backup_super_blkno(struct super_block *sb, int index)
1390{
1391	u64 offset = OCFS2_BACKUP_SB_START;
1392
1393	if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1394		offset <<= (2 * index);
1395		offset >>= sb->s_blocksize_bits;
1396		return offset;
1397	}
1398
1399	return 0;
1400
1401}
1402
1403static inline u16 ocfs2_xattr_recs_per_xb(struct super_block *sb)
1404{
1405	int size;
1406
1407	size = sb->s_blocksize -
1408		offsetof(struct ocfs2_xattr_block,
1409			 xb_attrs.xb_root.xt_list.l_recs);
1410
1411	return size / sizeof(struct ocfs2_extent_rec);
1412}
1413
1414static inline u16 ocfs2_extent_recs_per_rb(struct super_block *sb)
1415{
1416	int size;
1417
1418	size = sb->s_blocksize -
1419		offsetof(struct ocfs2_refcount_block, rf_list.l_recs);
1420
1421	return size / sizeof(struct ocfs2_extent_rec);
1422}
1423
1424static inline u16 ocfs2_refcount_recs_per_rb(struct super_block *sb)
1425{
1426	int size;
1427
1428	size = sb->s_blocksize -
1429		offsetof(struct ocfs2_refcount_block, rf_records.rl_recs);
1430
1431	return size / sizeof(struct ocfs2_refcount_rec);
1432}
1433
1434static inline u32
1435ocfs2_get_ref_rec_low_cpos(const struct ocfs2_refcount_rec *rec)
1436{
1437	return le64_to_cpu(rec->r_cpos) & OCFS2_32BIT_POS_MASK;
1438}
1439#else
1440static inline int ocfs2_fast_symlink_chars(int blocksize)
1441{
1442	return blocksize - offsetof(struct ocfs2_dinode, id2.i_symlink);
1443}
1444
1445static inline int ocfs2_max_inline_data_with_xattr(int blocksize,
1446						   struct ocfs2_dinode *di)
1447{
1448	if (di && (di->i_dyn_features & OCFS2_INLINE_XATTR_FL))
1449		return blocksize -
1450			offsetof(struct ocfs2_dinode, id2.i_data.id_data) -
1451			di->i_xattr_inline_size;
1452	else
1453		return blocksize -
1454			offsetof(struct ocfs2_dinode, id2.i_data.id_data);
1455}
1456
1457static inline int ocfs2_extent_recs_per_inode(int blocksize)
1458{
1459	int size;
1460
1461	size = blocksize -
1462		offsetof(struct ocfs2_dinode, id2.i_list.l_recs);
1463
1464	return size / sizeof(struct ocfs2_extent_rec);
1465}
1466
1467static inline int ocfs2_chain_recs_per_inode(int blocksize)
1468{
1469	int size;
1470
1471	size = blocksize -
1472		offsetof(struct ocfs2_dinode, id2.i_chain.cl_recs);
1473
1474	return size / sizeof(struct ocfs2_chain_rec);
1475}
1476
1477static inline int ocfs2_extent_recs_per_eb(int blocksize)
1478{
1479	int size;
1480
1481	size = blocksize -
1482		offsetof(struct ocfs2_extent_block, h_list.l_recs);
1483
1484	return size / sizeof(struct ocfs2_extent_rec);
1485}
1486
1487static inline int ocfs2_extent_recs_per_gd(int blocksize)
1488{
1489	int size;
1490
1491	size = blocksize -
1492		offsetof(struct ocfs2_group_desc, bg_list.l_recs);
1493
1494	return size / sizeof(struct ocfs2_extent_rec);
1495}
1496
1497static inline int ocfs2_local_alloc_size(int blocksize)
1498{
1499	int size;
1500
1501	size = blocksize -
1502		offsetof(struct ocfs2_dinode, id2.i_lab.la_bitmap);
1503
1504	return size;
1505}
1506
1507static inline int ocfs2_group_bitmap_size(int blocksize,
1508					  int suballocator,
1509					  uint32_t feature_incompat)
1510{
1511	int size = sb->s_blocksize -
1512		offsetof(struct ocfs2_group_desc, bg_bitmap);
1513
1514	/*
1515	 * The cluster allocator uses the entire block.  Suballocators have
1516	 * never used more than OCFS2_MAX_BG_BITMAP_SIZE.  Unfortunately, older
1517	 * code expects bg_size set to the maximum.  Thus we must keep
1518	 * bg_size as-is unless discontig_bg is enabled.
1519	 */
1520	if (suballocator &&
1521	    (feature_incompat & OCFS2_FEATURE_INCOMPAT_DISCONTIG_BG))
1522		size = OCFS2_MAX_BG_BITMAP_SIZE;
1523
1524	return size;
1525}
1526
1527static inline int ocfs2_truncate_recs_per_inode(int blocksize)
1528{
1529	int size;
1530
1531	size = blocksize -
1532		offsetof(struct ocfs2_dinode, id2.i_dealloc.tl_recs);
1533
1534	return size / sizeof(struct ocfs2_truncate_rec);
1535}
1536
1537static inline uint64_t ocfs2_backup_super_blkno(int blocksize, int index)
1538{
1539	uint64_t offset = OCFS2_BACKUP_SB_START;
1540
1541	if (index >= 0 && index < OCFS2_MAX_BACKUP_SUPERBLOCKS) {
1542		offset <<= (2 * index);
1543		offset /= blocksize;
1544		return offset;
1545	}
1546
1547	return 0;
1548}
1549
1550static inline int ocfs2_xattr_recs_per_xb(int blocksize)
1551{
1552	int size;
1553
1554	size = blocksize -
1555		offsetof(struct ocfs2_xattr_block,
1556			 xb_attrs.xb_root.xt_list.l_recs);
1557
1558	return size / sizeof(struct ocfs2_extent_rec);
1559}
1560#endif  /* __KERNEL__ */
1561
1562
1563static inline int ocfs2_system_inode_is_global(int type)
1564{
1565	return ((type >= 0) &&
1566		(type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE));
1567}
1568
1569static inline int ocfs2_sprintf_system_inode_name(char *buf, int len,
1570						  int type, int slot)
1571{
1572	int chars;
1573
1574        /*
1575         * Global system inodes can only have one copy.  Everything
1576         * after OCFS2_LAST_GLOBAL_SYSTEM_INODE in the system inode
1577         * list has a copy per slot.
1578         */
1579	if (type <= OCFS2_LAST_GLOBAL_SYSTEM_INODE)
1580		chars = snprintf(buf, len, "%s",
1581				 ocfs2_system_inodes[type].si_name);
1582	else
1583		chars = snprintf(buf, len,
1584				 ocfs2_system_inodes[type].si_name,
1585				 slot);
1586
1587	return chars;
1588}
1589
1590static inline void ocfs2_set_de_type(struct ocfs2_dir_entry *de,
1591				    umode_t mode)
1592{
1593	de->file_type = fs_umode_to_ftype(mode);
1594}
1595
1596static inline int ocfs2_gd_is_discontig(struct ocfs2_group_desc *gd)
1597{
1598	if ((offsetof(struct ocfs2_group_desc, bg_bitmap) +
1599	     le16_to_cpu(gd->bg_size)) !=
1600	    offsetof(struct ocfs2_group_desc, bg_list))
1601		return 0;
1602	/*
1603	 * Only valid to check l_next_free_rec if
1604	 * bg_bitmap + bg_size == bg_list.
1605	 */
1606	if (!gd->bg_list.l_next_free_rec)
1607		return 0;
1608	return 1;
1609}
1610#endif  /* _OCFS2_FS_H */
1611
1612