1/*
   2 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
   3 * All Rights Reserved.
   4 *
   5 * This program is free software; you can redistribute it and/or
   6 * modify it under the terms of the GNU General Public License as
   7 * published by the Free Software Foundation.
   8 *
   9 * This program is distributed in the hope that it would be useful,
  10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12 * GNU General Public License for more details.
  13 *
  14 * You should have received a copy of the GNU General Public License
  15 * along with this program; if not, write the Free Software Foundation,
  16 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17 */
  18#ifndef __XFS_FORMAT_H__
  19#define __XFS_FORMAT_H__
  20
  21/*
  22 * XFS On Disk Format Definitions
  23 *
  24 * This header file defines all the on-disk format definitions for 
  25 * general XFS objects. Directory and attribute related objects are defined in
  26 * xfs_da_format.h, which log and log item formats are defined in
  27 * xfs_log_format.h. Everything else goes here.
  28 */
  29
  30struct xfs_mount;
  31struct xfs_trans;
  32struct xfs_inode;
  33struct xfs_buf;
  34struct xfs_ifork;
  35
  36/*
  37 * Super block
  38 * Fits into a sector-sized buffer at address 0 of each allocation group.
  39 * Only the first of these is ever updated except during growfs.
  40 */
  41#define	XFS_SB_MAGIC		0x58465342	/* 'XFSB' */
  42#define	XFS_SB_VERSION_1	1		/* 5.3, 6.0.1, 6.1 */
  43#define	XFS_SB_VERSION_2	2		/* 6.2 - attributes */
  44#define	XFS_SB_VERSION_3	3		/* 6.2 - new inode version */
  45#define	XFS_SB_VERSION_4	4		/* 6.2+ - bitmask version */
  46#define	XFS_SB_VERSION_5	5		/* CRC enabled filesystem */
  47#define	XFS_SB_VERSION_NUMBITS		0x000f
  48#define	XFS_SB_VERSION_ALLFBITS		0xfff0
  49#define	XFS_SB_VERSION_ATTRBIT		0x0010
  50#define	XFS_SB_VERSION_NLINKBIT		0x0020
  51#define	XFS_SB_VERSION_QUOTABIT		0x0040
  52#define	XFS_SB_VERSION_ALIGNBIT		0x0080
  53#define	XFS_SB_VERSION_DALIGNBIT	0x0100
  54#define	XFS_SB_VERSION_SHAREDBIT	0x0200
  55#define XFS_SB_VERSION_LOGV2BIT		0x0400
  56#define XFS_SB_VERSION_SECTORBIT	0x0800
  57#define	XFS_SB_VERSION_EXTFLGBIT	0x1000
  58#define	XFS_SB_VERSION_DIRV2BIT		0x2000
  59#define	XFS_SB_VERSION_BORGBIT		0x4000	/* ASCII only case-insens. */
  60#define	XFS_SB_VERSION_MOREBITSBIT	0x8000
  61
  62/*
  63 * The size of a single extended attribute on disk is limited by
  64 * the size of index values within the attribute entries themselves.
  65 * These are be16 fields, so we can only support attribute data
  66 * sizes up to 2^16 bytes in length.
  67 */
  68#define XFS_XATTR_SIZE_MAX (1 << 16)
  69
  70/*
  71 * Supported feature bit list is just all bits in the versionnum field because
  72 * we've used them all up and understand them all. Except, of course, for the
  73 * shared superblock bit, which nobody knows what it does and so is unsupported.
  74 */
  75#define	XFS_SB_VERSION_OKBITS		\
  76	((XFS_SB_VERSION_NUMBITS | XFS_SB_VERSION_ALLFBITS) & \
  77		~XFS_SB_VERSION_SHAREDBIT)
  78
  79/*
  80 * There are two words to hold XFS "feature" bits: the original
  81 * word, sb_versionnum, and sb_features2.  Whenever a bit is set in
  82 * sb_features2, the feature bit XFS_SB_VERSION_MOREBITSBIT must be set.
  83 *
  84 * These defines represent bits in sb_features2.
  85 */
  86#define XFS_SB_VERSION2_RESERVED1BIT	0x00000001
  87#define XFS_SB_VERSION2_LAZYSBCOUNTBIT	0x00000002	/* Superblk counters */
  88#define XFS_SB_VERSION2_RESERVED4BIT	0x00000004
  89#define XFS_SB_VERSION2_ATTR2BIT	0x00000008	/* Inline attr rework */
  90#define XFS_SB_VERSION2_PARENTBIT	0x00000010	/* parent pointers */
  91#define XFS_SB_VERSION2_PROJID32BIT	0x00000080	/* 32 bit project id */
  92#define XFS_SB_VERSION2_CRCBIT		0x00000100	/* metadata CRCs */
  93#define XFS_SB_VERSION2_FTYPE		0x00000200	/* inode type in dir */
  94
  95#define	XFS_SB_VERSION2_OKBITS		\
  96	(XFS_SB_VERSION2_LAZYSBCOUNTBIT	| \
  97	 XFS_SB_VERSION2_ATTR2BIT	| \
  98	 XFS_SB_VERSION2_PROJID32BIT	| \
  99	 XFS_SB_VERSION2_FTYPE)
 100
 
 
 
 101/*
 102 * Superblock - in core version.  Must match the ondisk version below.
 103 * Must be padded to 64 bit alignment.
 104 */
 105typedef struct xfs_sb {
 106	__uint32_t	sb_magicnum;	/* magic number == XFS_SB_MAGIC */
 107	__uint32_t	sb_blocksize;	/* logical block size, bytes */
 108	xfs_rfsblock_t	sb_dblocks;	/* number of data blocks */
 109	xfs_rfsblock_t	sb_rblocks;	/* number of realtime blocks */
 110	xfs_rtblock_t	sb_rextents;	/* number of realtime extents */
 111	uuid_t		sb_uuid;	/* user-visible file system unique id */
 112	xfs_fsblock_t	sb_logstart;	/* starting block of log if internal */
 113	xfs_ino_t	sb_rootino;	/* root inode number */
 114	xfs_ino_t	sb_rbmino;	/* bitmap inode for realtime extents */
 115	xfs_ino_t	sb_rsumino;	/* summary inode for rt bitmap */
 116	xfs_agblock_t	sb_rextsize;	/* realtime extent size, blocks */
 117	xfs_agblock_t	sb_agblocks;	/* size of an allocation group */
 118	xfs_agnumber_t	sb_agcount;	/* number of allocation groups */
 119	xfs_extlen_t	sb_rbmblocks;	/* number of rt bitmap blocks */
 120	xfs_extlen_t	sb_logblocks;	/* number of log blocks */
 121	__uint16_t	sb_versionnum;	/* header version == XFS_SB_VERSION */
 122	__uint16_t	sb_sectsize;	/* volume sector size, bytes */
 123	__uint16_t	sb_inodesize;	/* inode size, bytes */
 124	__uint16_t	sb_inopblock;	/* inodes per block */
 125	char		sb_fname[12];	/* file system name */
 126	__uint8_t	sb_blocklog;	/* log2 of sb_blocksize */
 127	__uint8_t	sb_sectlog;	/* log2 of sb_sectsize */
 128	__uint8_t	sb_inodelog;	/* log2 of sb_inodesize */
 129	__uint8_t	sb_inopblog;	/* log2 of sb_inopblock */
 130	__uint8_t	sb_agblklog;	/* log2 of sb_agblocks (rounded up) */
 131	__uint8_t	sb_rextslog;	/* log2 of sb_rextents */
 132	__uint8_t	sb_inprogress;	/* mkfs is in progress, don't mount */
 133	__uint8_t	sb_imax_pct;	/* max % of fs for inode space */
 134					/* statistics */
 135	/*
 136	 * These fields must remain contiguous.  If you really
 137	 * want to change their layout, make sure you fix the
 138	 * code in xfs_trans_apply_sb_deltas().
 139	 */
 140	__uint64_t	sb_icount;	/* allocated inodes */
 141	__uint64_t	sb_ifree;	/* free inodes */
 142	__uint64_t	sb_fdblocks;	/* free data blocks */
 143	__uint64_t	sb_frextents;	/* free realtime extents */
 144	/*
 145	 * End contiguous fields.
 146	 */
 147	xfs_ino_t	sb_uquotino;	/* user quota inode */
 148	xfs_ino_t	sb_gquotino;	/* group quota inode */
 149	__uint16_t	sb_qflags;	/* quota flags */
 150	__uint8_t	sb_flags;	/* misc. flags */
 151	__uint8_t	sb_shared_vn;	/* shared version number */
 152	xfs_extlen_t	sb_inoalignmt;	/* inode chunk alignment, fsblocks */
 153	__uint32_t	sb_unit;	/* stripe or raid unit */
 154	__uint32_t	sb_width;	/* stripe or raid width */
 155	__uint8_t	sb_dirblklog;	/* log2 of dir block size (fsbs) */
 156	__uint8_t	sb_logsectlog;	/* log2 of the log sector size */
 157	__uint16_t	sb_logsectsize;	/* sector size for the log, bytes */
 158	__uint32_t	sb_logsunit;	/* stripe unit size for the log */
 159	__uint32_t	sb_features2;	/* additional feature bits */
 160
 161	/*
 162	 * bad features2 field as a result of failing to pad the sb structure to
 163	 * 64 bits. Some machines will be using this field for features2 bits.
 164	 * Easiest just to mark it bad and not use it for anything else.
 165	 *
 166	 * This is not kept up to date in memory; it is always overwritten by
 167	 * the value in sb_features2 when formatting the incore superblock to
 168	 * the disk buffer.
 169	 */
 170	__uint32_t	sb_bad_features2;
 171
 172	/* version 5 superblock fields start here */
 173
 174	/* feature masks */
 175	__uint32_t	sb_features_compat;
 176	__uint32_t	sb_features_ro_compat;
 177	__uint32_t	sb_features_incompat;
 178	__uint32_t	sb_features_log_incompat;
 179
 180	__uint32_t	sb_crc;		/* superblock crc */
 181	xfs_extlen_t	sb_spino_align;	/* sparse inode chunk alignment */
 182
 183	xfs_ino_t	sb_pquotino;	/* project quota inode */
 184	xfs_lsn_t	sb_lsn;		/* last write sequence */
 185	uuid_t		sb_meta_uuid;	/* metadata file system unique id */
 186
 187	/* must be padded to 64 bit alignment */
 188} xfs_sb_t;
 189
 190#define XFS_SB_CRC_OFF		offsetof(struct xfs_sb, sb_crc)
 191
 192/*
 193 * Superblock - on disk version.  Must match the in core version above.
 194 * Must be padded to 64 bit alignment.
 195 */
 196typedef struct xfs_dsb {
 197	__be32		sb_magicnum;	/* magic number == XFS_SB_MAGIC */
 198	__be32		sb_blocksize;	/* logical block size, bytes */
 199	__be64		sb_dblocks;	/* number of data blocks */
 200	__be64		sb_rblocks;	/* number of realtime blocks */
 201	__be64		sb_rextents;	/* number of realtime extents */
 202	uuid_t		sb_uuid;	/* user-visible file system unique id */
 203	__be64		sb_logstart;	/* starting block of log if internal */
 204	__be64		sb_rootino;	/* root inode number */
 205	__be64		sb_rbmino;	/* bitmap inode for realtime extents */
 206	__be64		sb_rsumino;	/* summary inode for rt bitmap */
 207	__be32		sb_rextsize;	/* realtime extent size, blocks */
 208	__be32		sb_agblocks;	/* size of an allocation group */
 209	__be32		sb_agcount;	/* number of allocation groups */
 210	__be32		sb_rbmblocks;	/* number of rt bitmap blocks */
 211	__be32		sb_logblocks;	/* number of log blocks */
 212	__be16		sb_versionnum;	/* header version == XFS_SB_VERSION */
 213	__be16		sb_sectsize;	/* volume sector size, bytes */
 214	__be16		sb_inodesize;	/* inode size, bytes */
 215	__be16		sb_inopblock;	/* inodes per block */
 216	char		sb_fname[12];	/* file system name */
 217	__u8		sb_blocklog;	/* log2 of sb_blocksize */
 218	__u8		sb_sectlog;	/* log2 of sb_sectsize */
 219	__u8		sb_inodelog;	/* log2 of sb_inodesize */
 220	__u8		sb_inopblog;	/* log2 of sb_inopblock */
 221	__u8		sb_agblklog;	/* log2 of sb_agblocks (rounded up) */
 222	__u8		sb_rextslog;	/* log2 of sb_rextents */
 223	__u8		sb_inprogress;	/* mkfs is in progress, don't mount */
 224	__u8		sb_imax_pct;	/* max % of fs for inode space */
 225					/* statistics */
 226	/*
 227	 * These fields must remain contiguous.  If you really
 228	 * want to change their layout, make sure you fix the
 229	 * code in xfs_trans_apply_sb_deltas().
 230	 */
 231	__be64		sb_icount;	/* allocated inodes */
 232	__be64		sb_ifree;	/* free inodes */
 233	__be64		sb_fdblocks;	/* free data blocks */
 234	__be64		sb_frextents;	/* free realtime extents */
 235	/*
 236	 * End contiguous fields.
 237	 */
 238	__be64		sb_uquotino;	/* user quota inode */
 239	__be64		sb_gquotino;	/* group quota inode */
 240	__be16		sb_qflags;	/* quota flags */
 241	__u8		sb_flags;	/* misc. flags */
 242	__u8		sb_shared_vn;	/* shared version number */
 243	__be32		sb_inoalignmt;	/* inode chunk alignment, fsblocks */
 244	__be32		sb_unit;	/* stripe or raid unit */
 245	__be32		sb_width;	/* stripe or raid width */
 246	__u8		sb_dirblklog;	/* log2 of dir block size (fsbs) */
 247	__u8		sb_logsectlog;	/* log2 of the log sector size */
 248	__be16		sb_logsectsize;	/* sector size for the log, bytes */
 249	__be32		sb_logsunit;	/* stripe unit size for the log */
 250	__be32		sb_features2;	/* additional feature bits */
 251	/*
 252	 * bad features2 field as a result of failing to pad the sb
 253	 * structure to 64 bits. Some machines will be using this field
 254	 * for features2 bits. Easiest just to mark it bad and not use
 255	 * it for anything else.
 256	 */
 257	__be32		sb_bad_features2;
 258
 259	/* version 5 superblock fields start here */
 260
 261	/* feature masks */
 262	__be32		sb_features_compat;
 263	__be32		sb_features_ro_compat;
 264	__be32		sb_features_incompat;
 265	__be32		sb_features_log_incompat;
 266
 267	__le32		sb_crc;		/* superblock crc */
 268	__be32		sb_spino_align;	/* sparse inode chunk alignment */
 269
 270	__be64		sb_pquotino;	/* project quota inode */
 271	__be64		sb_lsn;		/* last write sequence */
 272	uuid_t		sb_meta_uuid;	/* metadata file system unique id */
 273
 274	/* must be padded to 64 bit alignment */
 275} xfs_dsb_t;
 276
 277
 278/*
 279 * Misc. Flags - warning - these will be cleared by xfs_repair unless
 280 * a feature bit is set when the flag is used.
 281 */
 282#define XFS_SBF_NOFLAGS		0x00	/* no flags set */
 283#define XFS_SBF_READONLY	0x01	/* only read-only mounts allowed */
 284
 285/*
 286 * define max. shared version we can interoperate with
 287 */
 288#define XFS_SB_MAX_SHARED_VN	0
 289
 290#define	XFS_SB_VERSION_NUM(sbp)	((sbp)->sb_versionnum & XFS_SB_VERSION_NUMBITS)
 291
 292/*
 293 * The first XFS version we support is a v4 superblock with V2 directories.
 294 */
 295static inline bool xfs_sb_good_v4_features(struct xfs_sb *sbp)
 296{
 297	if (!(sbp->sb_versionnum & XFS_SB_VERSION_DIRV2BIT))
 298		return false;
 
 
 299
 300	/* check for unknown features in the fs */
 301	if ((sbp->sb_versionnum & ~XFS_SB_VERSION_OKBITS) ||
 302	    ((sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT) &&
 303	     (sbp->sb_features2 & ~XFS_SB_VERSION2_OKBITS)))
 304		return false;
 305
 306	return true;
 307}
 308
 309static inline bool xfs_sb_good_version(struct xfs_sb *sbp)
 310{
 311	if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5)
 312		return true;
 313	if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4)
 314		return xfs_sb_good_v4_features(sbp);
 315	return false;
 316}
 317
 
 
 
 
 
 318/*
 319 * Detect a mismatched features2 field.  Older kernels read/wrote
 320 * this into the wrong slot, so to be safe we keep them in sync.
 321 */
 322static inline bool xfs_sb_has_mismatched_features2(struct xfs_sb *sbp)
 323{
 324	return sbp->sb_bad_features2 != sbp->sb_features2;
 325}
 326
 327static inline bool xfs_sb_version_hasattr(struct xfs_sb *sbp)
 328{
 329	return (sbp->sb_versionnum & XFS_SB_VERSION_ATTRBIT);
 330}
 331
 332static inline void xfs_sb_version_addattr(struct xfs_sb *sbp)
 333{
 334	sbp->sb_versionnum |= XFS_SB_VERSION_ATTRBIT;
 335}
 336
 337static inline bool xfs_sb_version_hasquota(struct xfs_sb *sbp)
 338{
 339	return (sbp->sb_versionnum & XFS_SB_VERSION_QUOTABIT);
 340}
 341
 342static inline void xfs_sb_version_addquota(struct xfs_sb *sbp)
 343{
 344	sbp->sb_versionnum |= XFS_SB_VERSION_QUOTABIT;
 345}
 346
 347static inline bool xfs_sb_version_hasalign(struct xfs_sb *sbp)
 348{
 349	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 ||
 350		(sbp->sb_versionnum & XFS_SB_VERSION_ALIGNBIT));
 351}
 352
 353static inline bool xfs_sb_version_hasdalign(struct xfs_sb *sbp)
 354{
 355	return (sbp->sb_versionnum & XFS_SB_VERSION_DALIGNBIT);
 356}
 357
 358static inline bool xfs_sb_version_haslogv2(struct xfs_sb *sbp)
 359{
 360	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 ||
 361	       (sbp->sb_versionnum & XFS_SB_VERSION_LOGV2BIT);
 362}
 363
 364static inline bool xfs_sb_version_hasextflgbit(struct xfs_sb *sbp)
 365{
 366	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 ||
 367	       (sbp->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT);
 368}
 369
 370static inline bool xfs_sb_version_hassector(struct xfs_sb *sbp)
 371{
 372	return (sbp->sb_versionnum & XFS_SB_VERSION_SECTORBIT);
 373}
 374
 375static inline bool xfs_sb_version_hasasciici(struct xfs_sb *sbp)
 376{
 377	return (sbp->sb_versionnum & XFS_SB_VERSION_BORGBIT);
 378}
 379
 380static inline bool xfs_sb_version_hasmorebits(struct xfs_sb *sbp)
 381{
 382	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 ||
 383	       (sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT);
 384}
 385
 386/*
 387 * sb_features2 bit version macros.
 388 */
 389static inline bool xfs_sb_version_haslazysbcount(struct xfs_sb *sbp)
 390{
 391	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
 392	       (xfs_sb_version_hasmorebits(sbp) &&
 393		(sbp->sb_features2 & XFS_SB_VERSION2_LAZYSBCOUNTBIT));
 394}
 395
 396static inline bool xfs_sb_version_hasattr2(struct xfs_sb *sbp)
 397{
 398	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
 399	       (xfs_sb_version_hasmorebits(sbp) &&
 400		(sbp->sb_features2 & XFS_SB_VERSION2_ATTR2BIT));
 401}
 402
 403static inline void xfs_sb_version_addattr2(struct xfs_sb *sbp)
 404{
 405	sbp->sb_versionnum |= XFS_SB_VERSION_MOREBITSBIT;
 406	sbp->sb_features2 |= XFS_SB_VERSION2_ATTR2BIT;
 407}
 408
 409static inline void xfs_sb_version_removeattr2(struct xfs_sb *sbp)
 410{
 411	sbp->sb_features2 &= ~XFS_SB_VERSION2_ATTR2BIT;
 412	if (!sbp->sb_features2)
 413		sbp->sb_versionnum &= ~XFS_SB_VERSION_MOREBITSBIT;
 414}
 415
 416static inline bool xfs_sb_version_hasprojid32bit(struct xfs_sb *sbp)
 417{
 418	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
 419	       (xfs_sb_version_hasmorebits(sbp) &&
 420		(sbp->sb_features2 & XFS_SB_VERSION2_PROJID32BIT));
 421}
 422
 423static inline void xfs_sb_version_addprojid32bit(struct xfs_sb *sbp)
 424{
 425	sbp->sb_versionnum |= XFS_SB_VERSION_MOREBITSBIT;
 426	sbp->sb_features2 |= XFS_SB_VERSION2_PROJID32BIT;
 427}
 428
 429/*
 430 * Extended v5 superblock feature masks. These are to be used for new v5
 431 * superblock features only.
 432 *
 433 * Compat features are new features that old kernels will not notice or affect
 434 * and so can mount read-write without issues.
 435 *
 436 * RO-Compat (read only) are features that old kernels can read but will break
 437 * if they write. Hence only read-only mounts of such filesystems are allowed on
 438 * kernels that don't support the feature bit.
 439 *
 440 * InCompat features are features which old kernels will not understand and so
 441 * must not mount.
 442 *
 443 * Log-InCompat features are for changes to log formats or new transactions that
 444 * can't be replayed on older kernels. The fields are set when the filesystem is
 445 * mounted, and a clean unmount clears the fields.
 446 */
 447#define XFS_SB_FEAT_COMPAT_ALL 0
 448#define XFS_SB_FEAT_COMPAT_UNKNOWN	~XFS_SB_FEAT_COMPAT_ALL
 449static inline bool
 450xfs_sb_has_compat_feature(
 451	struct xfs_sb	*sbp,
 452	__uint32_t	feature)
 453{
 454	return (sbp->sb_features_compat & feature) != 0;
 455}
 456
 457#define XFS_SB_FEAT_RO_COMPAT_FINOBT   (1 << 0)		/* free inode btree */
 458#define XFS_SB_FEAT_RO_COMPAT_RMAPBT   (1 << 1)		/* reverse map btree */
 459#define XFS_SB_FEAT_RO_COMPAT_REFLINK  (1 << 2)		/* reflinked files */
 460#define XFS_SB_FEAT_RO_COMPAT_ALL \
 461		(XFS_SB_FEAT_RO_COMPAT_FINOBT | \
 462		 XFS_SB_FEAT_RO_COMPAT_RMAPBT | \
 463		 XFS_SB_FEAT_RO_COMPAT_REFLINK)
 464#define XFS_SB_FEAT_RO_COMPAT_UNKNOWN	~XFS_SB_FEAT_RO_COMPAT_ALL
 465static inline bool
 466xfs_sb_has_ro_compat_feature(
 467	struct xfs_sb	*sbp,
 468	__uint32_t	feature)
 469{
 470	return (sbp->sb_features_ro_compat & feature) != 0;
 471}
 472
 473#define XFS_SB_FEAT_INCOMPAT_FTYPE	(1 << 0)	/* filetype in dirent */
 474#define XFS_SB_FEAT_INCOMPAT_SPINODES	(1 << 1)	/* sparse inode chunks */
 475#define XFS_SB_FEAT_INCOMPAT_META_UUID	(1 << 2)	/* metadata UUID */
 476#define XFS_SB_FEAT_INCOMPAT_ALL \
 477		(XFS_SB_FEAT_INCOMPAT_FTYPE|	\
 478		 XFS_SB_FEAT_INCOMPAT_SPINODES|	\
 479		 XFS_SB_FEAT_INCOMPAT_META_UUID)
 480
 481#define XFS_SB_FEAT_INCOMPAT_UNKNOWN	~XFS_SB_FEAT_INCOMPAT_ALL
 482static inline bool
 483xfs_sb_has_incompat_feature(
 484	struct xfs_sb	*sbp,
 485	__uint32_t	feature)
 486{
 487	return (sbp->sb_features_incompat & feature) != 0;
 488}
 489
 490#define XFS_SB_FEAT_INCOMPAT_LOG_ALL 0
 491#define XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN	~XFS_SB_FEAT_INCOMPAT_LOG_ALL
 492static inline bool
 493xfs_sb_has_incompat_log_feature(
 494	struct xfs_sb	*sbp,
 495	__uint32_t	feature)
 496{
 497	return (sbp->sb_features_log_incompat & feature) != 0;
 498}
 499
 500/*
 501 * V5 superblock specific feature checks
 502 */
 503static inline int xfs_sb_version_hascrc(struct xfs_sb *sbp)
 504{
 505	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5;
 506}
 507
 508static inline int xfs_sb_version_has_pquotino(struct xfs_sb *sbp)
 509{
 510	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5;
 511}
 512
 513static inline int xfs_sb_version_hasftype(struct xfs_sb *sbp)
 514{
 515	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 &&
 516		xfs_sb_has_incompat_feature(sbp, XFS_SB_FEAT_INCOMPAT_FTYPE)) ||
 517	       (xfs_sb_version_hasmorebits(sbp) &&
 518		 (sbp->sb_features2 & XFS_SB_VERSION2_FTYPE));
 519}
 520
 521static inline int xfs_sb_version_hasfinobt(xfs_sb_t *sbp)
 522{
 523	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) &&
 524		(sbp->sb_features_ro_compat & XFS_SB_FEAT_RO_COMPAT_FINOBT);
 525}
 526
 527static inline bool xfs_sb_version_hassparseinodes(struct xfs_sb *sbp)
 528{
 529	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 &&
 530		xfs_sb_has_incompat_feature(sbp, XFS_SB_FEAT_INCOMPAT_SPINODES);
 531}
 532
 533/*
 534 * XFS_SB_FEAT_INCOMPAT_META_UUID indicates that the metadata UUID
 535 * is stored separately from the user-visible UUID; this allows the
 536 * user-visible UUID to be changed on V5 filesystems which have a
 537 * filesystem UUID stamped into every piece of metadata.
 538 */
 539static inline bool xfs_sb_version_hasmetauuid(struct xfs_sb *sbp)
 540{
 541	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) &&
 542		(sbp->sb_features_incompat & XFS_SB_FEAT_INCOMPAT_META_UUID);
 543}
 544
 545static inline bool xfs_sb_version_hasrmapbt(struct xfs_sb *sbp)
 546{
 547	return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) &&
 548		(sbp->sb_features_ro_compat & XFS_SB_FEAT_RO_COMPAT_RMAPBT);
 549}
 550
 551static inline bool xfs_sb_version_hasreflink(struct xfs_sb *sbp)
 552{
 553	return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5 &&
 554		(sbp->sb_features_ro_compat & XFS_SB_FEAT_RO_COMPAT_REFLINK);
 555}
 556
 557/*
 558 * end of superblock version macros
 559 */
 560
 561static inline bool
 562xfs_is_quota_inode(struct xfs_sb *sbp, xfs_ino_t ino)
 563{
 564	return (ino == sbp->sb_uquotino ||
 565		ino == sbp->sb_gquotino ||
 566		ino == sbp->sb_pquotino);
 567}
 568
 569#define XFS_SB_DADDR		((xfs_daddr_t)0) /* daddr in filesystem/ag */
 570#define	XFS_SB_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_SB_DADDR)
 571#define XFS_BUF_TO_SBP(bp)	((xfs_dsb_t *)((bp)->b_addr))
 572
 573#define	XFS_HDR_BLOCK(mp,d)	((xfs_agblock_t)XFS_BB_TO_FSBT(mp,d))
 574#define	XFS_DADDR_TO_FSB(mp,d)	XFS_AGB_TO_FSB(mp, \
 575			xfs_daddr_to_agno(mp,d), xfs_daddr_to_agbno(mp,d))
 576#define	XFS_FSB_TO_DADDR(mp,fsbno)	XFS_AGB_TO_DADDR(mp, \
 577			XFS_FSB_TO_AGNO(mp,fsbno), XFS_FSB_TO_AGBNO(mp,fsbno))
 578
 579/*
 580 * File system sector to basic block conversions.
 581 */
 582#define XFS_FSS_TO_BB(mp,sec)	((sec) << (mp)->m_sectbb_log)
 583
 584/*
 585 * File system block to basic block conversions.
 586 */
 587#define	XFS_FSB_TO_BB(mp,fsbno)	((fsbno) << (mp)->m_blkbb_log)
 588#define	XFS_BB_TO_FSB(mp,bb)	\
 589	(((bb) + (XFS_FSB_TO_BB(mp,1) - 1)) >> (mp)->m_blkbb_log)
 590#define	XFS_BB_TO_FSBT(mp,bb)	((bb) >> (mp)->m_blkbb_log)
 591
 592/*
 593 * File system block to byte conversions.
 594 */
 595#define XFS_FSB_TO_B(mp,fsbno)	((xfs_fsize_t)(fsbno) << (mp)->m_sb.sb_blocklog)
 596#define XFS_B_TO_FSB(mp,b)	\
 597	((((__uint64_t)(b)) + (mp)->m_blockmask) >> (mp)->m_sb.sb_blocklog)
 598#define XFS_B_TO_FSBT(mp,b)	(((__uint64_t)(b)) >> (mp)->m_sb.sb_blocklog)
 599#define XFS_B_FSB_OFFSET(mp,b)	((b) & (mp)->m_blockmask)
 600
 601/*
 602 * Allocation group header
 603 *
 604 * This is divided into three structures, placed in sequential 512-byte
 605 * buffers after a copy of the superblock (also in a 512-byte buffer).
 606 */
 607#define	XFS_AGF_MAGIC	0x58414746	/* 'XAGF' */
 608#define	XFS_AGI_MAGIC	0x58414749	/* 'XAGI' */
 609#define	XFS_AGFL_MAGIC	0x5841464c	/* 'XAFL' */
 610#define	XFS_AGF_VERSION	1
 611#define	XFS_AGI_VERSION	1
 612
 613#define	XFS_AGF_GOOD_VERSION(v)	((v) == XFS_AGF_VERSION)
 614#define	XFS_AGI_GOOD_VERSION(v)	((v) == XFS_AGI_VERSION)
 615
 616/*
 617 * Btree number 0 is bno, 1 is cnt, 2 is rmap. This value gives the size of the
 618 * arrays below.
 619 */
 620#define	XFS_BTNUM_AGF	((int)XFS_BTNUM_RMAPi + 1)
 621
 622/*
 623 * The second word of agf_levels in the first a.g. overlaps the EFS
 624 * superblock's magic number.  Since the magic numbers valid for EFS
 625 * are > 64k, our value cannot be confused for an EFS superblock's.
 626 */
 627
 628typedef struct xfs_agf {
 629	/*
 630	 * Common allocation group header information
 631	 */
 632	__be32		agf_magicnum;	/* magic number == XFS_AGF_MAGIC */
 633	__be32		agf_versionnum;	/* header version == XFS_AGF_VERSION */
 634	__be32		agf_seqno;	/* sequence # starting from 0 */
 635	__be32		agf_length;	/* size in blocks of a.g. */
 636	/*
 637	 * Freespace and rmap information
 638	 */
 639	__be32		agf_roots[XFS_BTNUM_AGF];	/* root blocks */
 640	__be32		agf_levels[XFS_BTNUM_AGF];	/* btree levels */
 641
 642	__be32		agf_flfirst;	/* first freelist block's index */
 643	__be32		agf_fllast;	/* last freelist block's index */
 644	__be32		agf_flcount;	/* count of blocks in freelist */
 645	__be32		agf_freeblks;	/* total free blocks */
 646
 647	__be32		agf_longest;	/* longest free space */
 648	__be32		agf_btreeblks;	/* # of blocks held in AGF btrees */
 649	uuid_t		agf_uuid;	/* uuid of filesystem */
 650
 651	__be32		agf_rmap_blocks;	/* rmapbt blocks used */
 652	__be32		agf_refcount_blocks;	/* refcountbt blocks used */
 653
 654	__be32		agf_refcount_root;	/* refcount tree root block */
 655	__be32		agf_refcount_level;	/* refcount btree levels */
 656
 657	/*
 658	 * reserve some contiguous space for future logged fields before we add
 659	 * the unlogged fields. This makes the range logging via flags and
 660	 * structure offsets much simpler.
 661	 */
 662	__be64		agf_spare64[14];
 663
 664	/* unlogged fields, written during buffer writeback. */
 665	__be64		agf_lsn;	/* last write sequence */
 666	__be32		agf_crc;	/* crc of agf sector */
 667	__be32		agf_spare2;
 668
 669	/* structure must be padded to 64 bit alignment */
 670} xfs_agf_t;
 671
 672#define XFS_AGF_CRC_OFF		offsetof(struct xfs_agf, agf_crc)
 673
 674#define	XFS_AGF_MAGICNUM	0x00000001
 675#define	XFS_AGF_VERSIONNUM	0x00000002
 676#define	XFS_AGF_SEQNO		0x00000004
 677#define	XFS_AGF_LENGTH		0x00000008
 678#define	XFS_AGF_ROOTS		0x00000010
 679#define	XFS_AGF_LEVELS		0x00000020
 680#define	XFS_AGF_FLFIRST		0x00000040
 681#define	XFS_AGF_FLLAST		0x00000080
 682#define	XFS_AGF_FLCOUNT		0x00000100
 683#define	XFS_AGF_FREEBLKS	0x00000200
 684#define	XFS_AGF_LONGEST		0x00000400
 685#define	XFS_AGF_BTREEBLKS	0x00000800
 686#define	XFS_AGF_UUID		0x00001000
 687#define	XFS_AGF_RMAP_BLOCKS	0x00002000
 688#define	XFS_AGF_REFCOUNT_BLOCKS	0x00004000
 689#define	XFS_AGF_REFCOUNT_ROOT	0x00008000
 690#define	XFS_AGF_REFCOUNT_LEVEL	0x00010000
 691#define	XFS_AGF_SPARE64		0x00020000
 692#define	XFS_AGF_NUM_BITS	18
 693#define	XFS_AGF_ALL_BITS	((1 << XFS_AGF_NUM_BITS) - 1)
 694
 695#define XFS_AGF_FLAGS \
 696	{ XFS_AGF_MAGICNUM,	"MAGICNUM" }, \
 697	{ XFS_AGF_VERSIONNUM,	"VERSIONNUM" }, \
 698	{ XFS_AGF_SEQNO,	"SEQNO" }, \
 699	{ XFS_AGF_LENGTH,	"LENGTH" }, \
 700	{ XFS_AGF_ROOTS,	"ROOTS" }, \
 701	{ XFS_AGF_LEVELS,	"LEVELS" }, \
 702	{ XFS_AGF_FLFIRST,	"FLFIRST" }, \
 703	{ XFS_AGF_FLLAST,	"FLLAST" }, \
 704	{ XFS_AGF_FLCOUNT,	"FLCOUNT" }, \
 705	{ XFS_AGF_FREEBLKS,	"FREEBLKS" }, \
 706	{ XFS_AGF_LONGEST,	"LONGEST" }, \
 707	{ XFS_AGF_BTREEBLKS,	"BTREEBLKS" }, \
 708	{ XFS_AGF_UUID,		"UUID" }, \
 709	{ XFS_AGF_RMAP_BLOCKS,	"RMAP_BLOCKS" }, \
 710	{ XFS_AGF_REFCOUNT_BLOCKS,	"REFCOUNT_BLOCKS" }, \
 711	{ XFS_AGF_REFCOUNT_ROOT,	"REFCOUNT_ROOT" }, \
 712	{ XFS_AGF_REFCOUNT_LEVEL,	"REFCOUNT_LEVEL" }, \
 713	{ XFS_AGF_SPARE64,	"SPARE64" }
 714
 715/* disk block (xfs_daddr_t) in the AG */
 716#define XFS_AGF_DADDR(mp)	((xfs_daddr_t)(1 << (mp)->m_sectbb_log))
 717#define	XFS_AGF_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGF_DADDR(mp))
 718#define	XFS_BUF_TO_AGF(bp)	((xfs_agf_t *)((bp)->b_addr))
 719
 720/*
 721 * Size of the unlinked inode hash table in the agi.
 722 */
 723#define	XFS_AGI_UNLINKED_BUCKETS	64
 724
 725typedef struct xfs_agi {
 726	/*
 727	 * Common allocation group header information
 728	 */
 729	__be32		agi_magicnum;	/* magic number == XFS_AGI_MAGIC */
 730	__be32		agi_versionnum;	/* header version == XFS_AGI_VERSION */
 731	__be32		agi_seqno;	/* sequence # starting from 0 */
 732	__be32		agi_length;	/* size in blocks of a.g. */
 733	/*
 734	 * Inode information
 735	 * Inodes are mapped by interpreting the inode number, so no
 736	 * mapping data is needed here.
 737	 */
 738	__be32		agi_count;	/* count of allocated inodes */
 739	__be32		agi_root;	/* root of inode btree */
 740	__be32		agi_level;	/* levels in inode btree */
 741	__be32		agi_freecount;	/* number of free inodes */
 742
 743	__be32		agi_newino;	/* new inode just allocated */
 744	__be32		agi_dirino;	/* last directory inode chunk */
 745	/*
 746	 * Hash table of inodes which have been unlinked but are
 747	 * still being referenced.
 748	 */
 749	__be32		agi_unlinked[XFS_AGI_UNLINKED_BUCKETS];
 750	/*
 751	 * This marks the end of logging region 1 and start of logging region 2.
 752	 */
 753	uuid_t		agi_uuid;	/* uuid of filesystem */
 754	__be32		agi_crc;	/* crc of agi sector */
 755	__be32		agi_pad32;
 756	__be64		agi_lsn;	/* last write sequence */
 757
 758	__be32		agi_free_root; /* root of the free inode btree */
 759	__be32		agi_free_level;/* levels in free inode btree */
 760
 761	/* structure must be padded to 64 bit alignment */
 762} xfs_agi_t;
 763
 764#define XFS_AGI_CRC_OFF		offsetof(struct xfs_agi, agi_crc)
 765
 766#define	XFS_AGI_MAGICNUM	(1 << 0)
 767#define	XFS_AGI_VERSIONNUM	(1 << 1)
 768#define	XFS_AGI_SEQNO		(1 << 2)
 769#define	XFS_AGI_LENGTH		(1 << 3)
 770#define	XFS_AGI_COUNT		(1 << 4)
 771#define	XFS_AGI_ROOT		(1 << 5)
 772#define	XFS_AGI_LEVEL		(1 << 6)
 773#define	XFS_AGI_FREECOUNT	(1 << 7)
 774#define	XFS_AGI_NEWINO		(1 << 8)
 775#define	XFS_AGI_DIRINO		(1 << 9)
 776#define	XFS_AGI_UNLINKED	(1 << 10)
 777#define	XFS_AGI_NUM_BITS_R1	11	/* end of the 1st agi logging region */
 778#define	XFS_AGI_ALL_BITS_R1	((1 << XFS_AGI_NUM_BITS_R1) - 1)
 779#define	XFS_AGI_FREE_ROOT	(1 << 11)
 780#define	XFS_AGI_FREE_LEVEL	(1 << 12)
 781#define	XFS_AGI_NUM_BITS_R2	13
 782
 783/* disk block (xfs_daddr_t) in the AG */
 784#define XFS_AGI_DADDR(mp)	((xfs_daddr_t)(2 << (mp)->m_sectbb_log))
 785#define	XFS_AGI_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGI_DADDR(mp))
 786#define	XFS_BUF_TO_AGI(bp)	((xfs_agi_t *)((bp)->b_addr))
 787
 788/*
 789 * The third a.g. block contains the a.g. freelist, an array
 790 * of block pointers to blocks owned by the allocation btree code.
 791 */
 792#define XFS_AGFL_DADDR(mp)	((xfs_daddr_t)(3 << (mp)->m_sectbb_log))
 793#define	XFS_AGFL_BLOCK(mp)	XFS_HDR_BLOCK(mp, XFS_AGFL_DADDR(mp))
 794#define	XFS_BUF_TO_AGFL(bp)	((xfs_agfl_t *)((bp)->b_addr))
 795
 796#define XFS_BUF_TO_AGFL_BNO(mp, bp) \
 797	(xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
 798		&(XFS_BUF_TO_AGFL(bp)->agfl_bno[0]) : \
 799		(__be32 *)(bp)->b_addr)
 800
 801/*
 802 * Size of the AGFL.  For CRC-enabled filesystes we steal a couple of
 803 * slots in the beginning of the block for a proper header with the
 804 * location information and CRC.
 805 */
 806#define XFS_AGFL_SIZE(mp) \
 807	(((mp)->m_sb.sb_sectsize - \
 808	 (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
 809		sizeof(struct xfs_agfl) : 0)) / \
 810	  sizeof(xfs_agblock_t))
 811
 812typedef struct xfs_agfl {
 813	__be32		agfl_magicnum;
 814	__be32		agfl_seqno;
 815	uuid_t		agfl_uuid;
 816	__be64		agfl_lsn;
 817	__be32		agfl_crc;
 818	__be32		agfl_bno[];	/* actually XFS_AGFL_SIZE(mp) */
 819} __attribute__((packed)) xfs_agfl_t;
 820
 821#define XFS_AGFL_CRC_OFF	offsetof(struct xfs_agfl, agfl_crc)
 822
 823#define XFS_AGB_TO_FSB(mp,agno,agbno)	\
 824	(((xfs_fsblock_t)(agno) << (mp)->m_sb.sb_agblklog) | (agbno))
 825#define	XFS_FSB_TO_AGNO(mp,fsbno)	\
 826	((xfs_agnumber_t)((fsbno) >> (mp)->m_sb.sb_agblklog))
 827#define	XFS_FSB_TO_AGBNO(mp,fsbno)	\
 828	((xfs_agblock_t)((fsbno) & xfs_mask32lo((mp)->m_sb.sb_agblklog)))
 829#define	XFS_AGB_TO_DADDR(mp,agno,agbno)	\
 830	((xfs_daddr_t)XFS_FSB_TO_BB(mp, \
 831		(xfs_fsblock_t)(agno) * (mp)->m_sb.sb_agblocks + (agbno)))
 832#define	XFS_AG_DADDR(mp,agno,d)		(XFS_AGB_TO_DADDR(mp, agno, 0) + (d))
 833
 834/*
 835 * For checking for bad ranges of xfs_daddr_t's, covering multiple
 836 * allocation groups or a single xfs_daddr_t that's a superblock copy.
 837 */
 838#define	XFS_AG_CHECK_DADDR(mp,d,len)	\
 839	((len) == 1 ? \
 840	    ASSERT((d) == XFS_SB_DADDR || \
 841		   xfs_daddr_to_agbno(mp, d) != XFS_SB_DADDR) : \
 842	    ASSERT(xfs_daddr_to_agno(mp, d) == \
 843		   xfs_daddr_to_agno(mp, (d) + (len) - 1)))
 844
 845typedef struct xfs_timestamp {
 846	__be32		t_sec;		/* timestamp seconds */
 847	__be32		t_nsec;		/* timestamp nanoseconds */
 848} xfs_timestamp_t;
 849
 850/*
 851 * On-disk inode structure.
 852 *
 853 * This is just the header or "dinode core", the inode is expanded to fill a
 854 * variable size the leftover area split into a data and an attribute fork.
 855 * The format of the data and attribute fork depends on the format of the
 856 * inode as indicated by di_format and di_aformat.  To access the data and
 857 * attribute use the XFS_DFORK_DPTR, XFS_DFORK_APTR, and XFS_DFORK_PTR macros
 858 * below.
 859 *
 860 * There is a very similar struct icdinode in xfs_inode which matches the
 861 * layout of the first 96 bytes of this structure, but is kept in native
 862 * format instead of big endian.
 863 *
 864 * Note: di_flushiter is only used by v1/2 inodes - it's effectively a zeroed
 865 * padding field for v3 inodes.
 866 */
 867#define	XFS_DINODE_MAGIC		0x494e	/* 'IN' */
 868typedef struct xfs_dinode {
 869	__be16		di_magic;	/* inode magic # = XFS_DINODE_MAGIC */
 870	__be16		di_mode;	/* mode and type of file */
 871	__u8		di_version;	/* inode version */
 872	__u8		di_format;	/* format of di_c data */
 873	__be16		di_onlink;	/* old number of links to file */
 874	__be32		di_uid;		/* owner's user id */
 875	__be32		di_gid;		/* owner's group id */
 876	__be32		di_nlink;	/* number of links to file */
 877	__be16		di_projid_lo;	/* lower part of owner's project id */
 878	__be16		di_projid_hi;	/* higher part owner's project id */
 879	__u8		di_pad[6];	/* unused, zeroed space */
 880	__be16		di_flushiter;	/* incremented on flush */
 881	xfs_timestamp_t	di_atime;	/* time last accessed */
 882	xfs_timestamp_t	di_mtime;	/* time last modified */
 883	xfs_timestamp_t	di_ctime;	/* time created/inode modified */
 884	__be64		di_size;	/* number of bytes in file */
 885	__be64		di_nblocks;	/* # of direct & btree blocks used */
 886	__be32		di_extsize;	/* basic/minimum extent size for file */
 887	__be32		di_nextents;	/* number of extents in data fork */
 888	__be16		di_anextents;	/* number of extents in attribute fork*/
 889	__u8		di_forkoff;	/* attr fork offs, <<3 for 64b align */
 890	__s8		di_aformat;	/* format of attr fork's data */
 891	__be32		di_dmevmask;	/* DMIG event mask */
 892	__be16		di_dmstate;	/* DMIG state info */
 893	__be16		di_flags;	/* random flags, XFS_DIFLAG_... */
 894	__be32		di_gen;		/* generation number */
 895
 896	/* di_next_unlinked is the only non-core field in the old dinode */
 897	__be32		di_next_unlinked;/* agi unlinked list ptr */
 898
 899	/* start of the extended dinode, writable fields */
 900	__le32		di_crc;		/* CRC of the inode */
 901	__be64		di_changecount;	/* number of attribute changes */
 902	__be64		di_lsn;		/* flush sequence */
 903	__be64		di_flags2;	/* more random flags */
 904	__be32		di_cowextsize;	/* basic cow extent size for file */
 905	__u8		di_pad2[12];	/* more padding for future expansion */
 906
 907	/* fields only written to during inode creation */
 908	xfs_timestamp_t	di_crtime;	/* time created */
 909	__be64		di_ino;		/* inode number */
 910	uuid_t		di_uuid;	/* UUID of the filesystem */
 911
 912	/* structure must be padded to 64 bit alignment */
 913} xfs_dinode_t;
 914
 915#define XFS_DINODE_CRC_OFF	offsetof(struct xfs_dinode, di_crc)
 916
 917#define DI_MAX_FLUSH 0xffff
 918
 919/*
 920 * Size of the core inode on disk.  Version 1 and 2 inodes have
 921 * the same size, but version 3 has grown a few additional fields.
 922 */
 923static inline uint xfs_dinode_size(int version)
 924{
 925	if (version == 3)
 926		return sizeof(struct xfs_dinode);
 927	return offsetof(struct xfs_dinode, di_crc);
 928}
 929
 930/*
 931 * The 32 bit link count in the inode theoretically maxes out at UINT_MAX.
 932 * Since the pathconf interface is signed, we use 2^31 - 1 instead.
 933 * The old inode format had a 16 bit link count, so its maximum is USHRT_MAX.
 934 */
 935#define	XFS_MAXLINK		((1U << 31) - 1U)
 936#define	XFS_MAXLINK_1		65535U
 937
 938/*
 939 * Values for di_format
 
 
 
 940 */
 941typedef enum xfs_dinode_fmt {
 942	XFS_DINODE_FMT_DEV,		/* xfs_dev_t */
 943	XFS_DINODE_FMT_LOCAL,		/* bulk data */
 944	XFS_DINODE_FMT_EXTENTS,		/* struct xfs_bmbt_rec */
 945	XFS_DINODE_FMT_BTREE,		/* struct xfs_bmdr_block */
 946	XFS_DINODE_FMT_UUID		/* uuid_t */
 947} xfs_dinode_fmt_t;
 948
 
 
 
 
 
 
 
 949/*
 950 * Inode minimum and maximum sizes.
 951 */
 952#define	XFS_DINODE_MIN_LOG	8
 953#define	XFS_DINODE_MAX_LOG	11
 954#define	XFS_DINODE_MIN_SIZE	(1 << XFS_DINODE_MIN_LOG)
 955#define	XFS_DINODE_MAX_SIZE	(1 << XFS_DINODE_MAX_LOG)
 956
 957/*
 958 * Inode size for given fs.
 959 */
 960#define XFS_LITINO(mp, version) \
 961	((int)(((mp)->m_sb.sb_inodesize) - xfs_dinode_size(version)))
 962
 963/*
 964 * Inode data & attribute fork sizes, per inode.
 965 */
 966#define XFS_DFORK_Q(dip)		((dip)->di_forkoff != 0)
 967#define XFS_DFORK_BOFF(dip)		((int)((dip)->di_forkoff << 3))
 968
 969#define XFS_DFORK_DSIZE(dip,mp) \
 970	(XFS_DFORK_Q(dip) ? \
 971		XFS_DFORK_BOFF(dip) : \
 972		XFS_LITINO(mp, (dip)->di_version))
 973#define XFS_DFORK_ASIZE(dip,mp) \
 974	(XFS_DFORK_Q(dip) ? \
 975		XFS_LITINO(mp, (dip)->di_version) - XFS_DFORK_BOFF(dip) : \
 976		0)
 977#define XFS_DFORK_SIZE(dip,mp,w) \
 978	((w) == XFS_DATA_FORK ? \
 979		XFS_DFORK_DSIZE(dip, mp) : \
 980		XFS_DFORK_ASIZE(dip, mp))
 981
 
 
 
 982/*
 983 * Return pointers to the data or attribute forks.
 984 */
 985#define XFS_DFORK_DPTR(dip) \
 986	((char *)dip + xfs_dinode_size(dip->di_version))
 987#define XFS_DFORK_APTR(dip)	\
 988	(XFS_DFORK_DPTR(dip) + XFS_DFORK_BOFF(dip))
 989#define XFS_DFORK_PTR(dip,w)	\
 990	((w) == XFS_DATA_FORK ? XFS_DFORK_DPTR(dip) : XFS_DFORK_APTR(dip))
 991
 992#define XFS_DFORK_FORMAT(dip,w) \
 993	((w) == XFS_DATA_FORK ? \
 994		(dip)->di_format : \
 995		(dip)->di_aformat)
 996#define XFS_DFORK_NEXTENTS(dip,w) \
 997	((w) == XFS_DATA_FORK ? \
 998		be32_to_cpu((dip)->di_nextents) : \
 999		be16_to_cpu((dip)->di_anextents))
1000
1001/*
1002 * For block and character special files the 32bit dev_t is stored at the
1003 * beginning of the data fork.
1004 */
1005static inline xfs_dev_t xfs_dinode_get_rdev(struct xfs_dinode *dip)
1006{
1007	return be32_to_cpu(*(__be32 *)XFS_DFORK_DPTR(dip));
1008}
1009
1010static inline void xfs_dinode_put_rdev(struct xfs_dinode *dip, xfs_dev_t rdev)
1011{
1012	*(__be32 *)XFS_DFORK_DPTR(dip) = cpu_to_be32(rdev);
1013}
1014
1015/*
1016 * Values for di_flags
1017 */
1018#define XFS_DIFLAG_REALTIME_BIT  0	/* file's blocks come from rt area */
1019#define XFS_DIFLAG_PREALLOC_BIT  1	/* file space has been preallocated */
1020#define XFS_DIFLAG_NEWRTBM_BIT   2	/* for rtbitmap inode, new format */
1021#define XFS_DIFLAG_IMMUTABLE_BIT 3	/* inode is immutable */
1022#define XFS_DIFLAG_APPEND_BIT    4	/* inode is append-only */
1023#define XFS_DIFLAG_SYNC_BIT      5	/* inode is written synchronously */
1024#define XFS_DIFLAG_NOATIME_BIT   6	/* do not update atime */
1025#define XFS_DIFLAG_NODUMP_BIT    7	/* do not dump */
1026#define XFS_DIFLAG_RTINHERIT_BIT 8	/* create with realtime bit set */
1027#define XFS_DIFLAG_PROJINHERIT_BIT   9	/* create with parents projid */
1028#define XFS_DIFLAG_NOSYMLINKS_BIT   10	/* disallow symlink creation */
1029#define XFS_DIFLAG_EXTSIZE_BIT      11	/* inode extent size allocator hint */
1030#define XFS_DIFLAG_EXTSZINHERIT_BIT 12	/* inherit inode extent size */
1031#define XFS_DIFLAG_NODEFRAG_BIT     13	/* do not reorganize/defragment */
1032#define XFS_DIFLAG_FILESTREAM_BIT   14  /* use filestream allocator */
 
 
1033#define XFS_DIFLAG_REALTIME      (1 << XFS_DIFLAG_REALTIME_BIT)
1034#define XFS_DIFLAG_PREALLOC      (1 << XFS_DIFLAG_PREALLOC_BIT)
1035#define XFS_DIFLAG_NEWRTBM       (1 << XFS_DIFLAG_NEWRTBM_BIT)
1036#define XFS_DIFLAG_IMMUTABLE     (1 << XFS_DIFLAG_IMMUTABLE_BIT)
1037#define XFS_DIFLAG_APPEND        (1 << XFS_DIFLAG_APPEND_BIT)
1038#define XFS_DIFLAG_SYNC          (1 << XFS_DIFLAG_SYNC_BIT)
1039#define XFS_DIFLAG_NOATIME       (1 << XFS_DIFLAG_NOATIME_BIT)
1040#define XFS_DIFLAG_NODUMP        (1 << XFS_DIFLAG_NODUMP_BIT)
1041#define XFS_DIFLAG_RTINHERIT     (1 << XFS_DIFLAG_RTINHERIT_BIT)
1042#define XFS_DIFLAG_PROJINHERIT   (1 << XFS_DIFLAG_PROJINHERIT_BIT)
1043#define XFS_DIFLAG_NOSYMLINKS    (1 << XFS_DIFLAG_NOSYMLINKS_BIT)
1044#define XFS_DIFLAG_EXTSIZE       (1 << XFS_DIFLAG_EXTSIZE_BIT)
1045#define XFS_DIFLAG_EXTSZINHERIT  (1 << XFS_DIFLAG_EXTSZINHERIT_BIT)
1046#define XFS_DIFLAG_NODEFRAG      (1 << XFS_DIFLAG_NODEFRAG_BIT)
1047#define XFS_DIFLAG_FILESTREAM    (1 << XFS_DIFLAG_FILESTREAM_BIT)
1048
1049#define XFS_DIFLAG_ANY \
1050	(XFS_DIFLAG_REALTIME | XFS_DIFLAG_PREALLOC | XFS_DIFLAG_NEWRTBM | \
1051	 XFS_DIFLAG_IMMUTABLE | XFS_DIFLAG_APPEND | XFS_DIFLAG_SYNC | \
1052	 XFS_DIFLAG_NOATIME | XFS_DIFLAG_NODUMP | XFS_DIFLAG_RTINHERIT | \
1053	 XFS_DIFLAG_PROJINHERIT | XFS_DIFLAG_NOSYMLINKS | XFS_DIFLAG_EXTSIZE | \
1054	 XFS_DIFLAG_EXTSZINHERIT | XFS_DIFLAG_NODEFRAG | XFS_DIFLAG_FILESTREAM)
1055
1056/*
1057 * Values for di_flags2 These start by being exposed to userspace in the upper
1058 * 16 bits of the XFS_XFLAG_s range.
1059 */
1060#define XFS_DIFLAG2_DAX_BIT	0	/* use DAX for this inode */
1061#define XFS_DIFLAG2_REFLINK_BIT	1	/* file's blocks may be shared */
1062#define XFS_DIFLAG2_COWEXTSIZE_BIT   2  /* copy on write extent size hint */
1063#define XFS_DIFLAG2_DAX		(1 << XFS_DIFLAG2_DAX_BIT)
1064#define XFS_DIFLAG2_REFLINK     (1 << XFS_DIFLAG2_REFLINK_BIT)
1065#define XFS_DIFLAG2_COWEXTSIZE  (1 << XFS_DIFLAG2_COWEXTSIZE_BIT)
1066
1067#define XFS_DIFLAG2_ANY \
1068	(XFS_DIFLAG2_DAX | XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE)
1069
1070/*
1071 * Inode number format:
1072 * low inopblog bits - offset in block
1073 * next agblklog bits - block number in ag
1074 * next agno_log bits - ag number
1075 * high agno_log-agblklog-inopblog bits - 0
1076 */
1077#define	XFS_INO_MASK(k)			(__uint32_t)((1ULL << (k)) - 1)
1078#define	XFS_INO_OFFSET_BITS(mp)		(mp)->m_sb.sb_inopblog
1079#define	XFS_INO_AGBNO_BITS(mp)		(mp)->m_sb.sb_agblklog
1080#define	XFS_INO_AGINO_BITS(mp)		(mp)->m_agino_log
1081#define	XFS_INO_AGNO_BITS(mp)		(mp)->m_agno_log
1082#define	XFS_INO_BITS(mp)		\
1083	XFS_INO_AGNO_BITS(mp) + XFS_INO_AGINO_BITS(mp)
1084#define	XFS_INO_TO_AGNO(mp,i)		\
1085	((xfs_agnumber_t)((i) >> XFS_INO_AGINO_BITS(mp)))
1086#define	XFS_INO_TO_AGINO(mp,i)		\
1087	((xfs_agino_t)(i) & XFS_INO_MASK(XFS_INO_AGINO_BITS(mp)))
1088#define	XFS_INO_TO_AGBNO(mp,i)		\
1089	(((xfs_agblock_t)(i) >> XFS_INO_OFFSET_BITS(mp)) & \
1090		XFS_INO_MASK(XFS_INO_AGBNO_BITS(mp)))
1091#define	XFS_INO_TO_OFFSET(mp,i)		\
1092	((int)(i) & XFS_INO_MASK(XFS_INO_OFFSET_BITS(mp)))
1093#define	XFS_INO_TO_FSB(mp,i)		\
1094	XFS_AGB_TO_FSB(mp, XFS_INO_TO_AGNO(mp,i), XFS_INO_TO_AGBNO(mp,i))
1095#define	XFS_AGINO_TO_INO(mp,a,i)	\
1096	(((xfs_ino_t)(a) << XFS_INO_AGINO_BITS(mp)) | (i))
1097#define	XFS_AGINO_TO_AGBNO(mp,i)	((i) >> XFS_INO_OFFSET_BITS(mp))
1098#define	XFS_AGINO_TO_OFFSET(mp,i)	\
1099	((i) & XFS_INO_MASK(XFS_INO_OFFSET_BITS(mp)))
1100#define	XFS_OFFBNO_TO_AGINO(mp,b,o)	\
1101	((xfs_agino_t)(((b) << XFS_INO_OFFSET_BITS(mp)) | (o)))
 
 
1102
1103#define	XFS_MAXINUMBER		((xfs_ino_t)((1ULL << 56) - 1ULL))
1104#define	XFS_MAXINUMBER_32	((xfs_ino_t)((1ULL << 32) - 1ULL))
1105
1106/*
1107 * RealTime Device format definitions
1108 */
1109
1110/* Min and max rt extent sizes, specified in bytes */
1111#define	XFS_MAX_RTEXTSIZE	(1024 * 1024 * 1024)	/* 1GB */
1112#define	XFS_DFL_RTEXTSIZE	(64 * 1024)	        /* 64kB */
1113#define	XFS_MIN_RTEXTSIZE	(4 * 1024)		/* 4kB */
1114
1115#define	XFS_BLOCKSIZE(mp)	((mp)->m_sb.sb_blocksize)
1116#define	XFS_BLOCKMASK(mp)	((mp)->m_blockmask)
1117#define	XFS_BLOCKWSIZE(mp)	((mp)->m_blockwsize)
1118#define	XFS_BLOCKWMASK(mp)	((mp)->m_blockwmask)
1119
1120/*
1121 * RT Summary and bit manipulation macros.
1122 */
1123#define	XFS_SUMOFFS(mp,ls,bb)	((int)((ls) * (mp)->m_sb.sb_rbmblocks + (bb)))
1124#define	XFS_SUMOFFSTOBLOCK(mp,s)	\
1125	(((s) * (uint)sizeof(xfs_suminfo_t)) >> (mp)->m_sb.sb_blocklog)
1126#define	XFS_SUMPTR(mp,bp,so)	\
1127	((xfs_suminfo_t *)((bp)->b_addr + \
1128		(((so) * (uint)sizeof(xfs_suminfo_t)) & XFS_BLOCKMASK(mp))))
1129
1130#define	XFS_BITTOBLOCK(mp,bi)	((bi) >> (mp)->m_blkbit_log)
1131#define	XFS_BLOCKTOBIT(mp,bb)	((bb) << (mp)->m_blkbit_log)
1132#define	XFS_BITTOWORD(mp,bi)	\
1133	((int)(((bi) >> XFS_NBWORDLOG) & XFS_BLOCKWMASK(mp)))
1134
1135#define	XFS_RTMIN(a,b)	((a) < (b) ? (a) : (b))
1136#define	XFS_RTMAX(a,b)	((a) > (b) ? (a) : (b))
1137
1138#define	XFS_RTLOBIT(w)	xfs_lowbit32(w)
1139#define	XFS_RTHIBIT(w)	xfs_highbit32(w)
1140
1141#define	XFS_RTBLOCKLOG(b)	xfs_highbit64(b)
1142
1143/*
1144 * Dquot and dquot block format definitions
1145 */
1146#define XFS_DQUOT_MAGIC		0x4451		/* 'DQ' */
1147#define XFS_DQUOT_VERSION	(u_int8_t)0x01	/* latest version number */
1148
1149/*
1150 * This is the main portion of the on-disk representation of quota
1151 * information for a user. This is the q_core of the xfs_dquot_t that
1152 * is kept in kernel memory. We pad this with some more expansion room
1153 * to construct the on disk structure.
1154 */
1155typedef struct	xfs_disk_dquot {
1156	__be16		d_magic;	/* dquot magic = XFS_DQUOT_MAGIC */
1157	__u8		d_version;	/* dquot version */
1158	__u8		d_flags;	/* XFS_DQ_USER/PROJ/GROUP */
1159	__be32		d_id;		/* user,project,group id */
1160	__be64		d_blk_hardlimit;/* absolute limit on disk blks */
1161	__be64		d_blk_softlimit;/* preferred limit on disk blks */
1162	__be64		d_ino_hardlimit;/* maximum # allocated inodes */
1163	__be64		d_ino_softlimit;/* preferred inode limit */
1164	__be64		d_bcount;	/* disk blocks owned by the user */
1165	__be64		d_icount;	/* inodes owned by the user */
1166	__be32		d_itimer;	/* zero if within inode limits if not,
1167					   this is when we refuse service */
1168	__be32		d_btimer;	/* similar to above; for disk blocks */
1169	__be16		d_iwarns;	/* warnings issued wrt num inodes */
1170	__be16		d_bwarns;	/* warnings issued wrt disk blocks */
1171	__be32		d_pad0;		/* 64 bit align */
1172	__be64		d_rtb_hardlimit;/* absolute limit on realtime blks */
1173	__be64		d_rtb_softlimit;/* preferred limit on RT disk blks */
1174	__be64		d_rtbcount;	/* realtime blocks owned */
1175	__be32		d_rtbtimer;	/* similar to above; for RT disk blocks */
1176	__be16		d_rtbwarns;	/* warnings issued wrt RT disk blocks */
1177	__be16		d_pad;
1178} xfs_disk_dquot_t;
1179
1180/*
1181 * This is what goes on disk. This is separated from the xfs_disk_dquot because
1182 * carrying the unnecessary padding would be a waste of memory.
1183 */
1184typedef struct xfs_dqblk {
1185	xfs_disk_dquot_t  dd_diskdq;	/* portion that lives incore as well */
1186	char		  dd_fill[4];	/* filling for posterity */
1187
1188	/*
1189	 * These two are only present on filesystems with the CRC bits set.
1190	 */
1191	__be32		  dd_crc;	/* checksum */
1192	__be64		  dd_lsn;	/* last modification in log */
1193	uuid_t		  dd_uuid;	/* location information */
1194} xfs_dqblk_t;
1195
1196#define XFS_DQUOT_CRC_OFF	offsetof(struct xfs_dqblk, dd_crc)
1197
1198/*
1199 * Remote symlink format and access functions.
1200 */
1201#define XFS_SYMLINK_MAGIC	0x58534c4d	/* XSLM */
1202
1203struct xfs_dsymlink_hdr {
1204	__be32	sl_magic;
1205	__be32	sl_offset;
1206	__be32	sl_bytes;
1207	__be32	sl_crc;
1208	uuid_t	sl_uuid;
1209	__be64	sl_owner;
1210	__be64	sl_blkno;
1211	__be64	sl_lsn;
1212};
1213
1214#define XFS_SYMLINK_CRC_OFF	offsetof(struct xfs_dsymlink_hdr, sl_crc)
1215
 
1216/*
1217 * The maximum pathlen is 1024 bytes. Since the minimum file system
1218 * blocksize is 512 bytes, we can get a max of 3 extents back from
1219 * bmapi when crc headers are taken into account.
1220 */
1221#define XFS_SYMLINK_MAPS 3
1222
1223#define XFS_SYMLINK_BUF_SPACE(mp, bufsize)	\
1224	((bufsize) - (xfs_sb_version_hascrc(&(mp)->m_sb) ? \
1225			sizeof(struct xfs_dsymlink_hdr) : 0))
1226
1227
1228/*
1229 * Allocation Btree format definitions
1230 *
1231 * There are two on-disk btrees, one sorted by blockno and one sorted
1232 * by blockcount and blockno.  All blocks look the same to make the code
1233 * simpler; if we have time later, we'll make the optimizations.
1234 */
1235#define	XFS_ABTB_MAGIC		0x41425442	/* 'ABTB' for bno tree */
1236#define	XFS_ABTB_CRC_MAGIC	0x41423342	/* 'AB3B' */
1237#define	XFS_ABTC_MAGIC		0x41425443	/* 'ABTC' for cnt tree */
1238#define	XFS_ABTC_CRC_MAGIC	0x41423343	/* 'AB3C' */
1239
1240/*
1241 * Data record/key structure
1242 */
1243typedef struct xfs_alloc_rec {
1244	__be32		ar_startblock;	/* starting block number */
1245	__be32		ar_blockcount;	/* count of free blocks */
1246} xfs_alloc_rec_t, xfs_alloc_key_t;
1247
1248typedef struct xfs_alloc_rec_incore {
1249	xfs_agblock_t	ar_startblock;	/* starting block number */
1250	xfs_extlen_t	ar_blockcount;	/* count of free blocks */
1251} xfs_alloc_rec_incore_t;
1252
1253/* btree pointer type */
1254typedef __be32 xfs_alloc_ptr_t;
1255
1256/*
1257 * Block numbers in the AG:
1258 * SB is sector 0, AGF is sector 1, AGI is sector 2, AGFL is sector 3.
1259 */
1260#define	XFS_BNO_BLOCK(mp)	((xfs_agblock_t)(XFS_AGFL_BLOCK(mp) + 1))
1261#define	XFS_CNT_BLOCK(mp)	((xfs_agblock_t)(XFS_BNO_BLOCK(mp) + 1))
1262
1263
1264/*
1265 * Inode Allocation Btree format definitions
1266 *
1267 * There is a btree for the inode map per allocation group.
1268 */
1269#define	XFS_IBT_MAGIC		0x49414254	/* 'IABT' */
1270#define	XFS_IBT_CRC_MAGIC	0x49414233	/* 'IAB3' */
1271#define	XFS_FIBT_MAGIC		0x46494254	/* 'FIBT' */
1272#define	XFS_FIBT_CRC_MAGIC	0x46494233	/* 'FIB3' */
1273
1274typedef	__uint64_t	xfs_inofree_t;
1275#define	XFS_INODES_PER_CHUNK		(NBBY * sizeof(xfs_inofree_t))
1276#define	XFS_INODES_PER_CHUNK_LOG	(XFS_NBBYLOG + 3)
1277#define	XFS_INOBT_ALL_FREE		((xfs_inofree_t)-1)
1278#define	XFS_INOBT_MASK(i)		((xfs_inofree_t)1 << (i))
1279
1280#define XFS_INOBT_HOLEMASK_FULL		0	/* holemask for full chunk */
1281#define XFS_INOBT_HOLEMASK_BITS		(NBBY * sizeof(__uint16_t))
1282#define XFS_INODES_PER_HOLEMASK_BIT	\
1283	(XFS_INODES_PER_CHUNK / (NBBY * sizeof(__uint16_t)))
1284
1285static inline xfs_inofree_t xfs_inobt_maskn(int i, int n)
1286{
1287	return ((n >= XFS_INODES_PER_CHUNK ? 0 : XFS_INOBT_MASK(n)) - 1) << i;
1288}
1289
1290/*
1291 * The on-disk inode record structure has two formats. The original "full"
1292 * format uses a 4-byte freecount. The "sparse" format uses a 1-byte freecount
1293 * and replaces the 3 high-order freecount bytes wth the holemask and inode
1294 * count.
1295 *
1296 * The holemask of the sparse record format allows an inode chunk to have holes
1297 * that refer to blocks not owned by the inode record. This facilitates inode
1298 * allocation in the event of severe free space fragmentation.
1299 */
1300typedef struct xfs_inobt_rec {
1301	__be32		ir_startino;	/* starting inode number */
1302	union {
1303		struct {
1304			__be32	ir_freecount;	/* count of free inodes */
1305		} f;
1306		struct {
1307			__be16	ir_holemask;/* hole mask for sparse chunks */
1308			__u8	ir_count;	/* total inode count */
1309			__u8	ir_freecount;	/* count of free inodes */
1310		} sp;
1311	} ir_u;
1312	__be64		ir_free;	/* free inode mask */
1313} xfs_inobt_rec_t;
1314
1315typedef struct xfs_inobt_rec_incore {
1316	xfs_agino_t	ir_startino;	/* starting inode number */
1317	__uint16_t	ir_holemask;	/* hole mask for sparse chunks */
1318	__uint8_t	ir_count;	/* total inode count */
1319	__uint8_t	ir_freecount;	/* count of free inodes (set bits) */
1320	xfs_inofree_t	ir_free;	/* free inode mask */
1321} xfs_inobt_rec_incore_t;
1322
1323static inline bool xfs_inobt_issparse(uint16_t holemask)
1324{
1325	/* non-zero holemask represents a sparse rec. */
1326	return holemask;
1327}
1328
1329/*
1330 * Key structure
1331 */
1332typedef struct xfs_inobt_key {
1333	__be32		ir_startino;	/* starting inode number */
1334} xfs_inobt_key_t;
1335
1336/* btree pointer type */
1337typedef __be32 xfs_inobt_ptr_t;
1338
1339/*
1340 * block numbers in the AG.
1341 */
1342#define	XFS_IBT_BLOCK(mp)		((xfs_agblock_t)(XFS_CNT_BLOCK(mp) + 1))
1343#define	XFS_FIBT_BLOCK(mp)		((xfs_agblock_t)(XFS_IBT_BLOCK(mp) + 1))
1344
1345/*
1346 * Reverse mapping btree format definitions
1347 *
1348 * There is a btree for the reverse map per allocation group
1349 */
1350#define	XFS_RMAP_CRC_MAGIC	0x524d4233	/* 'RMB3' */
1351
1352/*
1353 * Ownership info for an extent.  This is used to create reverse-mapping
1354 * entries.
1355 */
1356#define XFS_OWNER_INFO_ATTR_FORK	(1 << 0)
1357#define XFS_OWNER_INFO_BMBT_BLOCK	(1 << 1)
1358struct xfs_owner_info {
1359	uint64_t		oi_owner;
1360	xfs_fileoff_t		oi_offset;
1361	unsigned int		oi_flags;
1362};
1363
1364/*
1365 * Special owner types.
1366 *
1367 * Seeing as we only support up to 8EB, we have the upper bit of the owner field
1368 * to tell us we have a special owner value. We use these for static metadata
1369 * allocated at mkfs/growfs time, as well as for freespace management metadata.
1370 */
1371#define XFS_RMAP_OWN_NULL	(-1ULL)	/* No owner, for growfs */
1372#define XFS_RMAP_OWN_UNKNOWN	(-2ULL)	/* Unknown owner, for EFI recovery */
1373#define XFS_RMAP_OWN_FS		(-3ULL)	/* static fs metadata */
1374#define XFS_RMAP_OWN_LOG	(-4ULL)	/* static fs metadata */
1375#define XFS_RMAP_OWN_AG		(-5ULL)	/* AG freespace btree blocks */
1376#define XFS_RMAP_OWN_INOBT	(-6ULL)	/* Inode btree blocks */
1377#define XFS_RMAP_OWN_INODES	(-7ULL)	/* Inode chunk */
1378#define XFS_RMAP_OWN_REFC	(-8ULL) /* refcount tree */
1379#define XFS_RMAP_OWN_COW	(-9ULL) /* cow allocations */
1380#define XFS_RMAP_OWN_MIN	(-10ULL) /* guard */
1381
1382#define XFS_RMAP_NON_INODE_OWNER(owner)	(!!((owner) & (1ULL << 63)))
1383
1384/*
1385 * Data record structure
1386 */
1387struct xfs_rmap_rec {
1388	__be32		rm_startblock;	/* extent start block */
1389	__be32		rm_blockcount;	/* extent length */
1390	__be64		rm_owner;	/* extent owner */
1391	__be64		rm_offset;	/* offset within the owner */
1392};
1393
1394/*
1395 * rmap btree record
1396 *  rm_offset:63 is the attribute fork flag
1397 *  rm_offset:62 is the bmbt block flag
1398 *  rm_offset:61 is the unwritten extent flag (same as l0:63 in bmbt)
1399 *  rm_offset:54-60 aren't used and should be zero
1400 *  rm_offset:0-53 is the block offset within the inode
1401 */
1402#define XFS_RMAP_OFF_ATTR_FORK	((__uint64_t)1ULL << 63)
1403#define XFS_RMAP_OFF_BMBT_BLOCK	((__uint64_t)1ULL << 62)
1404#define XFS_RMAP_OFF_UNWRITTEN	((__uint64_t)1ULL << 61)
1405
1406#define XFS_RMAP_LEN_MAX	((__uint32_t)~0U)
1407#define XFS_RMAP_OFF_FLAGS	(XFS_RMAP_OFF_ATTR_FORK | \
1408				 XFS_RMAP_OFF_BMBT_BLOCK | \
1409				 XFS_RMAP_OFF_UNWRITTEN)
1410#define XFS_RMAP_OFF_MASK	((__uint64_t)0x3FFFFFFFFFFFFFULL)
1411
1412#define XFS_RMAP_OFF(off)		((off) & XFS_RMAP_OFF_MASK)
1413
1414#define XFS_RMAP_IS_BMBT_BLOCK(off)	(!!((off) & XFS_RMAP_OFF_BMBT_BLOCK))
1415#define XFS_RMAP_IS_ATTR_FORK(off)	(!!((off) & XFS_RMAP_OFF_ATTR_FORK))
1416#define XFS_RMAP_IS_UNWRITTEN(len)	(!!((off) & XFS_RMAP_OFF_UNWRITTEN))
1417
1418#define RMAPBT_STARTBLOCK_BITLEN	32
1419#define RMAPBT_BLOCKCOUNT_BITLEN	32
1420#define RMAPBT_OWNER_BITLEN		64
1421#define RMAPBT_ATTRFLAG_BITLEN		1
1422#define RMAPBT_BMBTFLAG_BITLEN		1
1423#define RMAPBT_EXNTFLAG_BITLEN		1
1424#define RMAPBT_UNUSED_OFFSET_BITLEN	7
1425#define RMAPBT_OFFSET_BITLEN		54
1426
1427#define XFS_RMAP_ATTR_FORK		(1 << 0)
1428#define XFS_RMAP_BMBT_BLOCK		(1 << 1)
1429#define XFS_RMAP_UNWRITTEN		(1 << 2)
1430#define XFS_RMAP_KEY_FLAGS		(XFS_RMAP_ATTR_FORK | \
1431					 XFS_RMAP_BMBT_BLOCK)
1432#define XFS_RMAP_REC_FLAGS		(XFS_RMAP_UNWRITTEN)
1433struct xfs_rmap_irec {
1434	xfs_agblock_t	rm_startblock;	/* extent start block */
1435	xfs_extlen_t	rm_blockcount;	/* extent length */
1436	__uint64_t	rm_owner;	/* extent owner */
1437	__uint64_t	rm_offset;	/* offset within the owner */
1438	unsigned int	rm_flags;	/* state flags */
1439};
1440
1441/*
1442 * Key structure
1443 *
1444 * We don't use the length for lookups
1445 */
1446struct xfs_rmap_key {
1447	__be32		rm_startblock;	/* extent start block */
1448	__be64		rm_owner;	/* extent owner */
1449	__be64		rm_offset;	/* offset within the owner */
1450} __attribute__((packed));
1451
1452/* btree pointer type */
1453typedef __be32 xfs_rmap_ptr_t;
1454
1455#define	XFS_RMAP_BLOCK(mp) \
1456	(xfs_sb_version_hasfinobt(&((mp)->m_sb)) ? \
1457	 XFS_FIBT_BLOCK(mp) + 1 : \
1458	 XFS_IBT_BLOCK(mp) + 1)
1459
1460/*
1461 * Reference Count Btree format definitions
1462 *
1463 */
1464#define	XFS_REFC_CRC_MAGIC	0x52334643	/* 'R3FC' */
1465
1466unsigned int xfs_refc_block(struct xfs_mount *mp);
1467
1468/*
1469 * Data record/key structure
1470 *
1471 * Each record associates a range of physical blocks (starting at
1472 * rc_startblock and ending rc_blockcount blocks later) with a reference
1473 * count (rc_refcount).  Extents that are being used to stage a copy on
1474 * write (CoW) operation are recorded in the refcount btree with a
1475 * refcount of 1.  All other records must have a refcount > 1 and must
1476 * track an extent mapped only by file data forks.
1477 *
1478 * Extents with a single owner (attributes, metadata, non-shared file
1479 * data) are not tracked here.  Free space is also not tracked here.
1480 * This is consistent with pre-reflink XFS.
1481 */
1482
1483/*
1484 * Extents that are being used to stage a copy on write are stored
1485 * in the refcount btree with a refcount of 1 and the upper bit set
1486 * on the startblock.  This speeds up mount time deletion of stale
1487 * staging extents because they're all at the right side of the tree.
1488 */
1489#define XFS_REFC_COW_START		((xfs_agblock_t)(1U << 31))
1490#define REFCNTBT_COWFLAG_BITLEN		1
1491#define REFCNTBT_AGBLOCK_BITLEN		31
1492
1493struct xfs_refcount_rec {
1494	__be32		rc_startblock;	/* starting block number */
1495	__be32		rc_blockcount;	/* count of blocks */
1496	__be32		rc_refcount;	/* number of inodes linked here */
1497};
1498
1499struct xfs_refcount_key {
1500	__be32		rc_startblock;	/* starting block number */
1501};
1502
1503struct xfs_refcount_irec {
1504	xfs_agblock_t	rc_startblock;	/* starting block number */
1505	xfs_extlen_t	rc_blockcount;	/* count of free blocks */
1506	xfs_nlink_t	rc_refcount;	/* number of inodes linked here */
1507};
1508
1509#define MAXREFCOUNT	((xfs_nlink_t)~0U)
1510#define MAXREFCEXTLEN	((xfs_extlen_t)~0U)
1511
1512/* btree pointer type */
1513typedef __be32 xfs_refcount_ptr_t;
1514
1515
1516/*
1517 * BMAP Btree format definitions
1518 *
1519 * This includes both the root block definition that sits inside an inode fork
1520 * and the record/pointer formats for the leaf/node in the blocks.
1521 */
1522#define XFS_BMAP_MAGIC		0x424d4150	/* 'BMAP' */
1523#define XFS_BMAP_CRC_MAGIC	0x424d4133	/* 'BMA3' */
1524
1525/*
1526 * Bmap root header, on-disk form only.
1527 */
1528typedef struct xfs_bmdr_block {
1529	__be16		bb_level;	/* 0 is a leaf */
1530	__be16		bb_numrecs;	/* current # of data records */
1531} xfs_bmdr_block_t;
1532
1533/*
1534 * Bmap btree record and extent descriptor.
1535 *  l0:63 is an extent flag (value 1 indicates non-normal).
1536 *  l0:9-62 are startoff.
1537 *  l0:0-8 and l1:21-63 are startblock.
1538 *  l1:0-20 are blockcount.
1539 */
1540#define BMBT_EXNTFLAG_BITLEN	1
1541#define BMBT_STARTOFF_BITLEN	54
1542#define BMBT_STARTBLOCK_BITLEN	52
1543#define BMBT_BLOCKCOUNT_BITLEN	21
1544
 
 
1545typedef struct xfs_bmbt_rec {
1546	__be64			l0, l1;
1547} xfs_bmbt_rec_t;
1548
1549typedef __uint64_t	xfs_bmbt_rec_base_t;	/* use this for casts */
1550typedef xfs_bmbt_rec_t xfs_bmdr_rec_t;
1551
1552typedef struct xfs_bmbt_rec_host {
1553	__uint64_t		l0, l1;
1554} xfs_bmbt_rec_host_t;
1555
1556/*
1557 * Values and macros for delayed-allocation startblock fields.
1558 */
1559#define STARTBLOCKVALBITS	17
1560#define STARTBLOCKMASKBITS	(15 + 20)
1561#define STARTBLOCKMASK		\
1562	(((((xfs_fsblock_t)1) << STARTBLOCKMASKBITS) - 1) << STARTBLOCKVALBITS)
1563
1564static inline int isnullstartblock(xfs_fsblock_t x)
1565{
1566	return ((x) & STARTBLOCKMASK) == STARTBLOCKMASK;
1567}
1568
1569static inline xfs_fsblock_t nullstartblock(int k)
1570{
1571	ASSERT(k < (1 << STARTBLOCKVALBITS));
1572	return STARTBLOCKMASK | (k);
1573}
1574
1575static inline xfs_filblks_t startblockval(xfs_fsblock_t x)
1576{
1577	return (xfs_filblks_t)((x) & ~STARTBLOCKMASK);
1578}
1579
1580/*
1581 * Possible extent formats.
1582 */
1583typedef enum {
1584	XFS_EXTFMT_NOSTATE = 0,
1585	XFS_EXTFMT_HASSTATE
1586} xfs_exntfmt_t;
1587
1588/*
1589 * Possible extent states.
1590 */
1591typedef enum {
1592	XFS_EXT_NORM, XFS_EXT_UNWRITTEN,
1593	XFS_EXT_DMAPI_OFFLINE, XFS_EXT_INVALID
1594} xfs_exntst_t;
1595
1596/*
1597 * Incore version of above.
1598 */
1599typedef struct xfs_bmbt_irec
1600{
1601	xfs_fileoff_t	br_startoff;	/* starting file offset */
1602	xfs_fsblock_t	br_startblock;	/* starting block number */
1603	xfs_filblks_t	br_blockcount;	/* number of blocks */
1604	xfs_exntst_t	br_state;	/* extent state */
1605} xfs_bmbt_irec_t;
1606
1607/*
1608 * Key structure for non-leaf levels of the tree.
1609 */
1610typedef struct xfs_bmbt_key {
1611	__be64		br_startoff;	/* starting file offset */
1612} xfs_bmbt_key_t, xfs_bmdr_key_t;
1613
1614/* btree pointer type */
1615typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
1616
1617
1618/*
1619 * Generic Btree block format definitions
1620 *
1621 * This is a combination of the actual format used on disk for short and long
1622 * format btrees.  The first three fields are shared by both format, but the
1623 * pointers are different and should be used with care.
1624 *
1625 * To get the size of the actual short or long form headers please use the size
1626 * macros below.  Never use sizeof(xfs_btree_block).
1627 *
1628 * The blkno, crc, lsn, owner and uuid fields are only available in filesystems
1629 * with the crc feature bit, and all accesses to them must be conditional on
1630 * that flag.
1631 */
1632/* short form block header */
1633struct xfs_btree_block_shdr {
1634	__be32		bb_leftsib;
1635	__be32		bb_rightsib;
1636
1637	__be64		bb_blkno;
1638	__be64		bb_lsn;
1639	uuid_t		bb_uuid;
1640	__be32		bb_owner;
1641	__le32		bb_crc;
1642};
1643
1644/* long form block header */
1645struct xfs_btree_block_lhdr {
1646	__be64		bb_leftsib;
1647	__be64		bb_rightsib;
1648
1649	__be64		bb_blkno;
1650	__be64		bb_lsn;
1651	uuid_t		bb_uuid;
1652	__be64		bb_owner;
1653	__le32		bb_crc;
1654	__be32		bb_pad; /* padding for alignment */
1655};
1656
1657struct xfs_btree_block {
1658	__be32		bb_magic;	/* magic number for block type */
1659	__be16		bb_level;	/* 0 is a leaf */
1660	__be16		bb_numrecs;	/* current # of data records */
1661	union {
1662		struct xfs_btree_block_shdr s;
1663		struct xfs_btree_block_lhdr l;
1664	} bb_u;				/* rest */
1665};
1666
1667/* size of a short form block */
1668#define XFS_BTREE_SBLOCK_LEN \
1669	(offsetof(struct xfs_btree_block, bb_u) + \
1670	 offsetof(struct xfs_btree_block_shdr, bb_blkno))
1671/* size of a long form block */
1672#define XFS_BTREE_LBLOCK_LEN \
1673	(offsetof(struct xfs_btree_block, bb_u) + \
1674	 offsetof(struct xfs_btree_block_lhdr, bb_blkno))
1675
1676/* sizes of CRC enabled btree blocks */
1677#define XFS_BTREE_SBLOCK_CRC_LEN \
1678	(offsetof(struct xfs_btree_block, bb_u) + \
1679	 sizeof(struct xfs_btree_block_shdr))
1680#define XFS_BTREE_LBLOCK_CRC_LEN \
1681	(offsetof(struct xfs_btree_block, bb_u) + \
1682	 sizeof(struct xfs_btree_block_lhdr))
1683
1684#define XFS_BTREE_SBLOCK_CRC_OFF \
1685	offsetof(struct xfs_btree_block, bb_u.s.bb_crc)
1686#define XFS_BTREE_LBLOCK_CRC_OFF \
1687	offsetof(struct xfs_btree_block, bb_u.l.bb_crc)
1688
1689/*
1690 * On-disk XFS access control list structure.
1691 */
1692struct xfs_acl_entry {
1693	__be32	ae_tag;
1694	__be32	ae_id;
1695	__be16	ae_perm;
1696	__be16	ae_pad;		/* fill the implicit hole in the structure */
1697};
1698
1699struct xfs_acl {
1700	__be32			acl_cnt;
1701	struct xfs_acl_entry	acl_entry[0];
1702};
1703
1704/*
1705 * The number of ACL entries allowed is defined by the on-disk format.
1706 * For v4 superblocks, that is limited to 25 entries. For v5 superblocks, it is
1707 * limited only by the maximum size of the xattr that stores the information.
1708 */
1709#define XFS_ACL_MAX_ENTRIES(mp)	\
1710	(xfs_sb_version_hascrc(&mp->m_sb) \
1711		?  (XFS_XATTR_SIZE_MAX - sizeof(struct xfs_acl)) / \
1712						sizeof(struct xfs_acl_entry) \
1713		: 25)
1714
1715#define XFS_ACL_SIZE(cnt) \
1716	(sizeof(struct xfs_acl) + \
1717		sizeof(struct xfs_acl_entry) * cnt)
1718
1719#define XFS_ACL_MAX_SIZE(mp) \
1720	XFS_ACL_SIZE(XFS_ACL_MAX_ENTRIES((mp)))
1721
1722
1723/* On-disk XFS extended attribute names */
1724#define SGI_ACL_FILE		"SGI_ACL_FILE"
1725#define SGI_ACL_DEFAULT		"SGI_ACL_DEFAULT"
1726#define SGI_ACL_FILE_SIZE	(sizeof(SGI_ACL_FILE)-1)
1727#define SGI_ACL_DEFAULT_SIZE	(sizeof(SGI_ACL_DEFAULT)-1)
1728
1729#endif /* __XFS_FORMAT_H__ */