xref: /freebsd/sys/ufs/ffs/fs.h (revision c66ec88fed842fbaad62c30d510644ceb7bd2d71)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1982, 1986, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the University nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  *
31  *	@(#)fs.h	8.13 (Berkeley) 3/21/95
32  * $FreeBSD$
33  */
34 
35 #ifndef _UFS_FFS_FS_H_
36 #define	_UFS_FFS_FS_H_
37 
38 #include <sys/mount.h>
39 #include <ufs/ufs/dinode.h>
40 
41 /*
42  * Each disk drive contains some number of filesystems.
43  * A filesystem consists of a number of cylinder groups.
44  * Each cylinder group has inodes and data.
45  *
46  * A filesystem is described by its super-block, which in turn
47  * describes the cylinder groups.  The super-block is critical
48  * data and is replicated in each cylinder group to protect against
49  * catastrophic loss.  This is done at `newfs' time and the critical
50  * super-block data does not change, so the copies need not be
51  * referenced further unless disaster strikes.
52  *
53  * For filesystem fs, the offsets of the various blocks of interest
54  * are given in the super block as:
55  *	[fs->fs_sblkno]		Super-block
56  *	[fs->fs_cblkno]		Cylinder group block
57  *	[fs->fs_iblkno]		Inode blocks
58  *	[fs->fs_dblkno]		Data blocks
59  * The beginning of cylinder group cg in fs, is given by
60  * the ``cgbase(fs, cg)'' macro.
61  *
62  * Depending on the architecture and the media, the superblock may
63  * reside in any one of four places. For tiny media where every block
64  * counts, it is placed at the very front of the partition. Historically,
65  * UFS1 placed it 8K from the front to leave room for the disk label and
66  * a small bootstrap. For UFS2 it got moved to 64K from the front to leave
67  * room for the disk label and a bigger bootstrap, and for really piggy
68  * systems we check at 256K from the front if the first three fail. In
69  * all cases the size of the superblock will be SBLOCKSIZE. All values are
70  * given in byte-offset form, so they do not imply a sector size. The
71  * SBLOCKSEARCH specifies the order in which the locations should be searched.
72  */
73 #define	SBLOCK_FLOPPY	     0
74 #define	SBLOCK_UFS1	  8192
75 #define	SBLOCK_UFS2	 65536
76 #define	SBLOCK_PIGGY	262144
77 #define	SBLOCKSIZE	  8192
78 #define	SBLOCKSEARCH \
79 	{ SBLOCK_UFS2, SBLOCK_UFS1, SBLOCK_FLOPPY, SBLOCK_PIGGY, -1 }
80 
81 /*
82  * Max number of fragments per block. This value is NOT tweakable.
83  */
84 #define	MAXFRAG 	8
85 
86 /*
87  * Addresses stored in inodes are capable of addressing fragments
88  * of `blocks'. File system blocks of at most size MAXBSIZE can
89  * be optionally broken into 2, 4, or 8 pieces, each of which is
90  * addressable; these pieces may be DEV_BSIZE, or some multiple of
91  * a DEV_BSIZE unit.
92  *
93  * Large files consist of exclusively large data blocks.  To avoid
94  * undue wasted disk space, the last data block of a small file may be
95  * allocated as only as many fragments of a large block as are
96  * necessary.  The filesystem format retains only a single pointer
97  * to such a fragment, which is a piece of a single large block that
98  * has been divided.  The size of such a fragment is determinable from
99  * information in the inode, using the ``blksize(fs, ip, lbn)'' macro.
100  *
101  * The filesystem records space availability at the fragment level;
102  * to determine block availability, aligned fragments are examined.
103  */
104 
105 /*
106  * MINBSIZE is the smallest allowable block size.
107  * In order to insure that it is possible to create files of size
108  * 2^32 with only two levels of indirection, MINBSIZE is set to 4096.
109  * MINBSIZE must be big enough to hold a cylinder group block,
110  * thus changes to (struct cg) must keep its size within MINBSIZE.
111  * Note that super blocks are always of size SBLOCKSIZE,
112  * and that both SBLOCKSIZE and MAXBSIZE must be >= MINBSIZE.
113  */
114 #define	MINBSIZE	4096
115 
116 /*
117  * The path name on which the filesystem is mounted is maintained
118  * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in
119  * the super block for this name.
120  */
121 #define	MAXMNTLEN	468
122 
123 /*
124  * The volume name for this filesystem is maintained in fs_volname.
125  * MAXVOLLEN defines the length of the buffer allocated.
126  */
127 #define	MAXVOLLEN	32
128 
129 /*
130  * There is a 128-byte region in the superblock reserved for in-core
131  * pointers to summary information. Originally this included an array
132  * of pointers to blocks of struct csum; now there are just a few
133  * pointers and the remaining space is padded with fs_ocsp[].
134  *
135  * NOCSPTRS determines the size of this padding. Historically this
136  * space was used to store pointers to structures that summaried
137  * filesystem usage and layout information. However, these pointers
138  * left various kernel pointers in the superblock which made otherwise
139  * identical superblocks appear to have differences. So, all the
140  * pointers in the superblock were moved to a fs_summary_info structure
141  * reducing the superblock to having only a single pointer to this
142  * structure. When writing the superblock to disk, this pointer is
143  * temporarily NULL'ed out so that the kernel pointer will not appear
144  * in the on-disk copy of the superblock.
145  */
146 #define	NOCSPTRS	((128 / sizeof(void *)) - 1)
147 
148 /*
149  * A summary of contiguous blocks of various sizes is maintained
150  * in each cylinder group. Normally this is set by the initial
151  * value of fs_maxcontig. To conserve space, a maximum summary size
152  * is set by FS_MAXCONTIG.
153  */
154 #define	FS_MAXCONTIG	16
155 
156 /*
157  * MINFREE gives the minimum acceptable percentage of filesystem
158  * blocks which may be free. If the freelist drops below this level
159  * only the superuser may continue to allocate blocks. This may
160  * be set to 0 if no reserve of free blocks is deemed necessary,
161  * however throughput drops by fifty percent if the filesystem
162  * is run at between 95% and 100% full; thus the minimum default
163  * value of fs_minfree is 5%. However, to get good clustering
164  * performance, 10% is a better choice. hence we use 10% as our
165  * default value. With 10% free space, fragmentation is not a
166  * problem, so we choose to optimize for time.
167  */
168 #define	MINFREE		8
169 #define	DEFAULTOPT	FS_OPTTIME
170 
171 /*
172  * Grigoriy Orlov <gluk@ptci.ru> has done some extensive work to fine
173  * tune the layout preferences for directories within a filesystem.
174  * His algorithm can be tuned by adjusting the following parameters
175  * which tell the system the average file size and the average number
176  * of files per directory. These defaults are well selected for typical
177  * filesystems, but may need to be tuned for odd cases like filesystems
178  * being used for squid caches or news spools.
179  */
180 #define	AVFILESIZ	16384	/* expected average file size */
181 #define	AFPDIR		64	/* expected number of files per directory */
182 
183 /*
184  * The maximum number of snapshot nodes that can be associated
185  * with each filesystem. This limit affects only the number of
186  * snapshot files that can be recorded within the superblock so
187  * that they can be found when the filesystem is mounted. However,
188  * maintaining too many will slow the filesystem performance, so
189  * having this limit is a good idea.
190  */
191 #define	FSMAXSNAP 20
192 
193 /*
194  * Used to identify special blocks in snapshots:
195  *
196  * BLK_NOCOPY - A block that was unallocated at the time the snapshot
197  *	was taken, hence does not need to be copied when written.
198  * BLK_SNAP - A block held by another snapshot that is not needed by this
199  *	snapshot. When the other snapshot is freed, the BLK_SNAP entries
200  *	are converted to BLK_NOCOPY. These are needed to allow fsck to
201  *	identify blocks that are in use by other snapshots (which are
202  *	expunged from this snapshot).
203  */
204 #define	BLK_NOCOPY ((ufs2_daddr_t)(1))
205 #define	BLK_SNAP ((ufs2_daddr_t)(2))
206 
207 /*
208  * Sysctl values for the fast filesystem.
209  */
210 #define	FFS_ADJ_REFCNT		 1	/* adjust inode reference count */
211 #define	FFS_ADJ_BLKCNT		 2	/* adjust inode used block count */
212 #define	FFS_BLK_FREE		 3	/* free range of blocks in map */
213 #define	FFS_DIR_FREE		 4	/* free specified dir inodes in map */
214 #define	FFS_FILE_FREE		 5	/* free specified file inodes in map */
215 #define	FFS_SET_FLAGS		 6	/* set filesystem flags */
216 #define	FFS_ADJ_NDIR		 7	/* adjust number of directories */
217 #define	FFS_ADJ_NBFREE		 8	/* adjust number of free blocks */
218 #define	FFS_ADJ_NIFREE		 9	/* adjust number of free inodes */
219 #define	FFS_ADJ_NFFREE		10 	/* adjust number of free frags */
220 #define	FFS_ADJ_NUMCLUSTERS	11	/* adjust number of free clusters */
221 #define	FFS_SET_CWD		12	/* set current directory */
222 #define	FFS_SET_DOTDOT		13	/* set inode number for ".." */
223 #define	FFS_UNLINK		14	/* remove a name in the filesystem */
224 /* Was FFS_SET_INODE		15 */
225 /* Was FFS_SET_BUFOUTPUT	16 */
226 #define	FFS_SET_SIZE		17	/* set inode size */
227 #define	FFS_MAXID		17	/* number of valid ffs ids */
228 
229 /*
230  * Command structure passed in to the filesystem to adjust filesystem values.
231  */
232 #define	FFS_CMD_VERSION		0x19790518	/* version ID */
233 struct fsck_cmd {
234 	int32_t	version;	/* version of command structure */
235 	int32_t	handle;		/* reference to filesystem to be changed */
236 	int64_t	value;		/* inode or block number to be affected */
237 	int64_t	size;		/* amount or range to be adjusted */
238 	int64_t	spare;		/* reserved for future use */
239 };
240 
241 /*
242  * A recovery structure placed at the end of the boot block area by newfs
243  * that can be used by fsck to search for alternate superblocks.
244  */
245 struct fsrecovery {
246 	int32_t	fsr_magic;	/* magic number */
247 	int32_t	fsr_fsbtodb;	/* fsbtodb and dbtofsb shift constant */
248 	int32_t	fsr_sblkno;	/* offset of super-block in filesys */
249 	int32_t	fsr_fpg;	/* blocks per group * fs_frag */
250 	u_int32_t fsr_ncg;	/* number of cylinder groups */
251 };
252 
253 /*
254  * Per cylinder group information; summarized in blocks allocated
255  * from first cylinder group data blocks.  These blocks have to be
256  * read in from fs_csaddr (size fs_cssize) in addition to the
257  * super block.
258  */
259 struct csum {
260 	int32_t	cs_ndir;		/* number of directories */
261 	int32_t	cs_nbfree;		/* number of free blocks */
262 	int32_t	cs_nifree;		/* number of free inodes */
263 	int32_t	cs_nffree;		/* number of free frags */
264 };
265 struct csum_total {
266 	int64_t	cs_ndir;		/* number of directories */
267 	int64_t	cs_nbfree;		/* number of free blocks */
268 	int64_t	cs_nifree;		/* number of free inodes */
269 	int64_t	cs_nffree;		/* number of free frags */
270 	int64_t	cs_numclusters;		/* number of free clusters */
271 	int64_t	cs_spare[3];		/* future expansion */
272 };
273 
274 /*
275  * Pointers to super block summary information. Placed in a separate
276  * structure so there is just one pointer in the superblock.
277  *
278  * The pointers in this structure are used as follows:
279  *   fs_contigdirs references an array that tracks the creation of new
280  *	directories
281  *   fs_csp references a contiguous array of struct csum for
282  *	all cylinder groups
283  *   fs_maxcluster references an array of cluster sizes that is computed
284  *	as cylinder groups are inspected
285  *   fs_active is used when creating snapshots; it points to a bitmap
286  *	of cylinder groups for which the free-block bitmap has changed
287  *	since the snapshot operation began.
288  */
289 struct fs_summary_info {
290 	uint8_t	*si_contigdirs;		/* (u) # of contig. allocated dirs */
291 	struct	csum *si_csp;		/* (u) cg summary info buffer */
292 	int32_t	*si_maxcluster;		/* (u) max cluster in each cyl group */
293 	u_int	*si_active;		/* (u) used by snapshots to track fs */
294 };
295 #define fs_contigdirs	fs_si->si_contigdirs
296 #define fs_csp		fs_si->si_csp
297 #define fs_maxcluster	fs_si->si_maxcluster
298 #define fs_active	fs_si->si_active
299 
300 /*
301  * Super block for an FFS filesystem.
302  */
303 struct fs {
304 	int32_t	 fs_firstfield;		/* historic filesystem linked list, */
305 	int32_t	 fs_unused_1;		/*     used for incore super blocks */
306 	int32_t	 fs_sblkno;		/* offset of super-block in filesys */
307 	int32_t	 fs_cblkno;		/* offset of cyl-block in filesys */
308 	int32_t	 fs_iblkno;		/* offset of inode-blocks in filesys */
309 	int32_t	 fs_dblkno;		/* offset of first data after cg */
310 	int32_t	 fs_old_cgoffset;	/* cylinder group offset in cylinder */
311 	int32_t	 fs_old_cgmask;		/* used to calc mod fs_ntrak */
312 	int32_t  fs_old_time;		/* last time written */
313 	int32_t	 fs_old_size;		/* number of blocks in fs */
314 	int32_t	 fs_old_dsize;		/* number of data blocks in fs */
315 	u_int32_t fs_ncg;		/* number of cylinder groups */
316 	int32_t	 fs_bsize;		/* size of basic blocks in fs */
317 	int32_t	 fs_fsize;		/* size of frag blocks in fs */
318 	int32_t	 fs_frag;		/* number of frags in a block in fs */
319 /* these are configuration parameters */
320 	int32_t	 fs_minfree;		/* minimum percentage of free blocks */
321 	int32_t	 fs_old_rotdelay;	/* num of ms for optimal next block */
322 	int32_t	 fs_old_rps;		/* disk revolutions per second */
323 /* these fields can be computed from the others */
324 	int32_t	 fs_bmask;		/* ``blkoff'' calc of blk offsets */
325 	int32_t	 fs_fmask;		/* ``fragoff'' calc of frag offsets */
326 	int32_t	 fs_bshift;		/* ``lblkno'' calc of logical blkno */
327 	int32_t	 fs_fshift;		/* ``numfrags'' calc number of frags */
328 /* these are configuration parameters */
329 	int32_t	 fs_maxcontig;		/* max number of contiguous blks */
330 	int32_t	 fs_maxbpg;		/* max number of blks per cyl group */
331 /* these fields can be computed from the others */
332 	int32_t	 fs_fragshift;		/* block to frag shift */
333 	int32_t	 fs_fsbtodb;		/* fsbtodb and dbtofsb shift constant */
334 	int32_t	 fs_sbsize;		/* actual size of super block */
335 	int32_t	 fs_spare1[2];		/* old fs_csmask */
336 					/* old fs_csshift */
337 	int32_t	 fs_nindir;		/* value of NINDIR */
338 	u_int32_t fs_inopb;		/* value of INOPB */
339 	int32_t	 fs_old_nspf;		/* value of NSPF */
340 /* yet another configuration parameter */
341 	int32_t	 fs_optim;		/* optimization preference, see below */
342 	int32_t	 fs_old_npsect;		/* # sectors/track including spares */
343 	int32_t	 fs_old_interleave;	/* hardware sector interleave */
344 	int32_t	 fs_old_trackskew;	/* sector 0 skew, per track */
345 	int32_t	 fs_id[2];		/* unique filesystem id */
346 /* sizes determined by number of cylinder groups and their sizes */
347 	int32_t	 fs_old_csaddr;		/* blk addr of cyl grp summary area */
348 	int32_t	 fs_cssize;		/* size of cyl grp summary area */
349 	int32_t	 fs_cgsize;		/* cylinder group size */
350 	int32_t	 fs_spare2;		/* old fs_ntrak */
351 	int32_t	 fs_old_nsect;		/* sectors per track */
352 	int32_t  fs_old_spc;		/* sectors per cylinder */
353 	int32_t	 fs_old_ncyl;		/* cylinders in filesystem */
354 	int32_t	 fs_old_cpg;		/* cylinders per group */
355 	u_int32_t fs_ipg;		/* inodes per group */
356 	int32_t	 fs_fpg;		/* blocks per group * fs_frag */
357 /* this data must be re-computed after crashes */
358 	struct	csum fs_old_cstotal;	/* cylinder summary information */
359 /* these fields are cleared at mount time */
360 	int8_t   fs_fmod;		/* super block modified flag */
361 	int8_t   fs_clean;		/* filesystem is clean flag */
362 	int8_t 	 fs_ronly;		/* mounted read-only flag */
363 	int8_t   fs_old_flags;		/* old FS_ flags */
364 	u_char	 fs_fsmnt[MAXMNTLEN];	/* name mounted on */
365 	u_char	 fs_volname[MAXVOLLEN];	/* volume name */
366 	u_int64_t fs_swuid;		/* system-wide uid */
367 	int32_t  fs_pad;		/* due to alignment of fs_swuid */
368 /* these fields retain the current block allocation info */
369 	int32_t	 fs_cgrotor;		/* last cg searched */
370 	void 	*fs_ocsp[NOCSPTRS];	/* padding; was list of fs_cs buffers */
371 	struct	 fs_summary_info *fs_si;/* In-core pointer to summary info */
372 	int32_t	 fs_old_cpc;		/* cyl per cycle in postbl */
373 	int32_t	 fs_maxbsize;		/* maximum blocking factor permitted */
374 	int64_t	 fs_unrefs;		/* number of unreferenced inodes */
375 	int64_t  fs_providersize;	/* size of underlying GEOM provider */
376 	int64_t	 fs_metaspace;		/* size of area reserved for metadata */
377 	int64_t	 fs_sparecon64[13];	/* old rotation block list head */
378 	int64_t	 fs_sblockactualloc;	/* byte offset of this superblock */
379 	int64_t	 fs_sblockloc;		/* byte offset of standard superblock */
380 	struct	csum_total fs_cstotal;	/* (u) cylinder summary information */
381 	ufs_time_t fs_time;		/* last time written */
382 	int64_t	 fs_size;		/* number of blocks in fs */
383 	int64_t	 fs_dsize;		/* number of data blocks in fs */
384 	ufs2_daddr_t fs_csaddr;		/* blk addr of cyl grp summary area */
385 	int64_t	 fs_pendingblocks;	/* (u) blocks being freed */
386 	u_int32_t fs_pendinginodes;	/* (u) inodes being freed */
387 	uint32_t fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */
388 	u_int32_t fs_avgfilesize;	/* expected average file size */
389 	u_int32_t fs_avgfpdir;		/* expected # of files per directory */
390 	int32_t	 fs_save_cgsize;	/* save real cg size to use fs_bsize */
391 	ufs_time_t fs_mtime;		/* Last mount or fsck time. */
392 	int32_t  fs_sujfree;		/* SUJ free list */
393 	int32_t	 fs_sparecon32[21];	/* reserved for future constants */
394 	u_int32_t fs_ckhash;		/* if CK_SUPERBLOCK, its check-hash */
395 	u_int32_t fs_metackhash;	/* metadata check-hash, see CK_ below */
396 	int32_t  fs_flags;		/* see FS_ flags below */
397 	int32_t	 fs_contigsumsize;	/* size of cluster summary array */
398 	int32_t	 fs_maxsymlinklen;	/* max length of an internal symlink */
399 	int32_t	 fs_old_inodefmt;	/* format of on-disk inodes */
400 	u_int64_t fs_maxfilesize;	/* maximum representable file size */
401 	int64_t	 fs_qbmask;		/* ~fs_bmask for use with 64-bit size */
402 	int64_t	 fs_qfmask;		/* ~fs_fmask for use with 64-bit size */
403 	int32_t	 fs_state;		/* validate fs_clean field */
404 	int32_t	 fs_old_postblformat;	/* format of positional layout tables */
405 	int32_t	 fs_old_nrpos;		/* number of rotational positions */
406 	int32_t	 fs_spare5[2];		/* old fs_postbloff */
407 					/* old fs_rotbloff */
408 	int32_t	 fs_magic;		/* magic number */
409 };
410 
411 /* Sanity checking. */
412 #ifdef CTASSERT
413 CTASSERT(sizeof(struct fs) == 1376);
414 #endif
415 
416 /*
417  * Filesystem identification
418  */
419 #define	FS_UFS1_MAGIC	0x011954	/* UFS1 fast filesystem magic number */
420 #define	FS_UFS2_MAGIC	0x19540119	/* UFS2 fast filesystem magic number */
421 #define	FS_BAD_MAGIC	0x19960408	/* UFS incomplete newfs magic number */
422 #define	FS_42INODEFMT	-1		/* 4.2BSD inode format */
423 #define	FS_44INODEFMT	2		/* 4.4BSD inode format */
424 
425 /*
426  * Preference for optimization.
427  */
428 #define	FS_OPTTIME	0	/* minimize allocation time */
429 #define	FS_OPTSPACE	1	/* minimize disk fragmentation */
430 
431 /*
432  * Filesystem flags.
433  *
434  * The FS_UNCLEAN flag is set by the kernel when the filesystem was
435  * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates
436  * that the filesystem should be managed by the soft updates code.
437  * Note that the FS_NEEDSFSCK flag is set and cleared by the fsck
438  * utility. It is set when background fsck finds an unexpected
439  * inconsistency which requires a traditional foreground fsck to be
440  * run. Such inconsistencies should only be found after an uncorrectable
441  * disk error. The FS_NEEDSFSCK can also be set when a mounted filesystem
442  * discovers an internal inconsistency such as freeing a freed inode.
443  * A foreground fsck will clear the FS_NEEDSFSCK flag when it has
444  * successfully cleaned up the filesystem. The kernel uses this
445  * flag to enforce that inconsistent filesystems be mounted read-only.
446  *
447  * The FS_METACKHASH flag when set indicates that the kernel maintains
448  * one or more check hashes. The actual set of supported check hashes
449  * is stored in the fs_metackhash field. Kernels that do not support
450  * check hashes clear the FS_METACKHASH flag to indicate that the
451  * check hashes need to be rebuilt (by fsck) before they can be used.
452  *
453  * When a filesystem is mounted, any flags not included in FS_SUPPORTED
454  * are cleared. This lets newer features know that the filesystem has
455  * been run on an older version of the filesystem and thus that data
456  * structures associated with those features are out-of-date and need
457  * to be rebuilt.
458  *
459  * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled
460  * for the file system, so they should be loaded from extended attributes,
461  * observed for access control purposes, and be administered by object
462  * owners.  FS_NFS4ACLS indicates that NFSv4 ACLs are administratively
463  * enabled.  This flag is mutually exclusive with FS_ACLS.  FS_MULTILABEL
464  * indicates that the TrustedBSD MAC Framework should attempt to back MAC
465  * labels into extended attributes on the file system rather than maintain
466  * a single mount label for all objects.
467  */
468 #define	FS_UNCLEAN	0x00000001 /* filesystem not clean at mount */
469 #define	FS_DOSOFTDEP	0x00000002 /* filesystem using soft dependencies */
470 #define	FS_NEEDSFSCK	0x00000004 /* filesystem needs sync fsck before mount */
471 #define	FS_SUJ       	0x00000008 /* Filesystem using softupdate journal */
472 #define	FS_ACLS		0x00000010 /* file system has POSIX.1e ACLs enabled */
473 #define	FS_MULTILABEL	0x00000020 /* file system is MAC multi-label */
474 #define	FS_GJOURNAL	0x00000040 /* gjournaled file system */
475 #define	FS_FLAGS_UPDATED 0x0000080 /* flags have been moved to new location */
476 #define	FS_NFS4ACLS	0x00000100 /* file system has NFSv4 ACLs enabled */
477 #define	FS_METACKHASH	0x00000200 /* kernel supports metadata check hashes */
478 #define	FS_TRIM		0x00000400 /* issue BIO_DELETE for deleted blocks */
479 #define	FS_SUPPORTED	0x00FFFFFF /* supported flags, others cleared at mount*/
480 /*
481  * Things that we may someday support, but currently do not.
482  * These flags are all cleared so we know if we ran on a kernel
483  * that does not support them.
484  */
485 #define	FS_INDEXDIRS	0x01000000 /* kernel supports indexed directories */
486 #define	FS_VARBLKSIZE	0x02000000 /* kernel supports variable block sizes */
487 #define	FS_COOLOPT1	0x04000000 /* kernel supports cool option 1 */
488 #define	FS_COOLOPT2	0x08000000 /* kernel supports cool option 2 */
489 #define	FS_COOLOPT3	0x10000000 /* kernel supports cool option 3 */
490 #define	FS_COOLOPT4	0x20000000 /* kernel supports cool option 4 */
491 #define	FS_COOLOPT5	0x40000000 /* kernel supports cool option 5 */
492 #define	FS_COOLOPT6	0x80000000 /* kernel supports cool option 6 */
493 
494 /*
495  * The fs_metackhash field indicates the types of metadata check-hash
496  * that are maintained for a filesystem. Not all filesystems check-hash
497  * all metadata.
498  */
499 #define	CK_SUPERBLOCK	0x0001	/* the superblock */
500 #define	CK_CYLGRP	0x0002	/* the cylinder groups */
501 #define	CK_INODE	0x0004	/* inodes */
502 #define	CK_INDIR	0x0008	/* indirect blocks */
503 #define	CK_DIR		0x0010	/* directory contents */
504 #define	CK_SUPPORTED	0x0007	/* supported flags, others cleared at mount */
505 /*
506  * The BX_FSPRIV buffer b_xflags are used to track types of data in buffers.
507  */
508 #define	BX_SUPERBLOCK	0x00010000	/* superblock */
509 #define	BX_CYLGRP	0x00020000	/* cylinder groups */
510 #define	BX_INODE	0x00040000	/* inodes */
511 #define	BX_INDIR	0x00080000	/* indirect blocks */
512 #define	BX_DIR		0x00100000	/* directory contents */
513 
514 #define	PRINT_UFS_BUF_XFLAGS "\20\25dir\24indir\23inode\22cylgrp\21superblock"
515 
516 /*
517  * Macros to access bits in the fs_active array.
518  */
519 #define	ACTIVECGNUM(fs, cg)	((fs)->fs_active[(cg) / (NBBY * sizeof(int))])
520 #define	ACTIVECGOFF(cg)		(1 << ((cg) % (NBBY * sizeof(int))))
521 #define	ACTIVESET(fs, cg)	do {					\
522 	if ((fs)->fs_active)						\
523 		ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg));		\
524 } while (0)
525 #define	ACTIVECLEAR(fs, cg)	do {					\
526 	if ((fs)->fs_active)						\
527 		ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg));		\
528 } while (0)
529 
530 /*
531  * The size of a cylinder group is calculated by CGSIZE. The maximum size
532  * is limited by the fact that cylinder groups are at most one block.
533  * Its size is derived from the size of the maps maintained in the
534  * cylinder group and the (struct cg) size.
535  */
536 #define	CGSIZE(fs) \
537     /* base cg */	(sizeof(struct cg) + sizeof(int32_t) + \
538     /* old btotoff */	(fs)->fs_old_cpg * sizeof(int32_t) + \
539     /* old boff */	(fs)->fs_old_cpg * sizeof(u_int16_t) + \
540     /* inode map */	howmany((fs)->fs_ipg, NBBY) + \
541     /* block map */	howmany((fs)->fs_fpg, NBBY) +\
542     /* if present */	((fs)->fs_contigsumsize <= 0 ? 0 : \
543     /* cluster sum */	(fs)->fs_contigsumsize * sizeof(int32_t) + \
544     /* cluster map */	howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY)))
545 
546 /*
547  * The minimal number of cylinder groups that should be created.
548  */
549 #define	MINCYLGRPS	4
550 
551 /*
552  * Convert cylinder group to base address of its global summary info.
553  */
554 #define	fs_cs(fs, indx) fs_csp[indx]
555 
556 /*
557  * Cylinder group block for a filesystem.
558  */
559 #define	CG_MAGIC	0x090255
560 struct cg {
561 	int32_t	 cg_firstfield;		/* historic cyl groups linked list */
562 	int32_t	 cg_magic;		/* magic number */
563 	int32_t  cg_old_time;		/* time last written */
564 	u_int32_t cg_cgx;		/* we are the cgx'th cylinder group */
565 	int16_t	 cg_old_ncyl;		/* number of cyl's this cg */
566 	int16_t  cg_old_niblk;		/* number of inode blocks this cg */
567 	u_int32_t cg_ndblk;		/* number of data blocks this cg */
568 	struct	 csum cg_cs;		/* cylinder summary information */
569 	u_int32_t cg_rotor;		/* position of last used block */
570 	u_int32_t cg_frotor;		/* position of last used frag */
571 	u_int32_t cg_irotor;		/* position of last used inode */
572 	u_int32_t cg_frsum[MAXFRAG];	/* counts of available frags */
573 	int32_t	 cg_old_btotoff;	/* (int32) block totals per cylinder */
574 	int32_t	 cg_old_boff;		/* (u_int16) free block positions */
575 	u_int32_t cg_iusedoff;		/* (u_int8) used inode map */
576 	u_int32_t cg_freeoff;		/* (u_int8) free block map */
577 	u_int32_t cg_nextfreeoff;	/* (u_int8) next available space */
578 	u_int32_t cg_clustersumoff;	/* (u_int32) counts of avail clusters */
579 	u_int32_t cg_clusteroff;		/* (u_int8) free cluster map */
580 	u_int32_t cg_nclusterblks;	/* number of clusters this cg */
581 	u_int32_t cg_niblk;		/* number of inode blocks this cg */
582 	u_int32_t cg_initediblk;		/* last initialized inode */
583 	u_int32_t cg_unrefs;		/* number of unreferenced inodes */
584 	int32_t	 cg_sparecon32[1];	/* reserved for future use */
585 	u_int32_t cg_ckhash;		/* check-hash of this cg */
586 	ufs_time_t cg_time;		/* time last written */
587 	int64_t	 cg_sparecon64[3];	/* reserved for future use */
588 	u_int8_t cg_space[1];		/* space for cylinder group maps */
589 /* actually longer */
590 };
591 
592 /*
593  * Macros for access to cylinder group array structures
594  */
595 #define	cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC)
596 #define	cg_inosused(cgp) \
597     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_iusedoff))
598 #define	cg_blksfree(cgp) \
599     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_freeoff))
600 #define	cg_clustersfree(cgp) \
601     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_clusteroff))
602 #define	cg_clustersum(cgp) \
603     ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff))
604 
605 /*
606  * Turn filesystem block numbers into disk block addresses.
607  * This maps filesystem blocks to device size blocks.
608  */
609 #define	fsbtodb(fs, b)	((daddr_t)(b) << (fs)->fs_fsbtodb)
610 #define	dbtofsb(fs, b)	((b) >> (fs)->fs_fsbtodb)
611 
612 /*
613  * Cylinder group macros to locate things in cylinder groups.
614  * They calc filesystem addresses of cylinder group data structures.
615  */
616 #define	cgbase(fs, c)	(((ufs2_daddr_t)(fs)->fs_fpg) * (c))
617 #define	cgdata(fs, c)	(cgdmin(fs, c) + (fs)->fs_metaspace)	/* data zone */
618 #define	cgmeta(fs, c)	(cgdmin(fs, c))				/* meta data */
619 #define	cgdmin(fs, c)	(cgstart(fs, c) + (fs)->fs_dblkno)	/* 1st data */
620 #define	cgimin(fs, c)	(cgstart(fs, c) + (fs)->fs_iblkno)	/* inode blk */
621 #define	cgsblock(fs, c)	(cgstart(fs, c) + (fs)->fs_sblkno)	/* super blk */
622 #define	cgtod(fs, c)	(cgstart(fs, c) + (fs)->fs_cblkno)	/* cg block */
623 #define	cgstart(fs, c)							\
624        ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) :		\
625        (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask))))
626 
627 /*
628  * Macros for handling inode numbers:
629  *     inode number to filesystem block offset.
630  *     inode number to cylinder group number.
631  *     inode number to filesystem block address.
632  */
633 #define	ino_to_cg(fs, x)	(((ino_t)(x)) / (fs)->fs_ipg)
634 #define	ino_to_fsba(fs, x)						\
635 	((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) +		\
636 	    (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs))))))
637 #define	ino_to_fsbo(fs, x)	(((ino_t)(x)) % INOPB(fs))
638 
639 /*
640  * Give cylinder group number for a filesystem block.
641  * Give cylinder group block number for a filesystem block.
642  */
643 #define	dtog(fs, d)	((d) / (fs)->fs_fpg)
644 #define	dtogd(fs, d)	((d) % (fs)->fs_fpg)
645 
646 /*
647  * Extract the bits for a block from a map.
648  * Compute the cylinder and rotational position of a cyl block addr.
649  */
650 #define	blkmap(fs, map, loc) \
651     (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag)))
652 
653 /*
654  * The following macros optimize certain frequently calculated
655  * quantities by using shifts and masks in place of divisions
656  * modulos and multiplications.
657  */
658 #define	blkoff(fs, loc)		/* calculates (loc % fs->fs_bsize) */ \
659 	((loc) & (fs)->fs_qbmask)
660 #define	fragoff(fs, loc)	/* calculates (loc % fs->fs_fsize) */ \
661 	((loc) & (fs)->fs_qfmask)
662 #define	lfragtosize(fs, frag)	/* calculates ((off_t)frag * fs->fs_fsize) */ \
663 	(((off_t)(frag)) << (fs)->fs_fshift)
664 #define	lblktosize(fs, blk)	/* calculates ((off_t)blk * fs->fs_bsize) */ \
665 	(((off_t)(blk)) << (fs)->fs_bshift)
666 /* Use this only when `blk' is known to be small, e.g., < UFS_NDADDR. */
667 #define	smalllblktosize(fs, blk)    /* calculates (blk * fs->fs_bsize) */ \
668 	((blk) << (fs)->fs_bshift)
669 #define	lblkno(fs, loc)		/* calculates (loc / fs->fs_bsize) */ \
670 	((loc) >> (fs)->fs_bshift)
671 #define	numfrags(fs, loc)	/* calculates (loc / fs->fs_fsize) */ \
672 	((loc) >> (fs)->fs_fshift)
673 #define	blkroundup(fs, size)	/* calculates roundup(size, fs->fs_bsize) */ \
674 	(((size) + (fs)->fs_qbmask) & (fs)->fs_bmask)
675 #define	fragroundup(fs, size)	/* calculates roundup(size, fs->fs_fsize) */ \
676 	(((size) + (fs)->fs_qfmask) & (fs)->fs_fmask)
677 #define	fragstoblks(fs, frags)	/* calculates (frags / fs->fs_frag) */ \
678 	((frags) >> (fs)->fs_fragshift)
679 #define	blkstofrags(fs, blks)	/* calculates (blks * fs->fs_frag) */ \
680 	((blks) << (fs)->fs_fragshift)
681 #define	fragnum(fs, fsb)	/* calculates (fsb % fs->fs_frag) */ \
682 	((fsb) & ((fs)->fs_frag - 1))
683 #define	blknum(fs, fsb)		/* calculates rounddown(fsb, fs->fs_frag) */ \
684 	((fsb) &~ ((fs)->fs_frag - 1))
685 
686 /*
687  * Determine the number of available frags given a
688  * percentage to hold in reserve.
689  */
690 #define	freespace(fs, percentreserved) \
691 	(blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \
692 	(fs)->fs_cstotal.cs_nffree - \
693 	(((off_t)((fs)->fs_dsize)) * (percentreserved) / 100))
694 
695 /*
696  * Determining the size of a file block in the filesystem.
697  */
698 #define	blksize(fs, ip, lbn) \
699 	(((lbn) >= UFS_NDADDR || (ip)->i_size >= \
700 	    (uint64_t)smalllblktosize(fs, (lbn) + 1)) \
701 	    ? (fs)->fs_bsize \
702 	    : (fragroundup(fs, blkoff(fs, (ip)->i_size))))
703 #define	sblksize(fs, size, lbn) \
704 	(((lbn) >= UFS_NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \
705 	  ? (fs)->fs_bsize \
706 	  : (fragroundup(fs, blkoff(fs, (size)))))
707 
708 /*
709  * Number of indirects in a filesystem block.
710  */
711 #define	NINDIR(fs)	((fs)->fs_nindir)
712 
713 /*
714  * Indirect lbns are aligned on UFS_NDADDR addresses where single indirects
715  * are the negated address of the lowest lbn reachable, double indirects
716  * are this lbn - 1 and triple indirects are this lbn - 2.  This yields
717  * an unusual bit order to determine level.
718  */
719 static inline int
720 lbn_level(ufs_lbn_t lbn)
721 {
722 	if (lbn >= 0)
723 		return 0;
724 	switch (lbn & 0x3) {
725 	case 0:
726 		return (0);
727 	case 1:
728 		break;
729 	case 2:
730 		return (2);
731 	case 3:
732 		return (1);
733 	default:
734 		break;
735 	}
736 	return (-1);
737 }
738 
739 static inline ufs_lbn_t
740 lbn_offset(struct fs *fs, int level)
741 {
742 	ufs_lbn_t res;
743 
744 	for (res = 1; level > 0; level--)
745 		res *= NINDIR(fs);
746 	return (res);
747 }
748 
749 /*
750  * Number of inodes in a secondary storage block/fragment.
751  */
752 #define	INOPB(fs)	((fs)->fs_inopb)
753 #define	INOPF(fs)	((fs)->fs_inopb >> (fs)->fs_fragshift)
754 
755 /*
756  * Softdep journal record format.
757  */
758 
759 #define	JOP_ADDREF	1	/* Add a reference to an inode. */
760 #define	JOP_REMREF	2	/* Remove a reference from an inode. */
761 #define	JOP_NEWBLK	3	/* Allocate a block. */
762 #define	JOP_FREEBLK	4	/* Free a block or a tree of blocks. */
763 #define	JOP_MVREF	5	/* Move a reference from one off to another. */
764 #define	JOP_TRUNC	6	/* Partial truncation record. */
765 #define	JOP_SYNC	7	/* fsync() complete record. */
766 
767 #define	JREC_SIZE	32	/* Record and segment header size. */
768 
769 #define	SUJ_MIN		(4 * 1024 * 1024)	/* Minimum journal size */
770 #define	SUJ_MAX		(32 * 1024 * 1024)	/* Maximum journal size */
771 #define	SUJ_FILE	".sujournal"		/* Journal file name */
772 
773 /*
774  * Size of the segment record header.  There is at most one for each disk
775  * block in the journal.  The segment header is followed by an array of
776  * records.  fsck depends on the first element in each record being 'op'
777  * and the second being 'ino'.  Segments may span multiple disk blocks but
778  * the header is present on each.
779  */
780 struct jsegrec {
781 	uint64_t	jsr_seq;	/* Our sequence number */
782 	uint64_t	jsr_oldest;	/* Oldest valid sequence number */
783 	uint16_t	jsr_cnt;	/* Count of valid records */
784 	uint16_t	jsr_blocks;	/* Count of device bsize blocks. */
785 	uint32_t	jsr_crc;	/* 32bit crc of the valid space */
786 	ufs_time_t	jsr_time;	/* timestamp for mount instance */
787 };
788 
789 /*
790  * Reference record.  Records a single link count modification.
791  */
792 struct jrefrec {
793 	uint32_t	jr_op;
794 	uint32_t	jr_ino;
795 	uint32_t	jr_parent;
796 	uint16_t	jr_nlink;
797 	uint16_t	jr_mode;
798 	int64_t		jr_diroff;
799 	uint64_t	jr_unused;
800 };
801 
802 /*
803  * Move record.  Records a reference moving within a directory block.  The
804  * nlink is unchanged but we must search both locations.
805  */
806 struct jmvrec {
807 	uint32_t	jm_op;
808 	uint32_t	jm_ino;
809 	uint32_t	jm_parent;
810 	uint16_t	jm_unused;
811 	int64_t		jm_oldoff;
812 	int64_t		jm_newoff;
813 };
814 
815 /*
816  * Block record.  A set of frags or tree of blocks starting at an indirect are
817  * freed or a set of frags are allocated.
818  */
819 struct jblkrec {
820 	uint32_t	jb_op;
821 	uint32_t	jb_ino;
822 	ufs2_daddr_t	jb_blkno;
823 	ufs_lbn_t	jb_lbn;
824 	uint16_t	jb_frags;
825 	uint16_t	jb_oldfrags;
826 	uint32_t	jb_unused;
827 };
828 
829 /*
830  * Truncation record.  Records a partial truncation so that it may be
831  * completed at check time.  Also used for sync records.
832  */
833 struct jtrncrec {
834 	uint32_t	jt_op;
835 	uint32_t	jt_ino;
836 	int64_t		jt_size;
837 	uint32_t	jt_extsize;
838 	uint32_t	jt_pad[3];
839 };
840 
841 union jrec {
842 	struct jsegrec	rec_jsegrec;
843 	struct jrefrec	rec_jrefrec;
844 	struct jmvrec	rec_jmvrec;
845 	struct jblkrec	rec_jblkrec;
846 	struct jtrncrec	rec_jtrncrec;
847 };
848 
849 #ifdef CTASSERT
850 CTASSERT(sizeof(struct jsegrec) == JREC_SIZE);
851 CTASSERT(sizeof(struct jrefrec) == JREC_SIZE);
852 CTASSERT(sizeof(struct jmvrec) == JREC_SIZE);
853 CTASSERT(sizeof(struct jblkrec) == JREC_SIZE);
854 CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE);
855 CTASSERT(sizeof(union jrec) == JREC_SIZE);
856 #endif
857 
858 extern int inside[], around[];
859 extern u_char *fragtbl[];
860 
861 /*
862  * IOCTLs used for filesystem write suspension.
863  */
864 #define	UFSSUSPEND	_IOW('U', 1, fsid_t)
865 #define	UFSRESUME	_IO('U', 2)
866 
867 #endif
868