xref: /freebsd/sys/ufs/ffs/fs.h (revision a02aba5f3c73d7ed377f88327fedd11f70f23353)
1 /*-
2  * Copyright (c) 1982, 1986, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 4. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)fs.h	8.13 (Berkeley) 3/21/95
30  * $FreeBSD$
31  */
32 
33 #ifndef _UFS_FFS_FS_H_
34 #define _UFS_FFS_FS_H_
35 
36 /*
37  * Each disk drive contains some number of filesystems.
38  * A filesystem consists of a number of cylinder groups.
39  * Each cylinder group has inodes and data.
40  *
41  * A filesystem is described by its super-block, which in turn
42  * describes the cylinder groups.  The super-block is critical
43  * data and is replicated in each cylinder group to protect against
44  * catastrophic loss.  This is done at `newfs' time and the critical
45  * super-block data does not change, so the copies need not be
46  * referenced further unless disaster strikes.
47  *
48  * For filesystem fs, the offsets of the various blocks of interest
49  * are given in the super block as:
50  *	[fs->fs_sblkno]		Super-block
51  *	[fs->fs_cblkno]		Cylinder group block
52  *	[fs->fs_iblkno]		Inode blocks
53  *	[fs->fs_dblkno]		Data blocks
54  * The beginning of cylinder group cg in fs, is given by
55  * the ``cgbase(fs, cg)'' macro.
56  *
57  * Depending on the architecture and the media, the superblock may
58  * reside in any one of four places. For tiny media where every block
59  * counts, it is placed at the very front of the partition. Historically,
60  * UFS1 placed it 8K from the front to leave room for the disk label and
61  * a small bootstrap. For UFS2 it got moved to 64K from the front to leave
62  * room for the disk label and a bigger bootstrap, and for really piggy
63  * systems we check at 256K from the front if the first three fail. In
64  * all cases the size of the superblock will be SBLOCKSIZE. All values are
65  * given in byte-offset form, so they do not imply a sector size. The
66  * SBLOCKSEARCH specifies the order in which the locations should be searched.
67  */
68 #define SBLOCK_FLOPPY	     0
69 #define SBLOCK_UFS1	  8192
70 #define SBLOCK_UFS2	 65536
71 #define SBLOCK_PIGGY	262144
72 #define SBLOCKSIZE	  8192
73 #define SBLOCKSEARCH \
74 	{ SBLOCK_UFS2, SBLOCK_UFS1, SBLOCK_FLOPPY, SBLOCK_PIGGY, -1 }
75 
76 /*
77  * Max number of fragments per block. This value is NOT tweakable.
78  */
79 #define MAXFRAG 	8
80 
81 /*
82  * Addresses stored in inodes are capable of addressing fragments
83  * of `blocks'. File system blocks of at most size MAXBSIZE can
84  * be optionally broken into 2, 4, or 8 pieces, each of which is
85  * addressable; these pieces may be DEV_BSIZE, or some multiple of
86  * a DEV_BSIZE unit.
87  *
88  * Large files consist of exclusively large data blocks.  To avoid
89  * undue wasted disk space, the last data block of a small file may be
90  * allocated as only as many fragments of a large block as are
91  * necessary.  The filesystem format retains only a single pointer
92  * to such a fragment, which is a piece of a single large block that
93  * has been divided.  The size of such a fragment is determinable from
94  * information in the inode, using the ``blksize(fs, ip, lbn)'' macro.
95  *
96  * The filesystem records space availability at the fragment level;
97  * to determine block availability, aligned fragments are examined.
98  */
99 
100 /*
101  * MINBSIZE is the smallest allowable block size.
102  * In order to insure that it is possible to create files of size
103  * 2^32 with only two levels of indirection, MINBSIZE is set to 4096.
104  * MINBSIZE must be big enough to hold a cylinder group block,
105  * thus changes to (struct cg) must keep its size within MINBSIZE.
106  * Note that super blocks are always of size SBLOCKSIZE,
107  * and that both SBLOCKSIZE and MAXBSIZE must be >= MINBSIZE.
108  */
109 #define MINBSIZE	4096
110 
111 /*
112  * The path name on which the filesystem is mounted is maintained
113  * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in
114  * the super block for this name.
115  */
116 #define MAXMNTLEN	468
117 
118 /*
119  * The volume name for this filesystem is maintained in fs_volname.
120  * MAXVOLLEN defines the length of the buffer allocated.
121  */
122 #define MAXVOLLEN	32
123 
124 /*
125  * There is a 128-byte region in the superblock reserved for in-core
126  * pointers to summary information. Originally this included an array
127  * of pointers to blocks of struct csum; now there are just a few
128  * pointers and the remaining space is padded with fs_ocsp[].
129  *
130  * NOCSPTRS determines the size of this padding. One pointer (fs_csp)
131  * is taken away to point to a contiguous array of struct csum for
132  * all cylinder groups; a second (fs_maxcluster) points to an array
133  * of cluster sizes that is computed as cylinder groups are inspected,
134  * and the third points to an array that tracks the creation of new
135  * directories. A fourth pointer, fs_active, is used when creating
136  * snapshots; it points to a bitmap of cylinder groups for which the
137  * free-block bitmap has changed since the snapshot operation began.
138  */
139 #define	NOCSPTRS	((128 / sizeof(void *)) - 4)
140 
141 /*
142  * A summary of contiguous blocks of various sizes is maintained
143  * in each cylinder group. Normally this is set by the initial
144  * value of fs_maxcontig. To conserve space, a maximum summary size
145  * is set by FS_MAXCONTIG.
146  */
147 #define FS_MAXCONTIG	16
148 
149 /*
150  * MINFREE gives the minimum acceptable percentage of filesystem
151  * blocks which may be free. If the freelist drops below this level
152  * only the superuser may continue to allocate blocks. This may
153  * be set to 0 if no reserve of free blocks is deemed necessary,
154  * however throughput drops by fifty percent if the filesystem
155  * is run at between 95% and 100% full; thus the minimum default
156  * value of fs_minfree is 5%. However, to get good clustering
157  * performance, 10% is a better choice. hence we use 10% as our
158  * default value. With 10% free space, fragmentation is not a
159  * problem, so we choose to optimize for time.
160  */
161 #define MINFREE		8
162 #define DEFAULTOPT	FS_OPTTIME
163 
164 /*
165  * Grigoriy Orlov <gluk@ptci.ru> has done some extensive work to fine
166  * tune the layout preferences for directories within a filesystem.
167  * His algorithm can be tuned by adjusting the following parameters
168  * which tell the system the average file size and the average number
169  * of files per directory. These defaults are well selected for typical
170  * filesystems, but may need to be tuned for odd cases like filesystems
171  * being used for squid caches or news spools.
172  */
173 #define AVFILESIZ	16384	/* expected average file size */
174 #define AFPDIR		64	/* expected number of files per directory */
175 
176 /*
177  * The maximum number of snapshot nodes that can be associated
178  * with each filesystem. This limit affects only the number of
179  * snapshot files that can be recorded within the superblock so
180  * that they can be found when the filesystem is mounted. However,
181  * maintaining too many will slow the filesystem performance, so
182  * having this limit is a good idea.
183  */
184 #define FSMAXSNAP 20
185 
186 /*
187  * Used to identify special blocks in snapshots:
188  *
189  * BLK_NOCOPY - A block that was unallocated at the time the snapshot
190  *	was taken, hence does not need to be copied when written.
191  * BLK_SNAP - A block held by another snapshot that is not needed by this
192  *	snapshot. When the other snapshot is freed, the BLK_SNAP entries
193  *	are converted to BLK_NOCOPY. These are needed to allow fsck to
194  *	identify blocks that are in use by other snapshots (which are
195  *	expunged from this snapshot).
196  */
197 #define BLK_NOCOPY ((ufs2_daddr_t)(1))
198 #define BLK_SNAP ((ufs2_daddr_t)(2))
199 
200 /*
201  * Sysctl values for the fast filesystem.
202  */
203 #define	FFS_ADJ_REFCNT		 1	/* adjust inode reference count */
204 #define	FFS_ADJ_BLKCNT		 2	/* adjust inode used block count */
205 #define	FFS_BLK_FREE		 3	/* free range of blocks in map */
206 #define	FFS_DIR_FREE		 4	/* free specified dir inodes in map */
207 #define	FFS_FILE_FREE		 5	/* free specified file inodes in map */
208 #define	FFS_SET_FLAGS		 6	/* set filesystem flags */
209 #define	FFS_ADJ_NDIR		 7	/* adjust number of directories */
210 #define	FFS_ADJ_NBFREE		 8	/* adjust number of free blocks */
211 #define	FFS_ADJ_NIFREE		 9	/* adjust number of free inodes */
212 #define	FFS_ADJ_NFFREE		10 	/* adjust number of free frags */
213 #define	FFS_ADJ_NUMCLUSTERS	11	/* adjust number of free clusters */
214 #define FFS_SET_CWD		12	/* set current directory */
215 #define	FFS_SET_DOTDOT		13	/* set inode number for ".." */
216 #define	FFS_UNLINK		14	/* remove a name in the filesystem */
217 #define	FFS_SET_INODE		15	/* update an on-disk inode */
218 #define	FFS_SET_BUFOUTPUT	16	/* set buffered writing on descriptor */
219 #define	FFS_MAXID		16	/* number of valid ffs ids */
220 
221 /*
222  * Command structure passed in to the filesystem to adjust filesystem values.
223  */
224 #define	FFS_CMD_VERSION		0x19790518	/* version ID */
225 struct fsck_cmd {
226 	int32_t	version;	/* version of command structure */
227 	int32_t	handle;		/* reference to filesystem to be changed */
228 	int64_t	value;		/* inode or block number to be affected */
229 	int64_t	size;		/* amount or range to be adjusted */
230 	int64_t	spare;		/* reserved for future use */
231 };
232 
233 /*
234  * Per cylinder group information; summarized in blocks allocated
235  * from first cylinder group data blocks.  These blocks have to be
236  * read in from fs_csaddr (size fs_cssize) in addition to the
237  * super block.
238  */
239 struct csum {
240 	int32_t	cs_ndir;		/* number of directories */
241 	int32_t	cs_nbfree;		/* number of free blocks */
242 	int32_t	cs_nifree;		/* number of free inodes */
243 	int32_t	cs_nffree;		/* number of free frags */
244 };
245 struct csum_total {
246 	int64_t	cs_ndir;		/* number of directories */
247 	int64_t	cs_nbfree;		/* number of free blocks */
248 	int64_t	cs_nifree;		/* number of free inodes */
249 	int64_t	cs_nffree;		/* number of free frags */
250 	int64_t	cs_numclusters;		/* number of free clusters */
251 	int64_t	cs_spare[3];		/* future expansion */
252 };
253 
254 /*
255  * Super block for an FFS filesystem.
256  */
257 struct fs {
258 	int32_t	 fs_firstfield;		/* historic filesystem linked list, */
259 	int32_t	 fs_unused_1;		/*     used for incore super blocks */
260 	int32_t	 fs_sblkno;		/* offset of super-block in filesys */
261 	int32_t	 fs_cblkno;		/* offset of cyl-block in filesys */
262 	int32_t	 fs_iblkno;		/* offset of inode-blocks in filesys */
263 	int32_t	 fs_dblkno;		/* offset of first data after cg */
264 	int32_t	 fs_old_cgoffset;	/* cylinder group offset in cylinder */
265 	int32_t	 fs_old_cgmask;		/* used to calc mod fs_ntrak */
266 	int32_t  fs_old_time;		/* last time written */
267 	int32_t	 fs_old_size;		/* number of blocks in fs */
268 	int32_t	 fs_old_dsize;		/* number of data blocks in fs */
269 	u_int32_t fs_ncg;		/* number of cylinder groups */
270 	int32_t	 fs_bsize;		/* size of basic blocks in fs */
271 	int32_t	 fs_fsize;		/* size of frag blocks in fs */
272 	int32_t	 fs_frag;		/* number of frags in a block in fs */
273 /* these are configuration parameters */
274 	int32_t	 fs_minfree;		/* minimum percentage of free blocks */
275 	int32_t	 fs_old_rotdelay;	/* num of ms for optimal next block */
276 	int32_t	 fs_old_rps;		/* disk revolutions per second */
277 /* these fields can be computed from the others */
278 	int32_t	 fs_bmask;		/* ``blkoff'' calc of blk offsets */
279 	int32_t	 fs_fmask;		/* ``fragoff'' calc of frag offsets */
280 	int32_t	 fs_bshift;		/* ``lblkno'' calc of logical blkno */
281 	int32_t	 fs_fshift;		/* ``numfrags'' calc number of frags */
282 /* these are configuration parameters */
283 	int32_t	 fs_maxcontig;		/* max number of contiguous blks */
284 	int32_t	 fs_maxbpg;		/* max number of blks per cyl group */
285 /* these fields can be computed from the others */
286 	int32_t	 fs_fragshift;		/* block to frag shift */
287 	int32_t	 fs_fsbtodb;		/* fsbtodb and dbtofsb shift constant */
288 	int32_t	 fs_sbsize;		/* actual size of super block */
289 	int32_t	 fs_spare1[2];		/* old fs_csmask */
290 					/* old fs_csshift */
291 	int32_t	 fs_nindir;		/* value of NINDIR */
292 	u_int32_t fs_inopb;		/* value of INOPB */
293 	int32_t	 fs_old_nspf;		/* value of NSPF */
294 /* yet another configuration parameter */
295 	int32_t	 fs_optim;		/* optimization preference, see below */
296 	int32_t	 fs_old_npsect;		/* # sectors/track including spares */
297 	int32_t	 fs_old_interleave;	/* hardware sector interleave */
298 	int32_t	 fs_old_trackskew;	/* sector 0 skew, per track */
299 	int32_t	 fs_id[2];		/* unique filesystem id */
300 /* sizes determined by number of cylinder groups and their sizes */
301 	int32_t	 fs_old_csaddr;		/* blk addr of cyl grp summary area */
302 	int32_t	 fs_cssize;		/* size of cyl grp summary area */
303 	int32_t	 fs_cgsize;		/* cylinder group size */
304 	int32_t	 fs_spare2;		/* old fs_ntrak */
305 	int32_t	 fs_old_nsect;		/* sectors per track */
306 	int32_t  fs_old_spc;		/* sectors per cylinder */
307 	int32_t	 fs_old_ncyl;		/* cylinders in filesystem */
308 	int32_t	 fs_old_cpg;		/* cylinders per group */
309 	u_int32_t fs_ipg;		/* inodes per group */
310 	int32_t	 fs_fpg;		/* blocks per group * fs_frag */
311 /* this data must be re-computed after crashes */
312 	struct	csum fs_old_cstotal;	/* cylinder summary information */
313 /* these fields are cleared at mount time */
314 	int8_t   fs_fmod;		/* super block modified flag */
315 	int8_t   fs_clean;		/* filesystem is clean flag */
316 	int8_t 	 fs_ronly;		/* mounted read-only flag */
317 	int8_t   fs_old_flags;		/* old FS_ flags */
318 	u_char	 fs_fsmnt[MAXMNTLEN];	/* name mounted on */
319 	u_char	 fs_volname[MAXVOLLEN];	/* volume name */
320 	u_int64_t fs_swuid;		/* system-wide uid */
321 	int32_t  fs_pad;		/* due to alignment of fs_swuid */
322 /* these fields retain the current block allocation info */
323 	int32_t	 fs_cgrotor;		/* last cg searched */
324 	void 	*fs_ocsp[NOCSPTRS];	/* padding; was list of fs_cs buffers */
325 	u_int8_t *fs_contigdirs;	/* (u) # of contig. allocated dirs */
326 	struct	csum *fs_csp;		/* (u) cg summary info buffer */
327 	int32_t	*fs_maxcluster;		/* (u) max cluster in each cyl group */
328 	u_int	*fs_active;		/* (u) used by snapshots to track fs */
329 	int32_t	 fs_old_cpc;		/* cyl per cycle in postbl */
330 	int32_t	 fs_maxbsize;		/* maximum blocking factor permitted */
331 	int64_t	 fs_unrefs;		/* number of unreferenced inodes */
332 	int64_t	 fs_sparecon64[16];	/* old rotation block list head */
333 	int64_t	 fs_sblockloc;		/* byte offset of standard superblock */
334 	struct	csum_total fs_cstotal;	/* (u) cylinder summary information */
335 	ufs_time_t fs_time;		/* last time written */
336 	int64_t	 fs_size;		/* number of blocks in fs */
337 	int64_t	 fs_dsize;		/* number of data blocks in fs */
338 	ufs2_daddr_t fs_csaddr;		/* blk addr of cyl grp summary area */
339 	int64_t	 fs_pendingblocks;	/* (u) blocks being freed */
340 	u_int32_t fs_pendinginodes;	/* (u) inodes being freed */
341 	ino_t	 fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */
342 	u_int32_t fs_avgfilesize;	/* expected average file size */
343 	u_int32_t fs_avgfpdir;		/* expected # of files per directory */
344 	int32_t	 fs_save_cgsize;	/* save real cg size to use fs_bsize */
345 	ufs_time_t fs_mtime;		/* Last mount or fsck time. */
346 	int32_t  fs_sujfree;		/* SUJ free list */
347 	int32_t	 fs_sparecon32[23];	/* reserved for future constants */
348 	int32_t  fs_flags;		/* see FS_ flags below */
349 	int32_t	 fs_contigsumsize;	/* size of cluster summary array */
350 	int32_t	 fs_maxsymlinklen;	/* max length of an internal symlink */
351 	int32_t	 fs_old_inodefmt;	/* format of on-disk inodes */
352 	u_int64_t fs_maxfilesize;	/* maximum representable file size */
353 	int64_t	 fs_qbmask;		/* ~fs_bmask for use with 64-bit size */
354 	int64_t	 fs_qfmask;		/* ~fs_fmask for use with 64-bit size */
355 	int32_t	 fs_state;		/* validate fs_clean field */
356 	int32_t	 fs_old_postblformat;	/* format of positional layout tables */
357 	int32_t	 fs_old_nrpos;		/* number of rotational positions */
358 	int32_t	 fs_spare5[2];		/* old fs_postbloff */
359 					/* old fs_rotbloff */
360 	int32_t	 fs_magic;		/* magic number */
361 };
362 
363 /* Sanity checking. */
364 #ifdef CTASSERT
365 CTASSERT(sizeof(struct fs) == 1376);
366 #endif
367 
368 /*
369  * Filesystem identification
370  */
371 #define	FS_UFS1_MAGIC	0x011954	/* UFS1 fast filesystem magic number */
372 #define	FS_UFS2_MAGIC	0x19540119	/* UFS2 fast filesystem magic number */
373 #define	FS_BAD_MAGIC	0x19960408	/* UFS incomplete newfs magic number */
374 #define	FS_OKAY		0x7c269d38	/* superblock checksum */
375 #define FS_42INODEFMT	-1		/* 4.2BSD inode format */
376 #define FS_44INODEFMT	2		/* 4.4BSD inode format */
377 
378 /*
379  * Preference for optimization.
380  */
381 #define FS_OPTTIME	0	/* minimize allocation time */
382 #define FS_OPTSPACE	1	/* minimize disk fragmentation */
383 
384 /*
385  * Filesystem flags.
386  *
387  * The FS_UNCLEAN flag is set by the kernel when the filesystem was
388  * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates
389  * that the filesystem should be managed by the soft updates code.
390  * Note that the FS_NEEDSFSCK flag is set and cleared only by the
391  * fsck utility. It is set when background fsck finds an unexpected
392  * inconsistency which requires a traditional foreground fsck to be
393  * run. Such inconsistencies should only be found after an uncorrectable
394  * disk error. A foreground fsck will clear the FS_NEEDSFSCK flag when
395  * it has successfully cleaned up the filesystem. The kernel uses this
396  * flag to enforce that inconsistent filesystems be mounted read-only.
397  * The FS_INDEXDIRS flag when set indicates that the kernel maintains
398  * on-disk auxiliary indexes (such as B-trees) for speeding directory
399  * accesses. Kernels that do not support auxiliary indicies clear the
400  * flag to indicate that the indicies need to be rebuilt (by fsck) before
401  * they can be used.
402  *
403  * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled
404  * for the file system, so they should be loaded from extended attributes,
405  * observed for access control purposes, and be administered by object
406  * owners.  FS_NFS4ACLS indicates that NFSv4 ACLs are administratively
407  * enabled.  This flag is mutually exclusive with FS_ACLS.  FS_MULTILABEL
408  * indicates that the TrustedBSD MAC Framework should attempt to back MAC
409  * labels into extended attributes on the file system rather than maintain
410  * a single mount label for all objects.
411  */
412 #define FS_UNCLEAN	0x0001	/* filesystem not clean at mount */
413 #define FS_DOSOFTDEP	0x0002	/* filesystem using soft dependencies */
414 #define FS_NEEDSFSCK	0x0004	/* filesystem needs sync fsck before mount */
415 #define	FS_SUJ       	0x0008	/* Filesystem using softupdate journal */
416 #define FS_ACLS		0x0010	/* file system has POSIX.1e ACLs enabled */
417 #define FS_MULTILABEL	0x0020	/* file system is MAC multi-label */
418 #define FS_GJOURNAL	0x0040	/* gjournaled file system */
419 #define FS_FLAGS_UPDATED 0x0080	/* flags have been moved to new location */
420 #define FS_NFS4ACLS	0x0100	/* file system has NFSv4 ACLs enabled */
421 #define FS_INDEXDIRS	0x0200	/* kernel supports indexed directories */
422 #define	FS_TRIM		0x0400	/* issue BIO_DELETE for deleted blocks */
423 
424 /*
425  * Macros to access bits in the fs_active array.
426  */
427 #define	ACTIVECGNUM(fs, cg)	((fs)->fs_active[(cg) / (NBBY * sizeof(int))])
428 #define	ACTIVECGOFF(cg)		(1 << ((cg) % (NBBY * sizeof(int))))
429 #define	ACTIVESET(fs, cg)	do {					\
430 	if ((fs)->fs_active)						\
431 		ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg));		\
432 } while (0)
433 #define	ACTIVECLEAR(fs, cg)	do {					\
434 	if ((fs)->fs_active)						\
435 		ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg));		\
436 } while (0)
437 
438 /*
439  * The size of a cylinder group is calculated by CGSIZE. The maximum size
440  * is limited by the fact that cylinder groups are at most one block.
441  * Its size is derived from the size of the maps maintained in the
442  * cylinder group and the (struct cg) size.
443  */
444 #define CGSIZE(fs) \
445     /* base cg */	(sizeof(struct cg) + sizeof(int32_t) + \
446     /* old btotoff */	(fs)->fs_old_cpg * sizeof(int32_t) + \
447     /* old boff */	(fs)->fs_old_cpg * sizeof(u_int16_t) + \
448     /* inode map */	howmany((fs)->fs_ipg, NBBY) + \
449     /* block map */	howmany((fs)->fs_fpg, NBBY) +\
450     /* if present */	((fs)->fs_contigsumsize <= 0 ? 0 : \
451     /* cluster sum */	(fs)->fs_contigsumsize * sizeof(int32_t) + \
452     /* cluster map */	howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY)))
453 
454 /*
455  * The minimal number of cylinder groups that should be created.
456  */
457 #define MINCYLGRPS	4
458 
459 /*
460  * Convert cylinder group to base address of its global summary info.
461  */
462 #define fs_cs(fs, indx) fs_csp[indx]
463 
464 /*
465  * Cylinder group block for a filesystem.
466  */
467 #define	CG_MAGIC	0x090255
468 struct cg {
469 	int32_t	 cg_firstfield;		/* historic cyl groups linked list */
470 	int32_t	 cg_magic;		/* magic number */
471 	int32_t  cg_old_time;		/* time last written */
472 	u_int32_t cg_cgx;		/* we are the cgx'th cylinder group */
473 	int16_t	 cg_old_ncyl;		/* number of cyl's this cg */
474 	int16_t  cg_old_niblk;		/* number of inode blocks this cg */
475 	u_int32_t cg_ndblk;		/* number of data blocks this cg */
476 	struct	 csum cg_cs;		/* cylinder summary information */
477 	u_int32_t cg_rotor;		/* position of last used block */
478 	u_int32_t cg_frotor;		/* position of last used frag */
479 	u_int32_t cg_irotor;		/* position of last used inode */
480 	u_int32_t cg_frsum[MAXFRAG];	/* counts of available frags */
481 	int32_t	 cg_old_btotoff;	/* (int32) block totals per cylinder */
482 	int32_t	 cg_old_boff;		/* (u_int16) free block positions */
483 	u_int32_t cg_iusedoff;		/* (u_int8) used inode map */
484 	u_int32_t cg_freeoff;		/* (u_int8) free block map */
485 	u_int32_t cg_nextfreeoff;	/* (u_int8) next available space */
486 	u_int32_t cg_clustersumoff;	/* (u_int32) counts of avail clusters */
487 	u_int32_t cg_clusteroff;		/* (u_int8) free cluster map */
488 	u_int32_t cg_nclusterblks;	/* number of clusters this cg */
489 	u_int32_t cg_niblk;		/* number of inode blocks this cg */
490 	u_int32_t cg_initediblk;		/* last initialized inode */
491 	u_int32_t cg_unrefs;		/* number of unreferenced inodes */
492 	int32_t	 cg_sparecon32[2];	/* reserved for future use */
493 	ufs_time_t cg_time;		/* time last written */
494 	int64_t	 cg_sparecon64[3];	/* reserved for future use */
495 	u_int8_t cg_space[1];		/* space for cylinder group maps */
496 /* actually longer */
497 };
498 
499 /*
500  * Macros for access to cylinder group array structures
501  */
502 #define cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC)
503 #define cg_inosused(cgp) \
504     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_iusedoff))
505 #define cg_blksfree(cgp) \
506     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_freeoff))
507 #define cg_clustersfree(cgp) \
508     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_clusteroff))
509 #define cg_clustersum(cgp) \
510     ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff))
511 
512 /*
513  * Turn filesystem block numbers into disk block addresses.
514  * This maps filesystem blocks to device size blocks.
515  */
516 #define	fsbtodb(fs, b)	((daddr_t)(b) << (fs)->fs_fsbtodb)
517 #define	dbtofsb(fs, b)	((b) >> (fs)->fs_fsbtodb)
518 
519 /*
520  * Cylinder group macros to locate things in cylinder groups.
521  * They calc filesystem addresses of cylinder group data structures.
522  */
523 #define	cgbase(fs, c)	(((ufs2_daddr_t)(fs)->fs_fpg) * (c))
524 #define	cgdmin(fs, c)	(cgstart(fs, c) + (fs)->fs_dblkno)	/* 1st data */
525 #define	cgimin(fs, c)	(cgstart(fs, c) + (fs)->fs_iblkno)	/* inode blk */
526 #define	cgsblock(fs, c)	(cgstart(fs, c) + (fs)->fs_sblkno)	/* super blk */
527 #define	cgtod(fs, c)	(cgstart(fs, c) + (fs)->fs_cblkno)	/* cg block */
528 #define cgstart(fs, c)							\
529        ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) :		\
530        (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask))))
531 
532 /*
533  * Macros for handling inode numbers:
534  *     inode number to filesystem block offset.
535  *     inode number to cylinder group number.
536  *     inode number to filesystem block address.
537  */
538 #define	ino_to_cg(fs, x)	(((ino_t)(x)) / (fs)->fs_ipg)
539 #define	ino_to_fsba(fs, x)						\
540 	((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) +		\
541 	    (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs))))))
542 #define	ino_to_fsbo(fs, x)	(((ino_t)(x)) % INOPB(fs))
543 
544 /*
545  * Give cylinder group number for a filesystem block.
546  * Give cylinder group block number for a filesystem block.
547  */
548 #define	dtog(fs, d)	((d) / (fs)->fs_fpg)
549 #define	dtogd(fs, d)	((d) % (fs)->fs_fpg)
550 
551 /*
552  * Extract the bits for a block from a map.
553  * Compute the cylinder and rotational position of a cyl block addr.
554  */
555 #define blkmap(fs, map, loc) \
556     (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag)))
557 
558 /*
559  * The following macros optimize certain frequently calculated
560  * quantities by using shifts and masks in place of divisions
561  * modulos and multiplications.
562  */
563 #define blkoff(fs, loc)		/* calculates (loc % fs->fs_bsize) */ \
564 	((loc) & (fs)->fs_qbmask)
565 #define fragoff(fs, loc)	/* calculates (loc % fs->fs_fsize) */ \
566 	((loc) & (fs)->fs_qfmask)
567 #define lfragtosize(fs, frag)	/* calculates ((off_t)frag * fs->fs_fsize) */ \
568 	(((off_t)(frag)) << (fs)->fs_fshift)
569 #define lblktosize(fs, blk)	/* calculates ((off_t)blk * fs->fs_bsize) */ \
570 	(((off_t)(blk)) << (fs)->fs_bshift)
571 /* Use this only when `blk' is known to be small, e.g., < NDADDR. */
572 #define smalllblktosize(fs, blk)    /* calculates (blk * fs->fs_bsize) */ \
573 	((blk) << (fs)->fs_bshift)
574 #define lblkno(fs, loc)		/* calculates (loc / fs->fs_bsize) */ \
575 	((loc) >> (fs)->fs_bshift)
576 #define numfrags(fs, loc)	/* calculates (loc / fs->fs_fsize) */ \
577 	((loc) >> (fs)->fs_fshift)
578 #define blkroundup(fs, size)	/* calculates roundup(size, fs->fs_bsize) */ \
579 	(((size) + (fs)->fs_qbmask) & (fs)->fs_bmask)
580 #define fragroundup(fs, size)	/* calculates roundup(size, fs->fs_fsize) */ \
581 	(((size) + (fs)->fs_qfmask) & (fs)->fs_fmask)
582 #define fragstoblks(fs, frags)	/* calculates (frags / fs->fs_frag) */ \
583 	((frags) >> (fs)->fs_fragshift)
584 #define blkstofrags(fs, blks)	/* calculates (blks * fs->fs_frag) */ \
585 	((blks) << (fs)->fs_fragshift)
586 #define fragnum(fs, fsb)	/* calculates (fsb % fs->fs_frag) */ \
587 	((fsb) & ((fs)->fs_frag - 1))
588 #define blknum(fs, fsb)		/* calculates rounddown(fsb, fs->fs_frag) */ \
589 	((fsb) &~ ((fs)->fs_frag - 1))
590 
591 /*
592  * Determine the number of available frags given a
593  * percentage to hold in reserve.
594  */
595 #define freespace(fs, percentreserved) \
596 	(blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \
597 	(fs)->fs_cstotal.cs_nffree - \
598 	(((off_t)((fs)->fs_dsize)) * (percentreserved) / 100))
599 
600 /*
601  * Determining the size of a file block in the filesystem.
602  */
603 #define blksize(fs, ip, lbn) \
604 	(((lbn) >= NDADDR || (ip)->i_size >= smalllblktosize(fs, (lbn) + 1)) \
605 	    ? (fs)->fs_bsize \
606 	    : (fragroundup(fs, blkoff(fs, (ip)->i_size))))
607 #define sblksize(fs, size, lbn) \
608 	(((lbn) >= NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \
609 	  ? (fs)->fs_bsize \
610 	  : (fragroundup(fs, blkoff(fs, (size)))))
611 
612 /*
613  * Number of indirects in a filesystem block.
614  */
615 #define	NINDIR(fs)	((fs)->fs_nindir)
616 
617 /*
618  * Indirect lbns are aligned on NDADDR addresses where single indirects
619  * are the negated address of the lowest lbn reachable, double indirects
620  * are this lbn - 1 and triple indirects are this lbn - 2.  This yields
621  * an unusual bit order to determine level.
622  */
623 static inline int
624 lbn_level(ufs_lbn_t lbn)
625 {
626 	if (lbn >= 0)
627 		return 0;
628 	switch (lbn & 0x3) {
629 	case 0:
630 		return (0);
631 	case 1:
632 		break;
633 	case 2:
634 		return (2);
635 	case 3:
636 		return (1);
637 	default:
638 		break;
639 	}
640 	return (-1);
641 }
642 
643 static inline ufs_lbn_t
644 lbn_offset(struct fs *fs, int level)
645 {
646 	ufs_lbn_t res;
647 
648 	for (res = 1; level > 0; level--)
649 		res *= NINDIR(fs);
650 	return (res);
651 }
652 
653 /*
654  * Number of inodes in a secondary storage block/fragment.
655  */
656 #define	INOPB(fs)	((fs)->fs_inopb)
657 #define	INOPF(fs)	((fs)->fs_inopb >> (fs)->fs_fragshift)
658 
659 /*
660  * Softdep journal record format.
661  */
662 
663 #define	JOP_ADDREF	1	/* Add a reference to an inode. */
664 #define	JOP_REMREF	2	/* Remove a reference from an inode. */
665 #define	JOP_NEWBLK	3	/* Allocate a block. */
666 #define	JOP_FREEBLK	4	/* Free a block or a tree of blocks. */
667 #define	JOP_MVREF	5	/* Move a reference from one off to another. */
668 #define	JOP_TRUNC	6	/* Partial truncation record. */
669 #define	JOP_SYNC	7	/* fsync() complete record. */
670 
671 #define	JREC_SIZE	32	/* Record and segment header size. */
672 
673 #define	SUJ_MIN		(4 * 1024 * 1024)	/* Minimum journal size */
674 #define	SUJ_MAX		(32 * 1024 * 1024)	/* Maximum journal size */
675 #define	SUJ_FILE	".sujournal"		/* Journal file name */
676 
677 /*
678  * Size of the segment record header.  There is at most one for each disk
679  * block in the journal.  The segment header is followed by an array of
680  * records.  fsck depends on the first element in each record being 'op'
681  * and the second being 'ino'.  Segments may span multiple disk blocks but
682  * the header is present on each.
683  */
684 struct jsegrec {
685 	uint64_t	jsr_seq;	/* Our sequence number */
686 	uint64_t	jsr_oldest;	/* Oldest valid sequence number */
687 	uint16_t	jsr_cnt;	/* Count of valid records */
688 	uint16_t	jsr_blocks;	/* Count of device bsize blocks. */
689 	uint32_t	jsr_crc;	/* 32bit crc of the valid space */
690 	ufs_time_t	jsr_time;	/* timestamp for mount instance */
691 };
692 
693 /*
694  * Reference record.  Records a single link count modification.
695  */
696 struct jrefrec {
697 	uint32_t	jr_op;
698 	ino_t		jr_ino;
699 	ino_t		jr_parent;
700 	uint16_t	jr_nlink;
701 	uint16_t	jr_mode;
702 	off_t		jr_diroff;
703 	uint64_t	jr_unused;
704 };
705 
706 /*
707  * Move record.  Records a reference moving within a directory block.  The
708  * nlink is unchanged but we must search both locations.
709  */
710 struct jmvrec {
711 	uint32_t	jm_op;
712 	ino_t		jm_ino;
713 	ino_t		jm_parent;
714 	uint16_t	jm_unused;
715 	off_t		jm_oldoff;
716 	off_t		jm_newoff;
717 };
718 
719 /*
720  * Block record.  A set of frags or tree of blocks starting at an indirect are
721  * freed or a set of frags are allocated.
722  */
723 struct jblkrec {
724 	uint32_t	jb_op;
725 	uint32_t	jb_ino;
726 	ufs2_daddr_t	jb_blkno;
727 	ufs_lbn_t	jb_lbn;
728 	uint16_t	jb_frags;
729 	uint16_t	jb_oldfrags;
730 	uint32_t	jb_unused;
731 };
732 
733 /*
734  * Truncation record.  Records a partial truncation so that it may be
735  * completed at check time.  Also used for sync records.
736  */
737 struct jtrncrec {
738 	uint32_t	jt_op;
739 	uint32_t	jt_ino;
740 	off_t		jt_size;
741 	uint32_t	jt_extsize;
742 	uint32_t	jt_pad[3];
743 };
744 
745 union jrec {
746 	struct jsegrec	rec_jsegrec;
747 	struct jrefrec	rec_jrefrec;
748 	struct jmvrec	rec_jmvrec;
749 	struct jblkrec	rec_jblkrec;
750 	struct jtrncrec	rec_jtrncrec;
751 };
752 
753 #ifdef CTASSERT
754 CTASSERT(sizeof(struct jsegrec) == JREC_SIZE);
755 CTASSERT(sizeof(struct jrefrec) == JREC_SIZE);
756 CTASSERT(sizeof(struct jmvrec) == JREC_SIZE);
757 CTASSERT(sizeof(struct jblkrec) == JREC_SIZE);
758 CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE);
759 CTASSERT(sizeof(union jrec) == JREC_SIZE);
760 #endif
761 
762 extern int inside[], around[];
763 extern u_char *fragtbl[];
764 
765 #endif
766