xref: /freebsd/sys/ufs/ffs/fs.h (revision 2e36db147e0edeb1b213f283f9b41a8d1a902a88)
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  * 3. 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 #include <sys/mount.h>
37 #include <ufs/ufs/dinode.h>
38 
39 /*
40  * Each disk drive contains some number of filesystems.
41  * A filesystem consists of a number of cylinder groups.
42  * Each cylinder group has inodes and data.
43  *
44  * A filesystem is described by its super-block, which in turn
45  * describes the cylinder groups.  The super-block is critical
46  * data and is replicated in each cylinder group to protect against
47  * catastrophic loss.  This is done at `newfs' time and the critical
48  * super-block data does not change, so the copies need not be
49  * referenced further unless disaster strikes.
50  *
51  * For filesystem fs, the offsets of the various blocks of interest
52  * are given in the super block as:
53  *	[fs->fs_sblkno]		Super-block
54  *	[fs->fs_cblkno]		Cylinder group block
55  *	[fs->fs_iblkno]		Inode blocks
56  *	[fs->fs_dblkno]		Data blocks
57  * The beginning of cylinder group cg in fs, is given by
58  * the ``cgbase(fs, cg)'' macro.
59  *
60  * Depending on the architecture and the media, the superblock may
61  * reside in any one of four places. For tiny media where every block
62  * counts, it is placed at the very front of the partition. Historically,
63  * UFS1 placed it 8K from the front to leave room for the disk label and
64  * a small bootstrap. For UFS2 it got moved to 64K from the front to leave
65  * room for the disk label and a bigger bootstrap, and for really piggy
66  * systems we check at 256K from the front if the first three fail. In
67  * all cases the size of the superblock will be SBLOCKSIZE. All values are
68  * given in byte-offset form, so they do not imply a sector size. The
69  * SBLOCKSEARCH specifies the order in which the locations should be searched.
70  */
71 #define	SBLOCK_FLOPPY	     0
72 #define	SBLOCK_UFS1	  8192
73 #define	SBLOCK_UFS2	 65536
74 #define	SBLOCK_PIGGY	262144
75 #define	SBLOCKSIZE	  8192
76 #define	SBLOCKSEARCH \
77 	{ SBLOCK_UFS2, SBLOCK_UFS1, SBLOCK_FLOPPY, SBLOCK_PIGGY, -1 }
78 
79 /*
80  * Max number of fragments per block. This value is NOT tweakable.
81  */
82 #define	MAXFRAG 	8
83 
84 /*
85  * Addresses stored in inodes are capable of addressing fragments
86  * of `blocks'. File system blocks of at most size MAXBSIZE can
87  * be optionally broken into 2, 4, or 8 pieces, each of which is
88  * addressable; these pieces may be DEV_BSIZE, or some multiple of
89  * a DEV_BSIZE unit.
90  *
91  * Large files consist of exclusively large data blocks.  To avoid
92  * undue wasted disk space, the last data block of a small file may be
93  * allocated as only as many fragments of a large block as are
94  * necessary.  The filesystem format retains only a single pointer
95  * to such a fragment, which is a piece of a single large block that
96  * has been divided.  The size of such a fragment is determinable from
97  * information in the inode, using the ``blksize(fs, ip, lbn)'' macro.
98  *
99  * The filesystem records space availability at the fragment level;
100  * to determine block availability, aligned fragments are examined.
101  */
102 
103 /*
104  * MINBSIZE is the smallest allowable block size.
105  * In order to insure that it is possible to create files of size
106  * 2^32 with only two levels of indirection, MINBSIZE is set to 4096.
107  * MINBSIZE must be big enough to hold a cylinder group block,
108  * thus changes to (struct cg) must keep its size within MINBSIZE.
109  * Note that super blocks are always of size SBLOCKSIZE,
110  * and that both SBLOCKSIZE and MAXBSIZE must be >= MINBSIZE.
111  */
112 #define	MINBSIZE	4096
113 
114 /*
115  * The path name on which the filesystem is mounted is maintained
116  * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in
117  * the super block for this name.
118  */
119 #define	MAXMNTLEN	468
120 
121 /*
122  * The volume name for this filesystem is maintained in fs_volname.
123  * MAXVOLLEN defines the length of the buffer allocated.
124  */
125 #define	MAXVOLLEN	32
126 
127 /*
128  * There is a 128-byte region in the superblock reserved for in-core
129  * pointers to summary information. Originally this included an array
130  * of pointers to blocks of struct csum; now there are just a few
131  * pointers and the remaining space is padded with fs_ocsp[].
132  *
133  * NOCSPTRS determines the size of this padding. One pointer (fs_csp)
134  * is taken away to point to a contiguous array of struct csum for
135  * all cylinder groups; a second (fs_maxcluster) points to an array
136  * of cluster sizes that is computed as cylinder groups are inspected,
137  * and the third points to an array that tracks the creation of new
138  * directories. A fourth pointer, fs_active, is used when creating
139  * snapshots; it points to a bitmap of cylinder groups for which the
140  * free-block bitmap has changed since the snapshot operation began.
141  */
142 #define	NOCSPTRS	((128 / sizeof(void *)) - 4)
143 
144 /*
145  * A summary of contiguous blocks of various sizes is maintained
146  * in each cylinder group. Normally this is set by the initial
147  * value of fs_maxcontig. To conserve space, a maximum summary size
148  * is set by FS_MAXCONTIG.
149  */
150 #define	FS_MAXCONTIG	16
151 
152 /*
153  * MINFREE gives the minimum acceptable percentage of filesystem
154  * blocks which may be free. If the freelist drops below this level
155  * only the superuser may continue to allocate blocks. This may
156  * be set to 0 if no reserve of free blocks is deemed necessary,
157  * however throughput drops by fifty percent if the filesystem
158  * is run at between 95% and 100% full; thus the minimum default
159  * value of fs_minfree is 5%. However, to get good clustering
160  * performance, 10% is a better choice. hence we use 10% as our
161  * default value. With 10% free space, fragmentation is not a
162  * problem, so we choose to optimize for time.
163  */
164 #define	MINFREE		8
165 #define	DEFAULTOPT	FS_OPTTIME
166 
167 /*
168  * Grigoriy Orlov <gluk@ptci.ru> has done some extensive work to fine
169  * tune the layout preferences for directories within a filesystem.
170  * His algorithm can be tuned by adjusting the following parameters
171  * which tell the system the average file size and the average number
172  * of files per directory. These defaults are well selected for typical
173  * filesystems, but may need to be tuned for odd cases like filesystems
174  * being used for squid caches or news spools.
175  */
176 #define	AVFILESIZ	16384	/* expected average file size */
177 #define	AFPDIR		64	/* expected number of files per directory */
178 
179 /*
180  * The maximum number of snapshot nodes that can be associated
181  * with each filesystem. This limit affects only the number of
182  * snapshot files that can be recorded within the superblock so
183  * that they can be found when the filesystem is mounted. However,
184  * maintaining too many will slow the filesystem performance, so
185  * having this limit is a good idea.
186  */
187 #define	FSMAXSNAP 20
188 
189 /*
190  * Used to identify special blocks in snapshots:
191  *
192  * BLK_NOCOPY - A block that was unallocated at the time the snapshot
193  *	was taken, hence does not need to be copied when written.
194  * BLK_SNAP - A block held by another snapshot that is not needed by this
195  *	snapshot. When the other snapshot is freed, the BLK_SNAP entries
196  *	are converted to BLK_NOCOPY. These are needed to allow fsck to
197  *	identify blocks that are in use by other snapshots (which are
198  *	expunged from this snapshot).
199  */
200 #define	BLK_NOCOPY ((ufs2_daddr_t)(1))
201 #define	BLK_SNAP ((ufs2_daddr_t)(2))
202 
203 /*
204  * Sysctl values for the fast filesystem.
205  */
206 #define	FFS_ADJ_REFCNT		 1	/* adjust inode reference count */
207 #define	FFS_ADJ_BLKCNT		 2	/* adjust inode used block count */
208 #define	FFS_BLK_FREE		 3	/* free range of blocks in map */
209 #define	FFS_DIR_FREE		 4	/* free specified dir inodes in map */
210 #define	FFS_FILE_FREE		 5	/* free specified file inodes in map */
211 #define	FFS_SET_FLAGS		 6	/* set filesystem flags */
212 #define	FFS_ADJ_NDIR		 7	/* adjust number of directories */
213 #define	FFS_ADJ_NBFREE		 8	/* adjust number of free blocks */
214 #define	FFS_ADJ_NIFREE		 9	/* adjust number of free inodes */
215 #define	FFS_ADJ_NFFREE		10 	/* adjust number of free frags */
216 #define	FFS_ADJ_NUMCLUSTERS	11	/* adjust number of free clusters */
217 #define	FFS_SET_CWD		12	/* set current directory */
218 #define	FFS_SET_DOTDOT		13	/* set inode number for ".." */
219 #define	FFS_UNLINK		14	/* remove a name in the filesystem */
220 #define	FFS_SET_INODE		15	/* update an on-disk inode */
221 #define	FFS_SET_BUFOUTPUT	16	/* set buffered writing on descriptor */
222 #define	FFS_MAXID		16	/* number of valid ffs ids */
223 
224 /*
225  * Command structure passed in to the filesystem to adjust filesystem values.
226  */
227 #define	FFS_CMD_VERSION		0x19790518	/* version ID */
228 struct fsck_cmd {
229 	int32_t	version;	/* version of command structure */
230 	int32_t	handle;		/* reference to filesystem to be changed */
231 	int64_t	value;		/* inode or block number to be affected */
232 	int64_t	size;		/* amount or range to be adjusted */
233 	int64_t	spare;		/* reserved for future use */
234 };
235 
236 /*
237  * A recovery structure placed at the end of the boot block area by newfs
238  * that can be used by fsck to search for alternate superblocks.
239  */
240 struct fsrecovery {
241 	int32_t	fsr_magic;	/* magic number */
242 	int32_t	fsr_fsbtodb;	/* fsbtodb and dbtofsb shift constant */
243 	int32_t	fsr_sblkno;	/* offset of super-block in filesys */
244 	int32_t	fsr_fpg;	/* blocks per group * fs_frag */
245 	u_int32_t fsr_ncg;	/* number of cylinder groups */
246 };
247 
248 /*
249  * Per cylinder group information; summarized in blocks allocated
250  * from first cylinder group data blocks.  These blocks have to be
251  * read in from fs_csaddr (size fs_cssize) in addition to the
252  * super block.
253  */
254 struct csum {
255 	int32_t	cs_ndir;		/* number of directories */
256 	int32_t	cs_nbfree;		/* number of free blocks */
257 	int32_t	cs_nifree;		/* number of free inodes */
258 	int32_t	cs_nffree;		/* number of free frags */
259 };
260 struct csum_total {
261 	int64_t	cs_ndir;		/* number of directories */
262 	int64_t	cs_nbfree;		/* number of free blocks */
263 	int64_t	cs_nifree;		/* number of free inodes */
264 	int64_t	cs_nffree;		/* number of free frags */
265 	int64_t	cs_numclusters;		/* number of free clusters */
266 	int64_t	cs_spare[3];		/* future expansion */
267 };
268 
269 /*
270  * Super block for an FFS filesystem.
271  */
272 struct fs {
273 	int32_t	 fs_firstfield;		/* historic filesystem linked list, */
274 	int32_t	 fs_unused_1;		/*     used for incore super blocks */
275 	int32_t	 fs_sblkno;		/* offset of super-block in filesys */
276 	int32_t	 fs_cblkno;		/* offset of cyl-block in filesys */
277 	int32_t	 fs_iblkno;		/* offset of inode-blocks in filesys */
278 	int32_t	 fs_dblkno;		/* offset of first data after cg */
279 	int32_t	 fs_old_cgoffset;	/* cylinder group offset in cylinder */
280 	int32_t	 fs_old_cgmask;		/* used to calc mod fs_ntrak */
281 	int32_t  fs_old_time;		/* last time written */
282 	int32_t	 fs_old_size;		/* number of blocks in fs */
283 	int32_t	 fs_old_dsize;		/* number of data blocks in fs */
284 	u_int32_t fs_ncg;		/* number of cylinder groups */
285 	int32_t	 fs_bsize;		/* size of basic blocks in fs */
286 	int32_t	 fs_fsize;		/* size of frag blocks in fs */
287 	int32_t	 fs_frag;		/* number of frags in a block in fs */
288 /* these are configuration parameters */
289 	int32_t	 fs_minfree;		/* minimum percentage of free blocks */
290 	int32_t	 fs_old_rotdelay;	/* num of ms for optimal next block */
291 	int32_t	 fs_old_rps;		/* disk revolutions per second */
292 /* these fields can be computed from the others */
293 	int32_t	 fs_bmask;		/* ``blkoff'' calc of blk offsets */
294 	int32_t	 fs_fmask;		/* ``fragoff'' calc of frag offsets */
295 	int32_t	 fs_bshift;		/* ``lblkno'' calc of logical blkno */
296 	int32_t	 fs_fshift;		/* ``numfrags'' calc number of frags */
297 /* these are configuration parameters */
298 	int32_t	 fs_maxcontig;		/* max number of contiguous blks */
299 	int32_t	 fs_maxbpg;		/* max number of blks per cyl group */
300 /* these fields can be computed from the others */
301 	int32_t	 fs_fragshift;		/* block to frag shift */
302 	int32_t	 fs_fsbtodb;		/* fsbtodb and dbtofsb shift constant */
303 	int32_t	 fs_sbsize;		/* actual size of super block */
304 	int32_t	 fs_spare1[2];		/* old fs_csmask */
305 					/* old fs_csshift */
306 	int32_t	 fs_nindir;		/* value of NINDIR */
307 	u_int32_t fs_inopb;		/* value of INOPB */
308 	int32_t	 fs_old_nspf;		/* value of NSPF */
309 /* yet another configuration parameter */
310 	int32_t	 fs_optim;		/* optimization preference, see below */
311 	int32_t	 fs_old_npsect;		/* # sectors/track including spares */
312 	int32_t	 fs_old_interleave;	/* hardware sector interleave */
313 	int32_t	 fs_old_trackskew;	/* sector 0 skew, per track */
314 	int32_t	 fs_id[2];		/* unique filesystem id */
315 /* sizes determined by number of cylinder groups and their sizes */
316 	int32_t	 fs_old_csaddr;		/* blk addr of cyl grp summary area */
317 	int32_t	 fs_cssize;		/* size of cyl grp summary area */
318 	int32_t	 fs_cgsize;		/* cylinder group size */
319 	int32_t	 fs_spare2;		/* old fs_ntrak */
320 	int32_t	 fs_old_nsect;		/* sectors per track */
321 	int32_t  fs_old_spc;		/* sectors per cylinder */
322 	int32_t	 fs_old_ncyl;		/* cylinders in filesystem */
323 	int32_t	 fs_old_cpg;		/* cylinders per group */
324 	u_int32_t fs_ipg;		/* inodes per group */
325 	int32_t	 fs_fpg;		/* blocks per group * fs_frag */
326 /* this data must be re-computed after crashes */
327 	struct	csum fs_old_cstotal;	/* cylinder summary information */
328 /* these fields are cleared at mount time */
329 	int8_t   fs_fmod;		/* super block modified flag */
330 	int8_t   fs_clean;		/* filesystem is clean flag */
331 	int8_t 	 fs_ronly;		/* mounted read-only flag */
332 	int8_t   fs_old_flags;		/* old FS_ flags */
333 	u_char	 fs_fsmnt[MAXMNTLEN];	/* name mounted on */
334 	u_char	 fs_volname[MAXVOLLEN];	/* volume name */
335 	u_int64_t fs_swuid;		/* system-wide uid */
336 	int32_t  fs_pad;		/* due to alignment of fs_swuid */
337 /* these fields retain the current block allocation info */
338 	int32_t	 fs_cgrotor;		/* last cg searched */
339 	void 	*fs_ocsp[NOCSPTRS];	/* padding; was list of fs_cs buffers */
340 	u_int8_t *fs_contigdirs;	/* (u) # of contig. allocated dirs */
341 	struct	csum *fs_csp;		/* (u) cg summary info buffer */
342 	int32_t	*fs_maxcluster;		/* (u) max cluster in each cyl group */
343 	u_int	*fs_active;		/* (u) used by snapshots to track fs */
344 	int32_t	 fs_old_cpc;		/* cyl per cycle in postbl */
345 	int32_t	 fs_maxbsize;		/* maximum blocking factor permitted */
346 	int64_t	 fs_unrefs;		/* number of unreferenced inodes */
347 	int64_t  fs_providersize;	/* size of underlying GEOM provider */
348 	int64_t	 fs_metaspace;		/* size of area reserved for metadata */
349 	int64_t	 fs_sparecon64[14];	/* old rotation block list head */
350 	int64_t	 fs_sblockloc;		/* byte offset of standard superblock */
351 	struct	csum_total fs_cstotal;	/* (u) cylinder summary information */
352 	ufs_time_t fs_time;		/* last time written */
353 	int64_t	 fs_size;		/* number of blocks in fs */
354 	int64_t	 fs_dsize;		/* number of data blocks in fs */
355 	ufs2_daddr_t fs_csaddr;		/* blk addr of cyl grp summary area */
356 	int64_t	 fs_pendingblocks;	/* (u) blocks being freed */
357 	u_int32_t fs_pendinginodes;	/* (u) inodes being freed */
358 	uint32_t fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */
359 	u_int32_t fs_avgfilesize;	/* expected average file size */
360 	u_int32_t fs_avgfpdir;		/* expected # of files per directory */
361 	int32_t	 fs_save_cgsize;	/* save real cg size to use fs_bsize */
362 	ufs_time_t fs_mtime;		/* Last mount or fsck time. */
363 	int32_t  fs_sujfree;		/* SUJ free list */
364 	int32_t	 fs_sparecon32[22];	/* reserved for future constants */
365 	u_int32_t fs_metackhash;	/* metadata check-hash, see CK_ below */
366 	int32_t  fs_flags;		/* see FS_ flags below */
367 	int32_t	 fs_contigsumsize;	/* size of cluster summary array */
368 	int32_t	 fs_maxsymlinklen;	/* max length of an internal symlink */
369 	int32_t	 fs_old_inodefmt;	/* format of on-disk inodes */
370 	u_int64_t fs_maxfilesize;	/* maximum representable file size */
371 	int64_t	 fs_qbmask;		/* ~fs_bmask for use with 64-bit size */
372 	int64_t	 fs_qfmask;		/* ~fs_fmask for use with 64-bit size */
373 	int32_t	 fs_state;		/* validate fs_clean field */
374 	int32_t	 fs_old_postblformat;	/* format of positional layout tables */
375 	int32_t	 fs_old_nrpos;		/* number of rotational positions */
376 	int32_t	 fs_spare5[2];		/* old fs_postbloff */
377 					/* old fs_rotbloff */
378 	int32_t	 fs_magic;		/* magic number */
379 };
380 
381 /* Sanity checking. */
382 #ifdef CTASSERT
383 CTASSERT(sizeof(struct fs) == 1376);
384 #endif
385 
386 /*
387  * Filesystem identification
388  */
389 #define	FS_UFS1_MAGIC	0x011954	/* UFS1 fast filesystem magic number */
390 #define	FS_UFS2_MAGIC	0x19540119	/* UFS2 fast filesystem magic number */
391 #define	FS_BAD_MAGIC	0x19960408	/* UFS incomplete newfs magic number */
392 #define	FS_42INODEFMT	-1		/* 4.2BSD inode format */
393 #define	FS_44INODEFMT	2		/* 4.4BSD inode format */
394 
395 /*
396  * Preference for optimization.
397  */
398 #define	FS_OPTTIME	0	/* minimize allocation time */
399 #define	FS_OPTSPACE	1	/* minimize disk fragmentation */
400 
401 /*
402  * Filesystem flags.
403  *
404  * The FS_UNCLEAN flag is set by the kernel when the filesystem was
405  * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates
406  * that the filesystem should be managed by the soft updates code.
407  * Note that the FS_NEEDSFSCK flag is set and cleared only by the
408  * fsck utility. It is set when background fsck finds an unexpected
409  * inconsistency which requires a traditional foreground fsck to be
410  * run. Such inconsistencies should only be found after an uncorrectable
411  * disk error. A foreground fsck will clear the FS_NEEDSFSCK flag when
412  * it has successfully cleaned up the filesystem. The kernel uses this
413  * flag to enforce that inconsistent filesystems be mounted read-only.
414  * The FS_INDEXDIRS flag when set indicates that the kernel maintains
415  * on-disk auxiliary indexes (such as B-trees) for speeding directory
416  * accesses. Kernels that do not support auxiliary indices clear the
417  * flag to indicate that the indices need to be rebuilt (by fsck) before
418  * they can be used. When a filesystem is mounted, any flags not
419  * included in FS_SUPPORTED are cleared. This lets newer features
420  * know that the filesystem has been run on an older version of the
421  * filesystem and thus that data structures associated with those
422  * features are out-of-date and need to be rebuilt.
423  *
424  * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled
425  * for the file system, so they should be loaded from extended attributes,
426  * observed for access control purposes, and be administered by object
427  * owners.  FS_NFS4ACLS indicates that NFSv4 ACLs are administratively
428  * enabled.  This flag is mutually exclusive with FS_ACLS.  FS_MULTILABEL
429  * indicates that the TrustedBSD MAC Framework should attempt to back MAC
430  * labels into extended attributes on the file system rather than maintain
431  * a single mount label for all objects.
432  */
433 #define	FS_UNCLEAN	0x0001	/* filesystem not clean at mount */
434 #define	FS_DOSOFTDEP	0x0002	/* filesystem using soft dependencies */
435 #define	FS_NEEDSFSCK	0x0004	/* filesystem needs sync fsck before mount */
436 #define	FS_SUJ       	0x0008	/* Filesystem using softupdate journal */
437 #define	FS_ACLS		0x0010	/* file system has POSIX.1e ACLs enabled */
438 #define	FS_MULTILABEL	0x0020	/* file system is MAC multi-label */
439 #define	FS_GJOURNAL	0x0040	/* gjournaled file system */
440 #define	FS_FLAGS_UPDATED 0x0080	/* flags have been moved to new location */
441 #define	FS_NFS4ACLS	0x0100	/* file system has NFSv4 ACLs enabled */
442 #define	FS_INDEXDIRS	0x0200	/* kernel supports indexed directories */
443 #define	FS_TRIM		0x0400	/* issue BIO_DELETE for deleted blocks */
444 #define	FS_SUPPORTED	0xFFFF	/* supported flags, others cleared at mount */
445 
446 /*
447  * The fs_metackhash field indicates the types of metadata check-hash
448  * that are maintained for a filesystem. Not all filesystems check-hash
449  * all metadata.
450  */
451 #define	CK_SUPERBLOCK	0x0001	/* the superblock */
452 #define	CK_CYLGRP	0x0002	/* the cylinder groups */
453 #define	CK_INODE	0x0004	/* inodes */
454 #define	CK_INDIR	0x0008	/* indirect blocks */
455 #define	CK_DIR		0x0010	/* directory contents */
456 /*
457  * The BX_FSPRIV buffer b_xflags are used to track types of data in buffers.
458  */
459 #define	BX_SUPERBLOCK	0x00010000	/* superblock */
460 #define	BX_CYLGRP	0x00020000	/* cylinder groups */
461 #define	BX_INODE	0x00040000	/* inodes */
462 #define	BX_INDIR	0x00080000	/* indirect blocks */
463 #define	BX_DIR		0x00100000	/* directory contents */
464 
465 #define	PRINT_UFS_BUF_XFLAGS "\20\25dir\24indir\23inode\22cylgrp\21superblock"
466 
467 /*
468  * Macros to access bits in the fs_active array.
469  */
470 #define	ACTIVECGNUM(fs, cg)	((fs)->fs_active[(cg) / (NBBY * sizeof(int))])
471 #define	ACTIVECGOFF(cg)		(1 << ((cg) % (NBBY * sizeof(int))))
472 #define	ACTIVESET(fs, cg)	do {					\
473 	if ((fs)->fs_active)						\
474 		ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg));		\
475 } while (0)
476 #define	ACTIVECLEAR(fs, cg)	do {					\
477 	if ((fs)->fs_active)						\
478 		ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg));		\
479 } while (0)
480 
481 /*
482  * The size of a cylinder group is calculated by CGSIZE. The maximum size
483  * is limited by the fact that cylinder groups are at most one block.
484  * Its size is derived from the size of the maps maintained in the
485  * cylinder group and the (struct cg) size.
486  */
487 #define	CGSIZE(fs) \
488     /* base cg */	(sizeof(struct cg) + sizeof(int32_t) + \
489     /* old btotoff */	(fs)->fs_old_cpg * sizeof(int32_t) + \
490     /* old boff */	(fs)->fs_old_cpg * sizeof(u_int16_t) + \
491     /* inode map */	howmany((fs)->fs_ipg, NBBY) + \
492     /* block map */	howmany((fs)->fs_fpg, NBBY) +\
493     /* if present */	((fs)->fs_contigsumsize <= 0 ? 0 : \
494     /* cluster sum */	(fs)->fs_contigsumsize * sizeof(int32_t) + \
495     /* cluster map */	howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY)))
496 
497 /*
498  * The minimal number of cylinder groups that should be created.
499  */
500 #define	MINCYLGRPS	4
501 
502 /*
503  * Convert cylinder group to base address of its global summary info.
504  */
505 #define	fs_cs(fs, indx) fs_csp[indx]
506 
507 /*
508  * Cylinder group block for a filesystem.
509  */
510 #define	CG_MAGIC	0x090255
511 struct cg {
512 	int32_t	 cg_firstfield;		/* historic cyl groups linked list */
513 	int32_t	 cg_magic;		/* magic number */
514 	int32_t  cg_old_time;		/* time last written */
515 	u_int32_t cg_cgx;		/* we are the cgx'th cylinder group */
516 	int16_t	 cg_old_ncyl;		/* number of cyl's this cg */
517 	int16_t  cg_old_niblk;		/* number of inode blocks this cg */
518 	u_int32_t cg_ndblk;		/* number of data blocks this cg */
519 	struct	 csum cg_cs;		/* cylinder summary information */
520 	u_int32_t cg_rotor;		/* position of last used block */
521 	u_int32_t cg_frotor;		/* position of last used frag */
522 	u_int32_t cg_irotor;		/* position of last used inode */
523 	u_int32_t cg_frsum[MAXFRAG];	/* counts of available frags */
524 	int32_t	 cg_old_btotoff;	/* (int32) block totals per cylinder */
525 	int32_t	 cg_old_boff;		/* (u_int16) free block positions */
526 	u_int32_t cg_iusedoff;		/* (u_int8) used inode map */
527 	u_int32_t cg_freeoff;		/* (u_int8) free block map */
528 	u_int32_t cg_nextfreeoff;	/* (u_int8) next available space */
529 	u_int32_t cg_clustersumoff;	/* (u_int32) counts of avail clusters */
530 	u_int32_t cg_clusteroff;		/* (u_int8) free cluster map */
531 	u_int32_t cg_nclusterblks;	/* number of clusters this cg */
532 	u_int32_t cg_niblk;		/* number of inode blocks this cg */
533 	u_int32_t cg_initediblk;		/* last initialized inode */
534 	u_int32_t cg_unrefs;		/* number of unreferenced inodes */
535 	int32_t	 cg_sparecon32[1];	/* reserved for future use */
536 	u_int32_t cg_ckhash;		/* check-hash of this cg */
537 	ufs_time_t cg_time;		/* time last written */
538 	int64_t	 cg_sparecon64[3];	/* reserved for future use */
539 	u_int8_t cg_space[1];		/* space for cylinder group maps */
540 /* actually longer */
541 };
542 
543 /*
544  * Macros for access to cylinder group array structures
545  */
546 #define	cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC)
547 #define	cg_inosused(cgp) \
548     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_iusedoff))
549 #define	cg_blksfree(cgp) \
550     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_freeoff))
551 #define	cg_clustersfree(cgp) \
552     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_clusteroff))
553 #define	cg_clustersum(cgp) \
554     ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff))
555 
556 /*
557  * Turn filesystem block numbers into disk block addresses.
558  * This maps filesystem blocks to device size blocks.
559  */
560 #define	fsbtodb(fs, b)	((daddr_t)(b) << (fs)->fs_fsbtodb)
561 #define	dbtofsb(fs, b)	((b) >> (fs)->fs_fsbtodb)
562 
563 /*
564  * Cylinder group macros to locate things in cylinder groups.
565  * They calc filesystem addresses of cylinder group data structures.
566  */
567 #define	cgbase(fs, c)	(((ufs2_daddr_t)(fs)->fs_fpg) * (c))
568 #define	cgdata(fs, c)	(cgdmin(fs, c) + (fs)->fs_metaspace)	/* data zone */
569 #define	cgmeta(fs, c)	(cgdmin(fs, c))				/* meta data */
570 #define	cgdmin(fs, c)	(cgstart(fs, c) + (fs)->fs_dblkno)	/* 1st data */
571 #define	cgimin(fs, c)	(cgstart(fs, c) + (fs)->fs_iblkno)	/* inode blk */
572 #define	cgsblock(fs, c)	(cgstart(fs, c) + (fs)->fs_sblkno)	/* super blk */
573 #define	cgtod(fs, c)	(cgstart(fs, c) + (fs)->fs_cblkno)	/* cg block */
574 #define	cgstart(fs, c)							\
575        ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) :		\
576        (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask))))
577 
578 /*
579  * Macros for handling inode numbers:
580  *     inode number to filesystem block offset.
581  *     inode number to cylinder group number.
582  *     inode number to filesystem block address.
583  */
584 #define	ino_to_cg(fs, x)	(((ino_t)(x)) / (fs)->fs_ipg)
585 #define	ino_to_fsba(fs, x)						\
586 	((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) +		\
587 	    (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs))))))
588 #define	ino_to_fsbo(fs, x)	(((ino_t)(x)) % INOPB(fs))
589 
590 /*
591  * Give cylinder group number for a filesystem block.
592  * Give cylinder group block number for a filesystem block.
593  */
594 #define	dtog(fs, d)	((d) / (fs)->fs_fpg)
595 #define	dtogd(fs, d)	((d) % (fs)->fs_fpg)
596 
597 /*
598  * Extract the bits for a block from a map.
599  * Compute the cylinder and rotational position of a cyl block addr.
600  */
601 #define	blkmap(fs, map, loc) \
602     (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag)))
603 
604 /*
605  * The following macros optimize certain frequently calculated
606  * quantities by using shifts and masks in place of divisions
607  * modulos and multiplications.
608  */
609 #define	blkoff(fs, loc)		/* calculates (loc % fs->fs_bsize) */ \
610 	((loc) & (fs)->fs_qbmask)
611 #define	fragoff(fs, loc)	/* calculates (loc % fs->fs_fsize) */ \
612 	((loc) & (fs)->fs_qfmask)
613 #define	lfragtosize(fs, frag)	/* calculates ((off_t)frag * fs->fs_fsize) */ \
614 	(((off_t)(frag)) << (fs)->fs_fshift)
615 #define	lblktosize(fs, blk)	/* calculates ((off_t)blk * fs->fs_bsize) */ \
616 	(((off_t)(blk)) << (fs)->fs_bshift)
617 /* Use this only when `blk' is known to be small, e.g., < UFS_NDADDR. */
618 #define	smalllblktosize(fs, blk)    /* calculates (blk * fs->fs_bsize) */ \
619 	((blk) << (fs)->fs_bshift)
620 #define	lblkno(fs, loc)		/* calculates (loc / fs->fs_bsize) */ \
621 	((loc) >> (fs)->fs_bshift)
622 #define	numfrags(fs, loc)	/* calculates (loc / fs->fs_fsize) */ \
623 	((loc) >> (fs)->fs_fshift)
624 #define	blkroundup(fs, size)	/* calculates roundup(size, fs->fs_bsize) */ \
625 	(((size) + (fs)->fs_qbmask) & (fs)->fs_bmask)
626 #define	fragroundup(fs, size)	/* calculates roundup(size, fs->fs_fsize) */ \
627 	(((size) + (fs)->fs_qfmask) & (fs)->fs_fmask)
628 #define	fragstoblks(fs, frags)	/* calculates (frags / fs->fs_frag) */ \
629 	((frags) >> (fs)->fs_fragshift)
630 #define	blkstofrags(fs, blks)	/* calculates (blks * fs->fs_frag) */ \
631 	((blks) << (fs)->fs_fragshift)
632 #define	fragnum(fs, fsb)	/* calculates (fsb % fs->fs_frag) */ \
633 	((fsb) & ((fs)->fs_frag - 1))
634 #define	blknum(fs, fsb)		/* calculates rounddown(fsb, fs->fs_frag) */ \
635 	((fsb) &~ ((fs)->fs_frag - 1))
636 
637 /*
638  * Determine the number of available frags given a
639  * percentage to hold in reserve.
640  */
641 #define	freespace(fs, percentreserved) \
642 	(blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \
643 	(fs)->fs_cstotal.cs_nffree - \
644 	(((off_t)((fs)->fs_dsize)) * (percentreserved) / 100))
645 
646 /*
647  * Determining the size of a file block in the filesystem.
648  */
649 #define	blksize(fs, ip, lbn) \
650 	(((lbn) >= UFS_NDADDR || (ip)->i_size >= \
651 	    (uint64_t)smalllblktosize(fs, (lbn) + 1)) \
652 	    ? (fs)->fs_bsize \
653 	    : (fragroundup(fs, blkoff(fs, (ip)->i_size))))
654 #define	sblksize(fs, size, lbn) \
655 	(((lbn) >= UFS_NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \
656 	  ? (fs)->fs_bsize \
657 	  : (fragroundup(fs, blkoff(fs, (size)))))
658 
659 /*
660  * Number of indirects in a filesystem block.
661  */
662 #define	NINDIR(fs)	((fs)->fs_nindir)
663 
664 /*
665  * Indirect lbns are aligned on UFS_NDADDR addresses where single indirects
666  * are the negated address of the lowest lbn reachable, double indirects
667  * are this lbn - 1 and triple indirects are this lbn - 2.  This yields
668  * an unusual bit order to determine level.
669  */
670 static inline int
671 lbn_level(ufs_lbn_t lbn)
672 {
673 	if (lbn >= 0)
674 		return 0;
675 	switch (lbn & 0x3) {
676 	case 0:
677 		return (0);
678 	case 1:
679 		break;
680 	case 2:
681 		return (2);
682 	case 3:
683 		return (1);
684 	default:
685 		break;
686 	}
687 	return (-1);
688 }
689 
690 static inline ufs_lbn_t
691 lbn_offset(struct fs *fs, int level)
692 {
693 	ufs_lbn_t res;
694 
695 	for (res = 1; level > 0; level--)
696 		res *= NINDIR(fs);
697 	return (res);
698 }
699 
700 /*
701  * Number of inodes in a secondary storage block/fragment.
702  */
703 #define	INOPB(fs)	((fs)->fs_inopb)
704 #define	INOPF(fs)	((fs)->fs_inopb >> (fs)->fs_fragshift)
705 
706 /*
707  * Softdep journal record format.
708  */
709 
710 #define	JOP_ADDREF	1	/* Add a reference to an inode. */
711 #define	JOP_REMREF	2	/* Remove a reference from an inode. */
712 #define	JOP_NEWBLK	3	/* Allocate a block. */
713 #define	JOP_FREEBLK	4	/* Free a block or a tree of blocks. */
714 #define	JOP_MVREF	5	/* Move a reference from one off to another. */
715 #define	JOP_TRUNC	6	/* Partial truncation record. */
716 #define	JOP_SYNC	7	/* fsync() complete record. */
717 
718 #define	JREC_SIZE	32	/* Record and segment header size. */
719 
720 #define	SUJ_MIN		(4 * 1024 * 1024)	/* Minimum journal size */
721 #define	SUJ_MAX		(32 * 1024 * 1024)	/* Maximum journal size */
722 #define	SUJ_FILE	".sujournal"		/* Journal file name */
723 
724 /*
725  * Size of the segment record header.  There is at most one for each disk
726  * block in the journal.  The segment header is followed by an array of
727  * records.  fsck depends on the first element in each record being 'op'
728  * and the second being 'ino'.  Segments may span multiple disk blocks but
729  * the header is present on each.
730  */
731 struct jsegrec {
732 	uint64_t	jsr_seq;	/* Our sequence number */
733 	uint64_t	jsr_oldest;	/* Oldest valid sequence number */
734 	uint16_t	jsr_cnt;	/* Count of valid records */
735 	uint16_t	jsr_blocks;	/* Count of device bsize blocks. */
736 	uint32_t	jsr_crc;	/* 32bit crc of the valid space */
737 	ufs_time_t	jsr_time;	/* timestamp for mount instance */
738 };
739 
740 /*
741  * Reference record.  Records a single link count modification.
742  */
743 struct jrefrec {
744 	uint32_t	jr_op;
745 	uint32_t	jr_ino;
746 	uint32_t	jr_parent;
747 	uint16_t	jr_nlink;
748 	uint16_t	jr_mode;
749 	int64_t		jr_diroff;
750 	uint64_t	jr_unused;
751 };
752 
753 /*
754  * Move record.  Records a reference moving within a directory block.  The
755  * nlink is unchanged but we must search both locations.
756  */
757 struct jmvrec {
758 	uint32_t	jm_op;
759 	uint32_t	jm_ino;
760 	uint32_t	jm_parent;
761 	uint16_t	jm_unused;
762 	int64_t		jm_oldoff;
763 	int64_t		jm_newoff;
764 };
765 
766 /*
767  * Block record.  A set of frags or tree of blocks starting at an indirect are
768  * freed or a set of frags are allocated.
769  */
770 struct jblkrec {
771 	uint32_t	jb_op;
772 	uint32_t	jb_ino;
773 	ufs2_daddr_t	jb_blkno;
774 	ufs_lbn_t	jb_lbn;
775 	uint16_t	jb_frags;
776 	uint16_t	jb_oldfrags;
777 	uint32_t	jb_unused;
778 };
779 
780 /*
781  * Truncation record.  Records a partial truncation so that it may be
782  * completed at check time.  Also used for sync records.
783  */
784 struct jtrncrec {
785 	uint32_t	jt_op;
786 	uint32_t	jt_ino;
787 	int64_t		jt_size;
788 	uint32_t	jt_extsize;
789 	uint32_t	jt_pad[3];
790 };
791 
792 union jrec {
793 	struct jsegrec	rec_jsegrec;
794 	struct jrefrec	rec_jrefrec;
795 	struct jmvrec	rec_jmvrec;
796 	struct jblkrec	rec_jblkrec;
797 	struct jtrncrec	rec_jtrncrec;
798 };
799 
800 #ifdef CTASSERT
801 CTASSERT(sizeof(struct jsegrec) == JREC_SIZE);
802 CTASSERT(sizeof(struct jrefrec) == JREC_SIZE);
803 CTASSERT(sizeof(struct jmvrec) == JREC_SIZE);
804 CTASSERT(sizeof(struct jblkrec) == JREC_SIZE);
805 CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE);
806 CTASSERT(sizeof(union jrec) == JREC_SIZE);
807 #endif
808 
809 extern int inside[], around[];
810 extern u_char *fragtbl[];
811 
812 /*
813  * IOCTLs used for filesystem write suspension.
814  */
815 #define	UFSSUSPEND	_IOW('U', 1, fsid_t)
816 #define	UFSRESUME	_IO('U', 2)
817 
818 #endif
819