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
32#ifndef _UFS_FFS_FS_H_
33#define _UFS_FFS_FS_H_
34
35#include <sys/mount.h>
36#include <ufs/ufs/dinode.h>
37
38/*
39 * Each disk drive contains some number of filesystems.
40 * A filesystem consists of a number of cylinder groups.
41 * Each cylinder group has inodes and data.
42 *
43 * A filesystem is described by its super-block, which in turn
44 * describes the cylinder groups. The super-block is critical
45 * data and is replicated in each cylinder group to protect against
46 * catastrophic loss. This is done at `newfs' time and the critical
47 * super-block data does not change, so the copies need not be
48 * referenced further unless disaster strikes.
49 *
50 * For filesystem fs, the offsets of the various blocks of interest
51 * are given in the super block as:
52 * [fs->fs_sblkno] Super-block
53 * [fs->fs_cblkno] Cylinder group block
54 * [fs->fs_iblkno] Inode blocks
55 * [fs->fs_dblkno] Data blocks
56 * The beginning of cylinder group cg in fs, is given by
57 * the ``cgbase(fs, cg)'' macro.
58 *
59 * Depending on the architecture and the media, the superblock may
60 * reside in any one of four places. For tiny media where every block
61 * counts, it is placed at the very front of the partition. Historically,
62 * UFS1 placed it 8K from the front to leave room for the disk label and
63 * a small bootstrap. For UFS2 it got moved to 64K from the front to leave
64 * room for the disk label and a bigger bootstrap, and for really piggy
65 * systems we check at 256K from the front if the first three fail. In
66 * all cases the size of the superblock will be SBLOCKSIZE. All values are
67 * given in byte-offset form, so they do not imply a sector size. The
68 * SBLOCKSEARCH specifies the order in which the locations should be searched.
69 */
70#define SBLOCK_FLOPPY 0
71#define SBLOCK_UFS1 8192
72#define SBLOCK_UFS2 65536
73#define SBLOCK_PIGGY 262144
74#define SBLOCKSIZE 8192
75#define SBLOCKSEARCH \
76 { SBLOCK_UFS2, SBLOCK_UFS1, SBLOCK_FLOPPY, SBLOCK_PIGGY, -1 }
77/*
78 * Request standard superblock location in ffs_sbget().
79 */
80#define UFS_STDSB -1 /* Search standard places for superblock */
81
82/*
83 * UFS_NOMSG indicates that superblock inconsistency error messages
84 * should not be printed. It is used by programs like fsck that
85 * want to print their own error message.
86 *
87 * UFS_NOCSUM causes only the superblock itself to be returned, but does
88 * not read in any auxiliary data structures like the cylinder group
89 * summary information. It is used by clients like glabel that just
90 * want to check for possible filesystem types. Using UFS_NOCSUM
91 * skips the superblock checks for csum data which allows superblocks
92 * that have corrupted csum data to be read and used.
93 *
94 * UFS_NOHASHFAIL will note that the check hash is wrong but will still
95 * return the superblock. This is used by the bootstrap code to
96 * give the system a chance to come up so that fsck can be run to
97 * correct the problem.
98 *
99 * UFS_NOWARNFAIL will warn about inconsistencies but still return the
100 * superblock. It includes UFS_NOHASHFAIL. UFS_NOWARNFAIL is used by
101 * programs like fsck_ffs(8) to debug broken filesystems.
102 *
103 * UFS_FSRONLY will only validate the superblock fields needed to
104 * calculate where the backup filesystem superblocks are located.
105 * If these values pass their validation tests, then the superblock
106 * is returned. This flag is used as part of the attempt to find
107 * alternate superblocks when using ffs_sbsearch().
108 */
109#define UFS_NOHASHFAIL 0x0001 /* Ignore check-hash failure */
110#define UFS_NOWARNFAIL 0x0003 /* Ignore non-fatal inconsistencies */
111#define UFS_NOMSG 0x0004 /* Print no error message */
112#define UFS_NOCSUM 0x0008 /* Read just the superblock without csum */
113#define UFS_FSRONLY 0x0010 /* Validate only values needed for recovery
114 of alternate superblocks */
115#define UFS_ALTSBLK 0x1000 /* Flag used internally */
116
117/*
118 * Max number of fragments per block. This value is NOT tweakable.
119 */
120#define MAXFRAG 8
121
122/*
123 * Addresses stored in inodes are capable of addressing fragments
124 * of `blocks'. File system blocks of at most size MAXBSIZE can
125 * be optionally broken into 2, 4, or 8 pieces, each of which is
126 * addressable; these pieces may be DEV_BSIZE, or some multiple of
127 * a DEV_BSIZE unit.
128 *
129 * Large files consist of exclusively large data blocks. To avoid
130 * undue wasted disk space, the last data block of a small file may be
131 * allocated as only as many fragments of a large block as are
132 * necessary. The filesystem format retains only a single pointer
133 * to such a fragment, which is a piece of a single large block that
134 * has been divided. The size of such a fragment is determinable from
135 * information in the inode, using the ``blksize(fs, ip, lbn)'' macro.
136 *
137 * The filesystem records space availability at the fragment level;
138 * to determine block availability, aligned fragments are examined.
139 */
140
141/*
142 * MINBSIZE is the smallest allowable block size.
143 * In order to insure that it is possible to create files of size
144 * 2^32 with only two levels of indirection, MINBSIZE is set to 4096.
145 * MINBSIZE must be big enough to hold a cylinder group block,
146 * thus changes to (struct cg) must keep its size within MINBSIZE.
147 * Note that super blocks are always of size SBLOCKSIZE,
148 * and that both SBLOCKSIZE and MAXBSIZE must be >= MINBSIZE.
149 */
150#define MINBSIZE 4096
151
152/*
153 * The path name on which the filesystem is mounted is maintained
154 * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in
155 * the super block for this name.
156 */
157#define MAXMNTLEN 468
158
159/*
160 * The volume name for this filesystem is maintained in fs_volname.
161 * MAXVOLLEN defines the length of the buffer allocated.
162 */
163#define MAXVOLLEN 32
164
165/*
166 * There is a 128-byte region in the superblock reserved for in-core
167 * pointers to summary information. Originally this included an array
168 * of pointers to blocks of struct csum; now there are just a few
169 * pointers and the remaining space is padded with fs_ocsp[].
170 *
171 * NOCSPTRS determines the size of this padding. Historically this
172 * space was used to store pointers to structures that summaried
173 * filesystem usage and layout information. However, these pointers
174 * left various kernel pointers in the superblock which made otherwise
175 * identical superblocks appear to have differences. So, all the
176 * pointers in the superblock were moved to a fs_summary_info structure
177 * reducing the superblock to having only a single pointer to this
178 * structure. When writing the superblock to disk, this pointer is
179 * temporarily NULL'ed out so that the kernel pointer will not appear
180 * in the on-disk copy of the superblock.
181 */
182#define NOCSPTRS ((128 / sizeof(void *)) - 1)
183
184/*
185 * A summary of contiguous blocks of various sizes is maintained
186 * in each cylinder group. Normally this is set by the initial
187 * value of fs_maxcontig. To conserve space, a maximum summary size
188 * is set by FS_MAXCONTIG.
189 */
190#define FS_MAXCONTIG 16
191
192/*
193 * MINFREE gives the minimum acceptable percentage of filesystem
194 * blocks which may be free. If the freelist drops below this level
195 * only the superuser may continue to allocate blocks. This may
196 * be set to 0 if no reserve of free blocks is deemed necessary,
197 * however throughput drops by fifty percent if the filesystem
198 * is run at between 95% and 100% full; thus the minimum default
199 * value of fs_minfree is 5%. However, to get good clustering
200 * performance, 10% is a better choice. hence we use 10% as our
201 * default value. With 10% free space, fragmentation is not a
202 * problem, so we choose to optimize for time.
203 */
204#define MINFREE 8
205#define DEFAULTOPT FS_OPTTIME
206
207/*
208 * Grigoriy Orlov <gluk@ptci.ru> has done some extensive work to fine
209 * tune the layout preferences for directories within a filesystem.
210 * His algorithm can be tuned by adjusting the following parameters
211 * which tell the system the average file size and the average number
212 * of files per directory. These defaults are well selected for typical
213 * filesystems, but may need to be tuned for odd cases like filesystems
214 * being used for squid caches or news spools.
215 */
216#define AVFILESIZ 16384 /* expected average file size */
217#define AFPDIR 64 /* expected number of files per directory */
218
219/*
220 * The maximum number of snapshot nodes that can be associated
221 * with each filesystem. This limit affects only the number of
222 * snapshot files that can be recorded within the superblock so
223 * that they can be found when the filesystem is mounted. However,
224 * maintaining too many will slow the filesystem performance, so
225 * having this limit is a good idea.
226 */
227#define FSMAXSNAP 20
228
229/*
230 * Used to identify special blocks in snapshots:
231 *
232 * BLK_NOCOPY - A block that was unallocated at the time the snapshot
233 * was taken, hence does not need to be copied when written.
234 * BLK_SNAP - A block held by another snapshot that is not needed by this
235 * snapshot. When the other snapshot is freed, the BLK_SNAP entries
236 * are converted to BLK_NOCOPY. These are needed to allow fsck to
237 * identify blocks that are in use by other snapshots (which are
238 * expunged from this snapshot).
239 */
240#define BLK_NOCOPY ((ufs2_daddr_t)(1))
241#define BLK_SNAP ((ufs2_daddr_t)(2))
242
243/*
244 * Sysctl values for the fast filesystem.
245 */
246#define FFS_ADJ_REFCNT 1 /* adjust inode reference count */
247#define FFS_ADJ_BLKCNT 2 /* adjust inode used block count */
248#define FFS_BLK_FREE 3 /* free range of blocks in map */
249#define FFS_DIR_FREE 4 /* free specified dir inodes in map */
250#define FFS_FILE_FREE 5 /* free specified file inodes in map */
251#define FFS_SET_FLAGS 6 /* set filesystem flags */
252#define FFS_ADJ_NDIR 7 /* adjust number of directories */
253#define FFS_ADJ_NBFREE 8 /* adjust number of free blocks */
254#define FFS_ADJ_NIFREE 9 /* adjust number of free inodes */
255#define FFS_ADJ_NFFREE 10 /* adjust number of free frags */
256#define FFS_ADJ_NUMCLUSTERS 11 /* adjust number of free clusters */
257#define FFS_SET_CWD 12 /* set current directory */
258#define FFS_SET_DOTDOT 13 /* set inode number for ".." */
259#define FFS_UNLINK 14 /* remove a name in the filesystem */
260/* Was FFS_SET_INODE 15 */
261/* Was FFS_SET_BUFOUTPUT 16 */
262#define FFS_SET_SIZE 17 /* set inode size */
263#define FFS_ADJ_DEPTH 18 /* adjust directory inode depth */
264#define FFS_MAXID 18 /* number of valid ffs ids */
265
266/*
267 * Command structure passed in to the filesystem to adjust filesystem values.
268 */
269#define FFS_CMD_VERSION 0x19790518 /* version ID */
270struct fsck_cmd {
271 int32_t version; /* version of command structure */
272 int32_t handle; /* reference to filesystem to be changed */
273 int64_t value; /* inode or block number to be affected */
274 int64_t size; /* amount or range to be adjusted */
275 int64_t spare; /* reserved for future use */
276};
277
278/*
279 * A recovery structure placed at the end of the boot block area by newfs
280 * that can be used by fsck to search for alternate superblocks.
281 */
282struct fsrecovery {
283 int32_t fsr_magic; /* magic number */
284 int32_t fsr_fsbtodb; /* fsbtodb and dbtofsb shift constant */
285 int32_t fsr_sblkno; /* offset of super-block in filesys */
286 int32_t fsr_fpg; /* blocks per group * fs_frag */
287 uint32_t fsr_ncg; /* number of cylinder groups */
288};
289
290/*
291 * Per cylinder group information; summarized in blocks allocated
292 * from first cylinder group data blocks. These blocks have to be
293 * read in from fs_csaddr (size fs_cssize) in addition to the
294 * super block.
295 */
296struct csum {
297 int32_t cs_ndir; /* number of directories */
298 int32_t cs_nbfree; /* number of free blocks */
299 int32_t cs_nifree; /* number of free inodes */
300 int32_t cs_nffree; /* number of free frags */
301};
302struct csum_total {
303 int64_t cs_ndir; /* number of directories */
304 int64_t cs_nbfree; /* number of free blocks */
305 int64_t cs_nifree; /* number of free inodes */
306 int64_t cs_nffree; /* number of free frags */
307 int64_t cs_numclusters; /* number of free clusters */
308 int64_t cs_spare[3]; /* future expansion */
309};
310
311/*
312 * Pointers to super block summary information. Placed in a separate
313 * structure so there is just one pointer in the superblock.
314 *
315 * The pointers in this structure are used as follows:
316 * fs_contigdirs references an array that tracks the creation of new
317 * directories
318 * fs_csp references a contiguous array of struct csum for
319 * all cylinder groups
320 * fs_maxcluster references an array of cluster sizes that is computed
321 * as cylinder groups are inspected
322 * fs_active is used when creating snapshots; it points to a bitmap
323 * of cylinder groups for which the free-block bitmap has changed
324 * since the snapshot operation began.
325 */
326struct fs_summary_info {
327 uint8_t *si_contigdirs; /* (u) # of contig. allocated dirs */
328 struct csum *si_csp; /* (u) cg summary info buffer */
329 int32_t *si_maxcluster; /* (u) max cluster in each cyl group */
330 uint64_t *si_active; /* (u) used by snapshots to track fs */
331};
332#define fs_contigdirs fs_si->si_contigdirs
333#define fs_csp fs_si->si_csp
334#define fs_maxcluster fs_si->si_maxcluster
335#define fs_active fs_si->si_active
336
337/*
338 * Super block for an FFS filesystem.
339 */
340struct fs {
341 int32_t fs_firstfield; /* historic filesystem linked list, */
342 int32_t fs_unused_1; /* used for incore super blocks */
343 int32_t fs_sblkno; /* offset of super-block in filesys */
344 int32_t fs_cblkno; /* offset of cyl-block in filesys */
345 int32_t fs_iblkno; /* offset of inode-blocks in filesys */
346 int32_t fs_dblkno; /* offset of first data after cg */
347 int32_t fs_old_cgoffset; /* cylinder group offset in cylinder */
348 int32_t fs_old_cgmask; /* used to calc mod fs_ntrak */
349 int32_t fs_old_time; /* last time written */
350 int32_t fs_old_size; /* number of blocks in fs */
351 int32_t fs_old_dsize; /* number of data blocks in fs */
352 uint32_t fs_ncg; /* number of cylinder groups */
353 int32_t fs_bsize; /* size of basic blocks in fs */
354 int32_t fs_fsize; /* size of frag blocks in fs */
355 int32_t fs_frag; /* number of frags in a block in fs */
356/* these are configuration parameters */
357 int32_t fs_minfree; /* minimum percentage of free blocks */
358 int32_t fs_old_rotdelay; /* num of ms for optimal next block */
359 int32_t fs_old_rps; /* disk revolutions per second */
360/* these fields can be computed from the others */
361 int32_t fs_bmask; /* ``blkoff'' calc of blk offsets */
362 int32_t fs_fmask; /* ``fragoff'' calc of frag offsets */
363 int32_t fs_bshift; /* ``lblkno'' calc of logical blkno */
364 int32_t fs_fshift; /* ``numfrags'' calc number of frags */
365/* these are configuration parameters */
366 int32_t fs_maxcontig; /* max number of contiguous blks */
367 int32_t fs_maxbpg; /* max number of blks per cyl group */
368/* these fields can be computed from the others */
369 int32_t fs_fragshift; /* block to frag shift */
370 int32_t fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */
371 int32_t fs_sbsize; /* actual size of super block */
372 int32_t fs_spare1[2]; /* old fs_csmask */
373 /* old fs_csshift */
374 int32_t fs_nindir; /* value of NINDIR */
375 uint32_t fs_inopb; /* value of INOPB */
376 int32_t fs_old_nspf; /* value of NSPF */
377/* yet another configuration parameter */
378 int32_t fs_optim; /* optimization preference, see below */
379 int32_t fs_old_npsect; /* # sectors/track including spares */
380 int32_t fs_old_interleave; /* hardware sector interleave */
381 int32_t fs_old_trackskew; /* sector 0 skew, per track */
382 int32_t fs_id[2]; /* unique filesystem id */
383/* sizes determined by number of cylinder groups and their sizes */
384 int32_t fs_old_csaddr; /* blk addr of cyl grp summary area */
385 int32_t fs_cssize; /* size of cyl grp summary area */
386 int32_t fs_cgsize; /* cylinder group size */
387 int32_t fs_spare2; /* old fs_ntrak */
388 int32_t fs_old_nsect; /* sectors per track */
389 int32_t fs_old_spc; /* sectors per cylinder */
390 int32_t fs_old_ncyl; /* cylinders in filesystem */
391 int32_t fs_old_cpg; /* cylinders per group */
392 uint32_t fs_ipg; /* inodes per group */
393 int32_t fs_fpg; /* blocks per group * fs_frag */
394/* this data must be re-computed after crashes */
395 struct csum fs_old_cstotal; /* cylinder summary information */
396/* these fields are cleared at mount time */
397 int8_t fs_fmod; /* super block modified flag */
398 int8_t fs_clean; /* filesystem is clean flag */
399 int8_t fs_ronly; /* mounted read-only flag */
400 int8_t fs_old_flags; /* old FS_ flags */
401 uint8_t fs_fsmnt[MAXMNTLEN]; /* name mounted on */
402 uint8_t fs_volname[MAXVOLLEN]; /* volume name */
403 uint64_t fs_swuid; /* system-wide uid */
404 int32_t fs_pad; /* due to alignment of fs_swuid */
405/* these fields retain the current block allocation info */
406 int32_t fs_cgrotor; /* last cg searched */
407 void *fs_ocsp[NOCSPTRS]; /* padding; was list of fs_cs buffers */
408 struct fs_summary_info *fs_si;/* In-core pointer to summary info */
409 int32_t fs_old_cpc; /* cyl per cycle in postbl */
410 int32_t fs_maxbsize; /* maximum blocking factor permitted */
411 int64_t fs_unrefs; /* number of unreferenced inodes */
412 int64_t fs_providersize; /* size of underlying GEOM provider */
413 int64_t fs_metaspace; /* size of area reserved for metadata */
414 uint64_t fs_save_maxfilesize; /* save old UFS1 maxfilesize */
415 int64_t fs_sparecon64[12]; /* old rotation block list head */
416 int64_t fs_sblockactualloc; /* byte offset of this superblock */
417 int64_t fs_sblockloc; /* byte offset of standard superblock */
418 struct csum_total fs_cstotal; /* (u) cylinder summary information */
419 ufs_time_t fs_time; /* last time written */
420 int64_t fs_size; /* number of blocks in fs */
421 int64_t fs_dsize; /* number of data blocks in fs */
422 ufs2_daddr_t fs_csaddr; /* blk addr of cyl grp summary area */
423 int64_t fs_pendingblocks; /* (u) blocks being freed */
424 uint32_t fs_pendinginodes; /* (u) inodes being freed */
425 uint32_t fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */
426 uint32_t fs_avgfilesize; /* expected average file size */
427 uint32_t fs_avgfpdir; /* expected # of files per directory */
428 uint32_t fs_available_spare; /* old scratch space */
429 ufs_time_t fs_mtime; /* Last mount or fsck time. */
430 int32_t fs_sujfree; /* SUJ free list */
431 int32_t fs_sparecon32[21]; /* reserved for future constants */
432 uint32_t fs_ckhash; /* if CK_SUPERBLOCK, its check-hash */
433 uint32_t fs_metackhash; /* metadata check-hash, see CK_ below */
434 int32_t fs_flags; /* see FS_ flags below */
435 int32_t fs_contigsumsize; /* size of cluster summary array */
436 int32_t fs_maxsymlinklen; /* max length of an internal symlink */
437 int32_t fs_old_inodefmt; /* format of on-disk inodes */
438 uint64_t fs_maxfilesize; /* maximum representable file size */
439 int64_t fs_qbmask; /* ~fs_bmask for use with 64-bit size */
440 int64_t fs_qfmask; /* ~fs_fmask for use with 64-bit size */
441 int32_t fs_state; /* validate fs_clean field */
442 int32_t fs_old_postblformat; /* format of positional layout tables */
443 int32_t fs_old_nrpos; /* number of rotational positions */
444 int32_t fs_spare5[2]; /* old fs_postbloff */
445 /* old fs_rotbloff */
446 int32_t fs_magic; /* magic number */
447};
448
449/* Sanity checking. */
450#ifdef CTASSERT
451CTASSERT(sizeof(struct fs) == 1376);
452#endif
453
454/*
455 * Filesystem identification
456 */
457#define FS_UFS1_MAGIC 0x011954 /* UFS1 fast filesystem magic number */
458#define FS_UFS2_MAGIC 0x19540119 /* UFS2 fast filesystem magic number */
459#define FS_BAD_MAGIC 0x19960408 /* UFS incomplete newfs magic number */
460#define FS_42INODEFMT -1 /* 4.2BSD inode format */
461#define FS_44INODEFMT 2 /* 4.4BSD inode format */
462
463/*
464 * Preference for optimization.
465 */
466#define FS_OPTTIME 0 /* minimize allocation time */
467#define FS_OPTSPACE 1 /* minimize disk fragmentation */
468
469/*
470 * Filesystem flags.
471 *
472 * The FS_UNCLEAN flag is set by the kernel when the filesystem was
473 * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates
474 * that the filesystem should be managed by the soft updates code.
475 * Note that the FS_NEEDSFSCK flag is set and cleared by the fsck
476 * utility. It is set when background fsck finds an unexpected
477 * inconsistency which requires a traditional foreground fsck to be
478 * run. Such inconsistencies should only be found after an uncorrectable
479 * disk error. The FS_NEEDSFSCK can also be set when a mounted filesystem
480 * discovers an internal inconsistency such as freeing a freed inode.
481 * A foreground fsck will clear the FS_NEEDSFSCK flag when it has
482 * successfully cleaned up the filesystem. The kernel uses this
483 * flag to enforce that inconsistent filesystems be mounted read-only.
484 *
485 * The FS_METACKHASH flag when set indicates that the kernel maintains
486 * one or more check hashes. The actual set of supported check hashes
487 * is stored in the fs_metackhash field. Kernels that do not support
488 * check hashes clear the FS_METACKHASH flag to indicate that the
489 * check hashes need to be rebuilt (by fsck) before they can be used.
490 *
491 * When a filesystem is mounted, any flags not included in FS_SUPPORTED
492 * are cleared. This lets newer features know that the filesystem has
493 * been run on an older version of the filesystem and thus that data
494 * structures associated with those features are out-of-date and need
495 * to be rebuilt.
496 *
497 * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled
498 * for the file system, so they should be loaded from extended attributes,
499 * observed for access control purposes, and be administered by object
500 * owners. FS_NFS4ACLS indicates that NFSv4 ACLs are administratively
501 * enabled. This flag is mutually exclusive with FS_ACLS. FS_MULTILABEL
502 * indicates that the TrustedBSD MAC Framework should attempt to back MAC
503 * labels into extended attributes on the file system rather than maintain
504 * a single mount label for all objects.
505 */
506#define FS_UNCLEAN 0x00000001 /* filesystem not clean at mount */
507#define FS_DOSOFTDEP 0x00000002 /* filesystem using soft dependencies */
508#define FS_NEEDSFSCK 0x00000004 /* filesystem needs sync fsck before mount */
509#define FS_SUJ 0x00000008 /* Filesystem using softupdate journal */
510#define FS_ACLS 0x00000010 /* file system has POSIX.1e ACLs enabled */
511#define FS_MULTILABEL 0x00000020 /* file system is MAC multi-label */
512#define FS_GJOURNAL 0x00000040 /* gjournaled file system */
513#define FS_FLAGS_UPDATED 0x0000080 /* flags have been moved to new location */
514#define FS_NFS4ACLS 0x00000100 /* file system has NFSv4 ACLs enabled */
515#define FS_METACKHASH 0x00000200 /* kernel supports metadata check hashes */
516#define FS_TRIM 0x00000400 /* issue BIO_DELETE for deleted blocks */
517#define FS_SUPPORTED 0x00FFFFFF /* supported flags, others cleared at mount*/
518/*
519 * Things that we may someday support, but currently do not.
520 * These flags are all cleared so we know if we ran on a kernel
521 * that does not support them.
522 */
523#define FS_INDEXDIRS 0x01000000 /* kernel supports indexed directories */
524#define FS_VARBLKSIZE 0x02000000 /* kernel supports variable block sizes */
525#define FS_COOLOPT1 0x04000000 /* kernel supports cool option 1 */
526#define FS_COOLOPT2 0x08000000 /* kernel supports cool option 2 */
527#define FS_COOLOPT3 0x10000000 /* kernel supports cool option 3 */
528#define FS_COOLOPT4 0x20000000 /* kernel supports cool option 4 */
529#define FS_COOLOPT5 0x40000000 /* kernel supports cool option 5 */
530#define FS_COOLOPT6 0x80000000 /* kernel supports cool option 6 */
531
532/*
533 * The fs_metackhash field indicates the types of metadata check-hash
534 * that are maintained for a filesystem. Not all filesystems check-hash
535 * all metadata.
536 */
537#define CK_SUPERBLOCK 0x0001 /* the superblock */
538#define CK_CYLGRP 0x0002 /* the cylinder groups */
539#define CK_INODE 0x0004 /* inodes */
540#define CK_INDIR 0x0008 /* indirect blocks */
541#define CK_DIR 0x0010 /* directory contents */
542#define CK_SUPPORTED 0x0007 /* supported flags, others cleared at mount */
543/*
544 * The BX_FSPRIV buffer b_xflags are used to track types of data in buffers.
545 */
546#define BX_SUPERBLOCK 0x00010000 /* superblock */
547#define BX_CYLGRP 0x00020000 /* cylinder groups */
548#define BX_INODE 0x00040000 /* inodes */
549#define BX_INDIR 0x00080000 /* indirect blocks */
550#define BX_DIR 0x00100000 /* directory contents */
551
552#define PRINT_UFS_BUF_XFLAGS "\20\25dir\24indir\23inode\22cylgrp\21superblock"
553
554/*
555 * Macros to access bits in the fs_active array.
556 */
557#define ACTIVECGNUM(fs, cg) ((fs)->fs_active[(cg) / (NBBY * sizeof(uint64_t))])
558#define ACTIVECGOFF(cg) (1 << ((cg) % (NBBY * sizeof(uint64_t))))
559#define ACTIVESET(fs, cg) do { \
560 if ((fs)->fs_active) \
561 ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg)); \
562} while (0)
563#define ACTIVECLEAR(fs, cg) do { \
564 if ((fs)->fs_active) \
565 ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg)); \
566} while (0)
567
568/*
569 * The size of a cylinder group is calculated by CGSIZE. The maximum size
570 * is limited by the fact that cylinder groups are at most one block.
571 * Its size is derived from the size of the maps maintained in the
572 * cylinder group and the (struct cg) size.
573 */
574#define CGSIZE(fs) \
575 /* base cg */ (sizeof(struct cg) + \
576 /* old btotoff */ (fs)->fs_old_cpg * sizeof(int32_t) + \
577 /* old boff */ (fs)->fs_old_cpg * sizeof(uint16_t) + \
578 /* inode map */ howmany((fs)->fs_ipg, NBBY) + \
579 /* block map */ howmany((fs)->fs_fpg, NBBY) + sizeof(int32_t) + \
580 /* if present */ ((fs)->fs_contigsumsize <= 0 ? 0 : \
581 /* cluster sum */ (fs)->fs_contigsumsize * sizeof(int32_t) + \
582 /* cluster map */ howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY)))
583
584/*
585 * The minimal number of cylinder groups that should be created.
586 */
587#define MINCYLGRPS 4
588
589/*
590 * Convert cylinder group to base address of its global summary info.
591 */
592#define fs_cs(fs, indx) fs_csp[indx]
593
594/*
595 * Cylinder group block for a filesystem.
596 */
597#define CG_MAGIC 0x090255
598struct cg {
599 int32_t cg_firstfield; /* historic cyl groups linked list */
600 int32_t cg_magic; /* magic number */
601 int32_t cg_old_time; /* time last written */
602 uint32_t cg_cgx; /* we are the cgx'th cylinder group */
603 int16_t cg_old_ncyl; /* number of cyl's this cg */
604 int16_t cg_old_niblk; /* number of inode blocks this cg */
605 uint32_t cg_ndblk; /* number of data blocks this cg */
606 struct csum cg_cs; /* cylinder summary information */
607 uint32_t cg_rotor; /* position of last used block */
608 uint32_t cg_frotor; /* position of last used frag */
609 uint32_t cg_irotor; /* position of last used inode */
610 uint32_t cg_frsum[MAXFRAG]; /* counts of available frags */
611 int32_t cg_old_btotoff; /* (int32) block totals per cylinder */
612 int32_t cg_old_boff; /* (uint16) free block positions */
613 uint32_t cg_iusedoff; /* (uint8) used inode map */
614 uint32_t cg_freeoff; /* (uint8) free block map */
615 uint32_t cg_nextfreeoff; /* (uint8) next available space */
616 uint32_t cg_clustersumoff; /* (uint32) counts of avail clusters */
617 uint32_t cg_clusteroff; /* (uint8) free cluster map */
618 uint32_t cg_nclusterblks; /* number of clusters this cg */
619 uint32_t cg_niblk; /* number of inode blocks this cg */
620 uint32_t cg_initediblk; /* last initialized inode */
621 uint32_t cg_unrefs; /* number of unreferenced inodes */
622 int32_t cg_sparecon32[1]; /* reserved for future use */
623 uint32_t cg_ckhash; /* check-hash of this cg */
624 ufs_time_t cg_time; /* time last written */
625 int64_t cg_sparecon64[3]; /* reserved for future use */
626 /* actually longer - space used for cylinder group maps */
627};
628
629/*
630 * Macros for access to cylinder group array structures
631 */
632#define cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC)
633#define cg_inosused(cgp) \
634 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_iusedoff))
635#define cg_blksfree(cgp) \
636 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_freeoff))
637#define cg_clustersfree(cgp) \
638 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_clusteroff))
639#define cg_clustersum(cgp) \
640 ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff))
641
642/*
643 * Turn filesystem block numbers into disk block addresses.
644 * This maps filesystem blocks to device size blocks.
645 */
646#define fsbtodb(fs, b) ((daddr_t)(b) << (fs)->fs_fsbtodb)
647#define dbtofsb(fs, b) ((b) >> (fs)->fs_fsbtodb)
648
649/*
650 * Cylinder group macros to locate things in cylinder groups.
651 * They calc filesystem addresses of cylinder group data structures.
652 */
653#define cgbase(fs, c) (((ufs2_daddr_t)(fs)->fs_fpg) * (c))
654#define cgdata(fs, c) (cgdmin(fs, c) + (fs)->fs_metaspace) /* data zone */
655#define cgmeta(fs, c) (cgdmin(fs, c)) /* meta data */
656#define cgdmin(fs, c) (cgstart(fs, c) + (fs)->fs_dblkno) /* 1st data */
657#define cgimin(fs, c) (cgstart(fs, c) + (fs)->fs_iblkno) /* inode blk */
658#define cgsblock(fs, c) (cgstart(fs, c) + (fs)->fs_sblkno) /* super blk */
659#define cgtod(fs, c) (cgstart(fs, c) + (fs)->fs_cblkno) /* cg block */
660#define cgstart(fs, c) \
661 ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) : \
662 (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask))))
663
664/*
665 * Macros for handling inode numbers:
666 * inode number to filesystem block offset.
667 * inode number to cylinder group number.
668 * inode number to filesystem block address.
669 */
670#define ino_to_cg(fs, x) (((ino_t)(x)) / (fs)->fs_ipg)
671#define ino_to_fsba(fs, x) \
672 ((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) + \
673 (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs))))))
674#define ino_to_fsbo(fs, x) (((ino_t)(x)) % INOPB(fs))
675
676/*
677 * Give cylinder group number for a filesystem block.
678 * Give cylinder group block number for a filesystem block.
679 */
680#define dtog(fs, d) ((d) / (fs)->fs_fpg)
681#define dtogd(fs, d) ((d) % (fs)->fs_fpg)
682
683/*
684 * Extract the bits for a block from a map.
685 * Compute the cylinder and rotational position of a cyl block addr.
686 */
687#define blkmap(fs, map, loc) \
688 (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag)))
689
690/*
691 * The following macros optimize certain frequently calculated
692 * quantities by using shifts and masks in place of divisions
693 * modulos and multiplications.
694 */
695#define blkoff(fs, loc) /* calculates (loc % fs->fs_bsize) */ \
696 ((loc) & (fs)->fs_qbmask)
697#define fragoff(fs, loc) /* calculates (loc % fs->fs_fsize) */ \
698 ((loc) & (fs)->fs_qfmask)
699#define lfragtosize(fs, frag) /* calculates ((off_t)frag * fs->fs_fsize) */ \
700 (((off_t)(frag)) << (fs)->fs_fshift)
701#define lblktosize(fs, blk) /* calculates ((off_t)blk * fs->fs_bsize) */ \
702 (((off_t)(blk)) << (fs)->fs_bshift)
703/* Use this only when `blk' is known to be small, e.g., < UFS_NDADDR. */
704#define smalllblktosize(fs, blk) /* calculates (blk * fs->fs_bsize) */ \
705 ((blk) << (fs)->fs_bshift)
706#define lblkno(fs, loc) /* calculates (loc / fs->fs_bsize) */ \
707 ((loc) >> (fs)->fs_bshift)
708#define numfrags(fs, loc) /* calculates (loc / fs->fs_fsize) */ \
709 ((loc) >> (fs)->fs_fshift)
710#define blkroundup(fs, size) /* calculates roundup(size, fs->fs_bsize) */ \
711 (((size) + (fs)->fs_qbmask) & (fs)->fs_bmask)
712#define fragroundup(fs, size) /* calculates roundup(size, fs->fs_fsize) */ \
713 (((size) + (fs)->fs_qfmask) & (fs)->fs_fmask)
714#define fragstoblks(fs, frags) /* calculates (frags / fs->fs_frag) */ \
715 ((frags) >> (fs)->fs_fragshift)
716#define blkstofrags(fs, blks) /* calculates (blks * fs->fs_frag) */ \
717 ((blks) << (fs)->fs_fragshift)
718#define fragnum(fs, fsb) /* calculates (fsb % fs->fs_frag) */ \
719 ((fsb) & ((fs)->fs_frag - 1))
720#define blknum(fs, fsb) /* calculates rounddown(fsb, fs->fs_frag) */ \
721 ((fsb) &~ ((fs)->fs_frag - 1))
722
723/*
724 * Determine the number of available frags given a
725 * percentage to hold in reserve.
726 */
727#define freespace(fs, percentreserved) \
728 (blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \
729 (fs)->fs_cstotal.cs_nffree - \
730 (((off_t)((fs)->fs_dsize)) * (percentreserved) / 100))
731
732/*
733 * Determining the size of a file block in the filesystem.
734 */
735#define blksize(fs, ip, lbn) \
736 (((lbn) >= UFS_NDADDR || (ip)->i_size >= \
737 (uint64_t)smalllblktosize(fs, (lbn) + 1)) \
738 ? (fs)->fs_bsize \
739 : (fragroundup(fs, blkoff(fs, (ip)->i_size))))
740#define sblksize(fs, size, lbn) \
741 (((lbn) >= UFS_NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \
742 ? (fs)->fs_bsize \
743 : (fragroundup(fs, blkoff(fs, (size)))))
744
745/*
746 * Number of indirects in a filesystem block.
747 */
748#define NINDIR(fs) ((fs)->fs_nindir)
749
750/*
751 * Indirect lbns are aligned on UFS_NDADDR addresses where single indirects
752 * are the negated address of the lowest lbn reachable, double indirects
753 * are this lbn - 1 and triple indirects are this lbn - 2. This yields
754 * an unusual bit order to determine level.
755 */
756static inline int
757lbn_level(ufs_lbn_t lbn)
758{
759 if (lbn >= 0)
760 return 0;
761 switch (lbn & 0x3) {
762 case 0:
763 return (0);
764 case 1:
765 break;
766 case 2:
767 return (2);
768 case 3:
769 return (1);
770 default:
771 break;
772 }
773 return (-1);
774}
775
776static inline ufs_lbn_t
777lbn_offset(struct fs *fs, int level)
778{
779 ufs_lbn_t res;
780
781 for (res = 1; level > 0; level--)
782 res *= NINDIR(fs);
783 return (res);
784}
785
786/*
787 * Number of inodes in a secondary storage block/fragment.
788 */
789#define INOPB(fs) ((fs)->fs_inopb)
790#define INOPF(fs) ((fs)->fs_inopb >> (fs)->fs_fragshift)
791
792/*
793 * Softdep journal record format.
794 */
795#define JOP_UNKNOWN 0 /* JOP operation is unknown */
796#define JOP_ADDREF 1 /* Add a reference to an inode. */
797#define JOP_REMREF 2 /* Remove a reference from an inode. */
798#define JOP_NEWBLK 3 /* Allocate a block. */
799#define JOP_FREEBLK 4 /* Free a block or a tree of blocks. */
800#define JOP_MVREF 5 /* Move a reference from one off to another. */
801#define JOP_TRUNC 6 /* Partial truncation record. */
802#define JOP_SYNC 7 /* fsync() complete record. */
803#define JOP_NUMJOPTYPES 8
804#define JOP_NAMES { \
805 "unknown", \
806 "JOP_ADDREF", \
807 "JOP_REMREF", \
808 "JOP_NEWBLK", \
809 "JOP_FREEBLK", \
810 "JOP_MVREF", \
811 "JOP_TRUNC", \
812 "JOP_SYNC" }
813#define JOP_OPTYPE(op) \
814 (op) < JOP_NUMJOPTYPES ? joptype[op] : joptype[JOP_UNKNOWN]
815
816#define JREC_SIZE 32 /* Record and segment header size. */
817
818#define SUJ_MIN (4 * 1024 * 1024) /* Minimum journal size */
819#define SUJ_FILE ".sujournal" /* Journal file name */
820
821/*
822 * Size of the segment record header. There is at most one for each disk
823 * block in the journal. The segment header is followed by an array of
824 * records. fsck depends on the first element in each record being 'op'
825 * and the second being 'ino'. Segments may span multiple disk blocks but
826 * the header is present on each.
827 */
828struct jsegrec {
829 uint64_t jsr_seq; /* Our sequence number */
830 uint64_t jsr_oldest; /* Oldest valid sequence number */
831 uint16_t jsr_cnt; /* Count of valid records */
832 uint16_t jsr_blocks; /* Count of device bsize blocks. */
833 uint32_t jsr_crc; /* 32bit crc of the valid space */
834 ufs_time_t jsr_time; /* timestamp for mount instance */
835};
836
837/*
838 * Reference record. Records a single link count modification.
839 */
840struct jrefrec {
841 uint32_t jr_op;
842 uint32_t jr_ino;
843 uint32_t jr_parent;
844 uint16_t jr_nlink;
845 uint16_t jr_mode;
846 int64_t jr_diroff;
847 uint64_t jr_unused;
848};
849
850/*
851 * Move record. Records a reference moving within a directory block. The
852 * nlink is unchanged but we must search both locations.
853 */
854struct jmvrec {
855 uint32_t jm_op;
856 uint32_t jm_ino;
857 uint32_t jm_parent;
858 uint16_t jm_unused;
859 int64_t jm_oldoff;
860 int64_t jm_newoff;
861};
862
863/*
864 * Block record. A set of frags or tree of blocks starting at an indirect are
865 * freed or a set of frags are allocated.
866 */
867struct jblkrec {
868 uint32_t jb_op;
869 uint32_t jb_ino;
870 ufs2_daddr_t jb_blkno;
871 ufs_lbn_t jb_lbn;
872 uint16_t jb_frags;
873 uint16_t jb_oldfrags;
874 uint32_t jb_unused;
875};
876
877/*
878 * Truncation record. Records a partial truncation so that it may be
879 * completed at check time. Also used for sync records.
880 */
881struct jtrncrec {
882 uint32_t jt_op;
883 uint32_t jt_ino;
884 int64_t jt_size;
885 uint32_t jt_extsize;
886 uint32_t jt_pad[3];
887};
888
889union jrec {
890 struct jsegrec rec_jsegrec;
891 struct jrefrec rec_jrefrec;
892 struct jmvrec rec_jmvrec;
893 struct jblkrec rec_jblkrec;
894 struct jtrncrec rec_jtrncrec;
895};
896
897#ifdef CTASSERT
898CTASSERT(sizeof(struct jsegrec) == JREC_SIZE);
899CTASSERT(sizeof(struct jrefrec) == JREC_SIZE);
900CTASSERT(sizeof(struct jmvrec) == JREC_SIZE);
901CTASSERT(sizeof(struct jblkrec) == JREC_SIZE);
902CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE);
903CTASSERT(sizeof(union jrec) == JREC_SIZE);
904#endif
905
906extern int inside[], around[];
907extern uint8_t *fragtbl[];
908
909/*
910 * IOCTLs used for filesystem write suspension.
911 */
912#define UFSSUSPEND _IOW('U', 1, fsid_t)
913#define UFSRESUME _IO('U', 2)
914
915#endif