1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1982, 1986, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the University nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 *
31 * @(#)fs.h 8.13 (Berkeley) 3/21/95
32 */
33
34 #ifndef _UFS_FFS_FS_H_
35 #define _UFS_FFS_FS_H_
36
37 #include <sys/mount.h>
38 #include <ufs/ufs/dinode.h>
39
40 /*
41 * Each disk drive contains some number of filesystems.
42 * A filesystem consists of a number of cylinder groups.
43 * Each cylinder group has inodes and data.
44 *
45 * A filesystem is described by its super-block, which in turn
46 * describes the cylinder groups. The super-block is critical
47 * data and is replicated in each cylinder group to protect against
48 * catastrophic loss. This is done at `newfs' time and the critical
49 * super-block data does not change, so the copies need not be
50 * referenced further unless disaster strikes.
51 *
52 * For filesystem fs, the offsets of the various blocks of interest
53 * are given in the super block as:
54 * [fs->fs_sblkno] Super-block
55 * [fs->fs_cblkno] Cylinder group block
56 * [fs->fs_iblkno] Inode blocks
57 * [fs->fs_dblkno] Data blocks
58 * The beginning of cylinder group cg in fs, is given by
59 * the ``cgbase(fs, cg)'' macro.
60 *
61 * Depending on the architecture and the media, the superblock may
62 * reside in any one of four places. For tiny media where every block
63 * counts, it is placed at the very front of the partition. Historically,
64 * UFS1 placed it 8K from the front to leave room for the disk label and
65 * a small bootstrap. For UFS2 it got moved to 64K from the front to leave
66 * room for the disk label and a bigger bootstrap, and for really piggy
67 * systems we check at 256K from the front if the first three fail. In
68 * all cases the size of the superblock will be SBLOCKSIZE. All values are
69 * given in byte-offset form, so they do not imply a sector size. The
70 * SBLOCKSEARCH specifies the order in which the locations should be searched.
71 */
72 #define SBLOCK_FLOPPY 0
73 #define SBLOCK_UFS1 8192
74 #define SBLOCK_UFS2 65536
75 #define SBLOCK_PIGGY 262144
76 #define SBLOCKSIZE 8192
77 #define SBLOCKSEARCH \
78 { SBLOCK_UFS2, SBLOCK_UFS1, SBLOCK_FLOPPY, SBLOCK_PIGGY, -1 }
79
80 /*
81 * Max number of fragments per block. This value is NOT tweakable.
82 */
83 #define MAXFRAG 8
84
85 /*
86 * Addresses stored in inodes are capable of addressing fragments
87 * of `blocks'. File system blocks of at most size MAXBSIZE can
88 * be optionally broken into 2, 4, or 8 pieces, each of which is
89 * addressable; these pieces may be DEV_BSIZE, or some multiple of
90 * a DEV_BSIZE unit.
91 *
92 * Large files consist of exclusively large data blocks. To avoid
93 * undue wasted disk space, the last data block of a small file may be
94 * allocated as only as many fragments of a large block as are
95 * necessary. The filesystem format retains only a single pointer
96 * to such a fragment, which is a piece of a single large block that
97 * has been divided. The size of such a fragment is determinable from
98 * information in the inode, using the ``blksize(fs, ip, lbn)'' macro.
99 *
100 * The filesystem records space availability at the fragment level;
101 * to determine block availability, aligned fragments are examined.
102 */
103
104 /*
105 * MINBSIZE is the smallest allowable block size.
106 * In order to insure that it is possible to create files of size
107 * 2^32 with only two levels of indirection, MINBSIZE is set to 4096.
108 * MINBSIZE must be big enough to hold a cylinder group block,
109 * thus changes to (struct cg) must keep its size within MINBSIZE.
110 * Note that super blocks are always of size SBLOCKSIZE,
111 * and that both SBLOCKSIZE and MAXBSIZE must be >= MINBSIZE.
112 */
113 #define MINBSIZE 4096
114
115 /*
116 * The path name on which the filesystem is mounted is maintained
117 * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in
118 * the super block for this name.
119 */
120 #define MAXMNTLEN 468
121
122 /*
123 * The volume name for this filesystem is maintained in fs_volname.
124 * MAXVOLLEN defines the length of the buffer allocated.
125 */
126 #define MAXVOLLEN 32
127
128 /*
129 * There is a 128-byte region in the superblock reserved for in-core
130 * pointers to summary information. Originally this included an array
131 * of pointers to blocks of struct csum; now there are just a few
132 * pointers and the remaining space is padded with fs_ocsp[].
133 *
134 * NOCSPTRS determines the size of this padding. Historically this
135 * space was used to store pointers to structures that summaried
136 * filesystem usage and layout information. However, these pointers
137 * left various kernel pointers in the superblock which made otherwise
138 * identical superblocks appear to have differences. So, all the
139 * pointers in the superblock were moved to a fs_summary_info structure
140 * reducing the superblock to having only a single pointer to this
141 * structure. When writing the superblock to disk, this pointer is
142 * temporarily NULL'ed out so that the kernel pointer will not appear
143 * in the on-disk copy of the superblock.
144 */
145 #define NOCSPTRS ((128 / sizeof(void *)) - 1)
146
147 /*
148 * A summary of contiguous blocks of various sizes is maintained
149 * in each cylinder group. Normally this is set by the initial
150 * value of fs_maxcontig. To conserve space, a maximum summary size
151 * is set by FS_MAXCONTIG.
152 */
153 #define FS_MAXCONTIG 16
154
155 /*
156 * MINFREE gives the minimum acceptable percentage of filesystem
157 * blocks which may be free. If the freelist drops below this level
158 * only the superuser may continue to allocate blocks. This may
159 * be set to 0 if no reserve of free blocks is deemed necessary,
160 * however throughput drops by fifty percent if the filesystem
161 * is run at between 95% and 100% full; thus the minimum default
162 * value of fs_minfree is 5%. However, to get good clustering
163 * performance, 10% is a better choice. hence we use 10% as our
164 * default value. With 10% free space, fragmentation is not a
165 * problem, so we choose to optimize for time.
166 */
167 #define MINFREE 8
168 #define DEFAULTOPT FS_OPTTIME
169
170 /*
171 * Grigoriy Orlov <gluk@ptci.ru> has done some extensive work to fine
172 * tune the layout preferences for directories within a filesystem.
173 * His algorithm can be tuned by adjusting the following parameters
174 * which tell the system the average file size and the average number
175 * of files per directory. These defaults are well selected for typical
176 * filesystems, but may need to be tuned for odd cases like filesystems
177 * being used for squid caches or news spools.
178 */
179 #define AVFILESIZ 16384 /* expected average file size */
180 #define AFPDIR 64 /* expected number of files per directory */
181
182 /*
183 * The maximum number of snapshot nodes that can be associated
184 * with each filesystem. This limit affects only the number of
185 * snapshot files that can be recorded within the superblock so
186 * that they can be found when the filesystem is mounted. However,
187 * maintaining too many will slow the filesystem performance, so
188 * having this limit is a good idea.
189 */
190 #define FSMAXSNAP 20
191
192 /*
193 * Used to identify special blocks in snapshots:
194 *
195 * BLK_NOCOPY - A block that was unallocated at the time the snapshot
196 * was taken, hence does not need to be copied when written.
197 * BLK_SNAP - A block held by another snapshot that is not needed by this
198 * snapshot. When the other snapshot is freed, the BLK_SNAP entries
199 * are converted to BLK_NOCOPY. These are needed to allow fsck to
200 * identify blocks that are in use by other snapshots (which are
201 * expunged from this snapshot).
202 */
203 #define BLK_NOCOPY ((ufs2_daddr_t)(1))
204 #define BLK_SNAP ((ufs2_daddr_t)(2))
205
206 /*
207 * Sysctl values for the fast filesystem.
208 */
209 #define FFS_ADJ_REFCNT 1 /* adjust inode reference count */
210 #define FFS_ADJ_BLKCNT 2 /* adjust inode used block count */
211 #define FFS_BLK_FREE 3 /* free range of blocks in map */
212 #define FFS_DIR_FREE 4 /* free specified dir inodes in map */
213 #define FFS_FILE_FREE 5 /* free specified file inodes in map */
214 #define FFS_SET_FLAGS 6 /* set filesystem flags */
215 #define FFS_ADJ_NDIR 7 /* adjust number of directories */
216 #define FFS_ADJ_NBFREE 8 /* adjust number of free blocks */
217 #define FFS_ADJ_NIFREE 9 /* adjust number of free inodes */
218 #define FFS_ADJ_NFFREE 10 /* adjust number of free frags */
219 #define FFS_ADJ_NUMCLUSTERS 11 /* adjust number of free clusters */
220 #define FFS_SET_CWD 12 /* set current directory */
221 #define FFS_SET_DOTDOT 13 /* set inode number for ".." */
222 #define FFS_UNLINK 14 /* remove a name in the filesystem */
223 /* Was FFS_SET_INODE 15 */
224 /* Was FFS_SET_BUFOUTPUT 16 */
225 #define FFS_SET_SIZE 17 /* set inode size */
226 #define FFS_ADJ_DEPTH 18 /* adjust directory inode depth */
227 #define FFS_MAXID 18 /* number of valid ffs ids */
228
229 /*
230 * Command structure passed in to the filesystem to adjust filesystem values.
231 */
232 #define FFS_CMD_VERSION 0x19790518 /* version ID */
233 struct fsck_cmd {
234 int32_t version; /* version of command structure */
235 int32_t handle; /* reference to filesystem to be changed */
236 int64_t value; /* inode or block number to be affected */
237 int64_t size; /* amount or range to be adjusted */
238 int64_t spare; /* reserved for future use */
239 };
240
241 /*
242 * A recovery structure placed at the end of the boot block area by newfs
243 * that can be used by fsck to search for alternate superblocks.
244 */
245 struct fsrecovery {
246 int32_t fsr_magic; /* magic number */
247 int32_t fsr_fsbtodb; /* fsbtodb and dbtofsb shift constant */
248 int32_t fsr_sblkno; /* offset of super-block in filesys */
249 int32_t fsr_fpg; /* blocks per group * fs_frag */
250 uint32_t fsr_ncg; /* number of cylinder groups */
251 };
252
253 /*
254 * Per cylinder group information; summarized in blocks allocated
255 * from first cylinder group data blocks. These blocks have to be
256 * read in from fs_csaddr (size fs_cssize) in addition to the
257 * super block.
258 */
259 struct csum {
260 int32_t cs_ndir; /* number of directories */
261 int32_t cs_nbfree; /* number of free blocks */
262 int32_t cs_nifree; /* number of free inodes */
263 int32_t cs_nffree; /* number of free frags */
264 };
265 struct csum_total {
266 int64_t cs_ndir; /* number of directories */
267 int64_t cs_nbfree; /* number of free blocks */
268 int64_t cs_nifree; /* number of free inodes */
269 int64_t cs_nffree; /* number of free frags */
270 int64_t cs_numclusters; /* number of free clusters */
271 int64_t cs_spare[3]; /* future expansion */
272 };
273
274 /*
275 * Pointers to super block summary information. Placed in a separate
276 * structure so there is just one pointer in the superblock.
277 *
278 * The pointers in this structure are used as follows:
279 * fs_contigdirs references an array that tracks the creation of new
280 * directories
281 * fs_csp references a contiguous array of struct csum for
282 * all cylinder groups
283 * fs_maxcluster references an array of cluster sizes that is computed
284 * as cylinder groups are inspected
285 * fs_active is used when creating snapshots; it points to a bitmap
286 * of cylinder groups for which the free-block bitmap has changed
287 * since the snapshot operation began.
288 */
289 struct fs_summary_info {
290 uint8_t *si_contigdirs; /* (u) # of contig. allocated dirs */
291 struct csum *si_csp; /* (u) cg summary info buffer */
292 int32_t *si_maxcluster; /* (u) max cluster in each cyl group */
293 uint64_t *si_active; /* (u) used by snapshots to track fs */
294 };
295 #define fs_contigdirs fs_si->si_contigdirs
296 #define fs_csp fs_si->si_csp
297 #define fs_maxcluster fs_si->si_maxcluster
298 #define fs_active fs_si->si_active
299
300 /*
301 * Super block for an FFS filesystem.
302 */
303 struct fs {
304 int32_t fs_firstfield; /* historic filesystem linked list, */
305 int32_t fs_unused_1; /* used for incore super blocks */
306 int32_t fs_sblkno; /* offset of super-block in filesys */
307 int32_t fs_cblkno; /* offset of cyl-block in filesys */
308 int32_t fs_iblkno; /* offset of inode-blocks in filesys */
309 int32_t fs_dblkno; /* offset of first data after cg */
310 int32_t fs_old_cgoffset; /* cylinder group offset in cylinder */
311 int32_t fs_old_cgmask; /* used to calc mod fs_ntrak */
312 int32_t fs_old_time; /* last time written */
313 int32_t fs_old_size; /* number of blocks in fs */
314 int32_t fs_old_dsize; /* number of data blocks in fs */
315 uint32_t fs_ncg; /* number of cylinder groups */
316 int32_t fs_bsize; /* size of basic blocks in fs */
317 int32_t fs_fsize; /* size of frag blocks in fs */
318 int32_t fs_frag; /* number of frags in a block in fs */
319 /* these are configuration parameters */
320 int32_t fs_minfree; /* minimum percentage of free blocks */
321 int32_t fs_old_rotdelay; /* num of ms for optimal next block */
322 int32_t fs_old_rps; /* disk revolutions per second */
323 /* these fields can be computed from the others */
324 int32_t fs_bmask; /* ``blkoff'' calc of blk offsets */
325 int32_t fs_fmask; /* ``fragoff'' calc of frag offsets */
326 int32_t fs_bshift; /* ``lblkno'' calc of logical blkno */
327 int32_t fs_fshift; /* ``numfrags'' calc number of frags */
328 /* these are configuration parameters */
329 int32_t fs_maxcontig; /* max number of contiguous blks */
330 int32_t fs_maxbpg; /* max number of blks per cyl group */
331 /* these fields can be computed from the others */
332 int32_t fs_fragshift; /* block to frag shift */
333 int32_t fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */
334 int32_t fs_sbsize; /* actual size of super block */
335 int32_t fs_spare1[2]; /* old fs_csmask */
336 /* old fs_csshift */
337 int32_t fs_nindir; /* value of NINDIR */
338 uint32_t fs_inopb; /* value of INOPB */
339 int32_t fs_old_nspf; /* value of NSPF */
340 /* yet another configuration parameter */
341 int32_t fs_optim; /* optimization preference, see below */
342 int32_t fs_old_npsect; /* # sectors/track including spares */
343 int32_t fs_old_interleave; /* hardware sector interleave */
344 int32_t fs_old_trackskew; /* sector 0 skew, per track */
345 int32_t fs_id[2]; /* unique filesystem id */
346 /* sizes determined by number of cylinder groups and their sizes */
347 int32_t fs_old_csaddr; /* blk addr of cyl grp summary area */
348 int32_t fs_cssize; /* size of cyl grp summary area */
349 int32_t fs_cgsize; /* cylinder group size */
350 int32_t fs_spare2; /* old fs_ntrak */
351 int32_t fs_old_nsect; /* sectors per track */
352 int32_t fs_old_spc; /* sectors per cylinder */
353 int32_t fs_old_ncyl; /* cylinders in filesystem */
354 int32_t fs_old_cpg; /* cylinders per group */
355 uint32_t fs_ipg; /* inodes per group */
356 int32_t fs_fpg; /* blocks per group * fs_frag */
357 /* this data must be re-computed after crashes */
358 struct csum fs_old_cstotal; /* cylinder summary information */
359 /* these fields are cleared at mount time */
360 int8_t fs_fmod; /* super block modified flag */
361 int8_t fs_clean; /* filesystem is clean flag */
362 int8_t fs_ronly; /* mounted read-only flag */
363 int8_t fs_old_flags; /* old FS_ flags */
364 uint8_t fs_fsmnt[MAXMNTLEN]; /* name mounted on */
365 uint8_t fs_volname[MAXVOLLEN]; /* volume name */
366 uint64_t fs_swuid; /* system-wide uid */
367 int32_t fs_pad; /* due to alignment of fs_swuid */
368 /* these fields retain the current block allocation info */
369 int32_t fs_cgrotor; /* last cg searched */
370 void *fs_ocsp[NOCSPTRS]; /* padding; was list of fs_cs buffers */
371 struct fs_summary_info *fs_si;/* In-core pointer to summary info */
372 int32_t fs_old_cpc; /* cyl per cycle in postbl */
373 int32_t fs_maxbsize; /* maximum blocking factor permitted */
374 int64_t fs_unrefs; /* number of unreferenced inodes */
375 int64_t fs_providersize; /* size of underlying GEOM provider */
376 int64_t fs_metaspace; /* size of area reserved for metadata */
377 uint64_t fs_save_maxfilesize; /* save old UFS1 maxfilesize */
378 int64_t fs_sparecon64[12]; /* old rotation block list head */
379 int64_t fs_sblockactualloc; /* byte offset of this superblock */
380 int64_t fs_sblockloc; /* byte offset of standard superblock */
381 struct csum_total fs_cstotal; /* (u) cylinder summary information */
382 ufs_time_t fs_time; /* last time written */
383 int64_t fs_size; /* number of blocks in fs */
384 int64_t fs_dsize; /* number of data blocks in fs */
385 ufs2_daddr_t fs_csaddr; /* blk addr of cyl grp summary area */
386 int64_t fs_pendingblocks; /* (u) blocks being freed */
387 uint32_t fs_pendinginodes; /* (u) inodes being freed */
388 uint32_t fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */
389 uint32_t fs_avgfilesize; /* expected average file size */
390 uint32_t fs_avgfpdir; /* expected # of files per directory */
391 uint32_t fs_available_spare; /* old scratch space */
392 ufs_time_t fs_mtime; /* Last mount or fsck time. */
393 int32_t fs_sujfree; /* SUJ free list */
394 int32_t fs_sparecon32[21]; /* reserved for future constants */
395 uint32_t fs_ckhash; /* if CK_SUPERBLOCK, its check-hash */
396 uint32_t fs_metackhash; /* metadata check-hash, see CK_ below */
397 int32_t fs_flags; /* see FS_ flags below */
398 int32_t fs_contigsumsize; /* size of cluster summary array */
399 int32_t fs_maxsymlinklen; /* max length of an internal symlink */
400 int32_t fs_old_inodefmt; /* format of on-disk inodes */
401 uint64_t fs_maxfilesize; /* maximum representable file size */
402 int64_t fs_qbmask; /* ~fs_bmask for use with 64-bit size */
403 int64_t fs_qfmask; /* ~fs_fmask for use with 64-bit size */
404 int32_t fs_state; /* validate fs_clean field */
405 int32_t fs_old_postblformat; /* format of positional layout tables */
406 int32_t fs_old_nrpos; /* number of rotational positions */
407 int32_t fs_spare5[2]; /* old fs_postbloff */
408 /* old fs_rotbloff */
409 int32_t fs_magic; /* magic number */
410 };
411
412 /* Sanity checking. */
413 #ifdef CTASSERT
414 CTASSERT(sizeof(struct fs) == 1376);
415 #endif
416
417 /*
418 * Filesystem identification
419 */
420 #define FS_UFS1_MAGIC 0x011954 /* UFS1 fast filesystem magic number */
421 #define FS_UFS2_MAGIC 0x19540119 /* UFS2 fast filesystem magic number */
422 #define FS_BAD_MAGIC 0x19960408 /* UFS incomplete newfs magic number */
423 #define FS_42INODEFMT -1 /* 4.2BSD inode format */
424 #define FS_44INODEFMT 2 /* 4.4BSD inode format */
425
426 /*
427 * Preference for optimization.
428 */
429 #define FS_OPTTIME 0 /* minimize allocation time */
430 #define FS_OPTSPACE 1 /* minimize disk fragmentation */
431
432 /*
433 * Filesystem flags.
434 *
435 * The FS_UNCLEAN flag is set by the kernel when the filesystem was
436 * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates
437 * that the filesystem should be managed by the soft updates code.
438 * Note that the FS_NEEDSFSCK flag is set and cleared by the fsck
439 * utility. It is set when background fsck finds an unexpected
440 * inconsistency which requires a traditional foreground fsck to be
441 * run. Such inconsistencies should only be found after an uncorrectable
442 * disk error. The FS_NEEDSFSCK can also be set when a mounted filesystem
443 * discovers an internal inconsistency such as freeing a freed inode.
444 * A foreground fsck will clear the FS_NEEDSFSCK flag when it has
445 * successfully cleaned up the filesystem. The kernel uses this
446 * flag to enforce that inconsistent filesystems be mounted read-only.
447 *
448 * The FS_METACKHASH flag when set indicates that the kernel maintains
449 * one or more check hashes. The actual set of supported check hashes
450 * is stored in the fs_metackhash field. Kernels that do not support
451 * check hashes clear the FS_METACKHASH flag to indicate that the
452 * check hashes need to be rebuilt (by fsck) before they can be used.
453 *
454 * When a filesystem is mounted, any flags not included in FS_SUPPORTED
455 * are cleared. This lets newer features know that the filesystem has
456 * been run on an older version of the filesystem and thus that data
457 * structures associated with those features are out-of-date and need
458 * to be rebuilt.
459 *
460 * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled
461 * for the file system, so they should be loaded from extended attributes,
462 * observed for access control purposes, and be administered by object
463 * owners. FS_NFS4ACLS indicates that NFSv4 ACLs are administratively
464 * enabled. This flag is mutually exclusive with FS_ACLS. FS_MULTILABEL
465 * indicates that the TrustedBSD MAC Framework should attempt to back MAC
466 * labels into extended attributes on the file system rather than maintain
467 * a single mount label for all objects.
468 */
469 #define FS_UNCLEAN 0x00000001 /* filesystem not clean at mount */
470 #define FS_DOSOFTDEP 0x00000002 /* filesystem using soft dependencies */
471 #define FS_NEEDSFSCK 0x00000004 /* filesystem needs sync fsck before mount */
472 #define FS_SUJ 0x00000008 /* Filesystem using softupdate journal */
473 #define FS_ACLS 0x00000010 /* file system has POSIX.1e ACLs enabled */
474 #define FS_MULTILABEL 0x00000020 /* file system is MAC multi-label */
475 #define FS_GJOURNAL 0x00000040 /* gjournaled file system */
476 #define FS_FLAGS_UPDATED 0x0000080 /* flags have been moved to new location */
477 #define FS_NFS4ACLS 0x00000100 /* file system has NFSv4 ACLs enabled */
478 #define FS_METACKHASH 0x00000200 /* kernel supports metadata check hashes */
479 #define FS_TRIM 0x00000400 /* issue BIO_DELETE for deleted blocks */
480 #define FS_SUPPORTED 0x00FFFFFF /* supported flags, others cleared at mount*/
481 /*
482 * Things that we may someday support, but currently do not.
483 * These flags are all cleared so we know if we ran on a kernel
484 * that does not support them.
485 */
486 #define FS_INDEXDIRS 0x01000000 /* kernel supports indexed directories */
487 #define FS_VARBLKSIZE 0x02000000 /* kernel supports variable block sizes */
488 #define FS_COOLOPT1 0x04000000 /* kernel supports cool option 1 */
489 #define FS_COOLOPT2 0x08000000 /* kernel supports cool option 2 */
490 #define FS_COOLOPT3 0x10000000 /* kernel supports cool option 3 */
491 #define FS_COOLOPT4 0x20000000 /* kernel supports cool option 4 */
492 #define FS_COOLOPT5 0x40000000 /* kernel supports cool option 5 */
493 #define FS_COOLOPT6 0x80000000 /* kernel supports cool option 6 */
494
495 /*
496 * The fs_metackhash field indicates the types of metadata check-hash
497 * that are maintained for a filesystem. Not all filesystems check-hash
498 * all metadata.
499 */
500 #define CK_SUPERBLOCK 0x0001 /* the superblock */
501 #define CK_CYLGRP 0x0002 /* the cylinder groups */
502 #define CK_INODE 0x0004 /* inodes */
503 #define CK_INDIR 0x0008 /* indirect blocks */
504 #define CK_DIR 0x0010 /* directory contents */
505 #define CK_SUPPORTED 0x0007 /* supported flags, others cleared at mount */
506 /*
507 * The BX_FSPRIV buffer b_xflags are used to track types of data in buffers.
508 */
509 #define BX_SUPERBLOCK 0x00010000 /* superblock */
510 #define BX_CYLGRP 0x00020000 /* cylinder groups */
511 #define BX_INODE 0x00040000 /* inodes */
512 #define BX_INDIR 0x00080000 /* indirect blocks */
513 #define BX_DIR 0x00100000 /* directory contents */
514
515 #define PRINT_UFS_BUF_XFLAGS "\20\25dir\24indir\23inode\22cylgrp\21superblock"
516
517 /*
518 * Macros to access bits in the fs_active array.
519 */
520 #define ACTIVECGNUM(fs, cg) ((fs)->fs_active[(cg) / (NBBY * sizeof(uint64_t))])
521 #define ACTIVECGOFF(cg) (1 << ((cg) % (NBBY * sizeof(uint64_t))))
522 #define ACTIVESET(fs, cg) do { \
523 if ((fs)->fs_active) \
524 ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg)); \
525 } while (0)
526 #define ACTIVECLEAR(fs, cg) do { \
527 if ((fs)->fs_active) \
528 ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg)); \
529 } while (0)
530
531 /*
532 * The size of a cylinder group is calculated by CGSIZE. The maximum size
533 * is limited by the fact that cylinder groups are at most one block.
534 * Its size is derived from the size of the maps maintained in the
535 * cylinder group and the (struct cg) size.
536 */
537 #define CGSIZE(fs) \
538 /* base cg */ (sizeof(struct cg) + \
539 /* old btotoff */ (fs)->fs_old_cpg * sizeof(int32_t) + \
540 /* old boff */ (fs)->fs_old_cpg * sizeof(uint16_t) + \
541 /* inode map */ howmany((fs)->fs_ipg, NBBY) + \
542 /* block map */ howmany((fs)->fs_fpg, NBBY) + sizeof(int32_t) + \
543 /* if present */ ((fs)->fs_contigsumsize <= 0 ? 0 : \
544 /* cluster sum */ (fs)->fs_contigsumsize * sizeof(int32_t) + \
545 /* cluster map */ howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY)))
546
547 /*
548 * The minimal number of cylinder groups that should be created.
549 */
550 #define MINCYLGRPS 4
551
552 /*
553 * Convert cylinder group to base address of its global summary info.
554 */
555 #define fs_cs(fs, indx) fs_csp[indx]
556
557 /*
558 * Cylinder group block for a filesystem.
559 */
560 #define CG_MAGIC 0x090255
561 struct cg {
562 int32_t cg_firstfield; /* historic cyl groups linked list */
563 int32_t cg_magic; /* magic number */
564 int32_t cg_old_time; /* time last written */
565 uint32_t cg_cgx; /* we are the cgx'th cylinder group */
566 int16_t cg_old_ncyl; /* number of cyl's this cg */
567 int16_t cg_old_niblk; /* number of inode blocks this cg */
568 uint32_t cg_ndblk; /* number of data blocks this cg */
569 struct csum cg_cs; /* cylinder summary information */
570 uint32_t cg_rotor; /* position of last used block */
571 uint32_t cg_frotor; /* position of last used frag */
572 uint32_t cg_irotor; /* position of last used inode */
573 uint32_t cg_frsum[MAXFRAG]; /* counts of available frags */
574 int32_t cg_old_btotoff; /* (int32) block totals per cylinder */
575 int32_t cg_old_boff; /* (uint16) free block positions */
576 uint32_t cg_iusedoff; /* (uint8) used inode map */
577 uint32_t cg_freeoff; /* (uint8) free block map */
578 uint32_t cg_nextfreeoff; /* (uint8) next available space */
579 uint32_t cg_clustersumoff; /* (uint32) counts of avail clusters */
580 uint32_t cg_clusteroff; /* (uint8) free cluster map */
581 uint32_t cg_nclusterblks; /* number of clusters this cg */
582 uint32_t cg_niblk; /* number of inode blocks this cg */
583 uint32_t cg_initediblk; /* last initialized inode */
584 uint32_t cg_unrefs; /* number of unreferenced inodes */
585 int32_t cg_sparecon32[1]; /* reserved for future use */
586 uint32_t cg_ckhash; /* check-hash of this cg */
587 ufs_time_t cg_time; /* time last written */
588 int64_t cg_sparecon64[3]; /* reserved for future use */
589 /* actually longer - space used for cylinder group maps */
590 };
591
592 /*
593 * Macros for access to cylinder group array structures
594 */
595 #define cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC)
596 #define cg_inosused(cgp) \
597 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_iusedoff))
598 #define cg_blksfree(cgp) \
599 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_freeoff))
600 #define cg_clustersfree(cgp) \
601 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_clusteroff))
602 #define cg_clustersum(cgp) \
603 ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff))
604
605 /*
606 * Turn filesystem block numbers into disk block addresses.
607 * This maps filesystem blocks to device size blocks.
608 */
609 #define fsbtodb(fs, b) ((daddr_t)(b) << (fs)->fs_fsbtodb)
610 #define dbtofsb(fs, b) ((b) >> (fs)->fs_fsbtodb)
611
612 /*
613 * Cylinder group macros to locate things in cylinder groups.
614 * They calc filesystem addresses of cylinder group data structures.
615 */
616 #define cgbase(fs, c) (((ufs2_daddr_t)(fs)->fs_fpg) * (c))
617 #define cgdata(fs, c) (cgdmin(fs, c) + (fs)->fs_metaspace) /* data zone */
618 #define cgmeta(fs, c) (cgdmin(fs, c)) /* meta data */
619 #define cgdmin(fs, c) (cgstart(fs, c) + (fs)->fs_dblkno) /* 1st data */
620 #define cgimin(fs, c) (cgstart(fs, c) + (fs)->fs_iblkno) /* inode blk */
621 #define cgsblock(fs, c) (cgstart(fs, c) + (fs)->fs_sblkno) /* super blk */
622 #define cgtod(fs, c) (cgstart(fs, c) + (fs)->fs_cblkno) /* cg block */
623 #define cgstart(fs, c) \
624 ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) : \
625 (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask))))
626
627 /*
628 * Macros for handling inode numbers:
629 * inode number to filesystem block offset.
630 * inode number to cylinder group number.
631 * inode number to filesystem block address.
632 */
633 #define ino_to_cg(fs, x) (((ino_t)(x)) / (fs)->fs_ipg)
634 #define ino_to_fsba(fs, x) \
635 ((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) + \
636 (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs))))))
637 #define ino_to_fsbo(fs, x) (((ino_t)(x)) % INOPB(fs))
638
639 /*
640 * Give cylinder group number for a filesystem block.
641 * Give cylinder group block number for a filesystem block.
642 */
643 #define dtog(fs, d) ((d) / (fs)->fs_fpg)
644 #define dtogd(fs, d) ((d) % (fs)->fs_fpg)
645
646 /*
647 * Extract the bits for a block from a map.
648 * Compute the cylinder and rotational position of a cyl block addr.
649 */
650 #define blkmap(fs, map, loc) \
651 (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag)))
652
653 /*
654 * The following macros optimize certain frequently calculated
655 * quantities by using shifts and masks in place of divisions
656 * modulos and multiplications.
657 */
658 #define blkoff(fs, loc) /* calculates (loc % fs->fs_bsize) */ \
659 ((loc) & (fs)->fs_qbmask)
660 #define fragoff(fs, loc) /* calculates (loc % fs->fs_fsize) */ \
661 ((loc) & (fs)->fs_qfmask)
662 #define lfragtosize(fs, frag) /* calculates ((off_t)frag * fs->fs_fsize) */ \
663 (((off_t)(frag)) << (fs)->fs_fshift)
664 #define lblktosize(fs, blk) /* calculates ((off_t)blk * fs->fs_bsize) */ \
665 (((off_t)(blk)) << (fs)->fs_bshift)
666 /* Use this only when `blk' is known to be small, e.g., < UFS_NDADDR. */
667 #define smalllblktosize(fs, blk) /* calculates (blk * fs->fs_bsize) */ \
668 ((blk) << (fs)->fs_bshift)
669 #define lblkno(fs, loc) /* calculates (loc / fs->fs_bsize) */ \
670 ((loc) >> (fs)->fs_bshift)
671 #define numfrags(fs, loc) /* calculates (loc / fs->fs_fsize) */ \
672 ((loc) >> (fs)->fs_fshift)
673 #define blkroundup(fs, size) /* calculates roundup(size, fs->fs_bsize) */ \
674 (((size) + (fs)->fs_qbmask) & (fs)->fs_bmask)
675 #define fragroundup(fs, size) /* calculates roundup(size, fs->fs_fsize) */ \
676 (((size) + (fs)->fs_qfmask) & (fs)->fs_fmask)
677 #define fragstoblks(fs, frags) /* calculates (frags / fs->fs_frag) */ \
678 ((frags) >> (fs)->fs_fragshift)
679 #define blkstofrags(fs, blks) /* calculates (blks * fs->fs_frag) */ \
680 ((blks) << (fs)->fs_fragshift)
681 #define fragnum(fs, fsb) /* calculates (fsb % fs->fs_frag) */ \
682 ((fsb) & ((fs)->fs_frag - 1))
683 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->fs_frag) */ \
684 ((fsb) &~ ((fs)->fs_frag - 1))
685
686 /*
687 * Determine the number of available frags given a
688 * percentage to hold in reserve.
689 */
690 #define freespace(fs, percentreserved) \
691 (blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \
692 (fs)->fs_cstotal.cs_nffree - \
693 (((off_t)((fs)->fs_dsize)) * (percentreserved) / 100))
694
695 /*
696 * Determining the size of a file block in the filesystem.
697 */
698 #define blksize(fs, ip, lbn) \
699 (((lbn) >= UFS_NDADDR || (ip)->i_size >= \
700 (uint64_t)smalllblktosize(fs, (lbn) + 1)) \
701 ? (fs)->fs_bsize \
702 : (fragroundup(fs, blkoff(fs, (ip)->i_size))))
703 #define sblksize(fs, size, lbn) \
704 (((lbn) >= UFS_NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \
705 ? (fs)->fs_bsize \
706 : (fragroundup(fs, blkoff(fs, (size)))))
707
708 /*
709 * Number of indirects in a filesystem block.
710 */
711 #define NINDIR(fs) ((fs)->fs_nindir)
712
713 /*
714 * Indirect lbns are aligned on UFS_NDADDR addresses where single indirects
715 * are the negated address of the lowest lbn reachable, double indirects
716 * are this lbn - 1 and triple indirects are this lbn - 2. This yields
717 * an unusual bit order to determine level.
718 */
719 static inline int
lbn_level(ufs_lbn_t lbn)720 lbn_level(ufs_lbn_t lbn)
721 {
722 if (lbn >= 0)
723 return 0;
724 switch (lbn & 0x3) {
725 case 0:
726 return (0);
727 case 1:
728 break;
729 case 2:
730 return (2);
731 case 3:
732 return (1);
733 default:
734 break;
735 }
736 return (-1);
737 }
738
739 static inline ufs_lbn_t
lbn_offset(struct fs * fs,int level)740 lbn_offset(struct fs *fs, int level)
741 {
742 ufs_lbn_t res;
743
744 for (res = 1; level > 0; level--)
745 res *= NINDIR(fs);
746 return (res);
747 }
748
749 /*
750 * Number of inodes in a secondary storage block/fragment.
751 */
752 #define INOPB(fs) ((fs)->fs_inopb)
753 #define INOPF(fs) ((fs)->fs_inopb >> (fs)->fs_fragshift)
754
755 /*
756 * Softdep journal record format.
757 */
758 #define JOP_UNKNOWN 0 /* JOP operation is unknown */
759 #define JOP_ADDREF 1 /* Add a reference to an inode. */
760 #define JOP_REMREF 2 /* Remove a reference from an inode. */
761 #define JOP_NEWBLK 3 /* Allocate a block. */
762 #define JOP_FREEBLK 4 /* Free a block or a tree of blocks. */
763 #define JOP_MVREF 5 /* Move a reference from one off to another. */
764 #define JOP_TRUNC 6 /* Partial truncation record. */
765 #define JOP_SYNC 7 /* fsync() complete record. */
766 #define JOP_NUMJOPTYPES 8
767 #define JOP_NAMES { \
768 "unknown", \
769 "JOP_ADDREF", \
770 "JOP_REMREF", \
771 "JOP_NEWBLK", \
772 "JOP_FREEBLK", \
773 "JOP_MVREF", \
774 "JOP_TRUNC", \
775 "JOP_SYNC" }
776 #define JOP_OPTYPE(op) \
777 (op) < JOP_NUMJOPTYPES ? joptype[op] : joptype[JOP_UNKNOWN]
778
779 #define JREC_SIZE 32 /* Record and segment header size. */
780
781 #define SUJ_MIN (4 * 1024 * 1024) /* Minimum journal size */
782 #define SUJ_FILE ".sujournal" /* Journal file name */
783
784 /*
785 * Size of the segment record header. There is at most one for each disk
786 * block in the journal. The segment header is followed by an array of
787 * records. fsck depends on the first element in each record being 'op'
788 * and the second being 'ino'. Segments may span multiple disk blocks but
789 * the header is present on each.
790 */
791 struct jsegrec {
792 uint64_t jsr_seq; /* Our sequence number */
793 uint64_t jsr_oldest; /* Oldest valid sequence number */
794 uint16_t jsr_cnt; /* Count of valid records */
795 uint16_t jsr_blocks; /* Count of device bsize blocks. */
796 uint32_t jsr_crc; /* 32bit crc of the valid space */
797 ufs_time_t jsr_time; /* timestamp for mount instance */
798 };
799
800 /*
801 * Reference record. Records a single link count modification.
802 */
803 struct jrefrec {
804 uint32_t jr_op;
805 uint32_t jr_ino;
806 uint32_t jr_parent;
807 uint16_t jr_nlink;
808 uint16_t jr_mode;
809 int64_t jr_diroff;
810 uint64_t jr_unused;
811 };
812
813 /*
814 * Move record. Records a reference moving within a directory block. The
815 * nlink is unchanged but we must search both locations.
816 */
817 struct jmvrec {
818 uint32_t jm_op;
819 uint32_t jm_ino;
820 uint32_t jm_parent;
821 uint16_t jm_unused;
822 int64_t jm_oldoff;
823 int64_t jm_newoff;
824 };
825
826 /*
827 * Block record. A set of frags or tree of blocks starting at an indirect are
828 * freed or a set of frags are allocated.
829 */
830 struct jblkrec {
831 uint32_t jb_op;
832 uint32_t jb_ino;
833 ufs2_daddr_t jb_blkno;
834 ufs_lbn_t jb_lbn;
835 uint16_t jb_frags;
836 uint16_t jb_oldfrags;
837 uint32_t jb_unused;
838 };
839
840 /*
841 * Truncation record. Records a partial truncation so that it may be
842 * completed at check time. Also used for sync records.
843 */
844 struct jtrncrec {
845 uint32_t jt_op;
846 uint32_t jt_ino;
847 int64_t jt_size;
848 uint32_t jt_extsize;
849 uint32_t jt_pad[3];
850 };
851
852 union jrec {
853 struct jsegrec rec_jsegrec;
854 struct jrefrec rec_jrefrec;
855 struct jmvrec rec_jmvrec;
856 struct jblkrec rec_jblkrec;
857 struct jtrncrec rec_jtrncrec;
858 };
859
860 #ifdef CTASSERT
861 CTASSERT(sizeof(struct jsegrec) == JREC_SIZE);
862 CTASSERT(sizeof(struct jrefrec) == JREC_SIZE);
863 CTASSERT(sizeof(struct jmvrec) == JREC_SIZE);
864 CTASSERT(sizeof(struct jblkrec) == JREC_SIZE);
865 CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE);
866 CTASSERT(sizeof(union jrec) == JREC_SIZE);
867 #endif
868
869 extern int inside[], around[];
870 extern uint8_t *fragtbl[];
871
872 /*
873 * IOCTLs used for filesystem write suspension.
874 */
875 #define UFSSUSPEND _IOW('U', 1, fsid_t)
876 #define UFSRESUME _IO('U', 2)
877
878 #endif
879