FreeBSD/Linux Kernel Cross Reference
sys/ufs/ffs/fs.h
1 /*-
2 * Copyright (c) 1982, 1986, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * @(#)fs.h 8.13 (Berkeley) 3/21/95
30 * $FreeBSD$
31 */
32
33 #ifndef _UFS_FFS_FS_H_
34 #define _UFS_FFS_FS_H_
35
36 #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 #define RESID (4096 - 20) /* disk sector size minus recovery area size */
241 struct fsrecovery {
242 char block[RESID]; /* unused part of sector */
243 int32_t fsr_magic; /* magic number */
244 int32_t fsr_fsbtodb; /* fsbtodb and dbtofsb shift constant */
245 int32_t fsr_sblkno; /* offset of super-block in filesys */
246 int32_t fsr_fpg; /* blocks per group * fs_frag */
247 u_int32_t fsr_ncg; /* number of cylinder groups */
248 };
249
250 /*
251 * Per cylinder group information; summarized in blocks allocated
252 * from first cylinder group data blocks. These blocks have to be
253 * read in from fs_csaddr (size fs_cssize) in addition to the
254 * super block.
255 */
256 struct csum {
257 int32_t cs_ndir; /* number of directories */
258 int32_t cs_nbfree; /* number of free blocks */
259 int32_t cs_nifree; /* number of free inodes */
260 int32_t cs_nffree; /* number of free frags */
261 };
262 struct csum_total {
263 int64_t cs_ndir; /* number of directories */
264 int64_t cs_nbfree; /* number of free blocks */
265 int64_t cs_nifree; /* number of free inodes */
266 int64_t cs_nffree; /* number of free frags */
267 int64_t cs_numclusters; /* number of free clusters */
268 int64_t cs_spare[3]; /* future expansion */
269 };
270
271 /*
272 * Super block for an FFS filesystem.
273 */
274 struct fs {
275 int32_t fs_firstfield; /* historic filesystem linked list, */
276 int32_t fs_unused_1; /* used for incore super blocks */
277 int32_t fs_sblkno; /* offset of super-block in filesys */
278 int32_t fs_cblkno; /* offset of cyl-block in filesys */
279 int32_t fs_iblkno; /* offset of inode-blocks in filesys */
280 int32_t fs_dblkno; /* offset of first data after cg */
281 int32_t fs_old_cgoffset; /* cylinder group offset in cylinder */
282 int32_t fs_old_cgmask; /* used to calc mod fs_ntrak */
283 int32_t fs_old_time; /* last time written */
284 int32_t fs_old_size; /* number of blocks in fs */
285 int32_t fs_old_dsize; /* number of data blocks in fs */
286 u_int32_t fs_ncg; /* number of cylinder groups */
287 int32_t fs_bsize; /* size of basic blocks in fs */
288 int32_t fs_fsize; /* size of frag blocks in fs */
289 int32_t fs_frag; /* number of frags in a block in fs */
290 /* these are configuration parameters */
291 int32_t fs_minfree; /* minimum percentage of free blocks */
292 int32_t fs_old_rotdelay; /* num of ms for optimal next block */
293 int32_t fs_old_rps; /* disk revolutions per second */
294 /* these fields can be computed from the others */
295 int32_t fs_bmask; /* ``blkoff'' calc of blk offsets */
296 int32_t fs_fmask; /* ``fragoff'' calc of frag offsets */
297 int32_t fs_bshift; /* ``lblkno'' calc of logical blkno */
298 int32_t fs_fshift; /* ``numfrags'' calc number of frags */
299 /* these are configuration parameters */
300 int32_t fs_maxcontig; /* max number of contiguous blks */
301 int32_t fs_maxbpg; /* max number of blks per cyl group */
302 /* these fields can be computed from the others */
303 int32_t fs_fragshift; /* block to frag shift */
304 int32_t fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */
305 int32_t fs_sbsize; /* actual size of super block */
306 int32_t fs_spare1[2]; /* old fs_csmask */
307 /* old fs_csshift */
308 int32_t fs_nindir; /* value of NINDIR */
309 u_int32_t fs_inopb; /* value of INOPB */
310 int32_t fs_old_nspf; /* value of NSPF */
311 /* yet another configuration parameter */
312 int32_t fs_optim; /* optimization preference, see below */
313 int32_t fs_old_npsect; /* # sectors/track including spares */
314 int32_t fs_old_interleave; /* hardware sector interleave */
315 int32_t fs_old_trackskew; /* sector 0 skew, per track */
316 int32_t fs_id[2]; /* unique filesystem id */
317 /* sizes determined by number of cylinder groups and their sizes */
318 int32_t fs_old_csaddr; /* blk addr of cyl grp summary area */
319 int32_t fs_cssize; /* size of cyl grp summary area */
320 int32_t fs_cgsize; /* cylinder group size */
321 int32_t fs_spare2; /* old fs_ntrak */
322 int32_t fs_old_nsect; /* sectors per track */
323 int32_t fs_old_spc; /* sectors per cylinder */
324 int32_t fs_old_ncyl; /* cylinders in filesystem */
325 int32_t fs_old_cpg; /* cylinders per group */
326 u_int32_t fs_ipg; /* inodes per group */
327 int32_t fs_fpg; /* blocks per group * fs_frag */
328 /* this data must be re-computed after crashes */
329 struct csum fs_old_cstotal; /* cylinder summary information */
330 /* these fields are cleared at mount time */
331 int8_t fs_fmod; /* super block modified flag */
332 int8_t fs_clean; /* filesystem is clean flag */
333 int8_t fs_ronly; /* mounted read-only flag */
334 int8_t fs_old_flags; /* old FS_ flags */
335 u_char fs_fsmnt[MAXMNTLEN]; /* name mounted on */
336 u_char fs_volname[MAXVOLLEN]; /* volume name */
337 u_int64_t fs_swuid; /* system-wide uid */
338 int32_t fs_pad; /* due to alignment of fs_swuid */
339 /* these fields retain the current block allocation info */
340 int32_t fs_cgrotor; /* last cg searched */
341 void *fs_ocsp[NOCSPTRS]; /* padding; was list of fs_cs buffers */
342 u_int8_t *fs_contigdirs; /* (u) # of contig. allocated dirs */
343 struct csum *fs_csp; /* (u) cg summary info buffer */
344 int32_t *fs_maxcluster; /* (u) max cluster in each cyl group */
345 u_int *fs_active; /* (u) used by snapshots to track fs */
346 int32_t fs_old_cpc; /* cyl per cycle in postbl */
347 int32_t fs_maxbsize; /* maximum blocking factor permitted */
348 int64_t fs_unrefs; /* number of unreferenced inodes */
349 int64_t fs_providersize; /* size of underlying GEOM provider */
350 int64_t fs_metaspace; /* size of area reserved for metadata */
351 int64_t fs_sparecon64[14]; /* old rotation block list head */
352 int64_t fs_sblockloc; /* byte offset of standard superblock */
353 struct csum_total fs_cstotal; /* (u) cylinder summary information */
354 ufs_time_t fs_time; /* last time written */
355 int64_t fs_size; /* number of blocks in fs */
356 int64_t fs_dsize; /* number of data blocks in fs */
357 ufs2_daddr_t fs_csaddr; /* blk addr of cyl grp summary area */
358 int64_t fs_pendingblocks; /* (u) blocks being freed */
359 u_int32_t fs_pendinginodes; /* (u) inodes being freed */
360 uint32_t fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */
361 u_int32_t fs_avgfilesize; /* expected average file size */
362 u_int32_t fs_avgfpdir; /* expected # of files per directory */
363 int32_t fs_save_cgsize; /* save real cg size to use fs_bsize */
364 ufs_time_t fs_mtime; /* Last mount or fsck time. */
365 int32_t fs_sujfree; /* SUJ free list */
366 int32_t fs_sparecon32[23]; /* reserved for future constants */
367 int32_t fs_flags; /* see FS_ flags below */
368 int32_t fs_contigsumsize; /* size of cluster summary array */
369 int32_t fs_maxsymlinklen; /* max length of an internal symlink */
370 int32_t fs_old_inodefmt; /* format of on-disk inodes */
371 u_int64_t fs_maxfilesize; /* maximum representable file size */
372 int64_t fs_qbmask; /* ~fs_bmask for use with 64-bit size */
373 int64_t fs_qfmask; /* ~fs_fmask for use with 64-bit size */
374 int32_t fs_state; /* validate fs_clean field */
375 int32_t fs_old_postblformat; /* format of positional layout tables */
376 int32_t fs_old_nrpos; /* number of rotational positions */
377 int32_t fs_spare5[2]; /* old fs_postbloff */
378 /* old fs_rotbloff */
379 int32_t fs_magic; /* magic number */
380 };
381
382 /* Sanity checking. */
383 #ifdef CTASSERT
384 CTASSERT(sizeof(struct fs) == 1376);
385 #endif
386
387 /*
388 * Filesystem identification
389 */
390 #define FS_UFS1_MAGIC 0x011954 /* UFS1 fast filesystem magic number */
391 #define FS_UFS2_MAGIC 0x19540119 /* UFS2 fast filesystem magic number */
392 #define FS_BAD_MAGIC 0x19960408 /* UFS incomplete newfs magic number */
393 #define FS_OKAY 0x7c269d38 /* superblock checksum */
394 #define FS_42INODEFMT -1 /* 4.2BSD inode format */
395 #define FS_44INODEFMT 2 /* 4.4BSD inode format */
396
397 /*
398 * Preference for optimization.
399 */
400 #define FS_OPTTIME 0 /* minimize allocation time */
401 #define FS_OPTSPACE 1 /* minimize disk fragmentation */
402
403 /*
404 * Filesystem flags.
405 *
406 * The FS_UNCLEAN flag is set by the kernel when the filesystem was
407 * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates
408 * that the filesystem should be managed by the soft updates code.
409 * Note that the FS_NEEDSFSCK flag is set and cleared only by the
410 * fsck utility. It is set when background fsck finds an unexpected
411 * inconsistency which requires a traditional foreground fsck to be
412 * run. Such inconsistencies should only be found after an uncorrectable
413 * disk error. A foreground fsck will clear the FS_NEEDSFSCK flag when
414 * it has successfully cleaned up the filesystem. The kernel uses this
415 * flag to enforce that inconsistent filesystems be mounted read-only.
416 * The FS_INDEXDIRS flag when set indicates that the kernel maintains
417 * on-disk auxiliary indexes (such as B-trees) for speeding directory
418 * accesses. Kernels that do not support auxiliary indicies clear the
419 * flag to indicate that the indicies need to be rebuilt (by fsck) before
420 * they can be used.
421 *
422 * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled
423 * for the file system, so they should be loaded from extended attributes,
424 * observed for access control purposes, and be administered by object
425 * owners. FS_NFS4ACLS indicates that NFSv4 ACLs are administratively
426 * enabled. This flag is mutually exclusive with FS_ACLS. FS_MULTILABEL
427 * indicates that the TrustedBSD MAC Framework should attempt to back MAC
428 * labels into extended attributes on the file system rather than maintain
429 * a single mount label for all objects.
430 */
431 #define FS_UNCLEAN 0x0001 /* filesystem not clean at mount */
432 #define FS_DOSOFTDEP 0x0002 /* filesystem using soft dependencies */
433 #define FS_NEEDSFSCK 0x0004 /* filesystem needs sync fsck before mount */
434 #define FS_SUJ 0x0008 /* Filesystem using softupdate journal */
435 #define FS_ACLS 0x0010 /* file system has POSIX.1e ACLs enabled */
436 #define FS_MULTILABEL 0x0020 /* file system is MAC multi-label */
437 #define FS_GJOURNAL 0x0040 /* gjournaled file system */
438 #define FS_FLAGS_UPDATED 0x0080 /* flags have been moved to new location */
439 #define FS_NFS4ACLS 0x0100 /* file system has NFSv4 ACLs enabled */
440 #define FS_INDEXDIRS 0x0200 /* kernel supports indexed directories */
441 #define FS_TRIM 0x0400 /* issue BIO_DELETE for deleted blocks */
442
443 /*
444 * Macros to access bits in the fs_active array.
445 */
446 #define ACTIVECGNUM(fs, cg) ((fs)->fs_active[(cg) / (NBBY * sizeof(int))])
447 #define ACTIVECGOFF(cg) (1 << ((cg) % (NBBY * sizeof(int))))
448 #define ACTIVESET(fs, cg) do { \
449 if ((fs)->fs_active) \
450 ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg)); \
451 } while (0)
452 #define ACTIVECLEAR(fs, cg) do { \
453 if ((fs)->fs_active) \
454 ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg)); \
455 } while (0)
456
457 /*
458 * The size of a cylinder group is calculated by CGSIZE. The maximum size
459 * is limited by the fact that cylinder groups are at most one block.
460 * Its size is derived from the size of the maps maintained in the
461 * cylinder group and the (struct cg) size.
462 */
463 #define CGSIZE(fs) \
464 /* base cg */ (sizeof(struct cg) + sizeof(int32_t) + \
465 /* old btotoff */ (fs)->fs_old_cpg * sizeof(int32_t) + \
466 /* old boff */ (fs)->fs_old_cpg * sizeof(u_int16_t) + \
467 /* inode map */ howmany((fs)->fs_ipg, NBBY) + \
468 /* block map */ howmany((fs)->fs_fpg, NBBY) +\
469 /* if present */ ((fs)->fs_contigsumsize <= 0 ? 0 : \
470 /* cluster sum */ (fs)->fs_contigsumsize * sizeof(int32_t) + \
471 /* cluster map */ howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY)))
472
473 /*
474 * The minimal number of cylinder groups that should be created.
475 */
476 #define MINCYLGRPS 4
477
478 /*
479 * Convert cylinder group to base address of its global summary info.
480 */
481 #define fs_cs(fs, indx) fs_csp[indx]
482
483 /*
484 * Cylinder group block for a filesystem.
485 */
486 #define CG_MAGIC 0x090255
487 struct cg {
488 int32_t cg_firstfield; /* historic cyl groups linked list */
489 int32_t cg_magic; /* magic number */
490 int32_t cg_old_time; /* time last written */
491 u_int32_t cg_cgx; /* we are the cgx'th cylinder group */
492 int16_t cg_old_ncyl; /* number of cyl's this cg */
493 int16_t cg_old_niblk; /* number of inode blocks this cg */
494 u_int32_t cg_ndblk; /* number of data blocks this cg */
495 struct csum cg_cs; /* cylinder summary information */
496 u_int32_t cg_rotor; /* position of last used block */
497 u_int32_t cg_frotor; /* position of last used frag */
498 u_int32_t cg_irotor; /* position of last used inode */
499 u_int32_t cg_frsum[MAXFRAG]; /* counts of available frags */
500 int32_t cg_old_btotoff; /* (int32) block totals per cylinder */
501 int32_t cg_old_boff; /* (u_int16) free block positions */
502 u_int32_t cg_iusedoff; /* (u_int8) used inode map */
503 u_int32_t cg_freeoff; /* (u_int8) free block map */
504 u_int32_t cg_nextfreeoff; /* (u_int8) next available space */
505 u_int32_t cg_clustersumoff; /* (u_int32) counts of avail clusters */
506 u_int32_t cg_clusteroff; /* (u_int8) free cluster map */
507 u_int32_t cg_nclusterblks; /* number of clusters this cg */
508 u_int32_t cg_niblk; /* number of inode blocks this cg */
509 u_int32_t cg_initediblk; /* last initialized inode */
510 u_int32_t cg_unrefs; /* number of unreferenced inodes */
511 int32_t cg_sparecon32[2]; /* reserved for future use */
512 ufs_time_t cg_time; /* time last written */
513 int64_t cg_sparecon64[3]; /* reserved for future use */
514 u_int8_t cg_space[1]; /* space for cylinder group maps */
515 /* actually longer */
516 };
517
518 /*
519 * Macros for access to cylinder group array structures
520 */
521 #define cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC)
522 #define cg_inosused(cgp) \
523 ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_iusedoff))
524 #define cg_blksfree(cgp) \
525 ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_freeoff))
526 #define cg_clustersfree(cgp) \
527 ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_clusteroff))
528 #define cg_clustersum(cgp) \
529 ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff))
530
531 /*
532 * Turn filesystem block numbers into disk block addresses.
533 * This maps filesystem blocks to device size blocks.
534 */
535 #define fsbtodb(fs, b) ((daddr_t)(b) << (fs)->fs_fsbtodb)
536 #define dbtofsb(fs, b) ((b) >> (fs)->fs_fsbtodb)
537
538 /*
539 * Cylinder group macros to locate things in cylinder groups.
540 * They calc filesystem addresses of cylinder group data structures.
541 */
542 #define cgbase(fs, c) (((ufs2_daddr_t)(fs)->fs_fpg) * (c))
543 #define cgdata(fs, c) (cgdmin(fs, c) + (fs)->fs_metaspace) /* data zone */
544 #define cgmeta(fs, c) (cgdmin(fs, c)) /* meta data */
545 #define cgdmin(fs, c) (cgstart(fs, c) + (fs)->fs_dblkno) /* 1st data */
546 #define cgimin(fs, c) (cgstart(fs, c) + (fs)->fs_iblkno) /* inode blk */
547 #define cgsblock(fs, c) (cgstart(fs, c) + (fs)->fs_sblkno) /* super blk */
548 #define cgtod(fs, c) (cgstart(fs, c) + (fs)->fs_cblkno) /* cg block */
549 #define cgstart(fs, c) \
550 ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) : \
551 (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask))))
552
553 /*
554 * Macros for handling inode numbers:
555 * inode number to filesystem block offset.
556 * inode number to cylinder group number.
557 * inode number to filesystem block address.
558 */
559 #define ino_to_cg(fs, x) (((ino_t)(x)) / (fs)->fs_ipg)
560 #define ino_to_fsba(fs, x) \
561 ((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) + \
562 (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs))))))
563 #define ino_to_fsbo(fs, x) (((ino_t)(x)) % INOPB(fs))
564
565 /*
566 * Give cylinder group number for a filesystem block.
567 * Give cylinder group block number for a filesystem block.
568 */
569 #define dtog(fs, d) ((d) / (fs)->fs_fpg)
570 #define dtogd(fs, d) ((d) % (fs)->fs_fpg)
571
572 /*
573 * Extract the bits for a block from a map.
574 * Compute the cylinder and rotational position of a cyl block addr.
575 */
576 #define blkmap(fs, map, loc) \
577 (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag)))
578
579 /*
580 * The following macros optimize certain frequently calculated
581 * quantities by using shifts and masks in place of divisions
582 * modulos and multiplications.
583 */
584 #define blkoff(fs, loc) /* calculates (loc % fs->fs_bsize) */ \
585 ((loc) & (fs)->fs_qbmask)
586 #define fragoff(fs, loc) /* calculates (loc % fs->fs_fsize) */ \
587 ((loc) & (fs)->fs_qfmask)
588 #define lfragtosize(fs, frag) /* calculates ((off_t)frag * fs->fs_fsize) */ \
589 (((off_t)(frag)) << (fs)->fs_fshift)
590 #define lblktosize(fs, blk) /* calculates ((off_t)blk * fs->fs_bsize) */ \
591 (((off_t)(blk)) << (fs)->fs_bshift)
592 /* Use this only when `blk' is known to be small, e.g., < NDADDR. */
593 #define smalllblktosize(fs, blk) /* calculates (blk * fs->fs_bsize) */ \
594 ((blk) << (fs)->fs_bshift)
595 #define lblkno(fs, loc) /* calculates (loc / fs->fs_bsize) */ \
596 ((loc) >> (fs)->fs_bshift)
597 #define numfrags(fs, loc) /* calculates (loc / fs->fs_fsize) */ \
598 ((loc) >> (fs)->fs_fshift)
599 #define blkroundup(fs, size) /* calculates roundup(size, fs->fs_bsize) */ \
600 (((size) + (fs)->fs_qbmask) & (fs)->fs_bmask)
601 #define fragroundup(fs, size) /* calculates roundup(size, fs->fs_fsize) */ \
602 (((size) + (fs)->fs_qfmask) & (fs)->fs_fmask)
603 #define fragstoblks(fs, frags) /* calculates (frags / fs->fs_frag) */ \
604 ((frags) >> (fs)->fs_fragshift)
605 #define blkstofrags(fs, blks) /* calculates (blks * fs->fs_frag) */ \
606 ((blks) << (fs)->fs_fragshift)
607 #define fragnum(fs, fsb) /* calculates (fsb % fs->fs_frag) */ \
608 ((fsb) & ((fs)->fs_frag - 1))
609 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->fs_frag) */ \
610 ((fsb) &~ ((fs)->fs_frag - 1))
611
612 /*
613 * Determine the number of available frags given a
614 * percentage to hold in reserve.
615 */
616 #define freespace(fs, percentreserved) \
617 (blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \
618 (fs)->fs_cstotal.cs_nffree - \
619 (((off_t)((fs)->fs_dsize)) * (percentreserved) / 100))
620
621 /*
622 * Determining the size of a file block in the filesystem.
623 */
624 #define blksize(fs, ip, lbn) \
625 (((lbn) >= NDADDR || (ip)->i_size >= smalllblktosize(fs, (lbn) + 1)) \
626 ? (fs)->fs_bsize \
627 : (fragroundup(fs, blkoff(fs, (ip)->i_size))))
628 #define sblksize(fs, size, lbn) \
629 (((lbn) >= NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \
630 ? (fs)->fs_bsize \
631 : (fragroundup(fs, blkoff(fs, (size)))))
632
633 /*
634 * Number of indirects in a filesystem block.
635 */
636 #define NINDIR(fs) ((fs)->fs_nindir)
637
638 /*
639 * Indirect lbns are aligned on NDADDR addresses where single indirects
640 * are the negated address of the lowest lbn reachable, double indirects
641 * are this lbn - 1 and triple indirects are this lbn - 2. This yields
642 * an unusual bit order to determine level.
643 */
644 static inline int
645 lbn_level(ufs_lbn_t lbn)
646 {
647 if (lbn >= 0)
648 return 0;
649 switch (lbn & 0x3) {
650 case 0:
651 return (0);
652 case 1:
653 break;
654 case 2:
655 return (2);
656 case 3:
657 return (1);
658 default:
659 break;
660 }
661 return (-1);
662 }
663
664 static inline ufs_lbn_t
665 lbn_offset(struct fs *fs, int level)
666 {
667 ufs_lbn_t res;
668
669 for (res = 1; level > 0; level--)
670 res *= NINDIR(fs);
671 return (res);
672 }
673
674 /*
675 * Number of inodes in a secondary storage block/fragment.
676 */
677 #define INOPB(fs) ((fs)->fs_inopb)
678 #define INOPF(fs) ((fs)->fs_inopb >> (fs)->fs_fragshift)
679
680 /*
681 * Softdep journal record format.
682 */
683
684 #define JOP_ADDREF 1 /* Add a reference to an inode. */
685 #define JOP_REMREF 2 /* Remove a reference from an inode. */
686 #define JOP_NEWBLK 3 /* Allocate a block. */
687 #define JOP_FREEBLK 4 /* Free a block or a tree of blocks. */
688 #define JOP_MVREF 5 /* Move a reference from one off to another. */
689 #define JOP_TRUNC 6 /* Partial truncation record. */
690 #define JOP_SYNC 7 /* fsync() complete record. */
691
692 #define JREC_SIZE 32 /* Record and segment header size. */
693
694 #define SUJ_MIN (4 * 1024 * 1024) /* Minimum journal size */
695 #define SUJ_MAX (32 * 1024 * 1024) /* Maximum journal size */
696 #define SUJ_FILE ".sujournal" /* Journal file name */
697
698 /*
699 * Size of the segment record header. There is at most one for each disk
700 * block in the journal. The segment header is followed by an array of
701 * records. fsck depends on the first element in each record being 'op'
702 * and the second being 'ino'. Segments may span multiple disk blocks but
703 * the header is present on each.
704 */
705 struct jsegrec {
706 uint64_t jsr_seq; /* Our sequence number */
707 uint64_t jsr_oldest; /* Oldest valid sequence number */
708 uint16_t jsr_cnt; /* Count of valid records */
709 uint16_t jsr_blocks; /* Count of device bsize blocks. */
710 uint32_t jsr_crc; /* 32bit crc of the valid space */
711 ufs_time_t jsr_time; /* timestamp for mount instance */
712 };
713
714 /*
715 * Reference record. Records a single link count modification.
716 */
717 struct jrefrec {
718 uint32_t jr_op;
719 uint32_t jr_ino;
720 uint32_t jr_parent;
721 uint16_t jr_nlink;
722 uint16_t jr_mode;
723 int64_t jr_diroff;
724 uint64_t jr_unused;
725 };
726
727 /*
728 * Move record. Records a reference moving within a directory block. The
729 * nlink is unchanged but we must search both locations.
730 */
731 struct jmvrec {
732 uint32_t jm_op;
733 uint32_t jm_ino;
734 uint32_t jm_parent;
735 uint16_t jm_unused;
736 int64_t jm_oldoff;
737 int64_t jm_newoff;
738 };
739
740 /*
741 * Block record. A set of frags or tree of blocks starting at an indirect are
742 * freed or a set of frags are allocated.
743 */
744 struct jblkrec {
745 uint32_t jb_op;
746 uint32_t jb_ino;
747 ufs2_daddr_t jb_blkno;
748 ufs_lbn_t jb_lbn;
749 uint16_t jb_frags;
750 uint16_t jb_oldfrags;
751 uint32_t jb_unused;
752 };
753
754 /*
755 * Truncation record. Records a partial truncation so that it may be
756 * completed at check time. Also used for sync records.
757 */
758 struct jtrncrec {
759 uint32_t jt_op;
760 uint32_t jt_ino;
761 int64_t jt_size;
762 uint32_t jt_extsize;
763 uint32_t jt_pad[3];
764 };
765
766 union jrec {
767 struct jsegrec rec_jsegrec;
768 struct jrefrec rec_jrefrec;
769 struct jmvrec rec_jmvrec;
770 struct jblkrec rec_jblkrec;
771 struct jtrncrec rec_jtrncrec;
772 };
773
774 #ifdef CTASSERT
775 CTASSERT(sizeof(struct jsegrec) == JREC_SIZE);
776 CTASSERT(sizeof(struct jrefrec) == JREC_SIZE);
777 CTASSERT(sizeof(struct jmvrec) == JREC_SIZE);
778 CTASSERT(sizeof(struct jblkrec) == JREC_SIZE);
779 CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE);
780 CTASSERT(sizeof(union jrec) == JREC_SIZE);
781 #endif
782
783 extern int inside[], around[];
784 extern u_char *fragtbl[];
785
786 /*
787 * IOCTLs used for filesystem write suspension.
788 */
789 #define UFSSUSPEND _IOW('U', 1, fsid_t)
790 #define UFSRESUME _IO('U', 2)
791
792 #endif
Cache object: 17e71a7627e3855c39d3e4d193562c5d
|