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 * Request standard superblock location in ffs_sbget(). 81 */ 82 #define UFS_STDSB -1 /* Search standard places for superblock */ 83 84 /* 85 * UFS_NOMSG indicates that superblock inconsistency error messages 86 * should not be printed. It is used by programs like fsck that 87 * want to print their own error message. 88 * 89 * UFS_NOCSUM causes only the superblock itself to be returned, but does 90 * not read in any auxiliary data structures like the cylinder group 91 * summary information. It is used by clients like glabel that just 92 * want to check for possible filesystem types. Using UFS_NOCSUM 93 * skips the superblock checks for csum data which allows superblocks 94 * that have corrupted csum data to be read and used. 95 * 96 * UFS_NOHASHFAIL will note that the check hash is wrong but will still 97 * return the superblock. This is used by the bootstrap code to 98 * give the system a chance to come up so that fsck can be run to 99 * correct the problem. 100 * 101 * UFS_NOWARNFAIL will warn about inconsistencies but still return the 102 * superblock. It includes UFS_NOHASHFAIL. UFS_NOWARNFAIL is used by 103 * programs like fsck_ffs(8) to debug broken filesystems. 104 * 105 * UFS_FSRONLY will only validate the superblock fields needed to 106 * calculate where the backup filesystem superblocks are located. 107 * If these values pass their validation tests, then the superblock 108 * is returned. This flag is used as part of the attempt to find 109 * alternate superblocks when using ffs_sbsearch(). 110 */ 111 #define UFS_NOHASHFAIL 0x0001 /* Ignore check-hash failure */ 112 #define UFS_NOWARNFAIL 0x0003 /* Ignore non-fatal inconsistencies */ 113 #define UFS_NOMSG 0x0004 /* Print no error message */ 114 #define UFS_NOCSUM 0x0008 /* Read just the superblock without csum */ 115 #define UFS_FSRONLY 0x0010 /* Validate only values needed for recovery 116 of alternate superblocks */ 117 #define UFS_ALTSBLK 0x1000 /* Flag used internally */ 118 119 /* 120 * Max number of fragments per block. This value is NOT tweakable. 121 */ 122 #define MAXFRAG 8 123 124 /* 125 * Addresses stored in inodes are capable of addressing fragments 126 * of `blocks'. File system blocks of at most size MAXBSIZE can 127 * be optionally broken into 2, 4, or 8 pieces, each of which is 128 * addressable; these pieces may be DEV_BSIZE, or some multiple of 129 * a DEV_BSIZE unit. 130 * 131 * Large files consist of exclusively large data blocks. To avoid 132 * undue wasted disk space, the last data block of a small file may be 133 * allocated as only as many fragments of a large block as are 134 * necessary. The filesystem format retains only a single pointer 135 * to such a fragment, which is a piece of a single large block that 136 * has been divided. The size of such a fragment is determinable from 137 * information in the inode, using the ``blksize(fs, ip, lbn)'' macro. 138 * 139 * The filesystem records space availability at the fragment level; 140 * to determine block availability, aligned fragments are examined. 141 */ 142 143 /* 144 * MINBSIZE is the smallest allowable block size. 145 * In order to insure that it is possible to create files of size 146 * 2^32 with only two levels of indirection, MINBSIZE is set to 4096. 147 * MINBSIZE must be big enough to hold a cylinder group block, 148 * thus changes to (struct cg) must keep its size within MINBSIZE. 149 * Note that super blocks are always of size SBLOCKSIZE, 150 * and that both SBLOCKSIZE and MAXBSIZE must be >= MINBSIZE. 151 */ 152 #define MINBSIZE 4096 153 154 /* 155 * The path name on which the filesystem is mounted is maintained 156 * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in 157 * the super block for this name. 158 */ 159 #define MAXMNTLEN 468 160 161 /* 162 * The volume name for this filesystem is maintained in fs_volname. 163 * MAXVOLLEN defines the length of the buffer allocated. 164 */ 165 #define MAXVOLLEN 32 166 167 /* 168 * There is a 128-byte region in the superblock reserved for in-core 169 * pointers to summary information. Originally this included an array 170 * of pointers to blocks of struct csum; now there are just a few 171 * pointers and the remaining space is padded with fs_ocsp[]. 172 * 173 * NOCSPTRS determines the size of this padding. Historically this 174 * space was used to store pointers to structures that summaried 175 * filesystem usage and layout information. However, these pointers 176 * left various kernel pointers in the superblock which made otherwise 177 * identical superblocks appear to have differences. So, all the 178 * pointers in the superblock were moved to a fs_summary_info structure 179 * reducing the superblock to having only a single pointer to this 180 * structure. When writing the superblock to disk, this pointer is 181 * temporarily NULL'ed out so that the kernel pointer will not appear 182 * in the on-disk copy of the superblock. 183 */ 184 #define NOCSPTRS ((128 / sizeof(void *)) - 1) 185 186 /* 187 * A summary of contiguous blocks of various sizes is maintained 188 * in each cylinder group. Normally this is set by the initial 189 * value of fs_maxcontig. To conserve space, a maximum summary size 190 * is set by FS_MAXCONTIG. 191 */ 192 #define FS_MAXCONTIG 16 193 194 /* 195 * MINFREE gives the minimum acceptable percentage of filesystem 196 * blocks which may be free. If the freelist drops below this level 197 * only the superuser may continue to allocate blocks. This may 198 * be set to 0 if no reserve of free blocks is deemed necessary, 199 * however throughput drops by fifty percent if the filesystem 200 * is run at between 95% and 100% full; thus the minimum default 201 * value of fs_minfree is 5%. However, to get good clustering 202 * performance, 10% is a better choice. hence we use 10% as our 203 * default value. With 10% free space, fragmentation is not a 204 * problem, so we choose to optimize for time. 205 */ 206 #define MINFREE 8 207 #define DEFAULTOPT FS_OPTTIME 208 209 /* 210 * Grigoriy Orlov <gluk@ptci.ru> has done some extensive work to fine 211 * tune the layout preferences for directories within a filesystem. 212 * His algorithm can be tuned by adjusting the following parameters 213 * which tell the system the average file size and the average number 214 * of files per directory. These defaults are well selected for typical 215 * filesystems, but may need to be tuned for odd cases like filesystems 216 * being used for squid caches or news spools. 217 */ 218 #define AVFILESIZ 16384 /* expected average file size */ 219 #define AFPDIR 64 /* expected number of files per directory */ 220 221 /* 222 * The maximum number of snapshot nodes that can be associated 223 * with each filesystem. This limit affects only the number of 224 * snapshot files that can be recorded within the superblock so 225 * that they can be found when the filesystem is mounted. However, 226 * maintaining too many will slow the filesystem performance, so 227 * having this limit is a good idea. 228 */ 229 #define FSMAXSNAP 20 230 231 /* 232 * Used to identify special blocks in snapshots: 233 * 234 * BLK_NOCOPY - A block that was unallocated at the time the snapshot 235 * was taken, hence does not need to be copied when written. 236 * BLK_SNAP - A block held by another snapshot that is not needed by this 237 * snapshot. When the other snapshot is freed, the BLK_SNAP entries 238 * are converted to BLK_NOCOPY. These are needed to allow fsck to 239 * identify blocks that are in use by other snapshots (which are 240 * expunged from this snapshot). 241 */ 242 #define BLK_NOCOPY ((ufs2_daddr_t)(1)) 243 #define BLK_SNAP ((ufs2_daddr_t)(2)) 244 245 /* 246 * Sysctl values for the fast filesystem. 247 */ 248 #define FFS_ADJ_REFCNT 1 /* adjust inode reference count */ 249 #define FFS_ADJ_BLKCNT 2 /* adjust inode used block count */ 250 #define FFS_BLK_FREE 3 /* free range of blocks in map */ 251 #define FFS_DIR_FREE 4 /* free specified dir inodes in map */ 252 #define FFS_FILE_FREE 5 /* free specified file inodes in map */ 253 #define FFS_SET_FLAGS 6 /* set filesystem flags */ 254 #define FFS_ADJ_NDIR 7 /* adjust number of directories */ 255 #define FFS_ADJ_NBFREE 8 /* adjust number of free blocks */ 256 #define FFS_ADJ_NIFREE 9 /* adjust number of free inodes */ 257 #define FFS_ADJ_NFFREE 10 /* adjust number of free frags */ 258 #define FFS_ADJ_NUMCLUSTERS 11 /* adjust number of free clusters */ 259 #define FFS_SET_CWD 12 /* set current directory */ 260 #define FFS_SET_DOTDOT 13 /* set inode number for ".." */ 261 #define FFS_UNLINK 14 /* remove a name in the filesystem */ 262 /* Was FFS_SET_INODE 15 */ 263 /* Was FFS_SET_BUFOUTPUT 16 */ 264 #define FFS_SET_SIZE 17 /* set inode size */ 265 #define FFS_ADJ_DEPTH 18 /* adjust directory inode depth */ 266 #define FFS_MAXID 18 /* number of valid ffs ids */ 267 268 /* 269 * Command structure passed in to the filesystem to adjust filesystem values. 270 */ 271 #define FFS_CMD_VERSION 0x19790518 /* version ID */ 272 struct fsck_cmd { 273 int32_t version; /* version of command structure */ 274 int32_t handle; /* reference to filesystem to be changed */ 275 int64_t value; /* inode or block number to be affected */ 276 int64_t size; /* amount or range to be adjusted */ 277 int64_t spare; /* reserved for future use */ 278 }; 279 280 /* 281 * A recovery structure placed at the end of the boot block area by newfs 282 * that can be used by fsck to search for alternate superblocks. 283 */ 284 struct fsrecovery { 285 int32_t fsr_magic; /* magic number */ 286 int32_t fsr_fsbtodb; /* fsbtodb and dbtofsb shift constant */ 287 int32_t fsr_sblkno; /* offset of super-block in filesys */ 288 int32_t fsr_fpg; /* blocks per group * fs_frag */ 289 uint32_t fsr_ncg; /* number of cylinder groups */ 290 }; 291 292 /* 293 * Per cylinder group information; summarized in blocks allocated 294 * from first cylinder group data blocks. These blocks have to be 295 * read in from fs_csaddr (size fs_cssize) in addition to the 296 * super block. 297 */ 298 struct csum { 299 int32_t cs_ndir; /* number of directories */ 300 int32_t cs_nbfree; /* number of free blocks */ 301 int32_t cs_nifree; /* number of free inodes */ 302 int32_t cs_nffree; /* number of free frags */ 303 }; 304 struct csum_total { 305 int64_t cs_ndir; /* number of directories */ 306 int64_t cs_nbfree; /* number of free blocks */ 307 int64_t cs_nifree; /* number of free inodes */ 308 int64_t cs_nffree; /* number of free frags */ 309 int64_t cs_numclusters; /* number of free clusters */ 310 int64_t cs_spare[3]; /* future expansion */ 311 }; 312 313 /* 314 * Pointers to super block summary information. Placed in a separate 315 * structure so there is just one pointer in the superblock. 316 * 317 * The pointers in this structure are used as follows: 318 * fs_contigdirs references an array that tracks the creation of new 319 * directories 320 * fs_csp references a contiguous array of struct csum for 321 * all cylinder groups 322 * fs_maxcluster references an array of cluster sizes that is computed 323 * as cylinder groups are inspected 324 * fs_active is used when creating snapshots; it points to a bitmap 325 * of cylinder groups for which the free-block bitmap has changed 326 * since the snapshot operation began. 327 */ 328 struct fs_summary_info { 329 uint8_t *si_contigdirs; /* (u) # of contig. allocated dirs */ 330 struct csum *si_csp; /* (u) cg summary info buffer */ 331 int32_t *si_maxcluster; /* (u) max cluster in each cyl group */ 332 uint64_t *si_active; /* (u) used by snapshots to track fs */ 333 }; 334 #define fs_contigdirs fs_si->si_contigdirs 335 #define fs_csp fs_si->si_csp 336 #define fs_maxcluster fs_si->si_maxcluster 337 #define fs_active fs_si->si_active 338 339 /* 340 * Super block for an FFS filesystem. 341 */ 342 struct fs { 343 int32_t fs_firstfield; /* historic filesystem linked list, */ 344 int32_t fs_unused_1; /* used for incore super blocks */ 345 int32_t fs_sblkno; /* offset of super-block in filesys */ 346 int32_t fs_cblkno; /* offset of cyl-block in filesys */ 347 int32_t fs_iblkno; /* offset of inode-blocks in filesys */ 348 int32_t fs_dblkno; /* offset of first data after cg */ 349 int32_t fs_old_cgoffset; /* cylinder group offset in cylinder */ 350 int32_t fs_old_cgmask; /* used to calc mod fs_ntrak */ 351 int32_t fs_old_time; /* last time written */ 352 int32_t fs_old_size; /* number of blocks in fs */ 353 int32_t fs_old_dsize; /* number of data blocks in fs */ 354 uint32_t fs_ncg; /* number of cylinder groups */ 355 int32_t fs_bsize; /* size of basic blocks in fs */ 356 int32_t fs_fsize; /* size of frag blocks in fs */ 357 int32_t fs_frag; /* number of frags in a block in fs */ 358 /* these are configuration parameters */ 359 int32_t fs_minfree; /* minimum percentage of free blocks */ 360 int32_t fs_old_rotdelay; /* num of ms for optimal next block */ 361 int32_t fs_old_rps; /* disk revolutions per second */ 362 /* these fields can be computed from the others */ 363 int32_t fs_bmask; /* ``blkoff'' calc of blk offsets */ 364 int32_t fs_fmask; /* ``fragoff'' calc of frag offsets */ 365 int32_t fs_bshift; /* ``lblkno'' calc of logical blkno */ 366 int32_t fs_fshift; /* ``numfrags'' calc number of frags */ 367 /* these are configuration parameters */ 368 int32_t fs_maxcontig; /* max number of contiguous blks */ 369 int32_t fs_maxbpg; /* max number of blks per cyl group */ 370 /* these fields can be computed from the others */ 371 int32_t fs_fragshift; /* block to frag shift */ 372 int32_t fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */ 373 int32_t fs_sbsize; /* actual size of super block */ 374 int32_t fs_spare1[2]; /* old fs_csmask */ 375 /* old fs_csshift */ 376 int32_t fs_nindir; /* value of NINDIR */ 377 uint32_t fs_inopb; /* value of INOPB */ 378 int32_t fs_old_nspf; /* value of NSPF */ 379 /* yet another configuration parameter */ 380 int32_t fs_optim; /* optimization preference, see below */ 381 int32_t fs_old_npsect; /* # sectors/track including spares */ 382 int32_t fs_old_interleave; /* hardware sector interleave */ 383 int32_t fs_old_trackskew; /* sector 0 skew, per track */ 384 int32_t fs_id[2]; /* unique filesystem id */ 385 /* sizes determined by number of cylinder groups and their sizes */ 386 int32_t fs_old_csaddr; /* blk addr of cyl grp summary area */ 387 int32_t fs_cssize; /* size of cyl grp summary area */ 388 int32_t fs_cgsize; /* cylinder group size */ 389 int32_t fs_spare2; /* old fs_ntrak */ 390 int32_t fs_old_nsect; /* sectors per track */ 391 int32_t fs_old_spc; /* sectors per cylinder */ 392 int32_t fs_old_ncyl; /* cylinders in filesystem */ 393 int32_t fs_old_cpg; /* cylinders per group */ 394 uint32_t fs_ipg; /* inodes per group */ 395 int32_t fs_fpg; /* blocks per group * fs_frag */ 396 /* this data must be re-computed after crashes */ 397 struct csum fs_old_cstotal; /* cylinder summary information */ 398 /* these fields are cleared at mount time */ 399 int8_t fs_fmod; /* super block modified flag */ 400 int8_t fs_clean; /* filesystem is clean flag */ 401 int8_t fs_ronly; /* mounted read-only flag */ 402 int8_t fs_old_flags; /* old FS_ flags */ 403 uint8_t fs_fsmnt[MAXMNTLEN]; /* name mounted on */ 404 uint8_t fs_volname[MAXVOLLEN]; /* volume name */ 405 uint64_t fs_swuid; /* system-wide uid */ 406 int32_t fs_pad; /* due to alignment of fs_swuid */ 407 /* these fields retain the current block allocation info */ 408 int32_t fs_cgrotor; /* last cg searched */ 409 void *fs_ocsp[NOCSPTRS]; /* padding; was list of fs_cs buffers */ 410 struct fs_summary_info *fs_si;/* In-core pointer to summary info */ 411 int32_t fs_old_cpc; /* cyl per cycle in postbl */ 412 int32_t fs_maxbsize; /* maximum blocking factor permitted */ 413 int64_t fs_unrefs; /* number of unreferenced inodes */ 414 int64_t fs_providersize; /* size of underlying GEOM provider */ 415 int64_t fs_metaspace; /* size of area reserved for metadata */ 416 int64_t fs_sparecon64[13]; /* old rotation block list head */ 417 int64_t fs_sblockactualloc; /* byte offset of this superblock */ 418 int64_t fs_sblockloc; /* byte offset of standard superblock */ 419 struct csum_total fs_cstotal; /* (u) cylinder summary information */ 420 ufs_time_t fs_time; /* last time written */ 421 int64_t fs_size; /* number of blocks in fs */ 422 int64_t fs_dsize; /* number of data blocks in fs */ 423 ufs2_daddr_t fs_csaddr; /* blk addr of cyl grp summary area */ 424 int64_t fs_pendingblocks; /* (u) blocks being freed */ 425 uint32_t fs_pendinginodes; /* (u) inodes being freed */ 426 uint32_t fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */ 427 uint32_t fs_avgfilesize; /* expected average file size */ 428 uint32_t fs_avgfpdir; /* expected # of files per directory */ 429 int32_t fs_save_cgsize; /* save real cg size to use fs_bsize */ 430 ufs_time_t fs_mtime; /* Last mount or fsck time. */ 431 int32_t fs_sujfree; /* SUJ free list */ 432 int32_t fs_sparecon32[21]; /* reserved for future constants */ 433 uint32_t fs_ckhash; /* if CK_SUPERBLOCK, its check-hash */ 434 uint32_t fs_metackhash; /* metadata check-hash, see CK_ below */ 435 int32_t fs_flags; /* see FS_ flags below */ 436 int32_t fs_contigsumsize; /* size of cluster summary array */ 437 int32_t fs_maxsymlinklen; /* max length of an internal symlink */ 438 int32_t fs_old_inodefmt; /* format of on-disk inodes */ 439 uint64_t fs_maxfilesize; /* maximum representable file size */ 440 int64_t fs_qbmask; /* ~fs_bmask for use with 64-bit size */ 441 int64_t fs_qfmask; /* ~fs_fmask for use with 64-bit size */ 442 int32_t fs_state; /* validate fs_clean field */ 443 int32_t fs_old_postblformat; /* format of positional layout tables */ 444 int32_t fs_old_nrpos; /* number of rotational positions */ 445 int32_t fs_spare5[2]; /* old fs_postbloff */ 446 /* old fs_rotbloff */ 447 int32_t fs_magic; /* magic number */ 448 }; 449 450 /* Sanity checking. */ 451 #ifdef CTASSERT 452 CTASSERT(sizeof(struct fs) == 1376); 453 #endif 454 455 /* 456 * Filesystem identification 457 */ 458 #define FS_UFS1_MAGIC 0x011954 /* UFS1 fast filesystem magic number */ 459 #define FS_UFS2_MAGIC 0x19540119 /* UFS2 fast filesystem magic number */ 460 #define FS_BAD_MAGIC 0x19960408 /* UFS incomplete newfs magic number */ 461 #define FS_42INODEFMT -1 /* 4.2BSD inode format */ 462 #define FS_44INODEFMT 2 /* 4.4BSD inode format */ 463 464 /* 465 * Preference for optimization. 466 */ 467 #define FS_OPTTIME 0 /* minimize allocation time */ 468 #define FS_OPTSPACE 1 /* minimize disk fragmentation */ 469 470 /* 471 * Filesystem flags. 472 * 473 * The FS_UNCLEAN flag is set by the kernel when the filesystem was 474 * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates 475 * that the filesystem should be managed by the soft updates code. 476 * Note that the FS_NEEDSFSCK flag is set and cleared by the fsck 477 * utility. It is set when background fsck finds an unexpected 478 * inconsistency which requires a traditional foreground fsck to be 479 * run. Such inconsistencies should only be found after an uncorrectable 480 * disk error. The FS_NEEDSFSCK can also be set when a mounted filesystem 481 * discovers an internal inconsistency such as freeing a freed inode. 482 * A foreground fsck will clear the FS_NEEDSFSCK flag when it has 483 * successfully cleaned up the filesystem. The kernel uses this 484 * flag to enforce that inconsistent filesystems be mounted read-only. 485 * 486 * The FS_METACKHASH flag when set indicates that the kernel maintains 487 * one or more check hashes. The actual set of supported check hashes 488 * is stored in the fs_metackhash field. Kernels that do not support 489 * check hashes clear the FS_METACKHASH flag to indicate that the 490 * check hashes need to be rebuilt (by fsck) before they can be used. 491 * 492 * When a filesystem is mounted, any flags not included in FS_SUPPORTED 493 * are cleared. This lets newer features know that the filesystem has 494 * been run on an older version of the filesystem and thus that data 495 * structures associated with those features are out-of-date and need 496 * to be rebuilt. 497 * 498 * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled 499 * for the file system, so they should be loaded from extended attributes, 500 * observed for access control purposes, and be administered by object 501 * owners. FS_NFS4ACLS indicates that NFSv4 ACLs are administratively 502 * enabled. This flag is mutually exclusive with FS_ACLS. FS_MULTILABEL 503 * indicates that the TrustedBSD MAC Framework should attempt to back MAC 504 * labels into extended attributes on the file system rather than maintain 505 * a single mount label for all objects. 506 */ 507 #define FS_UNCLEAN 0x00000001 /* filesystem not clean at mount */ 508 #define FS_DOSOFTDEP 0x00000002 /* filesystem using soft dependencies */ 509 #define FS_NEEDSFSCK 0x00000004 /* filesystem needs sync fsck before mount */ 510 #define FS_SUJ 0x00000008 /* Filesystem using softupdate journal */ 511 #define FS_ACLS 0x00000010 /* file system has POSIX.1e ACLs enabled */ 512 #define FS_MULTILABEL 0x00000020 /* file system is MAC multi-label */ 513 #define FS_GJOURNAL 0x00000040 /* gjournaled file system */ 514 #define FS_FLAGS_UPDATED 0x0000080 /* flags have been moved to new location */ 515 #define FS_NFS4ACLS 0x00000100 /* file system has NFSv4 ACLs enabled */ 516 #define FS_METACKHASH 0x00000200 /* kernel supports metadata check hashes */ 517 #define FS_TRIM 0x00000400 /* issue BIO_DELETE for deleted blocks */ 518 #define FS_SUPPORTED 0x00FFFFFF /* supported flags, others cleared at mount*/ 519 /* 520 * Things that we may someday support, but currently do not. 521 * These flags are all cleared so we know if we ran on a kernel 522 * that does not support them. 523 */ 524 #define FS_INDEXDIRS 0x01000000 /* kernel supports indexed directories */ 525 #define FS_VARBLKSIZE 0x02000000 /* kernel supports variable block sizes */ 526 #define FS_COOLOPT1 0x04000000 /* kernel supports cool option 1 */ 527 #define FS_COOLOPT2 0x08000000 /* kernel supports cool option 2 */ 528 #define FS_COOLOPT3 0x10000000 /* kernel supports cool option 3 */ 529 #define FS_COOLOPT4 0x20000000 /* kernel supports cool option 4 */ 530 #define FS_COOLOPT5 0x40000000 /* kernel supports cool option 5 */ 531 #define FS_COOLOPT6 0x80000000 /* kernel supports cool option 6 */ 532 533 /* 534 * The fs_metackhash field indicates the types of metadata check-hash 535 * that are maintained for a filesystem. Not all filesystems check-hash 536 * all metadata. 537 */ 538 #define CK_SUPERBLOCK 0x0001 /* the superblock */ 539 #define CK_CYLGRP 0x0002 /* the cylinder groups */ 540 #define CK_INODE 0x0004 /* inodes */ 541 #define CK_INDIR 0x0008 /* indirect blocks */ 542 #define CK_DIR 0x0010 /* directory contents */ 543 #define CK_SUPPORTED 0x0007 /* supported flags, others cleared at mount */ 544 /* 545 * The BX_FSPRIV buffer b_xflags are used to track types of data in buffers. 546 */ 547 #define BX_SUPERBLOCK 0x00010000 /* superblock */ 548 #define BX_CYLGRP 0x00020000 /* cylinder groups */ 549 #define BX_INODE 0x00040000 /* inodes */ 550 #define BX_INDIR 0x00080000 /* indirect blocks */ 551 #define BX_DIR 0x00100000 /* directory contents */ 552 553 #define PRINT_UFS_BUF_XFLAGS "\20\25dir\24indir\23inode\22cylgrp\21superblock" 554 555 /* 556 * Macros to access bits in the fs_active array. 557 */ 558 #define ACTIVECGNUM(fs, cg) ((fs)->fs_active[(cg) / (NBBY * sizeof(uint64_t))]) 559 #define ACTIVECGOFF(cg) (1 << ((cg) % (NBBY * sizeof(uint64_t)))) 560 #define ACTIVESET(fs, cg) do { \ 561 if ((fs)->fs_active) \ 562 ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg)); \ 563 } while (0) 564 #define ACTIVECLEAR(fs, cg) do { \ 565 if ((fs)->fs_active) \ 566 ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg)); \ 567 } while (0) 568 569 /* 570 * The size of a cylinder group is calculated by CGSIZE. The maximum size 571 * is limited by the fact that cylinder groups are at most one block. 572 * Its size is derived from the size of the maps maintained in the 573 * cylinder group and the (struct cg) size. 574 */ 575 #define CGSIZE(fs) \ 576 /* base cg */ (sizeof(struct cg) + \ 577 /* old btotoff */ (fs)->fs_old_cpg * sizeof(int32_t) + \ 578 /* old boff */ (fs)->fs_old_cpg * sizeof(uint16_t) + \ 579 /* inode map */ howmany((fs)->fs_ipg, NBBY) + \ 580 /* block map */ howmany((fs)->fs_fpg, NBBY) + sizeof(int32_t) + \ 581 /* if present */ ((fs)->fs_contigsumsize <= 0 ? 0 : \ 582 /* cluster sum */ (fs)->fs_contigsumsize * sizeof(int32_t) + \ 583 /* cluster map */ howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY))) 584 585 /* 586 * The minimal number of cylinder groups that should be created. 587 */ 588 #define MINCYLGRPS 4 589 590 /* 591 * Convert cylinder group to base address of its global summary info. 592 */ 593 #define fs_cs(fs, indx) fs_csp[indx] 594 595 /* 596 * Cylinder group block for a filesystem. 597 */ 598 #define CG_MAGIC 0x090255 599 struct cg { 600 int32_t cg_firstfield; /* historic cyl groups linked list */ 601 int32_t cg_magic; /* magic number */ 602 int32_t cg_old_time; /* time last written */ 603 uint32_t cg_cgx; /* we are the cgx'th cylinder group */ 604 int16_t cg_old_ncyl; /* number of cyl's this cg */ 605 int16_t cg_old_niblk; /* number of inode blocks this cg */ 606 uint32_t cg_ndblk; /* number of data blocks this cg */ 607 struct csum cg_cs; /* cylinder summary information */ 608 uint32_t cg_rotor; /* position of last used block */ 609 uint32_t cg_frotor; /* position of last used frag */ 610 uint32_t cg_irotor; /* position of last used inode */ 611 uint32_t cg_frsum[MAXFRAG]; /* counts of available frags */ 612 int32_t cg_old_btotoff; /* (int32) block totals per cylinder */ 613 int32_t cg_old_boff; /* (uint16) free block positions */ 614 uint32_t cg_iusedoff; /* (uint8) used inode map */ 615 uint32_t cg_freeoff; /* (uint8) free block map */ 616 uint32_t cg_nextfreeoff; /* (uint8) next available space */ 617 uint32_t cg_clustersumoff; /* (uint32) counts of avail clusters */ 618 uint32_t cg_clusteroff; /* (uint8) free cluster map */ 619 uint32_t cg_nclusterblks; /* number of clusters this cg */ 620 uint32_t cg_niblk; /* number of inode blocks this cg */ 621 uint32_t cg_initediblk; /* last initialized inode */ 622 uint32_t cg_unrefs; /* number of unreferenced inodes */ 623 int32_t cg_sparecon32[1]; /* reserved for future use */ 624 uint32_t cg_ckhash; /* check-hash of this cg */ 625 ufs_time_t cg_time; /* time last written */ 626 int64_t cg_sparecon64[3]; /* reserved for future use */ 627 /* actually longer - space used for cylinder group maps */ 628 }; 629 630 /* 631 * Macros for access to cylinder group array structures 632 */ 633 #define cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC) 634 #define cg_inosused(cgp) \ 635 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_iusedoff)) 636 #define cg_blksfree(cgp) \ 637 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_freeoff)) 638 #define cg_clustersfree(cgp) \ 639 ((uint8_t *)((uint8_t *)(cgp) + (cgp)->cg_clusteroff)) 640 #define cg_clustersum(cgp) \ 641 ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff)) 642 643 /* 644 * Turn filesystem block numbers into disk block addresses. 645 * This maps filesystem blocks to device size blocks. 646 */ 647 #define fsbtodb(fs, b) ((daddr_t)(b) << (fs)->fs_fsbtodb) 648 #define dbtofsb(fs, b) ((b) >> (fs)->fs_fsbtodb) 649 650 /* 651 * Cylinder group macros to locate things in cylinder groups. 652 * They calc filesystem addresses of cylinder group data structures. 653 */ 654 #define cgbase(fs, c) (((ufs2_daddr_t)(fs)->fs_fpg) * (c)) 655 #define cgdata(fs, c) (cgdmin(fs, c) + (fs)->fs_metaspace) /* data zone */ 656 #define cgmeta(fs, c) (cgdmin(fs, c)) /* meta data */ 657 #define cgdmin(fs, c) (cgstart(fs, c) + (fs)->fs_dblkno) /* 1st data */ 658 #define cgimin(fs, c) (cgstart(fs, c) + (fs)->fs_iblkno) /* inode blk */ 659 #define cgsblock(fs, c) (cgstart(fs, c) + (fs)->fs_sblkno) /* super blk */ 660 #define cgtod(fs, c) (cgstart(fs, c) + (fs)->fs_cblkno) /* cg block */ 661 #define cgstart(fs, c) \ 662 ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) : \ 663 (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask)))) 664 665 /* 666 * Macros for handling inode numbers: 667 * inode number to filesystem block offset. 668 * inode number to cylinder group number. 669 * inode number to filesystem block address. 670 */ 671 #define ino_to_cg(fs, x) (((ino_t)(x)) / (fs)->fs_ipg) 672 #define ino_to_fsba(fs, x) \ 673 ((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) + \ 674 (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs)))))) 675 #define ino_to_fsbo(fs, x) (((ino_t)(x)) % INOPB(fs)) 676 677 /* 678 * Give cylinder group number for a filesystem block. 679 * Give cylinder group block number for a filesystem block. 680 */ 681 #define dtog(fs, d) ((d) / (fs)->fs_fpg) 682 #define dtogd(fs, d) ((d) % (fs)->fs_fpg) 683 684 /* 685 * Extract the bits for a block from a map. 686 * Compute the cylinder and rotational position of a cyl block addr. 687 */ 688 #define blkmap(fs, map, loc) \ 689 (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag))) 690 691 /* 692 * The following macros optimize certain frequently calculated 693 * quantities by using shifts and masks in place of divisions 694 * modulos and multiplications. 695 */ 696 #define blkoff(fs, loc) /* calculates (loc % fs->fs_bsize) */ \ 697 ((loc) & (fs)->fs_qbmask) 698 #define fragoff(fs, loc) /* calculates (loc % fs->fs_fsize) */ \ 699 ((loc) & (fs)->fs_qfmask) 700 #define lfragtosize(fs, frag) /* calculates ((off_t)frag * fs->fs_fsize) */ \ 701 (((off_t)(frag)) << (fs)->fs_fshift) 702 #define lblktosize(fs, blk) /* calculates ((off_t)blk * fs->fs_bsize) */ \ 703 (((off_t)(blk)) << (fs)->fs_bshift) 704 /* Use this only when `blk' is known to be small, e.g., < UFS_NDADDR. */ 705 #define smalllblktosize(fs, blk) /* calculates (blk * fs->fs_bsize) */ \ 706 ((blk) << (fs)->fs_bshift) 707 #define lblkno(fs, loc) /* calculates (loc / fs->fs_bsize) */ \ 708 ((loc) >> (fs)->fs_bshift) 709 #define numfrags(fs, loc) /* calculates (loc / fs->fs_fsize) */ \ 710 ((loc) >> (fs)->fs_fshift) 711 #define blkroundup(fs, size) /* calculates roundup(size, fs->fs_bsize) */ \ 712 (((size) + (fs)->fs_qbmask) & (fs)->fs_bmask) 713 #define fragroundup(fs, size) /* calculates roundup(size, fs->fs_fsize) */ \ 714 (((size) + (fs)->fs_qfmask) & (fs)->fs_fmask) 715 #define fragstoblks(fs, frags) /* calculates (frags / fs->fs_frag) */ \ 716 ((frags) >> (fs)->fs_fragshift) 717 #define blkstofrags(fs, blks) /* calculates (blks * fs->fs_frag) */ \ 718 ((blks) << (fs)->fs_fragshift) 719 #define fragnum(fs, fsb) /* calculates (fsb % fs->fs_frag) */ \ 720 ((fsb) & ((fs)->fs_frag - 1)) 721 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->fs_frag) */ \ 722 ((fsb) &~ ((fs)->fs_frag - 1)) 723 724 /* 725 * Determine the number of available frags given a 726 * percentage to hold in reserve. 727 */ 728 #define freespace(fs, percentreserved) \ 729 (blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \ 730 (fs)->fs_cstotal.cs_nffree - \ 731 (((off_t)((fs)->fs_dsize)) * (percentreserved) / 100)) 732 733 /* 734 * Determining the size of a file block in the filesystem. 735 */ 736 #define blksize(fs, ip, lbn) \ 737 (((lbn) >= UFS_NDADDR || (ip)->i_size >= \ 738 (uint64_t)smalllblktosize(fs, (lbn) + 1)) \ 739 ? (fs)->fs_bsize \ 740 : (fragroundup(fs, blkoff(fs, (ip)->i_size)))) 741 #define sblksize(fs, size, lbn) \ 742 (((lbn) >= UFS_NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \ 743 ? (fs)->fs_bsize \ 744 : (fragroundup(fs, blkoff(fs, (size))))) 745 746 /* 747 * Number of indirects in a filesystem block. 748 */ 749 #define NINDIR(fs) ((fs)->fs_nindir) 750 751 /* 752 * Indirect lbns are aligned on UFS_NDADDR addresses where single indirects 753 * are the negated address of the lowest lbn reachable, double indirects 754 * are this lbn - 1 and triple indirects are this lbn - 2. This yields 755 * an unusual bit order to determine level. 756 */ 757 static inline int 758 lbn_level(ufs_lbn_t lbn) 759 { 760 if (lbn >= 0) 761 return 0; 762 switch (lbn & 0x3) { 763 case 0: 764 return (0); 765 case 1: 766 break; 767 case 2: 768 return (2); 769 case 3: 770 return (1); 771 default: 772 break; 773 } 774 return (-1); 775 } 776 777 static inline ufs_lbn_t 778 lbn_offset(struct fs *fs, int level) 779 { 780 ufs_lbn_t res; 781 782 for (res = 1; level > 0; level--) 783 res *= NINDIR(fs); 784 return (res); 785 } 786 787 /* 788 * Number of inodes in a secondary storage block/fragment. 789 */ 790 #define INOPB(fs) ((fs)->fs_inopb) 791 #define INOPF(fs) ((fs)->fs_inopb >> (fs)->fs_fragshift) 792 793 /* 794 * Softdep journal record format. 795 */ 796 #define JOP_UNKNOWN 0 /* JOP operation is unknown */ 797 #define JOP_ADDREF 1 /* Add a reference to an inode. */ 798 #define JOP_REMREF 2 /* Remove a reference from an inode. */ 799 #define JOP_NEWBLK 3 /* Allocate a block. */ 800 #define JOP_FREEBLK 4 /* Free a block or a tree of blocks. */ 801 #define JOP_MVREF 5 /* Move a reference from one off to another. */ 802 #define JOP_TRUNC 6 /* Partial truncation record. */ 803 #define JOP_SYNC 7 /* fsync() complete record. */ 804 #define JOP_NUMJOPTYPES 8 805 #define JOP_NAMES { \ 806 "unknown", \ 807 "JOP_ADDREF", \ 808 "JOP_REMREF", \ 809 "JOP_NEWBLK", \ 810 "JOP_FREEBLK", \ 811 "JOP_MVREF", \ 812 "JOP_TRUNC", \ 813 "JOP_SYNC" } 814 #define JOP_OPTYPE(op) \ 815 (op) < JOP_NUMJOPTYPES ? joptype[op] : joptype[JOP_UNKNOWN] 816 817 #define JREC_SIZE 32 /* Record and segment header size. */ 818 819 #define SUJ_MIN (4 * 1024 * 1024) /* Minimum journal size */ 820 #define SUJ_FILE ".sujournal" /* Journal file name */ 821 822 /* 823 * Size of the segment record header. There is at most one for each disk 824 * block in the journal. The segment header is followed by an array of 825 * records. fsck depends on the first element in each record being 'op' 826 * and the second being 'ino'. Segments may span multiple disk blocks but 827 * the header is present on each. 828 */ 829 struct jsegrec { 830 uint64_t jsr_seq; /* Our sequence number */ 831 uint64_t jsr_oldest; /* Oldest valid sequence number */ 832 uint16_t jsr_cnt; /* Count of valid records */ 833 uint16_t jsr_blocks; /* Count of device bsize blocks. */ 834 uint32_t jsr_crc; /* 32bit crc of the valid space */ 835 ufs_time_t jsr_time; /* timestamp for mount instance */ 836 }; 837 838 /* 839 * Reference record. Records a single link count modification. 840 */ 841 struct jrefrec { 842 uint32_t jr_op; 843 uint32_t jr_ino; 844 uint32_t jr_parent; 845 uint16_t jr_nlink; 846 uint16_t jr_mode; 847 int64_t jr_diroff; 848 uint64_t jr_unused; 849 }; 850 851 /* 852 * Move record. Records a reference moving within a directory block. The 853 * nlink is unchanged but we must search both locations. 854 */ 855 struct jmvrec { 856 uint32_t jm_op; 857 uint32_t jm_ino; 858 uint32_t jm_parent; 859 uint16_t jm_unused; 860 int64_t jm_oldoff; 861 int64_t jm_newoff; 862 }; 863 864 /* 865 * Block record. A set of frags or tree of blocks starting at an indirect are 866 * freed or a set of frags are allocated. 867 */ 868 struct jblkrec { 869 uint32_t jb_op; 870 uint32_t jb_ino; 871 ufs2_daddr_t jb_blkno; 872 ufs_lbn_t jb_lbn; 873 uint16_t jb_frags; 874 uint16_t jb_oldfrags; 875 uint32_t jb_unused; 876 }; 877 878 /* 879 * Truncation record. Records a partial truncation so that it may be 880 * completed at check time. Also used for sync records. 881 */ 882 struct jtrncrec { 883 uint32_t jt_op; 884 uint32_t jt_ino; 885 int64_t jt_size; 886 uint32_t jt_extsize; 887 uint32_t jt_pad[3]; 888 }; 889 890 union jrec { 891 struct jsegrec rec_jsegrec; 892 struct jrefrec rec_jrefrec; 893 struct jmvrec rec_jmvrec; 894 struct jblkrec rec_jblkrec; 895 struct jtrncrec rec_jtrncrec; 896 }; 897 898 #ifdef CTASSERT 899 CTASSERT(sizeof(struct jsegrec) == JREC_SIZE); 900 CTASSERT(sizeof(struct jrefrec) == JREC_SIZE); 901 CTASSERT(sizeof(struct jmvrec) == JREC_SIZE); 902 CTASSERT(sizeof(struct jblkrec) == JREC_SIZE); 903 CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE); 904 CTASSERT(sizeof(union jrec) == JREC_SIZE); 905 #endif 906 907 extern int inside[], around[]; 908 extern uint8_t *fragtbl[]; 909 910 /* 911 * IOCTLs used for filesystem write suspension. 912 */ 913 #define UFSSUSPEND _IOW('U', 1, fsid_t) 914 #define UFSRESUME _IO('U', 2) 915 916 #endif 917