1 /* 2 * Copyright (c) 1982, 1986, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * %sccs.include.redist.c% 6 * 7 * @(#)fs.h 8.6 (Berkeley) 04/19/94 8 */ 9 10 /* 11 * Each disk drive contains some number of file systems. 12 * A file system consists of a number of cylinder groups. 13 * Each cylinder group has inodes and data. 14 * 15 * A file system is described by its super-block, which in turn 16 * describes the cylinder groups. The super-block is critical 17 * data and is replicated in each cylinder group to protect against 18 * catastrophic loss. This is done at `newfs' time and the critical 19 * super-block data does not change, so the copies need not be 20 * referenced further unless disaster strikes. 21 * 22 * For file system fs, the offsets of the various blocks of interest 23 * are given in the super block as: 24 * [fs->fs_sblkno] Super-block 25 * [fs->fs_cblkno] Cylinder group block 26 * [fs->fs_iblkno] Inode blocks 27 * [fs->fs_dblkno] Data blocks 28 * The beginning of cylinder group cg in fs, is given by 29 * the ``cgbase(fs, cg)'' macro. 30 * 31 * The first boot and super blocks are given in absolute disk addresses. 32 * The byte-offset forms are preferred, as they don't imply a sector size. 33 */ 34 #define BBSIZE 8192 35 #define SBSIZE 8192 36 #define BBOFF ((off_t)(0)) 37 #define SBOFF ((off_t)(BBOFF + BBSIZE)) 38 #define BBLOCK ((daddr_t)(0)) 39 #define SBLOCK ((daddr_t)(BBLOCK + BBSIZE / DEV_BSIZE)) 40 41 /* 42 * Addresses stored in inodes are capable of addressing fragments 43 * of `blocks'. File system blocks of at most size MAXBSIZE can 44 * be optionally broken into 2, 4, or 8 pieces, each of which is 45 * addressible; these pieces may be DEV_BSIZE, or some multiple of 46 * a DEV_BSIZE unit. 47 * 48 * Large files consist of exclusively large data blocks. To avoid 49 * undue wasted disk space, the last data block of a small file may be 50 * allocated as only as many fragments of a large block as are 51 * necessary. The file system format retains only a single pointer 52 * to such a fragment, which is a piece of a single large block that 53 * has been divided. The size of such a fragment is determinable from 54 * information in the inode, using the ``blksize(fs, ip, lbn)'' macro. 55 * 56 * The file system records space availability at the fragment level; 57 * to determine block availability, aligned fragments are examined. 58 */ 59 60 /* 61 * MINBSIZE is the smallest allowable block size. 62 * In order to insure that it is possible to create files of size 63 * 2^32 with only two levels of indirection, MINBSIZE is set to 4096. 64 * MINBSIZE must be big enough to hold a cylinder group block, 65 * thus changes to (struct cg) must keep its size within MINBSIZE. 66 * Note that super blocks are always of size SBSIZE, 67 * and that both SBSIZE and MAXBSIZE must be >= MINBSIZE. 68 */ 69 #define MINBSIZE 4096 70 71 /* 72 * The path name on which the file system is mounted is maintained 73 * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in 74 * the super block for this name. 75 * The limit on the amount of summary information per file system 76 * is defined by MAXCSBUFS. It is currently parameterized for a 77 * maximum of two million cylinders. 78 */ 79 #define MAXMNTLEN 512 80 #define MAXCSBUFS 32 81 82 /* 83 * A summary of contiguous blocks of various sizes is maintained 84 * in each cylinder group. Normally this is set by the initial 85 * value of fs_maxcontig. To conserve space, a maximum summary size 86 * is set by FS_MAXCONTIG. 87 */ 88 #define FS_MAXCONTIG 16 89 90 /* 91 * MINFREE gives the minimum acceptable percentage of file system 92 * blocks which may be free. If the freelist drops below this level 93 * only the superuser may continue to allocate blocks. This may 94 * be set to 0 if no reserve of free blocks is deemed necessary, 95 * however throughput drops by fifty percent if the file system 96 * is run at between 95% and 100% full; thus the minimum default 97 * value of fs_minfree is 5%. However, to get good clustering 98 * performance, 10% is a better choice. hence we use 10% as our 99 * default value. With 10% free space, fragmentation is not a 100 * problem, so we choose to optimize for time. 101 */ 102 #define MINFREE 5 103 #define DEFAULTOPT FS_OPTTIME 104 105 /* 106 * Per cylinder group information; summarized in blocks allocated 107 * from first cylinder group data blocks. These blocks have to be 108 * read in from fs_csaddr (size fs_cssize) in addition to the 109 * super block. 110 * 111 * N.B. sizeof(struct csum) must be a power of two in order for 112 * the ``fs_cs'' macro to work (see below). 113 */ 114 struct csum { 115 long cs_ndir; /* number of directories */ 116 long cs_nbfree; /* number of free blocks */ 117 long cs_nifree; /* number of free inodes */ 118 long cs_nffree; /* number of free frags */ 119 }; 120 121 /* 122 * Super block for a file system. 123 */ 124 struct fs { 125 struct fs *fs_link; /* linked list of file systems */ 126 struct fs *fs_rlink; /* used for incore super blocks */ 127 daddr_t fs_sblkno; /* addr of super-block in filesys */ 128 daddr_t fs_cblkno; /* offset of cyl-block in filesys */ 129 daddr_t fs_iblkno; /* offset of inode-blocks in filesys */ 130 daddr_t fs_dblkno; /* offset of first data after cg */ 131 long fs_cgoffset; /* cylinder group offset in cylinder */ 132 long fs_cgmask; /* used to calc mod fs_ntrak */ 133 time_t fs_time; /* last time written */ 134 long fs_size; /* number of blocks in fs */ 135 long fs_dsize; /* number of data blocks in fs */ 136 long fs_ncg; /* number of cylinder groups */ 137 long fs_bsize; /* size of basic blocks in fs */ 138 long fs_fsize; /* size of frag blocks in fs */ 139 long fs_frag; /* number of frags in a block in fs */ 140 /* these are configuration parameters */ 141 long fs_minfree; /* minimum percentage of free blocks */ 142 long fs_rotdelay; /* num of ms for optimal next block */ 143 long fs_rps; /* disk revolutions per second */ 144 /* these fields can be computed from the others */ 145 long fs_bmask; /* ``blkoff'' calc of blk offsets */ 146 long fs_fmask; /* ``fragoff'' calc of frag offsets */ 147 long fs_bshift; /* ``lblkno'' calc of logical blkno */ 148 long fs_fshift; /* ``numfrags'' calc number of frags */ 149 /* these are configuration parameters */ 150 long fs_maxcontig; /* max number of contiguous blks */ 151 long fs_maxbpg; /* max number of blks per cyl group */ 152 /* these fields can be computed from the others */ 153 long fs_fragshift; /* block to frag shift */ 154 long fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */ 155 long fs_sbsize; /* actual size of super block */ 156 long fs_csmask; /* csum block offset */ 157 long fs_csshift; /* csum block number */ 158 long fs_nindir; /* value of NINDIR */ 159 long fs_inopb; /* value of INOPB */ 160 long fs_nspf; /* value of NSPF */ 161 /* yet another configuration parameter */ 162 long fs_optim; /* optimization preference, see below */ 163 /* these fields are derived from the hardware */ 164 long fs_npsect; /* # sectors/track including spares */ 165 long fs_interleave; /* hardware sector interleave */ 166 long fs_trackskew; /* sector 0 skew, per track */ 167 long fs_headswitch; /* head switch time, usec */ 168 long fs_trkseek; /* track-to-track seek, usec */ 169 /* sizes determined by number of cylinder groups and their sizes */ 170 daddr_t fs_csaddr; /* blk addr of cyl grp summary area */ 171 long fs_cssize; /* size of cyl grp summary area */ 172 long fs_cgsize; /* cylinder group size */ 173 /* these fields are derived from the hardware */ 174 long fs_ntrak; /* tracks per cylinder */ 175 long fs_nsect; /* sectors per track */ 176 long fs_spc; /* sectors per cylinder */ 177 /* this comes from the disk driver partitioning */ 178 long fs_ncyl; /* cylinders in file system */ 179 /* these fields can be computed from the others */ 180 long fs_cpg; /* cylinders per group */ 181 long fs_ipg; /* inodes per group */ 182 long fs_fpg; /* blocks per group * fs_frag */ 183 /* this data must be re-computed after crashes */ 184 struct csum fs_cstotal; /* cylinder summary information */ 185 /* these fields are cleared at mount time */ 186 char fs_fmod; /* super block modified flag */ 187 char fs_clean; /* file system is clean flag */ 188 char fs_ronly; /* mounted read-only flag */ 189 char fs_flags; /* currently unused flag */ 190 char fs_fsmnt[MAXMNTLEN]; /* name mounted on */ 191 /* these fields retain the current block allocation info */ 192 long fs_cgrotor; /* last cg searched */ 193 struct csum *fs_csp[MAXCSBUFS];/* list of fs_cs info buffers */ 194 long fs_cpc; /* cyl per cycle in postbl */ 195 short fs_opostbl[16][8]; /* old rotation block list head */ 196 long fs_sparecon[50]; /* reserved for future constants */ 197 long fs_contigsumsize; /* size of cluster summary array */ 198 long fs_maxsymlinklen; /* max length of an internal symlink */ 199 long fs_inodefmt; /* format of on-disk inodes */ 200 u_quad_t fs_maxfilesize; /* maximum representable file size */ 201 quad_t fs_qbmask; /* ~fs_bmask - for use with quad size */ 202 quad_t fs_qfmask; /* ~fs_fmask - for use with quad size */ 203 long fs_state; /* validate fs_clean field */ 204 long fs_postblformat; /* format of positional layout tables */ 205 long fs_nrpos; /* number of rotaional positions */ 206 long fs_postbloff; /* (short) rotation block list head */ 207 long fs_rotbloff; /* (u_char) blocks for each rotation */ 208 long fs_magic; /* magic number */ 209 u_char fs_space[1]; /* list of blocks for each rotation */ 210 /* actually longer */ 211 }; 212 /* 213 * Filesystem idetification 214 */ 215 #define FS_MAGIC 0x011954 /* the fast filesystem magic number */ 216 #define FS_OKAY 0x7c269d38 /* superblock checksum */ 217 #define FS_42INODEFMT -1 /* 4.2BSD inode format */ 218 #define FS_44INODEFMT 2 /* 4.4BSD inode format */ 219 /* 220 * Preference for optimization. 221 */ 222 #define FS_OPTTIME 0 /* minimize allocation time */ 223 #define FS_OPTSPACE 1 /* minimize disk fragmentation */ 224 225 /* 226 * Rotational layout table format types 227 */ 228 #define FS_42POSTBLFMT -1 /* 4.2BSD rotational table format */ 229 #define FS_DYNAMICPOSTBLFMT 1 /* dynamic rotational table format */ 230 /* 231 * Macros for access to superblock array structures 232 */ 233 #define fs_postbl(fs, cylno) \ 234 (((fs)->fs_postblformat == FS_42POSTBLFMT) \ 235 ? ((fs)->fs_opostbl[cylno]) \ 236 : ((short *)((char *)(fs) + (fs)->fs_postbloff) + (cylno) * (fs)->fs_nrpos)) 237 #define fs_rotbl(fs) \ 238 (((fs)->fs_postblformat == FS_42POSTBLFMT) \ 239 ? ((fs)->fs_space) \ 240 : ((u_char *)((char *)(fs) + (fs)->fs_rotbloff))) 241 242 /* 243 * The size of a cylinder group is calculated by CGSIZE. The maximum size 244 * is limited by the fact that cylinder groups are at most one block. 245 * Its size is derived from the size of the maps maintained in the 246 * cylinder group and the (struct cg) size. 247 */ 248 #define CGSIZE(fs) \ 249 /* base cg */ (sizeof(struct cg) + sizeof(long) + \ 250 /* blktot size */ (fs)->fs_cpg * sizeof(long) + \ 251 /* blks size */ (fs)->fs_cpg * (fs)->fs_nrpos * sizeof(short) + \ 252 /* inode map */ howmany((fs)->fs_ipg, NBBY) + \ 253 /* block map */ howmany((fs)->fs_cpg * (fs)->fs_spc / NSPF(fs), NBBY) +\ 254 /* if present */ ((fs)->fs_contigsumsize <= 0 ? 0 : \ 255 /* cluster sum */ (fs)->fs_contigsumsize * sizeof(long) + \ 256 /* cluster map */ howmany((fs)->fs_cpg * (fs)->fs_spc / NSPB(fs), NBBY))) 257 258 /* 259 * Convert cylinder group to base address of its global summary info. 260 * 261 * N.B. This macro assumes that sizeof(struct csum) is a power of two. 262 */ 263 #define fs_cs(fs, indx) \ 264 fs_csp[(indx) >> (fs)->fs_csshift][(indx) & ~(fs)->fs_csmask] 265 266 /* 267 * Cylinder group block for a file system. 268 */ 269 #define CG_MAGIC 0x090255 270 struct cg { 271 struct cg *cg_link; /* linked list of cyl groups */ 272 long cg_magic; /* magic number */ 273 time_t cg_time; /* time last written */ 274 long cg_cgx; /* we are the cgx'th cylinder group */ 275 short cg_ncyl; /* number of cyl's this cg */ 276 short cg_niblk; /* number of inode blocks this cg */ 277 long cg_ndblk; /* number of data blocks this cg */ 278 struct csum cg_cs; /* cylinder summary information */ 279 long cg_rotor; /* position of last used block */ 280 long cg_frotor; /* position of last used frag */ 281 long cg_irotor; /* position of last used inode */ 282 long cg_frsum[MAXFRAG]; /* counts of available frags */ 283 long cg_btotoff; /* (long) block totals per cylinder */ 284 long cg_boff; /* (short) free block positions */ 285 long cg_iusedoff; /* (char) used inode map */ 286 long cg_freeoff; /* (u_char) free block map */ 287 long cg_nextfreeoff; /* (u_char) next available space */ 288 long cg_clustersumoff; /* (long) counts of avail clusters */ 289 long cg_clusteroff; /* (char) free cluster map */ 290 long cg_nclusterblks; /* number of clusters this cg */ 291 long cg_sparecon[13]; /* reserved for future use */ 292 u_char cg_space[1]; /* space for cylinder group maps */ 293 /* actually longer */ 294 }; 295 /* 296 * Macros for access to cylinder group array structures 297 */ 298 #define cg_blktot(cgp) \ 299 (((cgp)->cg_magic != CG_MAGIC) \ 300 ? (((struct ocg *)(cgp))->cg_btot) \ 301 : ((long *)((char *)(cgp) + (cgp)->cg_btotoff))) 302 #define cg_blks(fs, cgp, cylno) \ 303 (((cgp)->cg_magic != CG_MAGIC) \ 304 ? (((struct ocg *)(cgp))->cg_b[cylno]) \ 305 : ((short *)((char *)(cgp) + (cgp)->cg_boff) + (cylno) * (fs)->fs_nrpos)) 306 #define cg_inosused(cgp) \ 307 (((cgp)->cg_magic != CG_MAGIC) \ 308 ? (((struct ocg *)(cgp))->cg_iused) \ 309 : ((char *)((char *)(cgp) + (cgp)->cg_iusedoff))) 310 #define cg_blksfree(cgp) \ 311 (((cgp)->cg_magic != CG_MAGIC) \ 312 ? (((struct ocg *)(cgp))->cg_free) \ 313 : ((u_char *)((char *)(cgp) + (cgp)->cg_freeoff))) 314 #define cg_chkmagic(cgp) \ 315 ((cgp)->cg_magic == CG_MAGIC || ((struct ocg *)(cgp))->cg_magic == CG_MAGIC) 316 #define cg_clustersfree(cgp) \ 317 ((u_char *)((char *)(cgp) + (cgp)->cg_clusteroff)) 318 #define cg_clustersum(cgp) \ 319 ((long *)((char *)(cgp) + (cgp)->cg_clustersumoff)) 320 321 /* 322 * The following structure is defined 323 * for compatibility with old file systems. 324 */ 325 struct ocg { 326 struct ocg *cg_link; /* linked list of cyl groups */ 327 struct ocg *cg_rlink; /* used for incore cyl groups */ 328 time_t cg_time; /* time last written */ 329 long cg_cgx; /* we are the cgx'th cylinder group */ 330 short cg_ncyl; /* number of cyl's this cg */ 331 short cg_niblk; /* number of inode blocks this cg */ 332 long cg_ndblk; /* number of data blocks this cg */ 333 struct csum cg_cs; /* cylinder summary information */ 334 long cg_rotor; /* position of last used block */ 335 long cg_frotor; /* position of last used frag */ 336 long cg_irotor; /* position of last used inode */ 337 long cg_frsum[8]; /* counts of available frags */ 338 long cg_btot[32]; /* block totals per cylinder */ 339 short cg_b[32][8]; /* positions of free blocks */ 340 char cg_iused[256]; /* used inode map */ 341 long cg_magic; /* magic number */ 342 u_char cg_free[1]; /* free block map */ 343 /* actually longer */ 344 }; 345 346 /* 347 * Turn file system block numbers into disk block addresses. 348 * This maps file system blocks to device size blocks. 349 */ 350 #define fsbtodb(fs, b) ((b) << (fs)->fs_fsbtodb) 351 #define dbtofsb(fs, b) ((b) >> (fs)->fs_fsbtodb) 352 353 /* 354 * Cylinder group macros to locate things in cylinder groups. 355 * They calc file system addresses of cylinder group data structures. 356 */ 357 #define cgbase(fs, c) ((daddr_t)((fs)->fs_fpg * (c))) 358 #define cgdmin(fs, c) (cgstart(fs, c) + (fs)->fs_dblkno) /* 1st data */ 359 #define cgimin(fs, c) (cgstart(fs, c) + (fs)->fs_iblkno) /* inode blk */ 360 #define cgsblock(fs, c) (cgstart(fs, c) + (fs)->fs_sblkno) /* super blk */ 361 #define cgtod(fs, c) (cgstart(fs, c) + (fs)->fs_cblkno) /* cg block */ 362 #define cgstart(fs, c) \ 363 (cgbase(fs, c) + (fs)->fs_cgoffset * ((c) & ~((fs)->fs_cgmask))) 364 365 /* 366 * Macros for handling inode numbers: 367 * inode number to file system block offset. 368 * inode number to cylinder group number. 369 * inode number to file system block address. 370 */ 371 #define ino_to_cg(fs, x) ((x) / (fs)->fs_ipg) 372 #define ino_to_fsba(fs, x) \ 373 ((daddr_t)(cgimin(fs, ino_to_cg(fs, x)) + \ 374 (blkstofrags((fs), (((x) % (fs)->fs_ipg) / INOPB(fs)))))) 375 #define ino_to_fsbo(fs, x) ((x) % INOPB(fs)) 376 377 /* 378 * Give cylinder group number for a file system block. 379 * Give cylinder group block number for a file system block. 380 */ 381 #define dtog(fs, d) ((d) / (fs)->fs_fpg) 382 #define dtogd(fs, d) ((d) % (fs)->fs_fpg) 383 384 /* 385 * Extract the bits for a block from a map. 386 * Compute the cylinder and rotational position of a cyl block addr. 387 */ 388 #define blkmap(fs, map, loc) \ 389 (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag))) 390 #define cbtocylno(fs, bno) \ 391 ((bno) * NSPF(fs) / (fs)->fs_spc) 392 #define cbtorpos(fs, bno) \ 393 (((bno) * NSPF(fs) % (fs)->fs_spc / (fs)->fs_nsect * (fs)->fs_trackskew + \ 394 (bno) * NSPF(fs) % (fs)->fs_spc % (fs)->fs_nsect * (fs)->fs_interleave) % \ 395 (fs)->fs_nsect * (fs)->fs_nrpos / (fs)->fs_npsect) 396 397 /* 398 * The following macros optimize certain frequently calculated 399 * quantities by using shifts and masks in place of divisions 400 * modulos and multiplications. 401 */ 402 #define blkoff(fs, loc) /* calculates (loc % fs->fs_bsize) */ \ 403 ((loc) & (fs)->fs_qbmask) 404 #define fragoff(fs, loc) /* calculates (loc % fs->fs_fsize) */ \ 405 ((loc) & (fs)->fs_qfmask) 406 #define lblktosize(fs, blk) /* calculates (blk * fs->fs_bsize) */ \ 407 ((blk) << (fs)->fs_bshift) 408 #define lblkno(fs, loc) /* calculates (loc / fs->fs_bsize) */ \ 409 ((loc) >> (fs)->fs_bshift) 410 #define numfrags(fs, loc) /* calculates (loc / fs->fs_fsize) */ \ 411 ((loc) >> (fs)->fs_fshift) 412 #define blkroundup(fs, size) /* calculates roundup(size, fs->fs_bsize) */ \ 413 (((size) + (fs)->fs_qbmask) & (fs)->fs_bmask) 414 #define fragroundup(fs, size) /* calculates roundup(size, fs->fs_fsize) */ \ 415 (((size) + (fs)->fs_qfmask) & (fs)->fs_fmask) 416 #define fragstoblks(fs, frags) /* calculates (frags / fs->fs_frag) */ \ 417 ((frags) >> (fs)->fs_fragshift) 418 #define blkstofrags(fs, blks) /* calculates (blks * fs->fs_frag) */ \ 419 ((blks) << (fs)->fs_fragshift) 420 #define fragnum(fs, fsb) /* calculates (fsb % fs->fs_frag) */ \ 421 ((fsb) & ((fs)->fs_frag - 1)) 422 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->fs_frag) */ \ 423 ((fsb) &~ ((fs)->fs_frag - 1)) 424 425 /* 426 * Determine the number of available frags given a 427 * percentage to hold in reserve 428 */ 429 #define freespace(fs, percentreserved) \ 430 (blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \ 431 (fs)->fs_cstotal.cs_nffree - ((fs)->fs_dsize * (percentreserved) / 100)) 432 433 /* 434 * Determining the size of a file block in the file system. 435 */ 436 #define blksize(fs, ip, lbn) \ 437 (((lbn) >= NDADDR || (ip)->i_size >= ((lbn) + 1) << (fs)->fs_bshift) \ 438 ? (fs)->fs_bsize \ 439 : (fragroundup(fs, blkoff(fs, (ip)->i_size)))) 440 #define dblksize(fs, dip, lbn) \ 441 (((lbn) >= NDADDR || (dip)->di_size >= ((lbn) + 1) << (fs)->fs_bshift) \ 442 ? (fs)->fs_bsize \ 443 : (fragroundup(fs, blkoff(fs, (dip)->di_size)))) 444 445 /* 446 * Number of disk sectors per block; assumes DEV_BSIZE byte sector size. 447 */ 448 #define NSPB(fs) ((fs)->fs_nspf << (fs)->fs_fragshift) 449 #define NSPF(fs) ((fs)->fs_nspf) 450 451 /* 452 * INOPB is the number of inodes in a secondary storage block. 453 */ 454 #define INOPB(fs) ((fs)->fs_inopb) 455 #define INOPF(fs) ((fs)->fs_inopb >> (fs)->fs_fragshift) 456 457 /* 458 * NINDIR is the number of indirects in a file system block. 459 */ 460 #define NINDIR(fs) ((fs)->fs_nindir) 461 462 extern int inside[], around[]; 463 extern u_char *fragtbl[]; 464