1 /* fs.h 6.3 85/02/08 */ 2 3 /* 4 * Each disk drive contains some number of file systems. 5 * A file system consists of a number of cylinder groups. 6 * Each cylinder group has inodes and data. 7 * 8 * A file system is described by its super-block, which in turn 9 * describes the cylinder groups. The super-block is critical 10 * data and is replicated in each cylinder group to protect against 11 * catastrophic loss. This is done at mkfs time and the critical 12 * super-block data does not change, so the copies need not be 13 * referenced further unless disaster strikes. 14 * 15 * For file system fs, the offsets of the various blocks of interest 16 * are given in the super block as: 17 * [fs->fs_sblkno] Super-block 18 * [fs->fs_cblkno] Cylinder group block 19 * [fs->fs_iblkno] Inode blocks 20 * [fs->fs_dblkno] Data blocks 21 * The beginning of cylinder group cg in fs, is given by 22 * the ``cgbase(fs, cg)'' macro. 23 * 24 * The first boot and super blocks are given in absolute disk addresses. 25 */ 26 #define BBSIZE 8192 27 #define SBSIZE 8192 28 #define BBLOCK ((daddr_t)(0)) 29 #define SBLOCK ((daddr_t)(BBLOCK + BBSIZE / DEV_BSIZE)) 30 31 /* 32 * Addresses stored in inodes are capable of addressing fragments 33 * of `blocks'. File system blocks of at most size MAXBSIZE can 34 * be optionally broken into 2, 4, or 8 pieces, each of which is 35 * addressible; these pieces may be DEV_BSIZE, or some multiple of 36 * a DEV_BSIZE unit. 37 * 38 * Large files consist of exclusively large data blocks. To avoid 39 * undue wasted disk space, the last data block of a small file may be 40 * allocated as only as many fragments of a large block as are 41 * necessary. The file system format retains only a single pointer 42 * to such a fragment, which is a piece of a single large block that 43 * has been divided. The size of such a fragment is determinable from 44 * information in the inode, using the ``blksize(fs, ip, lbn)'' macro. 45 * 46 * The file system records space availability at the fragment level; 47 * to determine block availability, aligned fragments are examined. 48 * 49 * The root inode is the root of the file system. 50 * Inode 0 can't be used for normal purposes and 51 * historically bad blocks were linked to inode 1, 52 * thus the root inode is 2. (inode 1 is no longer used for 53 * this purpose, however numerous dump tapes make this 54 * assumption, so we are stuck with it) 55 * The lost+found directory is given the next available 56 * inode when it is created by ``mkfs''. 57 */ 58 #define ROOTINO ((ino_t)2) /* i number of all roots */ 59 #define LOSTFOUNDINO (ROOTINO + 1) 60 61 /* 62 * Cylinder group related limits. 63 * 64 * For each cylinder we keep track of the availability of blocks at different 65 * rotational positions, so that we can lay out the data to be picked 66 * up with minimum rotational latency. NRPOS is the number of rotational 67 * positions which we distinguish. With NRPOS 8 the resolution of our 68 * summary information is 2ms for a typical 3600 rpm drive. 69 */ 70 #define NRPOS 8 /* number distinct rotational positions */ 71 72 /* 73 * MAXIPG bounds the number of inodes per cylinder group, and 74 * is needed only to keep the structure simpler by having the 75 * only a single variable size element (the free bit map). 76 * 77 * N.B.: MAXIPG must be a multiple of INOPB(fs). 78 */ 79 #define MAXIPG 2048 /* max number inodes/cyl group */ 80 81 /* 82 * MINBSIZE is the smallest allowable block size. 83 * In order to insure that it is possible to create files of size 84 * 2^32 with only two levels of indirection, MINBSIZE is set to 4096. 85 * MINBSIZE must be big enough to hold a cylinder group block, 86 * thus changes to (struct cg) must keep its size within MINBSIZE. 87 * MAXCPG is limited only to dimension an array in (struct cg); 88 * it can be made larger as long as that structures size remains 89 * within the bounds dictated by MINBSIZE. 90 * Note that super blocks are always of size SBSIZE, 91 * and that both SBSIZE and MAXBSIZE must be >= MINBSIZE. 92 */ 93 #define MINBSIZE 4096 94 #define MAXCPG 32 /* maximum fs_cpg */ 95 96 /* 97 * The path name on which the file system is mounted is maintained 98 * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in 99 * the super block for this name. 100 * The limit on the amount of summary information per file system 101 * is defined by MAXCSBUFS. It is currently parameterized for a 102 * maximum of two million cylinders. 103 */ 104 #define MAXMNTLEN 512 105 #define MAXCSBUFS 32 106 107 /* 108 * Per cylinder group information; summarized in blocks allocated 109 * from first cylinder group data blocks. These blocks have to be 110 * read in from fs_csaddr (size fs_cssize) in addition to the 111 * super block. 112 * 113 * N.B. sizeof(struct csum) must be a power of two in order for 114 * the ``fs_cs'' macro to work (see below). 115 */ 116 struct csum { 117 long cs_ndir; /* number of directories */ 118 long cs_nbfree; /* number of free blocks */ 119 long cs_nifree; /* number of free inodes */ 120 long cs_nffree; /* number of free frags */ 121 }; 122 123 /* 124 * Super block for a file system. 125 */ 126 #define FS_MAGIC 0x011954 127 struct fs 128 { 129 struct fs *fs_link; /* linked list of file systems */ 130 struct fs *fs_rlink; /* used for incore super blocks */ 131 daddr_t fs_sblkno; /* addr of super-block in filesys */ 132 daddr_t fs_cblkno; /* offset of cyl-block in filesys */ 133 daddr_t fs_iblkno; /* offset of inode-blocks in filesys */ 134 daddr_t fs_dblkno; /* offset of first data after cg */ 135 long fs_cgoffset; /* cylinder group offset in cylinder */ 136 long fs_cgmask; /* used to calc mod fs_ntrak */ 137 time_t fs_time; /* last time written */ 138 long fs_size; /* number of blocks in fs */ 139 long fs_dsize; /* number of data blocks in fs */ 140 long fs_ncg; /* number of cylinder groups */ 141 long fs_bsize; /* size of basic blocks in fs */ 142 long fs_fsize; /* size of frag blocks in fs */ 143 long fs_frag; /* number of frags in a block in fs */ 144 /* these are configuration parameters */ 145 long fs_minfree; /* minimum percentage of free blocks */ 146 long fs_rotdelay; /* num of ms for optimal next block */ 147 long fs_rps; /* disk revolutions per second */ 148 /* these fields can be computed from the others */ 149 long fs_bmask; /* ``blkoff'' calc of blk offsets */ 150 long fs_fmask; /* ``fragoff'' calc of frag offsets */ 151 long fs_bshift; /* ``lblkno'' calc of logical blkno */ 152 long fs_fshift; /* ``numfrags'' calc number of frags */ 153 /* these are configuration parameters */ 154 long fs_maxcontig; /* max number of contiguous blks */ 155 long fs_maxbpg; /* max number of blks per cyl group */ 156 /* these fields can be computed from the others */ 157 long fs_fragshift; /* block to frag shift */ 158 long fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */ 159 long fs_sbsize; /* actual size of super block */ 160 long fs_csmask; /* csum block offset */ 161 long fs_csshift; /* csum block number */ 162 long fs_nindir; /* value of NINDIR */ 163 long fs_inopb; /* value of INOPB */ 164 long fs_nspf; /* value of NSPF */ 165 long fs_sparecon[6]; /* reserved for future constants */ 166 /* sizes determined by number of cylinder groups and their sizes */ 167 daddr_t fs_csaddr; /* blk addr of cyl grp summary area */ 168 long fs_cssize; /* size of cyl grp summary area */ 169 long fs_cgsize; /* cylinder group size */ 170 /* these fields should be derived from the hardware */ 171 long fs_ntrak; /* tracks per cylinder */ 172 long fs_nsect; /* sectors per track */ 173 long fs_spc; /* sectors per cylinder */ 174 /* this comes from the disk driver partitioning */ 175 long fs_ncyl; /* cylinders in file system */ 176 /* these fields can be computed from the others */ 177 long fs_cpg; /* cylinders per group */ 178 long fs_ipg; /* inodes per group */ 179 long fs_fpg; /* blocks per group * fs_frag */ 180 /* this data must be re-computed after crashes */ 181 struct csum fs_cstotal; /* cylinder summary information */ 182 /* these fields are cleared at mount time */ 183 char fs_fmod; /* super block modified flag */ 184 char fs_clean; /* file system is clean flag */ 185 char fs_ronly; /* mounted read-only flag */ 186 char fs_flags; /* currently unused flag */ 187 char fs_fsmnt[MAXMNTLEN]; /* name mounted on */ 188 /* these fields retain the current block allocation info */ 189 long fs_cgrotor; /* last cg searched */ 190 struct csum *fs_csp[MAXCSBUFS];/* list of fs_cs info buffers */ 191 long fs_cpc; /* cyl per cycle in postbl */ 192 short fs_postbl[MAXCPG][NRPOS];/* head of blocks for each rotation */ 193 long fs_magic; /* magic number */ 194 u_char fs_rotbl[1]; /* list of blocks for each rotation */ 195 /* actually longer */ 196 }; 197 198 /* 199 * Convert cylinder group to base address of its global summary info. 200 * 201 * N.B. This macro assumes that sizeof(struct csum) is a power of two. 202 */ 203 #define fs_cs(fs, indx) \ 204 fs_csp[(indx) >> (fs)->fs_csshift][(indx) & ~(fs)->fs_csmask] 205 206 /* 207 * MAXBPC bounds the size of the rotational layout tables and 208 * is limited by the fact that the super block is of size SBSIZE. 209 * The size of these tables is INVERSELY proportional to the block 210 * size of the file system. It is aggravated by sector sizes that 211 * are not powers of two, as this increases the number of cylinders 212 * included before the rotational pattern repeats (fs_cpc). 213 * Its size is derived from the number of bytes remaining in (struct fs) 214 */ 215 #define MAXBPC (SBSIZE - sizeof (struct fs)) 216 217 /* 218 * Cylinder group block for a file system. 219 */ 220 #define CG_MAGIC 0x090255 221 struct cg { 222 struct cg *cg_link; /* linked list of cyl groups */ 223 struct cg *cg_rlink; /* used for incore cyl groups */ 224 time_t cg_time; /* time last written */ 225 long cg_cgx; /* we are the cgx'th cylinder group */ 226 short cg_ncyl; /* number of cyl's this cg */ 227 short cg_niblk; /* number of inode blocks this cg */ 228 long cg_ndblk; /* number of data blocks this cg */ 229 struct csum cg_cs; /* cylinder summary information */ 230 long cg_rotor; /* position of last used block */ 231 long cg_frotor; /* position of last used frag */ 232 long cg_irotor; /* position of last used inode */ 233 long cg_frsum[MAXFRAG]; /* counts of available frags */ 234 long cg_btot[MAXCPG]; /* block totals per cylinder */ 235 short cg_b[MAXCPG][NRPOS]; /* positions of free blocks */ 236 char cg_iused[MAXIPG/NBBY]; /* used inode map */ 237 long cg_magic; /* magic number */ 238 u_char cg_free[1]; /* free block map */ 239 /* actually longer */ 240 }; 241 242 /* 243 * MAXBPG bounds the number of blocks of data per cylinder group, 244 * and is limited by the fact that cylinder groups are at most one block. 245 * Its size is derived from the size of blocks and the (struct cg) size, 246 * by the number of remaining bits. 247 */ 248 #define MAXBPG(fs) \ 249 (fragstoblks((fs), (NBBY * ((fs)->fs_bsize - (sizeof (struct cg)))))) 250 251 /* 252 * Turn file system block numbers into disk block addresses. 253 * This maps file system blocks to device size blocks. 254 */ 255 #define fsbtodb(fs, b) ((b) << (fs)->fs_fsbtodb) 256 #define dbtofsb(fs, b) ((b) >> (fs)->fs_fsbtodb) 257 258 /* 259 * Cylinder group macros to locate things in cylinder groups. 260 * They calc file system addresses of cylinder group data structures. 261 */ 262 #define cgbase(fs, c) ((daddr_t)((fs)->fs_fpg * (c))) 263 #define cgstart(fs, c) \ 264 (cgbase(fs, c) + (fs)->fs_cgoffset * ((c) & ~((fs)->fs_cgmask))) 265 #define cgsblock(fs, c) (cgstart(fs, c) + (fs)->fs_sblkno) /* super blk */ 266 #define cgtod(fs, c) (cgstart(fs, c) + (fs)->fs_cblkno) /* cg block */ 267 #define cgimin(fs, c) (cgstart(fs, c) + (fs)->fs_iblkno) /* inode blk */ 268 #define cgdmin(fs, c) (cgstart(fs, c) + (fs)->fs_dblkno) /* 1st data */ 269 270 /* 271 * Macros for handling inode numbers: 272 * inode number to file system block offset. 273 * inode number to cylinder group number. 274 * inode number to file system block address. 275 */ 276 #define itoo(fs, x) ((x) % INOPB(fs)) 277 #define itog(fs, x) ((x) / (fs)->fs_ipg) 278 #define itod(fs, x) \ 279 ((daddr_t)(cgimin(fs, itog(fs, x)) + \ 280 (blkstofrags((fs), (((x) % (fs)->fs_ipg) / INOPB(fs)))))) 281 282 /* 283 * Give cylinder group number for a file system block. 284 * Give cylinder group block number for a file system block. 285 */ 286 #define dtog(fs, d) ((d) / (fs)->fs_fpg) 287 #define dtogd(fs, d) ((d) % (fs)->fs_fpg) 288 289 /* 290 * Extract the bits for a block from a map. 291 * Compute the cylinder and rotational position of a cyl block addr. 292 */ 293 #define blkmap(fs, map, loc) \ 294 (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag))) 295 #define cbtocylno(fs, bno) \ 296 ((bno) * NSPF(fs) / (fs)->fs_spc) 297 #define cbtorpos(fs, bno) \ 298 ((bno) * NSPF(fs) % (fs)->fs_nsect * NRPOS / (fs)->fs_nsect) 299 300 /* 301 * The following macros optimize certain frequently calculated 302 * quantities by using shifts and masks in place of divisions 303 * modulos and multiplications. 304 */ 305 #define blkoff(fs, loc) /* calculates (loc % fs->fs_bsize) */ \ 306 ((loc) & ~(fs)->fs_bmask) 307 #define fragoff(fs, loc) /* calculates (loc % fs->fs_fsize) */ \ 308 ((loc) & ~(fs)->fs_fmask) 309 #define lblkno(fs, loc) /* calculates (loc / fs->fs_bsize) */ \ 310 ((loc) >> (fs)->fs_bshift) 311 #define numfrags(fs, loc) /* calculates (loc / fs->fs_fsize) */ \ 312 ((loc) >> (fs)->fs_fshift) 313 #define blkroundup(fs, size) /* calculates roundup(size, fs->fs_bsize) */ \ 314 (((size) + (fs)->fs_bsize - 1) & (fs)->fs_bmask) 315 #define fragroundup(fs, size) /* calculates roundup(size, fs->fs_fsize) */ \ 316 (((size) + (fs)->fs_fsize - 1) & (fs)->fs_fmask) 317 #define fragstoblks(fs, frags) /* calculates (frags / fs->fs_frag) */ \ 318 ((frags) >> (fs)->fs_fragshift) 319 #define blkstofrags(fs, blks) /* calculates (blks * fs->fs_frag) */ \ 320 ((blks) << (fs)->fs_fragshift) 321 #define fragnum(fs, fsb) /* calculates (fsb % fs->fs_frag) */ \ 322 ((fsb) & ((fs)->fs_frag - 1)) 323 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->fs_frag) */ \ 324 ((fsb) &~ ((fs)->fs_frag - 1)) 325 326 /* 327 * Determine the number of available frags given a 328 * percentage to hold in reserve 329 */ 330 #define freespace(fs, percentreserved) \ 331 (blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \ 332 (fs)->fs_cstotal.cs_nffree - ((fs)->fs_dsize * (percentreserved) / 100)) 333 334 /* 335 * Determining the size of a file block in the file system. 336 */ 337 #define blksize(fs, ip, lbn) \ 338 (((lbn) >= NDADDR || (ip)->i_size >= ((lbn) + 1) << (fs)->fs_bshift) \ 339 ? (fs)->fs_bsize \ 340 : (fragroundup(fs, blkoff(fs, (ip)->i_size)))) 341 #define dblksize(fs, dip, lbn) \ 342 (((lbn) >= NDADDR || (dip)->di_size >= ((lbn) + 1) << (fs)->fs_bshift) \ 343 ? (fs)->fs_bsize \ 344 : (fragroundup(fs, blkoff(fs, (dip)->di_size)))) 345 346 /* 347 * Number of disk sectors per block; assumes DEV_BSIZE byte sector size. 348 */ 349 #define NSPB(fs) ((fs)->fs_nspf << (fs)->fs_fragshift) 350 #define NSPF(fs) ((fs)->fs_nspf) 351 352 /* 353 * INOPB is the number of inodes in a secondary storage block. 354 */ 355 #define INOPB(fs) ((fs)->fs_inopb) 356 #define INOPF(fs) ((fs)->fs_inopb >> (fs)->fs_fragshift) 357 358 /* 359 * NINDIR is the number of indirects in a file system block. 360 */ 361 #define NINDIR(fs) ((fs)->fs_nindir) 362 363 #ifdef KERNEL 364 struct fs *getfs(); 365 struct fs *mountfs(); 366 #endif 367