1.\" Copyright (c) 1983, 1991, 1993 2.\" The Regents of the University of California. All rights reserved. 3.\" 4.\" Redistribution and use in source and binary forms, with or without 5.\" modification, are permitted provided that the following conditions 6.\" are met: 7.\" 1. Redistributions of source code must retain the above copyright 8.\" notice, this list of conditions and the following disclaimer. 9.\" 2. Redistributions in binary form must reproduce the above copyright 10.\" notice, this list of conditions and the following disclaimer in the 11.\" documentation and/or other materials provided with the distribution. 12.\" 3. All advertising materials mentioning features or use of this software 13.\" must display the following acknowledgement: 14.\" This product includes software developed by the University of 15.\" California, Berkeley and its contributors. 16.\" 4. Neither the name of the University nor the names of its contributors 17.\" may be used to endorse or promote products derived from this software 18.\" without specific prior written permission. 19.\" 20.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 21.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23.\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 24.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30.\" SUCH DAMAGE. 31.\" 32.\" @(#)fs.5 8.2 (Berkeley) 4/19/94 33.\" 34.Dd April 19, 1994 35.Dt FS 5 36.Os BSD 4.2 37.Sh NAME 38.Nm fs , 39.Nm inode 40.Nd format of file system volume 41.Sh SYNOPSIS 42.Fd #include <sys/types.h> 43.Fd #include <ufs/ffs/fs.h> 44.Fd #include <ufs/ufs/inode.h> 45.Sh DESCRIPTION 46The files 47.Aq Pa fs.h 48and 49.Aq Pa inode.h 50declare several structures, defined variables and macros 51which are used to create and manage the underlying format of 52file system objects on random access devices (disks). 53.Pp 54The block size and number of blocks which 55comprise a file system are parameters of the file system. 56Sectors beginning at 57.Dv BBLOCK 58and continuing for 59.Dv BBSIZE 60are used 61for a disklabel and for some hardware primary 62and secondary bootstrapping programs. 63.Pp 64The actual file system begins at sector 65.Dv SBLOCK 66with the 67.Em super-block 68that is of size 69.Dv SBSIZE . 70The following structure described the super-block and is 71from the file 72.Aq Pa ufs/ffs/fs.h : 73.Bd -literal 74#define FS_MAGIC 0x011954 75struct fs { 76 struct fs *fs_link; /* linked list of file systems */ 77 struct fs *fs_rlink; /* used for incore super blocks */ 78 daddr_t fs_sblkno; /* addr of super-block in filesys */ 79 daddr_t fs_cblkno; /* offset of cyl-block in filesys */ 80 daddr_t fs_iblkno; /* offset of inode-blocks in filesys */ 81 daddr_t fs_dblkno; /* offset of first data after cg */ 82 long fs_cgoffset; /* cylinder group offset in cylinder */ 83 long fs_cgmask; /* used to calc mod fs_ntrak */ 84 time_t fs_time; /* last time written */ 85 long fs_size; /* number of blocks in fs */ 86 long fs_dsize; /* number of data blocks in fs */ 87 long fs_ncg; /* number of cylinder groups */ 88 long fs_bsize; /* size of basic blocks in fs */ 89 long fs_fsize; /* size of frag blocks in fs */ 90 long fs_frag; /* number of frags in a block in fs */ 91/* these are configuration parameters */ 92 long fs_minfree; /* minimum percentage of free blocks */ 93 long fs_rotdelay; /* num of ms for optimal next block */ 94 long fs_rps; /* disk revolutions per second */ 95/* these fields can be computed from the others */ 96 long fs_bmask; /* ``blkoff'' calc of blk offsets */ 97 long fs_fmask; /* ``fragoff'' calc of frag offsets */ 98 long fs_bshift; /* ``lblkno'' calc of logical blkno */ 99 long fs_fshift; /* ``numfrags'' calc number of frags */ 100/* these are configuration parameters */ 101 long fs_maxcontig; /* max number of contiguous blks */ 102 long fs_maxbpg; /* max number of blks per cyl group */ 103/* these fields can be computed from the others */ 104 long fs_fragshift; /* block to frag shift */ 105 long fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */ 106 long fs_sbsize; /* actual size of super block */ 107 long fs_csmask; /* csum block offset */ 108 long fs_csshift; /* csum block number */ 109 long fs_nindir; /* value of NINDIR */ 110 long fs_inopb; /* value of INOPB */ 111 long fs_nspf; /* value of NSPF */ 112/* yet another configuration parameter */ 113 long fs_optim; /* optimization preference, see below */ 114/* these fields are derived from the hardware */ 115 long fs_npsect; /* # sectors/track including spares */ 116 long fs_interleave; /* hardware sector interleave */ 117 long fs_trackskew; /* sector 0 skew, per track */ 118 long fs_headswitch; /* head switch time, usec */ 119 long fs_trkseek; /* track-to-track seek, usec */ 120/* sizes determined by number of cylinder groups and their sizes */ 121 daddr_t fs_csaddr; /* blk addr of cyl grp summary area */ 122 long fs_cssize; /* size of cyl grp summary area */ 123 long fs_cgsize; /* cylinder group size */ 124/* these fields are derived from the hardware */ 125 long fs_ntrak; /* tracks per cylinder */ 126 long fs_nsect; /* sectors per track */ 127 long fs_spc; /* sectors per cylinder */ 128/* this comes from the disk driver partitioning */ 129 long fs_ncyl; /* cylinders in file system */ 130/* these fields can be computed from the others */ 131 long fs_cpg; /* cylinders per group */ 132 long fs_ipg; /* inodes per group */ 133 long fs_fpg; /* blocks per group * fs_frag */ 134/* this data must be re-computed after crashes */ 135 struct csum fs_cstotal; /* cylinder summary information */ 136/* these fields are cleared at mount time */ 137 char fs_fmod; /* super block modified flag */ 138 char fs_clean; /* file system is clean flag */ 139 char fs_ronly; /* mounted read-only flag */ 140 char fs_flags; /* currently unused flag */ 141 char fs_fsmnt[MAXMNTLEN]; /* name mounted on */ 142/* these fields retain the current block allocation info */ 143 long fs_cgrotor; /* last cg searched */ 144 struct csum *fs_csp[MAXCSBUFS]; /* list of fs_cs info buffers */ 145 long fs_cpc; /* cyl per cycle in postbl */ 146 short fs_opostbl[16][8]; /* old rotation block list head */ 147 long fs_sparecon[56]; /* reserved for future constants */ 148 quad fs_qbmask; /* ~fs_bmask - for use with quad size */ 149 quad fs_qfmask; /* ~fs_fmask - for use with quad size */ 150 long fs_postblformat; /* format of positional layout tables */ 151 long fs_nrpos; /* number of rotational positions */ 152 long fs_postbloff; /* (short) rotation block list head */ 153 long fs_rotbloff; /* (u_char) blocks for each rotation */ 154 long fs_magic; /* magic number */ 155 u_char fs_space[1]; /* list of blocks for each rotation */ 156/* actually longer */ 157}; 158.Ed 159.Pp 160Each disk drive contains some number of file systems. 161A file system consists of a number of cylinder groups. 162Each cylinder group has inodes and data. 163.Pp 164A file system is described by its super-block, which in turn 165describes the cylinder groups. The super-block is critical 166data and is replicated in each cylinder group to protect against 167catastrophic loss. This is done at file system creation 168time and the critical 169super-block data does not change, so the copies need not be 170referenced further unless disaster strikes. 171.Pp 172Addresses stored in inodes are capable of addressing fragments 173of `blocks'. File system blocks of at most size 174.Dv MAXBSIZE 175can 176be optionally broken into 2, 4, or 8 pieces, each of which is 177addressable; these pieces may be 178.Dv DEV_BSIZE , 179or some multiple of 180a 181.Dv DEV_BSIZE 182unit. 183.Pp 184Large files consist of exclusively large data blocks. To avoid 185undue wasted disk space, the last data block of a small file is 186allocated as only as many fragments of a large block as are 187necessary. The file system format retains only a single pointer 188to such a fragment, which is a piece of a single large block that 189has been divided. The size of such a fragment is determinable from 190information in the inode, using the 191.Fn blksize fs ip lbn 192macro. 193.Pp 194The file system records space availability at the fragment level; 195to determine block availability, aligned fragments are examined. 196.Pp 197The root inode is the root of the file system. 198Inode 0 can't be used for normal purposes and 199historically bad blocks were linked to inode 1, 200thus the root inode is 2 (inode 1 is no longer used for 201this purpose, however numerous dump tapes make this 202assumption, so we are stuck with it). 203.Pp 204The 205.Fa fs_minfree 206element gives the minimum acceptable percentage of file system 207blocks that may be free. If the freelist drops below this level 208only the super-user may continue to allocate blocks. 209The 210.Fa fs_minfree 211element 212may be set to 0 if no reserve of free blocks is deemed necessary, 213however severe performance degradations will be observed if the 214file system is run at greater than 90% full; thus the default 215value of 216.Fa fs_minfree 217is 10%. 218.Pp 219Empirically the best trade-off between block fragmentation and 220overall disk utilization at a loading of 90% comes with a 221fragmentation of 8, thus the default fragment size is an eighth 222of the block size. 223.Pp 224The element 225.Fa fs_optim 226specifies whether the file system should try to minimize the time spent 227allocating blocks, or if it should attempt to minimize the space 228fragmentation on the disk. 229If the value of fs_minfree (see above) is less than 10%, 230then the file system defaults to optimizing for space to avoid 231running out of full sized blocks. 232If the value of minfree is greater than or equal to 10%, 233fragmentation is unlikely to be problematical, and 234the file system defaults to optimizing for time. 235.Pp 236.Em Cylinder group related limits : 237Each cylinder keeps track of the availability of blocks at different 238rotational positions, so that sequential blocks can be laid out 239with minimum rotational latency. With the default of 8 distinguished 240rotational positions, the resolution of the 241summary information is 2ms for a typical 3600 rpm drive. 242.Pp 243The element 244.Fa fs_rotdelay 245gives the minimum number of milliseconds to initiate 246another disk transfer on the same cylinder. 247It is used in determining the rotationally optimal 248layout for disk blocks within a file; 249the default value for 250.Fa fs_rotdelay 251is 2ms. 252.Pp 253Each file system has a statically allocated number of inodes. 254An inode is allocated for each 255.Dv NBPI 256bytes of disk space. 257The inode allocation strategy is extremely conservative. 258.Pp 259.Dv MINBSIZE 260is the smallest allowable block size. 261With a 262.Dv MINBSIZE 263of 4096 264it is possible to create files of size 2652^32 with only two levels of indirection. 266.Dv MINBSIZE 267must be big enough to hold a cylinder group block, 268thus changes to 269.Pq Fa struct cg 270must keep its size within 271.Dv MINBSIZE . 272Note that super-blocks are never more than size 273.Dv SBSIZE . 274.Pp 275The path name on which the file system is mounted is maintained in 276.Fa fs_fsmnt . 277.Dv MAXMNTLEN 278defines the amount of space allocated in 279the super-block for this name. 280The limit on the amount of summary information per file system 281is defined by 282.Dv MAXCSBUFS. 283For a 4096 byte block size, it is currently parameterized for a 284maximum of two million cylinders. 285.Pp 286Per cylinder group information is summarized in blocks allocated 287from the first cylinder group's data blocks. 288These blocks are read in from 289.Fa fs_csaddr 290(size 291.Fa fs_cssize ) 292in addition to the super-block. 293.Pp 294.Sy N.B.: 295.Fn sizeof "struct csum" 296must be a power of two in order for 297the 298.Fn fs_cs 299macro to work. 300.Pp 301The 302.Em "Super-block for a file system" : 303The size of the rotational layout tables 304is limited by the fact that the super-block is of size 305.Dv SBSIZE . 306The size of these tables is 307.Em inversely 308proportional to the block 309size of the file system. The size of the tables is 310increased when sector sizes are not powers of two, 311as this increases the number of cylinders 312included before the rotational pattern repeats 313.Pq Fa fs_cpc . 314The size of the rotational layout 315tables is derived from the number of bytes remaining in 316.Pq Fa struct fs . 317.Pp 318The number of blocks of data per cylinder group 319is limited because cylinder groups are at most one block. 320The inode and free block tables 321must fit into a single block after deducting space for 322the cylinder group structure 323.Pq Fa struct cg . 324.Pp 325The 326.Em Inode : 327The inode is the focus of all file activity in the 328.Tn UNIX 329file system. 330There is a unique inode allocated 331for each active file, 332each current directory, each mounted-on file, 333text file, and the root. 334An inode is `named' by its device/i-number pair. 335For further information, see the include file 336.Aq Pa sys/inode.h . 337.Sh HISTORY 338A super-block structure named filsys appeared in 339.At v6 . 340The file system described in this manual appeared 341in 342.Bx 4.2 . 343