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.\" $FreeBSD$ 34.\" 35.Dd October 31, 2006 36.Dt FS 5 37.Os 38.Sh NAME 39.Nm fs , 40.Nm inode 41.Nd format of file system volume 42.Sh SYNOPSIS 43.In sys/param.h 44.In ufs/ffs/fs.h 45.Pp 46.In sys/types.h 47.In sys/lock.h 48.In sys/extattr.h 49.In sys/acl.h 50.In ufs/ufs/quota.h 51.In ufs/ufs/dinode.h 52.In ufs/ufs/extattr.h 53.Sh DESCRIPTION 54The files 55.In fs.h 56and 57.In inode.h 58declare several structures, defined variables and macros 59which are used to create and manage the underlying format of 60file system objects on random access devices (disks). 61.Pp 62The block size and number of blocks which 63comprise a file system are parameters of the file system. 64Sectors beginning at 65.Dv BBLOCK 66and continuing for 67.Dv BBSIZE 68are used 69for a disklabel and for some hardware primary 70and secondary bootstrapping programs. 71.Pp 72The actual file system begins at sector 73.Dv SBLOCK 74with the 75.Em super-block 76that is of size 77.Dv SBLOCKSIZE . 78The following structure describes the super-block and is 79from the file 80.In ufs/ffs/fs.h : 81.Bd -literal 82/* 83 * Super block for an FFS filesystem. 84 */ 85struct fs { 86 int32_t fs_firstfield; /* historic filesystem linked list, */ 87 int32_t fs_unused_1; /* used for incore super blocks */ 88 int32_t fs_sblkno; /* offset of super-block in filesys */ 89 int32_t fs_cblkno; /* offset of cyl-block in filesys */ 90 int32_t fs_iblkno; /* offset of inode-blocks in filesys */ 91 int32_t fs_dblkno; /* offset of first data after cg */ 92 int32_t fs_old_cgoffset; /* cylinder group offset in cylinder */ 93 int32_t fs_old_cgmask; /* used to calc mod fs_ntrak */ 94 int32_t fs_old_time; /* last time written */ 95 int32_t fs_old_size; /* number of blocks in fs */ 96 int32_t fs_old_dsize; /* number of data blocks in fs */ 97 int32_t fs_ncg; /* number of cylinder groups */ 98 int32_t fs_bsize; /* size of basic blocks in fs */ 99 int32_t fs_fsize; /* size of frag blocks in fs */ 100 int32_t fs_frag; /* number of frags in a block in fs */ 101/* these are configuration parameters */ 102 int32_t fs_minfree; /* minimum percentage of free blocks */ 103 int32_t fs_old_rotdelay; /* num of ms for optimal next block */ 104 int32_t fs_old_rps; /* disk revolutions per second */ 105/* these fields can be computed from the others */ 106 int32_t fs_bmask; /* ``blkoff'' calc of blk offsets */ 107 int32_t fs_fmask; /* ``fragoff'' calc of frag offsets */ 108 int32_t fs_bshift; /* ``lblkno'' calc of logical blkno */ 109 int32_t fs_fshift; /* ``numfrags'' calc number of frags */ 110/* these are configuration parameters */ 111 int32_t fs_maxcontig; /* max number of contiguous blks */ 112 int32_t fs_maxbpg; /* max number of blks per cyl group */ 113/* these fields can be computed from the others */ 114 int32_t fs_fragshift; /* block to frag shift */ 115 int32_t fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */ 116 int32_t fs_sbsize; /* actual size of super block */ 117 int32_t fs_spare1[2]; /* old fs_csmask */ 118 /* old fs_csshift */ 119 int32_t fs_nindir; /* value of NINDIR */ 120 int32_t fs_inopb; /* value of INOPB */ 121 int32_t fs_old_nspf; /* value of NSPF */ 122/* yet another configuration parameter */ 123 int32_t fs_optim; /* optimization preference, see below */ 124 int32_t fs_old_npsect; /* # sectors/track including spares */ 125 int32_t fs_old_interleave; /* hardware sector interleave */ 126 int32_t fs_old_trackskew; /* sector 0 skew, per track */ 127 int32_t fs_id[2]; /* unique filesystem id */ 128/* sizes determined by number of cylinder groups and their sizes */ 129 int32_t fs_old_csaddr; /* blk addr of cyl grp summary area */ 130 int32_t fs_cssize; /* size of cyl grp summary area */ 131 int32_t fs_cgsize; /* cylinder group size */ 132 int32_t fs_spare2; /* old fs_ntrak */ 133 int32_t fs_old_nsect; /* sectors per track */ 134 int32_t fs_old_spc; /* sectors per cylinder */ 135 int32_t fs_old_ncyl; /* cylinders in filesystem */ 136 int32_t fs_old_cpg; /* cylinders per group */ 137 int32_t fs_ipg; /* inodes per group */ 138 int32_t fs_fpg; /* blocks per group * fs_frag */ 139/* this data must be re-computed after crashes */ 140 struct csum fs_old_cstotal; /* cylinder summary information */ 141/* these fields are cleared at mount time */ 142 int8_t fs_fmod; /* super block modified flag */ 143 int8_t fs_clean; /* filesystem is clean flag */ 144 int8_t fs_ronly; /* mounted read-only flag */ 145 int8_t fs_old_flags; /* old FS_ flags */ 146 u_char fs_fsmnt[MAXMNTLEN]; /* name mounted on */ 147 u_char fs_volname[MAXVOLLEN]; /* volume name */ 148 uint64_t fs_swuid; /* system-wide uid */ 149 int32_t fs_pad; /* due to alignment of fs_swuid */ 150/* these fields retain the current block allocation info */ 151 int32_t fs_cgrotor; /* last cg searched */ 152 void *fs_ocsp[NOCSPTRS]; /* padding; was list of fs_cs buffers */ 153 uint8_t *fs_contigdirs; /* # of contiguously allocated dirs */ 154 struct csum *fs_csp; /* cg summary info buffer for fs_cs */ 155 int32_t *fs_maxcluster; /* max cluster in each cyl group */ 156 u_int *fs_active; /* used by snapshots to track fs */ 157 int32_t fs_old_cpc; /* cyl per cycle in postbl */ 158 int32_t fs_maxbsize; /* maximum blocking factor permitted */ 159 int64_t fs_unrefs; /* number of unreferenced inodes */ 160 int64_t fs_sparecon64[16]; /* old rotation block list head */ 161 int64_t fs_sblockloc; /* byte offset of standard superblock */ 162 struct csum_total fs_cstotal; /* cylinder summary information */ 163 ufs_time_t fs_time; /* last time written */ 164 int64_t fs_size; /* number of blocks in fs */ 165 int64_t fs_dsize; /* number of data blocks in fs */ 166 ufs2_daddr_t fs_csaddr; /* blk addr of cyl grp summary area */ 167 int64_t fs_pendingblocks; /* blocks in process of being freed */ 168 int32_t fs_pendinginodes; /* inodes in process of being freed */ 169 int32_t fs_snapinum[FSMAXSNAP]; /* list of snapshot inode numbers */ 170 int32_t fs_avgfilesize; /* expected average file size */ 171 int32_t fs_avgfpdir; /* expected # of files per directory */ 172 int32_t fs_save_cgsize; /* save real cg size to use fs_bsize */ 173 int32_t fs_sparecon32[26]; /* reserved for future constants */ 174 int32_t fs_flags; /* see FS_ flags below */ 175 int32_t fs_contigsumsize; /* size of cluster summary array */ 176 int32_t fs_maxsymlinklen; /* max length of an internal symlink */ 177 int32_t fs_old_inodefmt; /* format of on-disk inodes */ 178 uint64_t fs_maxfilesize; /* maximum representable file size */ 179 int64_t fs_qbmask; /* ~fs_bmask for use with 64-bit size */ 180 int64_t fs_qfmask; /* ~fs_fmask for use with 64-bit size */ 181 int32_t fs_state; /* validate fs_clean field */ 182 int32_t fs_old_postblformat; /* format of positional layout tables */ 183 int32_t fs_old_nrpos; /* number of rotational positions */ 184 int32_t fs_spare5[2]; /* old fs_postbloff */ 185 /* old fs_rotbloff */ 186 int32_t fs_magic; /* magic number */ 187}; 188 189/* 190 * Filesystem identification 191 */ 192#define FS_UFS1_MAGIC 0x011954 /* UFS1 fast filesystem magic number */ 193#define FS_UFS2_MAGIC 0x19540119 /* UFS2 fast filesystem magic number */ 194#define FS_OKAY 0x7c269d38 /* superblock checksum */ 195#define FS_42INODEFMT -1 /* 4.2BSD inode format */ 196#define FS_44INODEFMT 2 /* 4.4BSD inode format */ 197 198/* 199 * Preference for optimization. 200 */ 201#define FS_OPTTIME 0 /* minimize allocation time */ 202#define FS_OPTSPACE 1 /* minimize disk fragmentation */ 203.Ed 204.Pp 205Each disk drive contains some number of file systems. 206A file system consists of a number of cylinder groups. 207Each cylinder group has inodes and data. 208.Pp 209A file system is described by its super-block, which in turn 210describes the cylinder groups. 211The super-block is critical 212data and is replicated in each cylinder group to protect against 213catastrophic loss. 214This is done at file system creation 215time and the critical 216super-block data does not change, so the copies need not be 217referenced further unless disaster strikes. 218.Pp 219Addresses stored in inodes are capable of addressing fragments 220of `blocks'. 221File system blocks of at most size 222.Dv MAXBSIZE 223can 224be optionally broken into 2, 4, or 8 pieces, each of which is 225addressable; these pieces may be 226.Dv DEV_BSIZE , 227or some multiple of 228a 229.Dv DEV_BSIZE 230unit. 231.Pp 232Large files consist of exclusively large data blocks. 233To avoid 234undue wasted disk space, the last data block of a small file is 235allocated as only as many fragments of a large block as are 236necessary. 237The file system format retains only a single pointer 238to such a fragment, which is a piece of a single large block that 239has been divided. 240The size of such a fragment is determinable from 241information in the inode, using the 242.Fn blksize fs ip lbn 243macro. 244.Pp 245The file system records space availability at the fragment level; 246to determine block availability, aligned fragments are examined. 247.Pp 248The root inode is the root of the file system. 249Inode 0 cannot be used for normal purposes and 250historically bad blocks were linked to inode 1, 251thus the root inode is 2 (inode 1 is no longer used for 252this purpose, however numerous dump tapes make this 253assumption, so we are stuck with it). 254.Pp 255The 256.Fa fs_minfree 257element gives the minimum acceptable percentage of file system 258blocks that may be free. 259If the freelist drops below this level 260only the super-user may continue to allocate blocks. 261The 262.Fa fs_minfree 263element 264may be set to 0 if no reserve of free blocks is deemed necessary, 265however severe performance degradations will be observed if the 266file system is run at greater than 90% full; thus the default 267value of 268.Fa fs_minfree 269is 10%. 270.Pp 271Empirically the best trade-off between block fragmentation and 272overall disk utilization at a loading of 90% comes with a 273fragmentation of 8, thus the default fragment size is an eighth 274of the block size. 275.Pp 276The element 277.Fa fs_optim 278specifies whether the file system should try to minimize the time spent 279allocating blocks, or if it should attempt to minimize the space 280fragmentation on the disk. 281If the value of fs_minfree (see above) is less than 10%, 282then the file system defaults to optimizing for space to avoid 283running out of full sized blocks. 284If the value of minfree is greater than or equal to 10%, 285fragmentation is unlikely to be problematical, and 286the file system defaults to optimizing for time. 287.Pp 288.Em Cylinder group related limits : 289Each cylinder keeps track of the availability of blocks at different 290rotational positions, so that sequential blocks can be laid out 291with minimum rotational latency. 292With the default of 8 distinguished 293rotational positions, the resolution of the 294summary information is 2ms for a typical 3600 rpm drive. 295.Pp 296The element 297.Fa fs_old_rotdelay 298gives the minimum number of milliseconds to initiate 299another disk transfer on the same cylinder. 300It is used in determining the rotationally optimal 301layout for disk blocks within a file; 302the default value for 303.Fa fs_old_rotdelay 304is 2ms. 305.Pp 306Each file system has a statically allocated number of inodes. 307An inode is allocated for each 308.Dv NBPI 309bytes of disk space. 310The inode allocation strategy is extremely conservative. 311.Pp 312.Dv MINBSIZE 313is the smallest allowable block size. 314With a 315.Dv MINBSIZE 316of 4096 317it is possible to create files of size 3182^32 with only two levels of indirection. 319.Dv MINBSIZE 320must be big enough to hold a cylinder group block, 321thus changes to 322.Pq Fa struct cg 323must keep its size within 324.Dv MINBSIZE . 325Note that super-blocks are never more than size 326.Dv SBLOCKSIZE . 327.Pp 328The path name on which the file system is mounted is maintained in 329.Fa fs_fsmnt . 330.Dv MAXMNTLEN 331defines the amount of space allocated in 332the super-block for this name. 333The limit on the amount of summary information per file system 334is defined by 335.Dv MAXCSBUFS . 336For a 4096 byte block size, it is currently parameterized for a 337maximum of two million cylinders. 338.Pp 339Per cylinder group information is summarized in blocks allocated 340from the first cylinder group's data blocks. 341These blocks are read in from 342.Fa fs_csaddr 343(size 344.Fa fs_cssize ) 345in addition to the super-block. 346.Pp 347.Sy N.B. : 348.Fn sizeof "struct csum" 349must be a power of two in order for 350the 351.Fn fs_cs 352macro to work. 353.Pp 354The 355.Em "Super-block for a file system" : 356The size of the rotational layout tables 357is limited by the fact that the super-block is of size 358.Dv SBLOCKSIZE . 359The size of these tables is 360.Em inversely 361proportional to the block 362size of the file system. 363The size of the tables is 364increased when sector sizes are not powers of two, 365as this increases the number of cylinders 366included before the rotational pattern repeats 367.Pq Fa fs_cpc . 368The size of the rotational layout 369tables is derived from the number of bytes remaining in 370.Pq Fa struct fs . 371.Pp 372The number of blocks of data per cylinder group 373is limited because cylinder groups are at most one block. 374The inode and free block tables 375must fit into a single block after deducting space for 376the cylinder group structure 377.Pq Fa struct cg . 378.Pp 379The 380.Em Inode : 381The inode is the focus of all file activity in the 382.Ux 383file system. 384There is a unique inode allocated 385for each active file, 386each current directory, each mounted-on file, 387text file, and the root. 388An inode is `named' by its device/i-number pair. 389For further information, see the include file 390.In ufs/ufs/inode.h . 391.Pp 392The format of an external attribute is defined by the extattr structure: 393.Bd -literal 394struct extattr { 395 int32_t ea_length; /* length of this attribute */ 396 int8_t ea_namespace; /* name space of this attribute */ 397 int8_t ea_contentpadlen; /* padding at end of attribute */ 398 int8_t ea_namelength; /* length of attribute name */ 399 char ea_name[1]; /* null-terminated attribute name */ 400 /* extended attribute content follows */ 401}; 402.Ed 403.Pp 404Several macros are defined to manipulate these structures. 405Each macro takes a pointer to an extattr structure. 406.Bl -tag -width ".Dv EXTATTR_SET_LENGTHS(eap, size)" 407.It Dv EXTATTR_NEXT(eap) 408Returns a pointer to the next extended attribute following 409.Fa eap . 410.It Dv EXTATTR_CONTENT(eap) 411Returns a pointer to the extended attribute content referenced by 412.Fa eap . 413.It Dv EXTATTR_CONTENT_SIZE(eap) 414Returns the size of the extended attribute content referenced by 415.Fa eap . 416.It Dv EXTATTR_SET_LENGTHS(eap, size) 417Called with the size of the attribute content after initializing 418the attribute name to calculate and set the 419.Fa ea_length , 420.Fa ea_namelength , 421and 422.Fa ea_contentpadlen 423fields of the extended attribute structure. 424.El 425.Pp 426The following code identifies an ACL: 427.Bd -literal 428 if (eap->ea_namespace == EXTATTR_NAMESPACE_SYSTEM && 429 !strcmp(eap->ea_name, POSIX1E_ACL_ACCESS_EXTATTR_NAME) { 430 aclp = EXTATTR_CONTENT(eap); 431 acllen = EXTATTR_CONTENT_SIZE(eap); 432 ... 433 } 434.Ed 435.Pp 436The following code creates an extended attribute 437containing a copy of a structure 438.Fa mygif : 439.Bd -literal 440 eap->ea_namespace = EXTATTR_NAMESPACE_USER; 441 strcpy(eap->ea_name, "filepic.gif"); 442 EXTATTR_SET_LENGTHS(eap, sizeof(struct mygif)); 443 memcpy(EXTATTR_CONTENT(eap), &mygif, sizeof(struct mygif)); 444.Ed 445.Sh HISTORY 446A super-block structure named filsys appeared in 447.At v6 . 448The file system described in this manual appeared 449in 450.Bx 4.2 . 451
