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. Neither the name of the University nor the names of its contributors 13.\" may be used to endorse or promote products derived from this software 14.\" without specific prior written permission. 15.\" 16.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 17.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 18.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 19.\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 20.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 21.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 22.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 23.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 24.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26.\" SUCH DAMAGE. 27.\" 28.\" @(#)fs.5 8.2 (Berkeley) 4/19/94 29.\" $FreeBSD$ 30.\" 31.Dd October 31, 2006 32.Dt FS 5 33.Os 34.Sh NAME 35.Nm fs , 36.Nm inode 37.Nd format of file system volume 38.Sh SYNOPSIS 39.In sys/param.h 40.In ufs/ffs/fs.h 41.Pp 42.In sys/types.h 43.In sys/lock.h 44.In sys/extattr.h 45.In sys/acl.h 46.In ufs/ufs/quota.h 47.In ufs/ufs/dinode.h 48.In ufs/ufs/extattr.h 49.Sh DESCRIPTION 50The files 51.In fs.h 52and 53.In inode.h 54declare several structures, defined variables and macros 55which are used to create and manage the underlying format of 56file system objects on random access devices (disks). 57.Pp 58The block size and number of blocks which 59comprise a file system are parameters of the file system. 60Sectors beginning at 61.Dv BBLOCK 62and continuing for 63.Dv BBSIZE 64are used 65for a disklabel and for some hardware primary 66and secondary bootstrapping programs. 67.Pp 68The actual file system begins at sector 69.Dv SBLOCK 70with the 71.Em super-block 72that is of size 73.Dv SBLOCKSIZE . 74The following structure describes the super-block and is 75from the file 76.In ufs/ffs/fs.h : 77.Bd -literal 78/* 79 * Super block for an FFS filesystem. 80 */ 81struct fs { 82 int32_t fs_firstfield; /* historic filesystem linked list, */ 83 int32_t fs_unused_1; /* used for incore super blocks */ 84 int32_t fs_sblkno; /* offset of super-block in filesys */ 85 int32_t fs_cblkno; /* offset of cyl-block in filesys */ 86 int32_t fs_iblkno; /* offset of inode-blocks in filesys */ 87 int32_t fs_dblkno; /* offset of first data after cg */ 88 int32_t fs_old_cgoffset; /* cylinder group offset in cylinder */ 89 int32_t fs_old_cgmask; /* used to calc mod fs_ntrak */ 90 int32_t fs_old_time; /* last time written */ 91 int32_t fs_old_size; /* number of blocks in fs */ 92 int32_t fs_old_dsize; /* number of data blocks in fs */ 93 int32_t fs_ncg; /* number of cylinder groups */ 94 int32_t fs_bsize; /* size of basic blocks in fs */ 95 int32_t fs_fsize; /* size of frag blocks in fs */ 96 int32_t fs_frag; /* number of frags in a block in fs */ 97/* these are configuration parameters */ 98 int32_t fs_minfree; /* minimum percentage of free blocks */ 99 int32_t fs_old_rotdelay; /* num of ms for optimal next block */ 100 int32_t fs_old_rps; /* disk revolutions per second */ 101/* these fields can be computed from the others */ 102 int32_t fs_bmask; /* ``blkoff'' calc of blk offsets */ 103 int32_t fs_fmask; /* ``fragoff'' calc of frag offsets */ 104 int32_t fs_bshift; /* ``lblkno'' calc of logical blkno */ 105 int32_t fs_fshift; /* ``numfrags'' calc number of frags */ 106/* these are configuration parameters */ 107 int32_t fs_maxcontig; /* max number of contiguous blks */ 108 int32_t fs_maxbpg; /* max number of blks per cyl group */ 109/* these fields can be computed from the others */ 110 int32_t fs_fragshift; /* block to frag shift */ 111 int32_t fs_fsbtodb; /* fsbtodb and dbtofsb shift constant */ 112 int32_t fs_sbsize; /* actual size of super block */ 113 int32_t fs_spare1[2]; /* old fs_csmask */ 114 /* old fs_csshift */ 115 int32_t fs_nindir; /* value of NINDIR */ 116 int32_t fs_inopb; /* value of INOPB */ 117 int32_t fs_old_nspf; /* value of NSPF */ 118/* yet another configuration parameter */ 119 int32_t fs_optim; /* optimization preference, see below */ 120 int32_t fs_old_npsect; /* # sectors/track including spares */ 121 int32_t fs_old_interleave; /* hardware sector interleave */ 122 int32_t fs_old_trackskew; /* sector 0 skew, per track */ 123 int32_t fs_id[2]; /* unique filesystem id */ 124/* sizes determined by number of cylinder groups and their sizes */ 125 int32_t fs_old_csaddr; /* blk addr of cyl grp summary area */ 126 int32_t fs_cssize; /* size of cyl grp summary area */ 127 int32_t fs_cgsize; /* cylinder group size */ 128 int32_t fs_spare2; /* old fs_ntrak */ 129 int32_t fs_old_nsect; /* sectors per track */ 130 int32_t fs_old_spc; /* sectors per cylinder */ 131 int32_t fs_old_ncyl; /* cylinders in filesystem */ 132 int32_t fs_old_cpg; /* cylinders per group */ 133 int32_t fs_ipg; /* inodes per group */ 134 int32_t fs_fpg; /* blocks per group * fs_frag */ 135/* this data must be re-computed after crashes */ 136 struct csum fs_old_cstotal; /* cylinder summary information */ 137/* these fields are cleared at mount time */ 138 int8_t fs_fmod; /* super block modified flag */ 139 int8_t fs_clean; /* filesystem is clean flag */ 140 int8_t fs_ronly; /* mounted read-only flag */ 141 int8_t fs_old_flags; /* old FS_ flags */ 142 u_char fs_fsmnt[MAXMNTLEN]; /* name mounted on */ 143 u_char fs_volname[MAXVOLLEN]; /* volume name */ 144 uint64_t fs_swuid; /* system-wide uid */ 145 int32_t fs_pad; /* due to alignment of fs_swuid */ 146/* these fields retain the current block allocation info */ 147 int32_t fs_cgrotor; /* last cg searched */ 148 void *fs_ocsp[NOCSPTRS]; /* padding; was list of fs_cs buffers */ 149 uint8_t *fs_contigdirs; /* # of contiguously allocated dirs */ 150 struct csum *fs_csp; /* cg summary info buffer for fs_cs */ 151 int32_t *fs_maxcluster; /* max cluster in each cyl group */ 152 u_int *fs_active; /* used by snapshots to track fs */ 153 int32_t fs_old_cpc; /* cyl per cycle in postbl */ 154 int32_t fs_maxbsize; /* maximum blocking factor permitted */ 155 int64_t fs_unrefs; /* number of unreferenced inodes */ 156 int64_t fs_sparecon64[16]; /* old rotation block list head */ 157 int64_t fs_sblockloc; /* byte offset of standard superblock */ 158 struct csum_total fs_cstotal; /* cylinder summary information */ 159 ufs_time_t fs_time; /* last time written */ 160 int64_t fs_size; /* number of blocks in fs */ 161 int64_t fs_dsize; /* number of data blocks in fs */ 162 ufs2_daddr_t fs_csaddr; /* blk addr of cyl grp summary area */ 163 int64_t fs_pendingblocks; /* blocks in process of being freed */ 164 int32_t fs_pendinginodes; /* inodes in process of being freed */ 165 int32_t fs_snapinum[FSMAXSNAP]; /* list of snapshot inode numbers */ 166 int32_t fs_avgfilesize; /* expected average file size */ 167 int32_t fs_avgfpdir; /* expected # of files per directory */ 168 int32_t fs_save_cgsize; /* save real cg size to use fs_bsize */ 169 int32_t fs_sparecon32[26]; /* reserved for future constants */ 170 int32_t fs_flags; /* see FS_ flags below */ 171 int32_t fs_contigsumsize; /* size of cluster summary array */ 172 int32_t fs_maxsymlinklen; /* max length of an internal symlink */ 173 int32_t fs_old_inodefmt; /* format of on-disk inodes */ 174 uint64_t fs_maxfilesize; /* maximum representable file size */ 175 int64_t fs_qbmask; /* ~fs_bmask for use with 64-bit size */ 176 int64_t fs_qfmask; /* ~fs_fmask for use with 64-bit size */ 177 int32_t fs_state; /* validate fs_clean field */ 178 int32_t fs_old_postblformat; /* format of positional layout tables */ 179 int32_t fs_old_nrpos; /* number of rotational positions */ 180 int32_t fs_spare5[2]; /* old fs_postbloff */ 181 /* old fs_rotbloff */ 182 int32_t fs_magic; /* magic number */ 183}; 184 185/* 186 * Filesystem identification 187 */ 188#define FS_UFS1_MAGIC 0x011954 /* UFS1 fast filesystem magic number */ 189#define FS_UFS2_MAGIC 0x19540119 /* UFS2 fast filesystem magic number */ 190#define FS_OKAY 0x7c269d38 /* superblock checksum */ 191#define FS_42INODEFMT -1 /* 4.2BSD inode format */ 192#define FS_44INODEFMT 2 /* 4.4BSD inode format */ 193 194/* 195 * Preference for optimization. 196 */ 197#define FS_OPTTIME 0 /* minimize allocation time */ 198#define FS_OPTSPACE 1 /* minimize disk fragmentation */ 199.Ed 200.Pp 201Each disk drive contains some number of file systems. 202A file system consists of a number of cylinder groups. 203Each cylinder group has inodes and data. 204.Pp 205A file system is described by its super-block, which in turn 206describes the cylinder groups. 207The super-block is critical 208data and is replicated in each cylinder group to protect against 209catastrophic loss. 210This is done at file system creation 211time and the critical 212super-block data does not change, so the copies need not be 213referenced further unless disaster strikes. 214.Pp 215Addresses stored in inodes are capable of addressing fragments 216of `blocks'. 217File system blocks of at most size 218.Dv MAXBSIZE 219can 220be optionally broken into 2, 4, or 8 pieces, each of which is 221addressable; these pieces may be 222.Dv DEV_BSIZE , 223or some multiple of 224a 225.Dv DEV_BSIZE 226unit. 227.Pp 228Large files consist of exclusively large data blocks. 229To avoid 230undue wasted disk space, the last data block of a small file is 231allocated as only as many fragments of a large block as are 232necessary. 233The file system format retains only a single pointer 234to such a fragment, which is a piece of a single large block that 235has been divided. 236The size of such a fragment is determinable from 237information in the inode, using the 238.Fn blksize fs ip lbn 239macro. 240.Pp 241The file system records space availability at the fragment level; 242to determine block availability, aligned fragments are examined. 243.Pp 244The root inode is the root of the file system. 245Inode 0 cannot be used for normal purposes and 246historically bad blocks were linked to inode 1, 247thus the root inode is 2 (inode 1 is no longer used for 248this purpose, however numerous dump tapes make this 249assumption, so we are stuck with it). 250.Pp 251The 252.Fa fs_minfree 253element gives the minimum acceptable percentage of file system 254blocks that may be free. 255If the freelist drops below this level 256only the super-user may continue to allocate blocks. 257The 258.Fa fs_minfree 259element 260may be set to 0 if no reserve of free blocks is deemed necessary, 261however severe performance degradations will be observed if the 262file system is run at greater than 90% full; thus the default 263value of 264.Fa fs_minfree 265is 10%. 266.Pp 267Empirically the best trade-off between block fragmentation and 268overall disk utilization at a loading of 90% comes with a 269fragmentation of 8, thus the default fragment size is an eighth 270of the block size. 271.Pp 272The element 273.Fa fs_optim 274specifies whether the file system should try to minimize the time spent 275allocating blocks, or if it should attempt to minimize the space 276fragmentation on the disk. 277If the value of fs_minfree (see above) is less than 10%, 278then the file system defaults to optimizing for space to avoid 279running out of full sized blocks. 280If the value of minfree is greater than or equal to 10%, 281fragmentation is unlikely to be problematical, and 282the file system defaults to optimizing for time. 283.Pp 284.Em Cylinder group related limits : 285Each cylinder keeps track of the availability of blocks at different 286rotational positions, so that sequential blocks can be laid out 287with minimum rotational latency. 288With the default of 8 distinguished 289rotational positions, the resolution of the 290summary information is 2ms for a typical 3600 rpm drive. 291.Pp 292The element 293.Fa fs_old_rotdelay 294gives the minimum number of milliseconds to initiate 295another disk transfer on the same cylinder. 296It is used in determining the rotationally optimal 297layout for disk blocks within a file; 298the default value for 299.Fa fs_old_rotdelay 300is 2ms. 301.Pp 302Each file system has a statically allocated number of inodes. 303An inode is allocated for each 304.Dv NBPI 305bytes of disk space. 306The inode allocation strategy is extremely conservative. 307.Pp 308.Dv MINBSIZE 309is the smallest allowable block size. 310With a 311.Dv MINBSIZE 312of 4096 313it is possible to create files of size 3142^32 with only two levels of indirection. 315.Dv MINBSIZE 316must be big enough to hold a cylinder group block, 317thus changes to 318.Pq Fa struct cg 319must keep its size within 320.Dv MINBSIZE . 321Note that super-blocks are never more than size 322.Dv SBLOCKSIZE . 323.Pp 324The path name on which the file system is mounted is maintained in 325.Fa fs_fsmnt . 326.Dv MAXMNTLEN 327defines the amount of space allocated in 328the super-block for this name. 329The limit on the amount of summary information per file system 330is defined by 331.Dv MAXCSBUFS . 332For a 4096 byte block size, it is currently parameterized for a 333maximum of two million cylinders. 334.Pp 335Per cylinder group information is summarized in blocks allocated 336from the first cylinder group's data blocks. 337These blocks are read in from 338.Fa fs_csaddr 339(size 340.Fa fs_cssize ) 341in addition to the super-block. 342.Pp 343.Sy N.B. : 344.Fn sizeof "struct csum" 345must be a power of two in order for 346the 347.Fn fs_cs 348macro to work. 349.Pp 350The 351.Em "Super-block for a file system" : 352The size of the rotational layout tables 353is limited by the fact that the super-block is of size 354.Dv SBLOCKSIZE . 355The size of these tables is 356.Em inversely 357proportional to the block 358size of the file system. 359The size of the tables is 360increased when sector sizes are not powers of two, 361as this increases the number of cylinders 362included before the rotational pattern repeats 363.Pq Fa fs_cpc . 364The size of the rotational layout 365tables is derived from the number of bytes remaining in 366.Pq Fa struct fs . 367.Pp 368The number of blocks of data per cylinder group 369is limited because cylinder groups are at most one block. 370The inode and free block tables 371must fit into a single block after deducting space for 372the cylinder group structure 373.Pq Fa struct cg . 374.Pp 375The 376.Em Inode : 377The inode is the focus of all file activity in the 378.Ux 379file system. 380There is a unique inode allocated 381for each active file, 382each current directory, each mounted-on file, 383text file, and the root. 384An inode is `named' by its device/i-number pair. 385For further information, see the include file 386.In ufs/ufs/inode.h . 387.Pp 388The format of an external attribute is defined by the extattr structure: 389.Bd -literal 390struct extattr { 391 int32_t ea_length; /* length of this attribute */ 392 int8_t ea_namespace; /* name space of this attribute */ 393 int8_t ea_contentpadlen; /* padding at end of attribute */ 394 int8_t ea_namelength; /* length of attribute name */ 395 char ea_name[1]; /* null-terminated attribute name */ 396 /* extended attribute content follows */ 397}; 398.Ed 399.Pp 400Several macros are defined to manipulate these structures. 401Each macro takes a pointer to an extattr structure. 402.Bl -tag -width ".Dv EXTATTR_SET_LENGTHS(eap, size)" 403.It Dv EXTATTR_NEXT(eap) 404Returns a pointer to the next extended attribute following 405.Fa eap . 406.It Dv EXTATTR_CONTENT(eap) 407Returns a pointer to the extended attribute content referenced by 408.Fa eap . 409.It Dv EXTATTR_CONTENT_SIZE(eap) 410Returns the size of the extended attribute content referenced by 411.Fa eap . 412.It Dv EXTATTR_SET_LENGTHS(eap, size) 413Called with the size of the attribute content after initializing 414the attribute name to calculate and set the 415.Fa ea_length , 416.Fa ea_namelength , 417and 418.Fa ea_contentpadlen 419fields of the extended attribute structure. 420.El 421.Pp 422The following code identifies an ACL: 423.Bd -literal 424 if (eap->ea_namespace == EXTATTR_NAMESPACE_SYSTEM && 425 !strcmp(eap->ea_name, POSIX1E_ACL_ACCESS_EXTATTR_NAME) { 426 aclp = EXTATTR_CONTENT(eap); 427 acllen = EXTATTR_CONTENT_SIZE(eap); 428 ... 429 } 430.Ed 431.Pp 432The following code creates an extended attribute 433containing a copy of a structure 434.Fa mygif : 435.Bd -literal 436 eap->ea_namespace = EXTATTR_NAMESPACE_USER; 437 strcpy(eap->ea_name, "filepic.gif"); 438 EXTATTR_SET_LENGTHS(eap, sizeof(struct mygif)); 439 memcpy(EXTATTR_CONTENT(eap), &mygif, sizeof(struct mygif)); 440.Ed 441.Sh HISTORY 442A super-block structure named filsys appeared in 443.At v6 . 444The file system described in this manual appeared 445in 446.Bx 4.2 . 447
