1 /* $NetBSD: lfs.h,v 1.211 2025/10/20 04:20:36 perseant Exp $ */ 2 3 /* from NetBSD: dinode.h,v 1.25 2016/01/22 23:06:10 dholland Exp */ 4 /* from NetBSD: dir.h,v 1.25 2015/09/01 06:16:03 dholland Exp */ 5 6 /*- 7 * Copyright (c) 1999, 2000, 2001, 2002, 2003 The NetBSD Foundation, Inc. 8 * All rights reserved. 9 * 10 * This code is derived from software contributed to The NetBSD Foundation 11 * by Konrad E. Schroder <perseant (at) hhhh.org>. 12 * 13 * Redistribution and use in source and binary forms, with or without 14 * modification, are permitted provided that the following conditions 15 * are met: 16 * 1. Redistributions of source code must retain the above copyright 17 * notice, this list of conditions and the following disclaimer. 18 * 2. Redistributions in binary form must reproduce the above copyright 19 * notice, this list of conditions and the following disclaimer in the 20 * documentation and/or other materials provided with the distribution. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 23 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 24 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 25 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 26 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 27 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 28 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 29 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 30 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 31 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 32 * POSSIBILITY OF SUCH DAMAGE. 33 */ 34 /*- 35 * Copyright (c) 1991, 1993 36 * The Regents of the University of California. All rights reserved. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. Neither the name of the University nor the names of its contributors 47 * may be used to endorse or promote products derived from this software 48 * without specific prior written permission. 49 * 50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 53 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 60 * SUCH DAMAGE. 61 * 62 * @(#)lfs.h 8.9 (Berkeley) 5/8/95 63 */ 64 /* 65 * Copyright (c) 2002 Networks Associates Technology, Inc. 66 * All rights reserved. 67 * 68 * This software was developed for the FreeBSD Project by Marshall 69 * Kirk McKusick and Network Associates Laboratories, the Security 70 * Research Division of Network Associates, Inc. under DARPA/SPAWAR 71 * contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS 72 * research program 73 * 74 * Copyright (c) 1982, 1989, 1993 75 * The Regents of the University of California. All rights reserved. 76 * (c) UNIX System Laboratories, Inc. 77 * All or some portions of this file are derived from material licensed 78 * to the University of California by American Telephone and Telegraph 79 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 80 * the permission of UNIX System Laboratories, Inc. 81 * 82 * Redistribution and use in source and binary forms, with or without 83 * modification, are permitted provided that the following conditions 84 * are met: 85 * 1. Redistributions of source code must retain the above copyright 86 * notice, this list of conditions and the following disclaimer. 87 * 2. Redistributions in binary form must reproduce the above copyright 88 * notice, this list of conditions and the following disclaimer in the 89 * documentation and/or other materials provided with the distribution. 90 * 3. Neither the name of the University nor the names of its contributors 91 * may be used to endorse or promote products derived from this software 92 * without specific prior written permission. 93 * 94 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 95 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 96 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 97 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 98 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 99 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 100 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 101 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 102 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 103 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 104 * SUCH DAMAGE. 105 * 106 * @(#)dinode.h 8.9 (Berkeley) 3/29/95 107 */ 108 /* 109 * Copyright (c) 1982, 1986, 1989, 1993 110 * The Regents of the University of California. All rights reserved. 111 * (c) UNIX System Laboratories, Inc. 112 * All or some portions of this file are derived from material licensed 113 * to the University of California by American Telephone and Telegraph 114 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 115 * the permission of UNIX System Laboratories, Inc. 116 * 117 * Redistribution and use in source and binary forms, with or without 118 * modification, are permitted provided that the following conditions 119 * are met: 120 * 1. Redistributions of source code must retain the above copyright 121 * notice, this list of conditions and the following disclaimer. 122 * 2. Redistributions in binary form must reproduce the above copyright 123 * notice, this list of conditions and the following disclaimer in the 124 * documentation and/or other materials provided with the distribution. 125 * 3. Neither the name of the University nor the names of its contributors 126 * may be used to endorse or promote products derived from this software 127 * without specific prior written permission. 128 * 129 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 130 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 131 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 132 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 133 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 134 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 135 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 136 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 137 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 138 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 139 * SUCH DAMAGE. 140 * 141 * @(#)dir.h 8.5 (Berkeley) 4/27/95 142 */ 143 144 /* 145 * NOTE: COORDINATE ON-DISK FORMAT CHANGES WITH THE FREEBSD PROJECT. 146 */ 147 148 #ifndef _UFS_LFS_LFS_H_ 149 #define _UFS_LFS_LFS_H_ 150 151 #if !defined(_KERNEL) && !defined(_STANDALONE) 152 #include <stddef.h> /* for offsetof */ 153 #endif 154 155 #include <sys/rwlock.h> 156 #include <sys/mutex.h> 157 #include <sys/queue.h> 158 #include <sys/condvar.h> 159 #include <sys/mount.h> 160 #include <sys/pool.h> 161 162 /* 163 * Compile-time options for LFS. 164 */ 165 #define LFS_IFIND_RETRIES 16 166 #define LFS_LOGLENGTH 1024 /* size of debugging log */ 167 #define LFS_MAX_ACTIVE 10 /* Dirty segments before ckp forced */ 168 169 /* 170 * Fixed filesystem layout parameters 171 */ 172 #define LFS_LABELPAD 8192 /* LFS label size */ 173 #define LFS_SBPAD 8192 /* LFS superblock size */ 174 175 #define LFS_UNUSED_INUM 0 /* 0: out of band inode number */ 176 #define LFS_IFILE_INUM 1 /* 1: IFILE inode number */ 177 /* 2: Root inode number */ 178 #define LFS_LOSTFOUNDINO 3 /* 3: lost+found inode number */ 179 #define LFS_FIRST_INUM 4 /* 4: first free inode number */ 180 181 /* 182 * The root inode is the root of the file system. Inode 0 can't be used for 183 * normal purposes and historically bad blocks were linked to inode 1, thus 184 * the root inode is 2. (Inode 1 is no longer used for this purpose, however 185 * numerous dump tapes make this assumption, so we are stuck with it). 186 */ 187 #define ULFS_ROOTINO ((ino_t)2) 188 189 /* 190 * The Whiteout inode# is a dummy non-zero inode number which will 191 * never be allocated to a real file. It is used as a place holder 192 * in the directory entry which has been tagged as a LFS_DT_WHT entry. 193 * See the comments about ULFS_ROOTINO above. 194 */ 195 #define ULFS_WINO ((ino_t)1) 196 197 198 #define LFS_V1_SUMMARY_SIZE 512 /* V1 fixed summary size */ 199 #define LFS_DFL_SUMMARY_SIZE 512 /* Default summary size */ 200 201 #define LFS_MAXNAMLEN 255 /* maximum name length in a dir */ 202 203 #define ULFS_NXADDR 2 204 #define ULFS_NDADDR 12 /* Direct addresses in inode. */ 205 #define ULFS_NIADDR 3 /* Indirect addresses in inode. */ 206 207 /* 208 * Adjustable filesystem parameters 209 */ 210 #ifndef LFS_ATIME_IFILE 211 # define LFS_ATIME_IFILE 0 /* Store atime info in ifile (optional in LFSv1) */ 212 #endif 213 #define LFS_MARKV_MAXBLKCNT 65536 /* Max block count for lfs_markv() */ 214 #define LFS_FILESTATS_MAXCNT 65536 /* Max entry count for LFCNFILESTATS */ 215 #define LFS_REWRITE_MAXCNT 65536 /* Max entry count for LFCNREWRITE* */ 216 #define LFS_SEGUSE_MAXCNT 65536 /* Max entry count for LFCNSEGUSE */ 217 218 /* 219 * Directories 220 */ 221 222 /* 223 * Directories in LFS are files; they use the same inode and block 224 * mapping structures that regular files do. The directory per se is 225 * manifested in the file contents: an unordered, unstructured 226 * sequence of variable-size directory entries. 227 * 228 * This format and structure is taken (via what was originally shared 229 * ufs-level code) from FFS. Each directory entry is a fixed header 230 * followed by a string, the total length padded to a 4-byte boundary. 231 * All strings include a null terminator; the maximum string length 232 * is LFS_MAXNAMLEN, which is 255. 233 * 234 * The directory entry header structure (struct lfs_dirheader) is just 235 * the header information. A complete entry is this plus a null- 236 * terminated name following it, plus some amount of padding. The 237 * length of the name (not including the null terminator) is given by 238 * the namlen field of the header; the complete record length, 239 * including the null terminator and padding, is given by the reclen 240 * field of the header. The record length is always 4-byte aligned. 241 * (Even on 64-bit volumes, the record length is only 4-byte aligned, 242 * not 8-byte.) 243 * 244 * Historically, FFS directories were/are organized into blocks of 245 * size DIRBLKSIZE that can be written atomically to disk at the 246 * hardware level. Directory entries are not allowed to cross the 247 * boundaries of these blocks. The resulting atomicity is important 248 * for the integrity of FFS volumes; however, for LFS it's irrelevant. 249 * All we have to care about is not writing out directories that 250 * confuse earlier ufs-based versions of the LFS code. 251 * 252 * This means [to be determined]. (XXX) 253 * 254 * As DIRBLKSIZE in its FFS sense is hardware-dependent, and file 255 * system images do from time to time move to different hardware, code 256 * that reads directories should be prepared to handle directories 257 * written in a context where DIRBLKSIZE was different (smaller or 258 * larger) than its current value. Note however that it is not 259 * sensible for DIRBLKSIZE to be larger than the volume fragment size, 260 * and not practically possible for it to be larger than the volume 261 * block size. 262 * 263 * Some further notes: 264 * - the LFS_DIRSIZ macro provides the minimum space needed to hold 265 * a directory entry. 266 * - any particular entry may be arbitrarily larger (which is why the 267 * header stores both the entry size and the name size) to pad out 268 * unused space. 269 * - historically the padding in an entry is not necessarily zeroed 270 * but may contain trash. 271 * - dp->d_reclen is the size of the entry. This is always 4-byte 272 * aligned. 273 * - dp->d_namlen is the length of the string, and should always be 274 * the same as strlen(dp->d_name). 275 * - in particular, space available in an entry is given by 276 * dp->d_reclen - LFS_DIRSIZ(dp), and all space available within a 277 * directory block is tucked away within an existing entry. 278 * - all space within a directory block is part of some entry. 279 * - therefore, inserting a new entry requires finding and 280 * splitting a suitable existing entry, and when entries are 281 * removed their space is merged into the entry ahead of them. 282 * - an empty/unused entry has d_ino set to 0. This normally only 283 * appears in the first entry in a block, as elsewhere the unused 284 * entry should have been merged into the one before it. However, 285 * fsck leaves such entries behind so they must be tolerated 286 * elsewhere. 287 * - a completely empty directory block has one entry whose 288 * d_reclen is DIRBLKSIZ and whose d_ino is 0. 289 * 290 * The "old directory format" referenced by the fs->lfs_isolddirfmt 291 * flag (and some other things) refers to when the type field was 292 * added to directory entries. This change was made to FFS in the 80s, 293 * well before LFS was first written; there should be no LFS volumes 294 * (and certainly no LFS v2-format volumes or LFS64 volumes) where the 295 * old format pertains. All of the related logic should probably be 296 * removed; however, it hasn't been yet, and we get to carry it around 297 * until we can be conclusively sure it isn't needed. 298 * 299 * In the "old directory format" there is no type field and the namlen 300 * field is correspondingly 16 bits wide. On big-endian volumes this 301 * has no effect: namlen cannot exceed 255, so the upper byte is 302 * always 0 and this reads back from the type field as LFS_DT_UNKNOWN. 303 * On little-endian volumes, the namlen field will always be 0 and 304 * the namlen value needs to be read out of the type field. (The type 305 * is always LFS_DT_UNKNOWN.) The directory accessor functions take 306 * care of this so nothing else needs to be aware of it. 307 * 308 * LFS_OLDDIRFMT and LFS_NEWDIRFMT are code numbers for the old and 309 * new directory format respectively. These codes do not appear on 310 * disk; they're generated from a runtime macro called FSFMT() that's 311 * cued by other things. This is why (confusingly) LFS_OLDDIRFMT is 1 312 * and LFS_NEWDIRFMT is 0. 313 * 314 * FSFMT(), LFS_OLDDIRFMT, and LFS_NEWDIRFMT should be removed. (XXX) 315 */ 316 317 /* 318 * Directory block size. 319 */ 320 #undef LFS_DIRBLKSIZ 321 #define LFS_DIRBLKSIZ DEV_BSIZE 322 323 /* 324 * Convert between stat structure type codes and directory entry type codes. 325 */ 326 #define LFS_IFTODT(mode) (((mode) & 0170000) >> 12) 327 #define LFS_DTTOIF(dirtype) ((dirtype) << 12) 328 329 /* 330 * Theoretically, directories can be more than 2Gb in length; however, in 331 * practice this seems unlikely. So, we define the type doff_t as a 32-bit 332 * quantity to keep down the cost of doing lookup on a 32-bit machine. 333 */ 334 #define doff_t int32_t 335 #define lfs_doff_t int32_t 336 #define LFS_MAXDIRSIZE (0x7fffffff) 337 338 /* 339 * File types for d_type 340 */ 341 #define LFS_DT_UNKNOWN 0 342 #define LFS_DT_FIFO 1 343 #define LFS_DT_CHR 2 344 #define LFS_DT_DIR 4 345 #define LFS_DT_BLK 6 346 #define LFS_DT_REG 8 347 #define LFS_DT_LNK 10 348 #define LFS_DT_SOCK 12 349 #define LFS_DT_WHT 14 350 351 /* 352 * (See notes above) 353 */ 354 355 struct lfs_dirheader32 { 356 uint32_t dh_ino; /* inode number of entry */ 357 uint16_t dh_reclen; /* length of this record */ 358 uint8_t dh_type; /* file type, see below */ 359 uint8_t dh_namlen; /* length of string in d_name */ 360 }; 361 __CTASSERT(sizeof(struct lfs_dirheader32) == 8); 362 363 struct lfs_dirheader64 { 364 uint64_t dh_ino; /* inode number of entry */ 365 uint16_t dh_reclen; /* length of this record */ 366 uint8_t dh_type; /* file type, see below */ 367 uint8_t dh_namlen; /* length of string in d_name */ 368 } __aligned(4) __packed; 369 __CTASSERT(sizeof(struct lfs_dirheader64) == 12); 370 371 union lfs_dirheader { 372 struct lfs_dirheader64 u_64; 373 struct lfs_dirheader32 u_32; 374 }; 375 __CTASSERT(__alignof(union lfs_dirheader) == __alignof(struct lfs_dirheader64)); 376 #ifndef __lint__ 377 __CTASSERT(__alignof(union lfs_dirheader) == __alignof(struct lfs_dirheader32)); 378 #endif 379 380 typedef union lfs_dirheader LFS_DIRHEADER; 381 382 /* 383 * Template for manipulating directories. 384 */ 385 386 struct lfs_dirtemplate32 { 387 struct lfs_dirheader32 dot_header; 388 char dot_name[4]; /* must be multiple of 4 */ 389 struct lfs_dirheader32 dotdot_header; 390 char dotdot_name[4]; /* ditto */ 391 }; 392 __CTASSERT(sizeof(struct lfs_dirtemplate32) == 2*(8 + 4)); 393 394 struct lfs_dirtemplate64 { 395 struct lfs_dirheader64 dot_header; 396 char dot_name[4]; /* must be multiple of 4 */ 397 struct lfs_dirheader64 dotdot_header; 398 char dotdot_name[4]; /* ditto */ 399 }; 400 __CTASSERT(sizeof(struct lfs_dirtemplate64) == 2*(12 + 4)); 401 402 union lfs_dirtemplate { 403 struct lfs_dirtemplate64 u_64; 404 struct lfs_dirtemplate32 u_32; 405 }; 406 407 #if 0 408 /* 409 * This is the old format of directories, sans type element. 410 */ 411 struct lfs_odirtemplate { 412 uint32_t dot_ino; 413 int16_t dot_reclen; 414 uint16_t dot_namlen; 415 char dot_name[4]; /* must be multiple of 4 */ 416 uint32_t dotdot_ino; 417 int16_t dotdot_reclen; 418 uint16_t dotdot_namlen; 419 char dotdot_name[4]; /* ditto */ 420 }; 421 __CTASSERT(sizeof(struct lfs_odirtemplate) == 2*(8 + 4)); 422 #endif 423 424 /* 425 * Inodes 426 */ 427 428 /* 429 * A dinode contains all the meta-data associated with a LFS file. 430 * This structure defines the on-disk format of a dinode. Since 431 * this structure describes an on-disk structure, all its fields 432 * are defined by types with precise widths. 433 */ 434 435 struct lfs32_dinode { 436 uint16_t di_mode; /* 0: IFMT, permissions; see below. */ 437 int16_t di_nlink; /* 2: File link count. */ 438 uint32_t di_inumber; /* 4: Inode number. */ 439 uint64_t di_size; /* 8: File byte count. */ 440 int32_t di_atime; /* 16: Last access time. */ 441 int32_t di_atimensec; /* 20: Last access time. */ 442 int32_t di_mtime; /* 24: Last modified time. */ 443 int32_t di_mtimensec; /* 28: Last modified time. */ 444 int32_t di_ctime; /* 32: Last inode change time. */ 445 int32_t di_ctimensec; /* 36: Last inode change time. */ 446 int32_t di_db[ULFS_NDADDR]; /* 40: Direct disk blocks. */ 447 int32_t di_ib[ULFS_NIADDR]; /* 88: Indirect disk blocks. */ 448 uint32_t di_flags; /* 100: Status flags (chflags). */ 449 uint32_t di_blocks; /* 104: Blocks actually held. */ 450 int32_t di_gen; /* 108: Generation number. */ 451 uint32_t di_uid; /* 112: File owner. */ 452 uint32_t di_gid; /* 116: File group. */ 453 uint64_t di_modrev; /* 120: i_modrev for NFSv4 */ 454 }; 455 __CTASSERT(sizeof(struct lfs32_dinode) == 128); 456 457 struct lfs64_dinode { 458 uint16_t di_mode; /* 0: IFMT, permissions; see below. */ 459 int16_t di_nlink; /* 2: File link count. */ 460 uint32_t di_uid; /* 4: File owner. */ 461 uint32_t di_gid; /* 8: File group. */ 462 uint32_t di_blksize; /* 12: Inode blocksize. */ 463 uint64_t di_size; /* 16: File byte count. */ 464 uint64_t di_blocks; /* 24: Bytes actually held. */ 465 int64_t di_atime; /* 32: Last access time. */ 466 int64_t di_mtime; /* 40: Last modified time. */ 467 int64_t di_ctime; /* 48: Last inode change time. */ 468 int64_t di_birthtime; /* 56: Inode creation time. */ 469 int32_t di_mtimensec; /* 64: Last modified time. */ 470 int32_t di_atimensec; /* 68: Last access time. */ 471 int32_t di_ctimensec; /* 72: Last inode change time. */ 472 int32_t di_birthnsec; /* 76: Inode creation time. */ 473 int32_t di_gen; /* 80: Generation number. */ 474 uint32_t di_kernflags; /* 84: Kernel flags. */ 475 uint32_t di_flags; /* 88: Status flags (chflags). */ 476 int32_t di_extsize; /* 92: External attributes block. */ 477 int64_t di_extb[ULFS_NXADDR];/* 96: External attributes block. */ 478 int64_t di_db[ULFS_NDADDR]; /* 112: Direct disk blocks. */ 479 int64_t di_ib[ULFS_NIADDR]; /* 208: Indirect disk blocks. */ 480 uint64_t di_modrev; /* 232: i_modrev for NFSv4 */ 481 uint64_t di_inumber; /* 240: Inode number */ 482 uint64_t di_spare[1]; /* 248: Reserved; currently unused */ 483 }; 484 __CTASSERT(sizeof(struct lfs64_dinode) == 256); 485 486 union lfs_dinode { 487 struct lfs64_dinode u_64; 488 struct lfs32_dinode u_32; 489 }; 490 __CTASSERT(__alignof(union lfs_dinode) == __alignof(struct lfs64_dinode)); 491 __CTASSERT(__alignof(union lfs_dinode) == __alignof(struct lfs32_dinode)); 492 493 /* 494 * The di_db fields may be overlaid with other information for 495 * file types that do not have associated disk storage. Block 496 * and character devices overlay the first data block with their 497 * dev_t value. Short symbolic links place their path in the 498 * di_db area. 499 */ 500 #define di_rdev di_db[0] 501 502 /* Size of the on-disk inode. */ 503 //#define LFS_DINODE1_SIZE (sizeof(struct ulfs1_dinode)) /* 128 */ 504 //#define LFS_DINODE2_SIZE (sizeof(struct ulfs2_dinode)) 505 506 /* File types, found in the upper bits of di_mode. */ 507 #define LFS_IFMT 0170000 /* Mask of file type. */ 508 #define LFS_IFIFO 0010000 /* Named pipe (fifo). */ 509 #define LFS_IFCHR 0020000 /* Character device. */ 510 #define LFS_IFDIR 0040000 /* Directory file. */ 511 #define LFS_IFBLK 0060000 /* Block device. */ 512 #define LFS_IFREG 0100000 /* Regular file. */ 513 #define LFS_IFLNK 0120000 /* Symbolic link. */ 514 #define LFS_IFSOCK 0140000 /* UNIX domain socket. */ 515 #define LFS_IFWHT 0160000 /* Whiteout. */ 516 517 /* 518 * "struct buf" associated definitions 519 */ 520 521 /* Unassigned disk addresses. */ 522 #define UNASSIGNED -1 523 #define UNWRITTEN -2 524 525 /* Unused logical block number */ 526 #define LFS_UNUSED_LBN -1 527 528 /* 529 * On-disk and in-memory checkpoint segment usage structure. 530 */ 531 typedef struct segusage SEGUSE; 532 struct segusage { 533 uint32_t su_nbytes; /* 0: number of live bytes */ 534 uint32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 535 uint16_t su_nsums; /* 8: number of summaries in segment */ 536 uint16_t su_ninos; /* 10: number of inode blocks in seg */ 537 538 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 539 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 540 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 541 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 542 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 543 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 544 uint32_t su_flags; /* 12: segment flags */ 545 uint64_t su_lastmod; /* 16: last modified timestamp */ 546 }; 547 __CTASSERT(sizeof(struct segusage) == 24); 548 549 typedef struct segusage_v1 SEGUSE_V1; 550 struct segusage_v1 { 551 uint32_t su_nbytes; /* 0: number of live bytes */ 552 uint32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 553 uint16_t su_nsums; /* 8: number of summaries in segment */ 554 uint16_t su_ninos; /* 10: number of inode blocks in seg */ 555 uint32_t su_flags; /* 12: segment flags */ 556 }; 557 __CTASSERT(sizeof(struct segusage_v1) == 16); 558 559 /* 560 * On-disk file information. One per file with data blocks in the segment. 561 * 562 * The FINFO structure is a header; it is followed by fi_nblocks block 563 * pointers, which are logical block numbers of the file. (These are the 564 * blocks of the file present in this segment.) 565 */ 566 567 typedef struct finfo64 FINFO64; 568 struct finfo64 { 569 uint32_t fi_nblocks; /* number of blocks */ 570 uint32_t fi_version; /* version number */ 571 uint64_t fi_ino; /* inode number */ 572 uint32_t fi_lastlength; /* length of last block in array */ 573 uint32_t fi_pad; /* unused */ 574 } __aligned(4) __packed; 575 __CTASSERT(sizeof(struct finfo64) == 24); 576 577 typedef struct finfo32 FINFO32; 578 struct finfo32 { 579 uint32_t fi_nblocks; /* number of blocks */ 580 uint32_t fi_version; /* version number */ 581 uint32_t fi_ino; /* inode number */ 582 uint32_t fi_lastlength; /* length of last block in array */ 583 }; 584 __CTASSERT(sizeof(struct finfo32) == 16); 585 586 typedef union finfo { 587 struct finfo64 u_64; 588 struct finfo32 u_32; 589 } FINFO; 590 __CTASSERT(__alignof(union finfo) == __alignof(struct finfo64)); 591 #ifndef __lint__ 592 __CTASSERT(__alignof(union finfo) == __alignof(struct finfo32)); 593 #endif 594 595 /* 596 * inode info (part of the segment summary) 597 * 598 * Each one of these is just a block number; wrapping the structure 599 * around it gives more contextual information in the code about 600 * what's going on. 601 */ 602 603 typedef struct iinfo64 { 604 uint64_t ii_block; /* block number */ 605 } __aligned(4) __packed IINFO64; 606 __CTASSERT(sizeof(struct iinfo64) == 8); 607 608 typedef struct iinfo32 { 609 uint32_t ii_block; /* block number */ 610 } IINFO32; 611 __CTASSERT(sizeof(struct iinfo32) == 4); 612 613 typedef union iinfo { 614 struct iinfo64 u_64; 615 struct iinfo32 u_32; 616 } IINFO; 617 __CTASSERT(__alignof(union iinfo) == __alignof(struct iinfo64)); 618 #ifndef __lint__ 619 __CTASSERT(__alignof(union iinfo) == __alignof(struct iinfo32)); 620 #endif 621 622 /* 623 * Index file inode entries. 624 */ 625 626 /* magic values for daddrs */ 627 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 628 #ifdef DEBUG 629 # define LFS_ILLEGAL_DADDR -1 /* out-of-band daddr */ 630 # define DADDR_IS_BAD(daddr) \ 631 ((daddr) == LFS_UNUSED_DADDR || (daddr) == LFS_ILLEGAL_DADDR) 632 #else /* ! DEBUG */ 633 # define LFS_ILLEGAL_DADDR LFS_UNUSED_DADDR 634 # define DADDR_IS_BAD(daddr) ((daddr) == LFS_UNUSED_DADDR) 635 #endif /* ! DEBUG */ 636 637 /* magic value for if_nextfree -- indicate orphaned file */ 638 #define LFS_ORPHAN_NEXTFREE(fs) \ 639 ((fs)->lfs_is64 ? ~(uint64_t)0 : ~(uint32_t)0) 640 641 #define INUM_IS_BAD(fs, ino) \ 642 (ino == LFS_UNUSED_INUM || ino == LFS_ORPHAN_NEXTFREE(fs)) 643 644 typedef struct ifile64 IFILE64; 645 struct ifile64 { 646 uint32_t if_version; /* inode version number */ 647 uint32_t if_atime_nsec; /* and nanoseconds */ 648 uint64_t if_atime_sec; /* Last access time, seconds */ 649 int64_t if_daddr; /* inode disk address */ 650 uint64_t if_nextfree; /* next-unallocated inode */ 651 } __aligned(4) __packed; 652 __CTASSERT(sizeof(struct ifile64) == 32); 653 654 typedef struct ifile32 IFILE32; 655 struct ifile32 { 656 uint32_t if_version; /* inode version number */ 657 int32_t if_daddr; /* inode disk address */ 658 uint32_t if_nextfree; /* next-unallocated inode */ 659 uint32_t if_atime_sec; /* Last access time, seconds */ 660 uint32_t if_atime_nsec; /* and nanoseconds */ 661 }; 662 __CTASSERT(sizeof(struct ifile32) == 20); 663 664 typedef struct ifile_v1 IFILE_V1; 665 struct ifile_v1 { 666 uint32_t if_version; /* inode version number */ 667 int32_t if_daddr; /* inode disk address */ 668 uint32_t if_nextfree; /* next-unallocated inode */ 669 #if LFS_ATIME_IFILE 670 #error "this cannot work" 671 struct timespec if_atime; /* Last access time */ 672 #endif 673 }; 674 __CTASSERT(sizeof(struct ifile_v1) == 12); 675 676 /* 677 * Note: struct ifile_v1 is often handled by accessing the first three 678 * fields of struct ifile32. (XXX: Blah. This should be cleaned up as 679 * it may in some cases violate the strict-aliasing rules.) 680 */ 681 typedef union ifile { 682 struct ifile64 u_64; 683 struct ifile32 u_32; 684 struct ifile_v1 u_v1; 685 } IFILE; 686 __CTASSERT(__alignof(union ifile) == __alignof(struct ifile64)); 687 #ifndef __lint__ 688 __CTASSERT(__alignof(union ifile) == __alignof(struct ifile32)); 689 __CTASSERT(__alignof(union ifile) == __alignof(struct ifile_v1)); 690 #endif 691 692 /* 693 * Cleaner information structure. This resides in the ifile and is used 694 * to pass information from the kernel to the cleaner. 695 */ 696 697 /* flags for ->flags */ 698 #define LFS_CLEANER_MUST_CLEAN 0x01 699 700 typedef struct _cleanerinfo32 { 701 uint32_t clean; /* 0: number of clean segments */ 702 uint32_t dirty; /* 4: number of dirty segments */ 703 int32_t bfree; /* 8: disk blocks free */ 704 int32_t avail; /* 12: disk blocks available */ 705 uint32_t free_head; /* 16: head of the inode free list */ 706 uint32_t free_tail; /* 20: tail of the inode free list */ 707 uint32_t flags; /* 24: status word from the kernel */ 708 } CLEANERINFO32; 709 __CTASSERT(sizeof(struct _cleanerinfo32) == 28); 710 711 typedef struct _cleanerinfo64 { 712 uint32_t clean; /* 0: number of clean segments */ 713 uint32_t dirty; /* 4: number of dirty segments */ 714 int64_t bfree; /* 8: disk blocks free */ 715 int64_t avail; /* 16: disk blocks available */ 716 uint64_t free_head; /* 24: head of the inode free list */ 717 uint64_t free_tail; /* 32: tail of the inode free list */ 718 uint32_t flags; /* 40: status word from the kernel */ 719 uint32_t pad; /* 44: must be 64-bit aligned */ 720 } __aligned(4) __packed CLEANERINFO64; 721 __CTASSERT(sizeof(struct _cleanerinfo64) == 48); 722 723 /* this must not go to disk directly of course */ 724 typedef union _cleanerinfo { 725 CLEANERINFO32 u_32; 726 CLEANERINFO64 u_64; 727 } CLEANERINFO; 728 #ifndef __lint__ 729 __CTASSERT(__alignof(union _cleanerinfo) == __alignof(struct _cleanerinfo32)); 730 __CTASSERT(__alignof(union _cleanerinfo) == __alignof(struct _cleanerinfo64)); 731 #endif 732 733 /* 734 * On-disk segment summary information 735 */ 736 737 /* magic value for ss_magic */ 738 #define SS_MAGIC 0x061561 739 740 /* flags for ss_flags */ 741 #define SS_DIROP 0x01 /* segment begins a dirop */ 742 #define SS_CONT 0x02 /* more partials to finish this write*/ 743 #define SS_CLEAN 0x04 /* written by the cleaner */ 744 #define SS_RFW 0x08 /* written by the roll-forward agent */ 745 #define SS_RECLAIM 0x10 /* written by the roll-forward agent */ 746 747 /* type used for reading checksum signatures from metadata structures */ 748 typedef uint32_t lfs_checkword; 749 750 typedef struct segsum_v1 SEGSUM_V1; 751 struct segsum_v1 { 752 uint32_t ss_sumsum; /* 0: check sum of summary block */ 753 uint32_t ss_datasum; /* 4: check sum of data */ 754 uint32_t ss_magic; /* 8: segment summary magic number */ 755 int32_t ss_next; /* 12: next segment */ 756 uint32_t ss_create; /* 16: creation time stamp */ 757 uint16_t ss_nfinfo; /* 20: number of file info structures */ 758 uint16_t ss_ninos; /* 22: number of inodes in summary */ 759 uint16_t ss_flags; /* 24: used for directory operations */ 760 uint16_t ss_pad; /* 26: extra space */ 761 /* FINFO's and inode daddr's... */ 762 }; 763 __CTASSERT(sizeof(struct segsum_v1) == 28); 764 765 typedef struct segsum32 SEGSUM32; 766 struct segsum32 { 767 uint32_t ss_sumsum; /* 0: check sum of summary block */ 768 uint32_t ss_datasum; /* 4: check sum of data */ 769 uint32_t ss_magic; /* 8: segment summary magic number */ 770 int32_t ss_next; /* 12: next segment (disk address) */ 771 uint32_t ss_ident; /* 16: roll-forward fsid */ 772 uint16_t ss_nfinfo; /* 20: number of file info structures */ 773 uint16_t ss_ninos; /* 22: number of inodes in summary */ 774 uint16_t ss_flags; /* 24: used for directory operations */ 775 uint8_t ss_pad[2]; /* 26: extra space */ 776 uint32_t ss_reclino; /* 28: inode being reclaimed */ 777 uint64_t ss_serial; /* 32: serial number */ 778 uint64_t ss_create; /* 40: time stamp */ 779 /* FINFO's and inode daddr's... */ 780 } __aligned(4) __packed; 781 __CTASSERT(sizeof(struct segsum32) == 48); 782 783 typedef struct segsum64 SEGSUM64; 784 struct segsum64 { 785 uint32_t ss_sumsum; /* 0: check sum of summary block */ 786 uint32_t ss_datasum; /* 4: check sum of data */ 787 uint32_t ss_magic; /* 8: segment summary magic number */ 788 uint32_t ss_ident; /* 12: roll-forward fsid */ 789 int64_t ss_next; /* 16: next segment (disk address) */ 790 uint16_t ss_nfinfo; /* 24: number of file info structures */ 791 uint16_t ss_ninos; /* 26: number of inodes in summary */ 792 uint16_t ss_flags; /* 28: used for directory operations */ 793 uint8_t ss_pad[2]; /* 30: extra space */ 794 uint64_t ss_reclino; /* 32: inode being reclaimed */ 795 uint64_t ss_serial; /* 40: serial number */ 796 uint64_t ss_create; /* 48: time stamp */ 797 /* FINFO's and inode daddr's... */ 798 } __aligned(4) __packed; 799 __CTASSERT(sizeof(struct segsum64) == 56); 800 801 typedef union segsum SEGSUM; 802 union segsum { 803 struct segsum64 u_64; 804 struct segsum32 u_32; 805 struct segsum_v1 u_v1; 806 }; 807 __CTASSERT(__alignof(union segsum) == __alignof(struct segsum64)); 808 __CTASSERT(__alignof(union segsum) == __alignof(struct segsum32)); 809 #ifndef __lint__ 810 __CTASSERT(__alignof(union segsum) == __alignof(struct segsum_v1)); 811 #endif 812 813 /* 814 * On-disk super block. 815 * 816 * We have separate superblock structures for the 32-bit and 64-bit 817 * LFS, and accessor functions to hide the differences. 818 * 819 * For lfs64, the format version is always 2; version 1 lfs is old. 820 * For both, the inode format version is 0; for lfs32 this selects the 821 * same 32-bit inode as always, and for lfs64 this selects the larger 822 * 64-bit inode structure we got from ffsv2. 823 * 824 * In lfs64: 825 * - inode numbers are 64 bit now 826 * - segments may not be larger than 4G (counted in bytes) 827 * - there may not be more than 2^32 (or perhaps 2^31) segments 828 * - the total volume size is limited to 2^63 frags and/or 2^63 829 * disk blocks, and probably in practice 2^63 bytes. 830 */ 831 832 #define LFS_MAGIC 0x070162 833 #define LFS_MAGIC_SWAPPED 0x62010700 834 835 #define LFS64_MAGIC (0x19620701 ^ 0xffffffff) 836 #define LFS64_MAGIC_SWAPPED (0x01076219 ^ 0xffffffff) 837 838 #define LFS_VERSION 2 839 840 #define LFS_MIN_SBINTERVAL 5 /* min superblock segment spacing */ 841 #define LFS_MAXNUMSB 10 /* max number of superblocks */ 842 843 /* flags for dlfs_pflags */ 844 #define LFS_PF_CLEAN 0x1 845 846 /* Inode format versions */ 847 #define LFS_44INODEFMT 0 848 #define LFS_MAXINODEFMT 0 849 850 struct dlfs { 851 uint32_t dlfs_magic; /* 0: magic number */ 852 uint32_t dlfs_version; /* 4: version number */ 853 854 uint32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 855 /* number of frags in fs (v2) */ 856 uint32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 857 /* number of bytes per segment (v2) */ 858 uint32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 859 uint32_t dlfs_bsize; /* 20: file system block size */ 860 uint32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 861 uint32_t dlfs_frag; /* 28: number of frags in a block in fs */ 862 863 /* Checkpoint region. */ 864 uint32_t dlfs_freehd; /* 32: start of the free inode list */ 865 int32_t dlfs_bfree; /* 36: number of free frags */ 866 uint32_t dlfs_nfiles; /* 40: number of allocated inodes */ 867 int32_t dlfs_avail; /* 44: blocks available for writing */ 868 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 869 int32_t dlfs_idaddr; /* 52: inode file disk address */ 870 uint32_t dlfs_ifile; /* 56: inode file inode number */ 871 int32_t dlfs_lastseg; /* 60: address of last segment written */ 872 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 873 int32_t dlfs_curseg; /* 68: current segment being written */ 874 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 875 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 876 uint32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 877 878 /* These are configuration parameters. */ 879 uint32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 880 881 /* These fields can be computed from the others. */ 882 uint64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 883 uint32_t dlfs_fsbpseg; /* 96: frags (fsb) per segment */ 884 uint32_t dlfs_inopb; /* 100: inodes per block */ 885 uint32_t dlfs_ifpb; /* 104: IFILE entries per block */ 886 uint32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 887 uint32_t dlfs_nindir; /* 112: indirect pointers per block */ 888 uint32_t dlfs_nseg; /* 116: number of segments */ 889 uint32_t dlfs_nspf; /* 120: number of sectors per fragment */ 890 uint32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 891 uint32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 892 uint32_t dlfs_segmask; /* 132: calculate offset within a segment */ 893 uint32_t dlfs_segshift; /* 136: fast mult/div for segments */ 894 uint32_t dlfs_bshift; /* 140: calc block number from file offset */ 895 uint32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 896 uint32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 897 uint64_t dlfs_bmask; /* 152: calc block offset from file offset */ 898 uint64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 899 uint64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 900 uint32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 901 uint32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 902 903 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 904 /* 188: superblock disk offsets */ 905 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 906 907 uint32_t dlfs_nclean; /* 228: Number of clean segments */ 908 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 909 uint16_t dlfs_pflags; /* 322: file system persistent flags */ 910 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 911 uint32_t dlfs_minfreeseg; /* 328: segments not counted in bfree */ 912 uint32_t dlfs_sumsize; /* 332: size of summary blocks */ 913 uint64_t dlfs_serial; /* 336: serial number */ 914 uint32_t dlfs_ibsize; /* 344: size of inode blocks */ 915 int32_t dlfs_s0addr; /* 348: start of segment 0 */ 916 uint64_t dlfs_tstamp; /* 352: time stamp */ 917 uint32_t dlfs_inodefmt; /* 360: inode format version */ 918 uint32_t dlfs_interleave; /* 364: segment interleave */ 919 uint32_t dlfs_ident; /* 368: per-fs identifier */ 920 uint32_t dlfs_fsbtodb; /* 372: fsbtodb and dbtodsb shift constant */ 921 uint32_t dlfs_resvseg; /* 376: segments reserved for the cleaner */ 922 int8_t dlfs_pad[128]; /* 380: round to 512 bytes */ 923 /* Checksum -- last valid disk field. */ 924 uint32_t dlfs_cksum; /* 508: checksum for superblock checking */ 925 }; 926 927 struct dlfs64 { 928 uint32_t dlfs_magic; /* 0: magic number */ 929 uint32_t dlfs_version; /* 4: version number (2) */ 930 931 uint64_t dlfs_size; /* 8: number of frags in fs (v2) */ 932 uint64_t dlfs_dsize; /* 16: number of disk blocks in fs */ 933 uint32_t dlfs_ssize; /* 24: number of bytes per segment (v2) */ 934 uint32_t dlfs_bsize; /* 28: file system block size */ 935 uint32_t dlfs_fsize; /* 32: size of frag blocks in fs */ 936 uint32_t dlfs_frag; /* 36: number of frags in a block in fs */ 937 938 /* Checkpoint region. */ 939 uint64_t dlfs_freehd; /* 40: start of the free inode list */ 940 uint64_t dlfs_nfiles; /* 48: number of allocated inodes */ 941 int64_t dlfs_bfree; /* 56: number of free frags */ 942 int64_t dlfs_avail; /* 64: blocks available for writing */ 943 int64_t dlfs_idaddr; /* 72: inode file disk address */ 944 int32_t dlfs_uinodes; /* 80: inodes in cache not yet on disk */ 945 uint32_t dlfs_unused_0; /* 84: not used */ 946 int64_t dlfs_lastseg; /* 88: address of last segment written */ 947 int64_t dlfs_nextseg; /* 96: address of next segment to write */ 948 int64_t dlfs_curseg; /* 104: current segment being written */ 949 int64_t dlfs_offset; /* 112: offset in curseg for next partial */ 950 int64_t dlfs_lastpseg; /* 120: address of last partial written */ 951 uint32_t dlfs_inopf; /* 128: inodes per frag */ 952 953 /* These are configuration parameters. */ 954 uint32_t dlfs_minfree; /* 132: minimum percentage of free blocks */ 955 956 /* These fields can be computed from the others. */ 957 uint64_t dlfs_maxfilesize; /* 136: maximum representable file size */ 958 uint32_t dlfs_fsbpseg; /* 144: frags (fsb) per segment */ 959 uint32_t dlfs_inopb; /* 148: inodes per block */ 960 uint32_t dlfs_ifpb; /* 152: IFILE entries per block */ 961 uint32_t dlfs_sepb; /* 156: SEGUSE entries per block */ 962 uint32_t dlfs_nindir; /* 160: indirect pointers per block */ 963 uint32_t dlfs_nseg; /* 164: number of segments */ 964 uint32_t dlfs_nspf; /* 168: number of sectors per fragment */ 965 uint32_t dlfs_cleansz; /* 172: cleaner info size in blocks */ 966 uint32_t dlfs_segtabsz; /* 176: segment table size in blocks */ 967 uint32_t dlfs_bshift; /* 180: calc block number from file offset */ 968 uint32_t dlfs_ffshift; /* 184: fast mult/div for frag from file */ 969 uint32_t dlfs_fbshift; /* 188: fast mult/div for frag from block */ 970 uint64_t dlfs_bmask; /* 192: calc block offset from file offset */ 971 uint64_t dlfs_ffmask; /* 200: calc frag offset from file offset */ 972 uint64_t dlfs_fbmask; /* 208: calc frag offset from block offset */ 973 uint32_t dlfs_blktodb; /* 216: blktodb and dbtoblk shift constant */ 974 uint32_t dlfs_sushift; /* 220: fast mult/div for segusage table */ 975 976 /* 224: superblock disk offsets */ 977 int64_t dlfs_sboffs[LFS_MAXNUMSB]; 978 979 int32_t dlfs_maxsymlinklen; /* 304: max len of an internal symlink */ 980 uint32_t dlfs_nclean; /* 308: Number of clean segments */ 981 u_char dlfs_fsmnt[MNAMELEN]; /* 312: name mounted on */ 982 uint16_t dlfs_pflags; /* 402: file system persistent flags */ 983 int32_t dlfs_dmeta; /* 404: total number of dirty summaries */ 984 uint32_t dlfs_minfreeseg; /* 408: segments not counted in bfree */ 985 uint32_t dlfs_sumsize; /* 412: size of summary blocks */ 986 uint32_t dlfs_ibsize; /* 416: size of inode blocks */ 987 uint32_t dlfs_inodefmt; /* 420: inode format version */ 988 uint64_t dlfs_serial; /* 424: serial number */ 989 int64_t dlfs_s0addr; /* 432: start of segment 0 */ 990 uint64_t dlfs_tstamp; /* 440: time stamp */ 991 uint32_t dlfs_interleave; /* 448: segment interleave */ 992 uint32_t dlfs_ident; /* 452: per-fs identifier */ 993 uint32_t dlfs_fsbtodb; /* 456: fsbtodb and dbtodsb shift constant */ 994 uint32_t dlfs_resvseg; /* 460: segments reserved for the cleaner */ 995 int8_t dlfs_pad[44]; /* 464: round to 512 bytes */ 996 /* Checksum -- last valid disk field. */ 997 uint32_t dlfs_cksum; /* 508: checksum for superblock checking */ 998 }; 999 1000 __CTASSERT(__alignof(struct dlfs) == __alignof(struct dlfs64)); 1001 1002 /* Type used for the inode bitmap */ 1003 typedef uint32_t lfs_bm_t; 1004 1005 /* 1006 * Linked list of segments whose byte count needs updating following a 1007 * file truncation. 1008 */ 1009 struct segdelta { 1010 long segnum; 1011 size_t num; 1012 LIST_ENTRY(segdelta) list; 1013 }; 1014 1015 /* 1016 * In-memory super block. 1017 */ 1018 struct lfs { 1019 union { /* on-disk parameters */ 1020 struct dlfs u_32; 1021 struct dlfs64 u_64; 1022 } lfs_dlfs_u; 1023 1024 /* These fields are set at mount time and are meaningless on disk. */ 1025 unsigned lfs_is64 : 1, /* are we lfs64 or lfs32? */ 1026 lfs_dobyteswap : 1, /* are we opposite-endian? */ 1027 lfs_hasolddirfmt : 1; /* dir entries have no d_type */ 1028 1029 struct segment *lfs_sp; /* current segment being written */ 1030 struct vnode *lfs_ivnode; /* vnode for the ifile */ 1031 uint32_t lfs_seglock; /* single-thread the segment writer */ 1032 pid_t lfs_lockpid; /* pid of lock holder */ 1033 lwpid_t lfs_locklwp; /* lwp of lock holder */ 1034 uint32_t lfs_iocount; /* number of ios pending */ 1035 uint32_t lfs_writer; /* don't allow any dirops to start */ 1036 uint32_t lfs_dirops; /* count of active directory ops */ 1037 kcondvar_t lfs_diropscv; /* condvar of active directory ops */ 1038 uint32_t lfs_dirvcount; /* count of VDIROP nodes in this fs */ 1039 uint32_t lfs_doifile; /* Write ifile blocks on next write */ 1040 uint32_t lfs_nactive; /* Number of segments since last ckp */ 1041 int8_t lfs_fmod; /* super block modified flag */ 1042 int8_t lfs_ronly; /* mounted read-only flag */ 1043 #define LFS_NOTYET 0x01 1044 #define LFS_IFDIRTY 0x02 1045 #define LFS_WARNED 0x04 1046 #define LFS_UNDIROP 0x08 1047 int8_t lfs_flags; /* currently unused flag */ 1048 uint16_t lfs_activesb; /* toggle between superblocks */ 1049 daddr_t lfs_sbactive; /* disk address of current sb write */ 1050 struct vnode *lfs_flushvp; /* vnode being flushed */ 1051 int lfs_flushvp_fakevref; /* fake vref count for flushvp */ 1052 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 1053 uint32_t lfs_diropwait; /* # procs waiting on dirop flush */ 1054 size_t lfs_devbsize; /* Device block size */ 1055 size_t lfs_devbshift; /* Device block shift */ 1056 krwlock_t lfs_fraglock; 1057 krwlock_t lfs_iflock; /* Ifile lock */ 1058 kcondvar_t lfs_stopcv; /* Wrap lock */ 1059 struct lwp *lfs_stoplwp; 1060 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 1061 int lfs_nadirop; /* number of active dirop nodes */ 1062 long lfs_ravail; /* blocks pre-reserved for writing */ 1063 long lfs_favail; /* blocks pre-reserved for writing */ 1064 struct lfs_res_blk *lfs_resblk; /* Reserved memory for pageout */ 1065 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 1066 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 1067 TAILQ_HEAD(, inode) lfs_cleanhd; /* vnodes being cleaned */ 1068 #define LFS_RESHASH_WIDTH 17 1069 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 1070 int lfs_pdflush; /* pagedaemon wants us to flush */ 1071 uint32_t **lfs_suflags; /* Segment use flags */ 1072 #ifdef _KERNEL 1073 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 1074 struct pool lfs_bpppool; /* Pool for bpp */ 1075 struct pool lfs_segpool; /* Pool for struct segment */ 1076 #endif /* _KERNEL */ 1077 #define LFS_MAX_CLEANIND 64 1078 daddr_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 1079 int lfs_cleanind; /* Index into intervals */ 1080 int lfs_sleepers; /* # procs sleeping this fs */ 1081 kcondvar_t lfs_sleeperscv; 1082 int lfs_pages; /* dirty pages blaming this fs */ 1083 lfs_bm_t *lfs_ino_bitmap; /* Inuse inodes bitmap */ 1084 int lfs_nowrap; /* Suspend log wrap */ 1085 int lfs_wrappass; /* Allow first log wrap requester to pass */ 1086 int lfs_wrapstatus; /* Wrap status */ 1087 int lfs_reclino; /* Inode being reclaimed */ 1088 daddr_t lfs_startseg; /* Segment we started writing at */ 1089 LIST_HEAD(, segdelta) lfs_segdhd; /* List of pending trunc accounting events */ 1090 1091 #ifdef _KERNEL 1092 /* The block device we're mounted on. */ 1093 dev_t lfs_dev; 1094 struct vnode *lfs_devvp; 1095 1096 /* ULFS-level information */ 1097 uint32_t um_flags; /* ULFS flags (below) */ 1098 u_long um_nindir; /* indirect ptrs per block */ 1099 u_long um_lognindir; /* log2 of um_nindir */ 1100 u_long um_bptrtodb; /* indir ptr to disk block */ 1101 u_long um_seqinc; /* inc between seq blocks */ 1102 int um_maxsymlinklen; 1103 int um_dirblksiz; 1104 uint64_t um_maxfilesize; 1105 1106 /* Stuff used by quota2 code, not currently operable */ 1107 unsigned lfs_use_quota2 : 1; 1108 uint32_t lfs_quota_magic; 1109 uint8_t lfs_quota_flags; 1110 uint64_t lfs_quotaino[2]; 1111 1112 /* Sleep address replacing &lfs_avail inside the on-disk superblock */ 1113 /* XXX: should be replaced with a condvar */ 1114 int lfs_availsleep; 1115 /* This one replaces &lfs_nextseg... all ditto */ 1116 kcondvar_t lfs_nextsegsleep; 1117 1118 /* Cleaner lwp, set on first bmapv syscall. */ 1119 struct lwp *lfs_cleaner_thread; 1120 1121 /* Hint from cleaner, only valid if curlwp == um_cleaner_thread. */ 1122 /* XXX change this to BLOCK_INFO after resorting this file */ 1123 struct block_info *lfs_cleaner_hint; 1124 1125 /* Cleaner lock */ 1126 struct lwp *lfs_cleanlock; /* LWP of lock holder */ 1127 kcondvar_t lfs_cleanercv; /* condvar for cleaner */ 1128 #endif 1129 }; 1130 1131 /* 1132 * Structures used to pass information between the partial-segment 1133 * parser and its callback functions. 1134 */ 1135 struct lfs_inofuncarg { 1136 struct lfs *fs; /* The filesystem */ 1137 daddr_t offset; /* Address of the inode block */ 1138 kauth_cred_t cred; /* Credential */ 1139 struct lwp *l; /* lwp */ 1140 char *buf; /* Pre-allocated buffer space */ 1141 void *arg; /* Opaque argument to function */ 1142 }; 1143 1144 struct lfs_finfofuncarg { 1145 struct lfs *fs; /* The file system */ 1146 FINFO *finfop; /* The FINFO structure to parse */ 1147 daddr_t *offsetp; /* Address of the first block */ 1148 kauth_cred_t cred; /* Credential */ 1149 struct lwp *l; /* lwp */ 1150 void *arg; /* Opaque argument to function */ 1151 }; 1152 1153 /* 1154 * Structures used by lfs_bmapv and lfs_markv to communicate information 1155 * about inodes and data blocks. 1156 */ 1157 typedef struct block_info { 1158 uint64_t bi_inode; /* inode # */ 1159 int64_t bi_lbn; /* logical block w/in file */ 1160 int64_t bi_daddr; /* disk address of block */ 1161 uint64_t bi_segcreate; /* origin segment create time */ 1162 int bi_version; /* file version number */ 1163 int bi_size; /* size of the block (if fragment) */ 1164 void *bi_bp; /* data buffer */ 1165 } BLOCK_INFO; 1166 1167 /* Compatibility for 7.0 binaries */ 1168 typedef struct block_info_70 { 1169 uint32_t bi_inode; /* inode # */ 1170 int32_t bi_lbn; /* logical block w/in file */ 1171 int32_t bi_daddr; /* disk address of block */ 1172 uint64_t bi_segcreate; /* origin segment create time */ 1173 int bi_version; /* file version number */ 1174 void *bi_bp; /* data buffer */ 1175 int bi_size; /* size of the block (if fragment) */ 1176 } BLOCK_INFO_70; 1177 1178 /* Compatibility for 1.5 binaries */ 1179 typedef struct block_info_15 { 1180 uint32_t bi_inode; /* inode # */ 1181 int32_t bi_lbn; /* logical block w/in file */ 1182 int32_t bi_daddr; /* disk address of block */ 1183 uint32_t bi_segcreate; /* origin segment create time */ 1184 int bi_version; /* file version number */ 1185 void *bi_bp; /* data buffer */ 1186 int bi_size; /* size of the block (if fragment) */ 1187 } BLOCK_INFO_15; 1188 1189 /* 1190 * 32/64-bit-clean pointer to block pointers. This points into 1191 * already-existing storage; it is mostly used to access the block 1192 * pointers following a FINFO. 1193 */ 1194 union lfs_blocks { 1195 int64_t *b64; 1196 int32_t *b32; 1197 }; 1198 1199 /* In-memory description of a segment about to be written. */ 1200 struct segment { 1201 struct lfs *fs; /* file system pointer */ 1202 struct buf **bpp; /* pointer to buffer array */ 1203 struct buf **cbpp; /* pointer to next available bp */ 1204 struct buf **start_bpp; /* pointer to first bp in this set */ 1205 struct buf *ibp; /* buffer pointer to inode page */ 1206 union lfs_dinode *idp; /* pointer to ifile dinode */ 1207 FINFO *fip; /* current fileinfo pointer */ 1208 struct vnode *vp; /* vnode being gathered */ 1209 void *segsum; /* segment summary info */ 1210 uint32_t ninodes; /* number of inodes in this segment */ 1211 int32_t seg_bytes_left; /* bytes left in segment */ 1212 int32_t sum_bytes_left; /* bytes left in summary block */ 1213 uint32_t seg_number; /* number of this segment */ 1214 union lfs_blocks start_lbp; /* beginning lbn for this set */ 1215 1216 #define SEGM_CKP 0x0001 /* doing a checkpoint */ 1217 #define SEGM_CLEAN 0x0002 /* cleaner call; don't sort */ 1218 #define SEGM_SYNC 0x0004 /* wait for segment */ 1219 #define SEGM_PROT 0x0008 /* don't inactivate at segunlock */ 1220 #define SEGM_PAGEDAEMON 0x0010 /* pagedaemon called us */ 1221 #define SEGM_WRITERD 0x0020 /* LFS writed called us */ 1222 #define SEGM_FORCE_CKP 0x0040 /* Force checkpoint right away */ 1223 #define SEGM_RECLAIM 0x0080 /* Writing to reclaim vnode */ 1224 #define SEGM_SINGLE 0x0100 /* Opportunistic writevnodes */ 1225 uint16_t seg_flags; /* run-time flags for this segment */ 1226 uint32_t seg_iocount; /* number of ios pending */ 1227 int ndupino; /* number of duplicate inodes */ 1228 uint64_t bytes_written; /* bytes written */ 1229 int gatherblock_loopcount; /* lfs_gatherblock loop count */ 1230 }; 1231 1232 /* Statistics Counters */ 1233 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1234 u_int segsused; 1235 u_int psegwrites; 1236 u_int psyncwrites; 1237 u_int pcleanwrites; 1238 u_int blocktot; 1239 u_int cleanblocks; 1240 u_int ncheckpoints; 1241 u_int nwrites; 1242 u_int nsync_writes; 1243 u_int wait_exceeded; 1244 u_int write_exceeded; 1245 u_int flush_invoked; 1246 u_int vflush_invoked; 1247 u_int clean_inlocked; 1248 u_int clean_vnlocked; 1249 u_int segs_reclaimed; 1250 }; 1251 1252 struct lfs_write_stats { 1253 int direct; /* Direct data written, in frags */ 1254 int offset; /* Offset difference, in frags */ 1255 }; 1256 #define LFCNRECLAIM _FCNW_FSPRIV('L', 4, struct lfs_write_stats) 1257 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1258 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1259 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1260 #define LFCNWRAPSTOP _FCNR_FSPRIV('L', 9, int) 1261 #define LFCNWRAPGO _FCNR_FSPRIV('L', 10, int) 1262 struct lfs_fhandle { 1263 char space[28]; /* FHANDLE_SIZE_COMPAT (but used from userland too) */ 1264 }; 1265 #define LFCNIFILEFH _FCNW_FSPRIV('L', 11, struct lfs_fhandle) 1266 #define LFCNWRAPPASS _FCNR_FSPRIV('L', 12, int) 1267 # define LFS_WRAP_GOING 0x0 1268 # define LFS_WRAP_WAITING 0x1 1269 #define LFCNWRAPSTATUS _FCNW_FSPRIV('L', 13, int) 1270 1271 /* Fcntls to take the place of the lfs syscalls */ 1272 struct lfs_fcntl_markv { 1273 BLOCK_INFO *blkiov; /* blocks to relocate */ 1274 int blkcnt; /* number of blocks (limited to 65536) */ 1275 }; 1276 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 14, struct timeval) 1277 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 15, struct timeval) 1278 #define LFCNBMAPV _FCNRW_FSPRIV('L', 16, struct lfs_fcntl_markv) 1279 #define LFCNMARKV _FCNRW_FSPRIV('L', 17, struct lfs_fcntl_markv) 1280 1281 /* File defragmentation */ 1282 struct lfs_filestats { 1283 ino_t ino; /* Inode number */ 1284 off_t nblk; /* Block count */ 1285 int dc_count; /* Count of discontinuities (gaps) */ 1286 daddr_t dc_sum; /* Sum of absolute values of gap lengths */ 1287 }; 1288 struct lfs_filestat_req { 1289 ino_t ino; /* Starting inode number */ 1290 int len; /* Number of inodes to report */ 1291 struct lfs_filestats *fss; /* Out: file statistics */ 1292 }; 1293 #define LFCNFILESTATS _FCNRW_FSPRIV('L', 18, struct lfs_filestat_req) 1294 struct lfs_inode_array { 1295 size_t len; /* Length of array */ 1296 ino_t *inodes; /* Array of inode numbers */ 1297 struct lfs_write_stats stats; /* Out: frags written */ 1298 }; 1299 #define LFCNREWRITEFILE _FCNRW_FSPRIV('L', 19, struct lfs_inode_array) 1300 #define LFCNSCRAMBLE _FCNRW_FSPRIV('L', 20, struct lfs_inode_array) 1301 /* Segment rewriting */ 1302 struct lfs_segnum_array { 1303 size_t len; /* Length of array */ 1304 int *segments; /* Array of segment numbers */ 1305 struct lfs_write_stats stats; /* Out: frags written */ 1306 }; 1307 #define LFCNREWRITESEGS _FCNRW_FSPRIV('L', 21, struct lfs_segnum_array) 1308 #define LFCNCLEANERINFO _FCNW_FSPRIV('L', 22, CLEANERINFO64) 1309 struct lfs_seguse_array { 1310 int start; /* Start segment */ 1311 size_t len; /* Length of the array */ 1312 SEGUSE *seguse; /* Out: array of seguse */ 1313 }; 1314 #define LFCNSEGUSE _FCNRW_FSPRIV('L', 23, struct lfs_seguse_array) 1315 1316 /* Debug segment lock */ 1317 #ifdef notyet 1318 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1319 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1320 # define ASSERT_DUNNO_SEGLOCK(fs) 1321 # define ASSERT_MAYBE_SEGLOCK(fs) 1322 #else /* !notyet */ 1323 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1324 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1325 LFS_SEGLOCK_HELD(fs))) 1326 # define ASSERT_SEGLOCK(fs) do { \ 1327 if (!LFS_SEGLOCK_HELD(fs)) { \ 1328 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1329 } \ 1330 } while(0) 1331 # define ASSERT_NO_SEGLOCK(fs) do { \ 1332 if (LFS_SEGLOCK_HELD(fs)) { \ 1333 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1334 } \ 1335 } while(0) 1336 # define ASSERT_MAYBE_SEGLOCK(x) 1337 #endif /* !notyet */ 1338 1339 /* 1340 * Arguments to mount LFS filesystems 1341 */ 1342 struct ulfs_args { 1343 char *fspec; /* block special device to mount */ 1344 }; 1345 1346 __BEGIN_DECLS 1347 void lfs_itimes(struct inode *, const struct timespec *, 1348 const struct timespec *, const struct timespec *); 1349 __END_DECLS 1350 1351 #endif /* !_UFS_LFS_LFS_H_ */ 1352
Indexes created Mon Oct 20 20:10:13 GMT 2025