lfs.h revision 1.150
1 /* $NetBSD: lfs.h,v 1.150 2013/06/08 22:23:52 dholland Exp $ */ 2 3 /* from NetBSD: dinode.h,v 1.22 2013/01/22 09:39:18 dholland Exp */ 4 /* from NetBSD: dir.h,v 1.21 2009/07/22 04:49:19 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 #include <sys/rwlock.h> 152 #include <sys/mutex.h> 153 #include <sys/queue.h> 154 #include <sys/condvar.h> 155 #include <sys/mount.h> 156 #include <sys/pool.h> 157 158 /* 159 * Compile-time options for LFS. 160 */ 161 #define LFS_IFIND_RETRIES 16 162 #define LFS_LOGLENGTH 1024 /* size of debugging log */ 163 #define LFS_MAX_ACTIVE 10 /* Dirty segments before ckp forced */ 164 165 /* 166 * Fixed filesystem layout parameters 167 */ 168 #define LFS_LABELPAD 8192 /* LFS label size */ 169 #define LFS_SBPAD 8192 /* LFS superblock size */ 170 171 #define LFS_UNUSED_INUM 0 /* 0: out of band inode number */ 172 #define LFS_IFILE_INUM 1 /* 1: IFILE inode number */ 173 /* 2: Root inode number */ 174 #define LFS_LOSTFOUNDINO 3 /* 3: lost+found inode number */ 175 #define LFS_FIRST_INUM 4 /* 4: first free inode number */ 176 177 /* 178 * The root inode is the root of the file system. Inode 0 can't be used for 179 * normal purposes and historically bad blocks were linked to inode 1, thus 180 * the root inode is 2. (Inode 1 is no longer used for this purpose, however 181 * numerous dump tapes make this assumption, so we are stuck with it). 182 */ 183 #define ULFS_ROOTINO ((ino_t)2) 184 185 /* 186 * The Whiteout inode# is a dummy non-zero inode number which will 187 * never be allocated to a real file. It is used as a place holder 188 * in the directory entry which has been tagged as a LFS_DT_WHT entry. 189 * See the comments about ULFS_ROOTINO above. 190 */ 191 #define ULFS_WINO ((ino_t)1) 192 193 194 #define LFS_V1_SUMMARY_SIZE 512 /* V1 fixed summary size */ 195 #define LFS_DFL_SUMMARY_SIZE 512 /* Default summary size */ 196 197 #define LFS_MAX_DADDR 0x7fffffff /* Highest addressable fsb */ 198 199 #define LFS_MAXNAMLEN 255 /* maximum name length in a dir */ 200 201 #define ULFS_NXADDR 2 202 #define ULFS_NDADDR 12 /* Direct addresses in inode. */ 203 #define ULFS_NIADDR 3 /* Indirect addresses in inode. */ 204 205 /* 206 * Adjustable filesystem parameters 207 */ 208 #define MIN_FREE_SEGS 20 209 #define MIN_RESV_SEGS 15 210 #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() */ 214 215 /* Misc. definitions */ 216 #define BW_CLEAN 1 /* Flag for lfs_bwrite_ext() */ 217 #define PG_DELWRI PG_PAGER1 /* Local def for delayed pageout */ 218 219 /* Resource limits */ 220 #define LFS_MAX_RESOURCE(x, u) (((x) >> 2) - 10 * (u)) 221 #define LFS_WAIT_RESOURCE(x, u) (((x) >> 1) - ((x) >> 3) - 10 * (u)) 222 #define LFS_INVERSE_MAX_RESOURCE(x, u) (((x) + 10 * (u)) << 2) 223 #define LFS_MAX_BUFS LFS_MAX_RESOURCE(nbuf, 1) 224 #define LFS_WAIT_BUFS LFS_WAIT_RESOURCE(nbuf, 1) 225 #define LFS_INVERSE_MAX_BUFS(n) LFS_INVERSE_MAX_RESOURCE(n, 1) 226 #define LFS_MAX_BYTES LFS_MAX_RESOURCE(bufmem_lowater, PAGE_SIZE) 227 #define LFS_INVERSE_MAX_BYTES(n) LFS_INVERSE_MAX_RESOURCE(n, PAGE_SIZE) 228 #define LFS_WAIT_BYTES LFS_WAIT_RESOURCE(bufmem_lowater, PAGE_SIZE) 229 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3)) 230 #define SIZEOF_DIROP(fs) (2 * ((fs)->lfs_bsize + LFS_DINODE1_SIZE)) 231 #define LFS_MAX_FSDIROP(fs) \ 232 ((fs)->lfs_nclean <= (fs)->lfs_resvseg ? 0 : \ 233 (((fs)->lfs_nclean - (fs)->lfs_resvseg) * (fs)->lfs_ssize) / \ 234 (2 * SIZEOF_DIROP(fs))) 235 #define LFS_MAX_PAGES lfs_max_pages() 236 #define LFS_WAIT_PAGES lfs_wait_pages() 237 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */ 238 239 #ifdef _KERNEL 240 int lfs_wait_pages(void); 241 int lfs_max_pages(void); 242 #endif /* _KERNEL */ 243 244 /* How starved can we be before we start holding back page writes */ 245 #define LFS_STARVED_FOR_SEGS(fs) ((fs)->lfs_nclean < (fs)->lfs_resvseg) 246 247 /* 248 * Reserved blocks for lfs_malloc 249 */ 250 251 /* Structure to keep reserved blocks */ 252 typedef struct lfs_res_blk { 253 void *p; 254 LIST_ENTRY(lfs_res_blk) res; 255 int size; 256 char inuse; 257 } res_t; 258 259 /* Types for lfs_newbuf and lfs_malloc */ 260 #define LFS_NB_UNKNOWN -1 261 #define LFS_NB_SUMMARY 0 262 #define LFS_NB_SBLOCK 1 263 #define LFS_NB_IBLOCK 2 264 #define LFS_NB_CLUSTER 3 265 #define LFS_NB_CLEAN 4 266 #define LFS_NB_BLKIOV 5 267 #define LFS_NB_COUNT 6 /* always last */ 268 269 /* Number of reserved memory blocks of each type */ 270 #define LFS_N_SUMMARIES 2 271 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 272 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 273 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 274 #define LFS_N_CLEAN 0 275 #define LFS_N_BLKIOV 1 276 277 /* Total count of "large" (non-pool) types */ 278 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 279 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV) 280 281 /* Counts for pool types */ 282 #define LFS_N_CL LFS_N_CLUSTERS 283 #define LFS_N_BPP 2 284 #define LFS_N_SEG 2 285 286 /* 287 * Directories 288 */ 289 290 /* 291 * A directory consists of some number of blocks of LFS_DIRBLKSIZ 292 * bytes, where LFS_DIRBLKSIZ is chosen such that it can be transferred 293 * to disk in a single atomic operation (e.g. 512 bytes on most machines). 294 * 295 * Each LFS_DIRBLKSIZ byte block contains some number of directory entry 296 * structures, which are of variable length. Each directory entry has 297 * a struct lfs_direct at the front of it, containing its inode number, 298 * the length of the entry, and the length of the name contained in 299 * the entry. These are followed by the name padded to a 4 byte boundary. 300 * All names are guaranteed null terminated. 301 * The maximum length of a name in a directory is LFS_MAXNAMLEN. 302 * 303 * The macro DIRSIZ(fmt, dp) gives the amount of space required to represent 304 * a directory entry. Free space in a directory is represented by 305 * entries which have dp->d_reclen > DIRSIZ(fmt, dp). All LFS_DIRBLKSIZ bytes 306 * in a directory block are claimed by the directory entries. This 307 * usually results in the last entry in a directory having a large 308 * dp->d_reclen. When entries are deleted from a directory, the 309 * space is returned to the previous entry in the same directory 310 * block by increasing its dp->d_reclen. If the first entry of 311 * a directory block is free, then its dp->d_ino is set to 0. 312 * Entries other than the first in a directory do not normally have 313 * dp->d_ino set to 0. 314 */ 315 316 /* 317 * Directory block size. 318 */ 319 #undef LFS_DIRBLKSIZ 320 #define LFS_DIRBLKSIZ DEV_BSIZE 321 322 /* 323 * Convert between stat structure types and directory types. 324 */ 325 #define LFS_IFTODT(mode) (((mode) & 0170000) >> 12) 326 #define LFS_DTTOIF(dirtype) ((dirtype) << 12) 327 328 /* 329 * The LFS_DIRSIZ macro gives the minimum record length which will hold 330 * the directory entry. This requires the amount of space in struct lfs_direct 331 * without the d_name field, plus enough space for the name with a terminating 332 * null byte (dp->d_namlen+1), rounded up to a 4 byte boundary. 333 */ 334 #define LFS_DIRECTSIZ(namlen) \ 335 ((sizeof(struct lfs_direct) - (LFS_MAXNAMLEN+1)) + (((namlen)+1 + 3) &~ 3)) 336 337 #if (BYTE_ORDER == LITTLE_ENDIAN) 338 #define LFS_DIRSIZ(oldfmt, dp, needswap) \ 339 (((oldfmt) && !(needswap)) ? \ 340 LFS_DIRECTSIZ((dp)->d_type) : LFS_DIRECTSIZ((dp)->d_namlen)) 341 #else 342 #define LFS_DIRSIZ(oldfmt, dp, needswap) \ 343 (((oldfmt) && (needswap)) ? \ 344 LFS_DIRECTSIZ((dp)->d_type) : LFS_DIRECTSIZ((dp)->d_namlen)) 345 #endif 346 347 /* Constants for the first argument of LFS_DIRSIZ */ 348 #define LFS_OLDDIRFMT 1 349 #define LFS_NEWDIRFMT 0 350 351 /* 352 * Theoretically, directories can be more than 2Gb in length; however, in 353 * practice this seems unlikely. So, we define the type doff_t as a 32-bit 354 * quantity to keep down the cost of doing lookup on a 32-bit machine. 355 */ 356 #define doff_t int32_t 357 #define lfs_doff_t int32_t 358 #define MAXDIRSIZE (0x7fffffff) 359 360 #define LFS_MAXNAMLEN 255 361 362 /* 363 * File types for d_type 364 */ 365 #define LFS_DT_UNKNOWN 0 366 #define LFS_DT_FIFO 1 367 #define LFS_DT_CHR 2 368 #define LFS_DT_DIR 4 369 #define LFS_DT_BLK 6 370 #define LFS_DT_REG 8 371 #define LFS_DT_LNK 10 372 #define LFS_DT_SOCK 12 373 #define LFS_DT_WHT 14 374 375 /* 376 * (See notes above) 377 */ 378 #define d_ino d_fileno 379 struct lfs_direct { 380 u_int32_t d_fileno; /* inode number of entry */ 381 u_int16_t d_reclen; /* length of this record */ 382 u_int8_t d_type; /* file type, see below */ 383 u_int8_t d_namlen; /* length of string in d_name */ 384 char d_name[LFS_MAXNAMLEN + 1];/* name with length <= LFS_MAXNAMLEN */ 385 }; 386 387 /* 388 * Template for manipulating directories. Should use struct lfs_direct's, 389 * but the name field is LFS_MAXNAMLEN - 1, and this just won't do. 390 */ 391 struct lfs_dirtemplate { 392 u_int32_t dot_ino; 393 int16_t dot_reclen; 394 u_int8_t dot_type; 395 u_int8_t dot_namlen; 396 char dot_name[4]; /* must be multiple of 4 */ 397 u_int32_t dotdot_ino; 398 int16_t dotdot_reclen; 399 u_int8_t dotdot_type; 400 u_int8_t dotdot_namlen; 401 char dotdot_name[4]; /* ditto */ 402 }; 403 404 /* 405 * This is the old format of directories, sans type element. 406 */ 407 struct lfs_odirtemplate { 408 u_int32_t dot_ino; 409 int16_t dot_reclen; 410 u_int16_t dot_namlen; 411 char dot_name[4]; /* must be multiple of 4 */ 412 u_int32_t dotdot_ino; 413 int16_t dotdot_reclen; 414 u_int16_t dotdot_namlen; 415 char dotdot_name[4]; /* ditto */ 416 }; 417 418 /* 419 * Inodes 420 */ 421 422 /* 423 * A dinode contains all the meta-data associated with a ULFS file. 424 * This structure defines the on-disk format of a dinode. Since 425 * this structure describes an on-disk structure, all its fields 426 * are defined by types with precise widths. 427 */ 428 429 struct ulfs1_dinode { 430 u_int16_t di_mode; /* 0: IFMT, permissions; see below. */ 431 int16_t di_nlink; /* 2: File link count. */ 432 union { 433 u_int16_t oldids[2]; /* 4: Ffs: old user and group ids. */ 434 u_int32_t inumber; /* 4: Lfs: inode number. */ 435 } di_u; 436 u_int64_t di_size; /* 8: File byte count. */ 437 int32_t di_atime; /* 16: Last access time. */ 438 int32_t di_atimensec; /* 20: Last access time. */ 439 int32_t di_mtime; /* 24: Last modified time. */ 440 int32_t di_mtimensec; /* 28: Last modified time. */ 441 int32_t di_ctime; /* 32: Last inode change time. */ 442 int32_t di_ctimensec; /* 36: Last inode change time. */ 443 int32_t di_db[ULFS_NDADDR]; /* 40: Direct disk blocks. */ 444 int32_t di_ib[ULFS_NIADDR]; /* 88: Indirect disk blocks. */ 445 u_int32_t di_flags; /* 100: Status flags (chflags). */ 446 u_int32_t di_blocks; /* 104: Blocks actually held. */ 447 int32_t di_gen; /* 108: Generation number. */ 448 u_int32_t di_uid; /* 112: File owner. */ 449 u_int32_t di_gid; /* 116: File group. */ 450 u_int64_t di_modrev; /* 120: i_modrev for NFSv4 */ 451 }; 452 453 struct ulfs2_dinode { 454 u_int16_t di_mode; /* 0: IFMT, permissions; see below. */ 455 int16_t di_nlink; /* 2: File link count. */ 456 u_int32_t di_uid; /* 4: File owner. */ 457 u_int32_t di_gid; /* 8: File group. */ 458 u_int32_t di_blksize; /* 12: Inode blocksize. */ 459 u_int64_t di_size; /* 16: File byte count. */ 460 u_int64_t di_blocks; /* 24: Bytes actually held. */ 461 int64_t di_atime; /* 32: Last access time. */ 462 int64_t di_mtime; /* 40: Last modified time. */ 463 int64_t di_ctime; /* 48: Last inode change time. */ 464 int64_t di_birthtime; /* 56: Inode creation time. */ 465 int32_t di_mtimensec; /* 64: Last modified time. */ 466 int32_t di_atimensec; /* 68: Last access time. */ 467 int32_t di_ctimensec; /* 72: Last inode change time. */ 468 int32_t di_birthnsec; /* 76: Inode creation time. */ 469 int32_t di_gen; /* 80: Generation number. */ 470 u_int32_t di_kernflags; /* 84: Kernel flags. */ 471 u_int32_t di_flags; /* 88: Status flags (chflags). */ 472 int32_t di_extsize; /* 92: External attributes block. */ 473 int64_t di_extb[ULFS_NXADDR];/* 96: External attributes block. */ 474 int64_t di_db[ULFS_NDADDR]; /* 112: Direct disk blocks. */ 475 int64_t di_ib[ULFS_NIADDR]; /* 208: Indirect disk blocks. */ 476 u_int64_t di_modrev; /* 232: i_modrev for NFSv4 */ 477 int64_t di_spare[2]; /* 240: Reserved; currently unused */ 478 }; 479 480 /* 481 * The di_db fields may be overlaid with other information for 482 * file types that do not have associated disk storage. Block 483 * and character devices overlay the first data block with their 484 * dev_t value. Short symbolic links place their path in the 485 * di_db area. 486 */ 487 #define di_inumber di_u.inumber 488 #define di_ogid di_u.oldids[1] 489 #define di_ouid di_u.oldids[0] 490 #define di_rdev di_db[0] 491 492 /* Size of the on-disk inode. */ 493 #define LFS_DINODE1_SIZE (sizeof(struct ulfs1_dinode)) /* 128 */ 494 #define LFS_DINODE2_SIZE (sizeof(struct ulfs2_dinode)) 495 496 /* File types, found in the upper bits of di_mode. */ 497 #define LFS_IFMT 0170000 /* Mask of file type. */ 498 #define LFS_IFIFO 0010000 /* Named pipe (fifo). */ 499 #define LFS_IFCHR 0020000 /* Character device. */ 500 #define LFS_IFDIR 0040000 /* Directory file. */ 501 #define LFS_IFBLK 0060000 /* Block device. */ 502 #define LFS_IFREG 0100000 /* Regular file. */ 503 #define LFS_IFLNK 0120000 /* Symbolic link. */ 504 #define LFS_IFSOCK 0140000 /* UNIX domain socket. */ 505 #define LFS_IFWHT 0160000 /* Whiteout. */ 506 507 /* 508 * Maximum length of a symlink that can be stored within the inode. 509 */ 510 #define ULFS1_MAXSYMLINKLEN ((ULFS_NDADDR + ULFS_NIADDR) * sizeof(int32_t)) 511 #define ULFS2_MAXSYMLINKLEN ((ULFS_NDADDR + ULFS_NIADDR) * sizeof(int64_t)) 512 513 #define ULFS_MAXSYMLINKLEN(ip) \ 514 ((ip)->i_ump->um_fstype == ULFS1) ? \ 515 ULFS1_MAXSYMLINKLEN : ULFS2_MAXSYMLINKLEN 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 /* Determine if a buffer belongs to the ifile */ 529 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 530 531 # define LFS_LOCK_BUF(bp) do { \ 532 if (((bp)->b_flags & B_LOCKED) == 0 && bp->b_iodone == NULL) { \ 533 mutex_enter(&lfs_lock); \ 534 ++locked_queue_count; \ 535 locked_queue_bytes += bp->b_bufsize; \ 536 mutex_exit(&lfs_lock); \ 537 } \ 538 (bp)->b_flags |= B_LOCKED; \ 539 } while (0) 540 541 # define LFS_UNLOCK_BUF(bp) do { \ 542 if (((bp)->b_flags & B_LOCKED) != 0 && bp->b_iodone == NULL) { \ 543 mutex_enter(&lfs_lock); \ 544 --locked_queue_count; \ 545 locked_queue_bytes -= bp->b_bufsize; \ 546 if (locked_queue_count < LFS_WAIT_BUFS && \ 547 locked_queue_bytes < LFS_WAIT_BYTES) \ 548 cv_broadcast(&locked_queue_cv); \ 549 mutex_exit(&lfs_lock); \ 550 } \ 551 (bp)->b_flags &= ~B_LOCKED; \ 552 } while (0) 553 554 #ifdef _KERNEL 555 556 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 557 558 # define LFS_IS_MALLOC_BUF(bp) ((bp)->b_iodone == lfs_callback) 559 560 # ifdef DEBUG 561 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 562 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 563 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 564 cv_broadcast(&locked_queue_cv); \ 565 } \ 566 } while (0) 567 # else 568 # define LFS_DEBUG_COUNTLOCKED(m) 569 # endif 570 571 /* log for debugging writes to the Ifile */ 572 # ifdef DEBUG 573 struct lfs_log_entry { 574 const char *op; 575 const char *file; 576 int pid; 577 int line; 578 daddr_t block; 579 unsigned long flags; 580 }; 581 extern int lfs_lognum; 582 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 583 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 584 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 585 int _s; \ 586 \ 587 mutex_enter(&lfs_lock); \ 588 _s = splbio(); \ 589 lfs_log[lfs_lognum].op = theop; \ 590 lfs_log[lfs_lognum].file = thefile; \ 591 lfs_log[lfs_lognum].line = (theline); \ 592 lfs_log[lfs_lognum].pid = (thepid); \ 593 lfs_log[lfs_lognum].block = (lbn); \ 594 lfs_log[lfs_lognum].flags = (theflags); \ 595 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 596 splx(_s); \ 597 mutex_exit(&lfs_lock); \ 598 } while (0) 599 600 # define LFS_BCLEAN_LOG(fs, bp) do { \ 601 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 602 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 603 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 604 } while (0) 605 606 /* Must match list in lfs_vfsops.c ! */ 607 # define DLOG_RF 0 /* roll forward */ 608 # define DLOG_ALLOC 1 /* inode alloc */ 609 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 610 # define DLOG_FLUSH 3 /* flush */ 611 # define DLOG_LLIST 4 /* locked list accounting */ 612 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 613 # define DLOG_VNODE 6 /* vflush/writevnodes */ 614 # define DLOG_SEG 7 /* segwrite */ 615 # define DLOG_SU 8 /* seguse accounting */ 616 # define DLOG_CLEAN 9 /* cleaner routines */ 617 # define DLOG_MOUNT 10 /* mount/unmount */ 618 # define DLOG_PAGE 11 /* putpages/gop_write */ 619 # define DLOG_DIROP 12 /* dirop accounting */ 620 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 621 # define DLOG_MAX 14 /* The terminator */ 622 # define DLOG(a) lfs_debug_log a 623 # else /* ! DEBUG */ 624 # define LFS_BCLEAN_LOG(fs, bp) 625 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)->b_vp, (bp)) 626 # define DLOG(a) 627 # endif /* ! DEBUG */ 628 #else /* ! _KERNEL */ 629 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 630 #endif /* _KERNEL */ 631 632 #ifdef _KERNEL 633 /* This overlays the fid structure (see fstypes.h). */ 634 struct ulfs_ufid { 635 u_int16_t ufid_len; /* Length of structure. */ 636 u_int16_t ufid_pad; /* Force 32-bit alignment. */ 637 u_int32_t ufid_ino; /* File number (ino). */ 638 int32_t ufid_gen; /* Generation number. */ 639 }; 640 /* Filehandle structure for exported LFSes */ 641 struct lfid { 642 struct ulfs_ufid lfid_ufid; 643 #define lfid_len lfid_ufid.ufid_len 644 #define lfid_ino lfid_ufid.ufid_ino 645 #define lfid_gen lfid_ufid.ufid_gen 646 uint32_t lfid_ident; 647 }; 648 #endif /* _KERNEL */ 649 650 /* 651 * "struct inode" associated definitions 652 */ 653 654 /* Address calculations for metadata located in the inode */ 655 #define S_INDIR(fs) -ULFS_NDADDR 656 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 657 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 658 659 /* For convenience */ 660 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 661 662 #define LFS_SET_UINO(ip, flags) do { \ 663 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 664 ++(ip)->i_lfs->lfs_uinodes; \ 665 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 666 ++(ip)->i_lfs->lfs_uinodes; \ 667 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 668 ++(ip)->i_lfs->lfs_uinodes; \ 669 (ip)->i_flag |= (flags); \ 670 } while (0) 671 672 #define LFS_CLR_UINO(ip, flags) do { \ 673 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 674 --(ip)->i_lfs->lfs_uinodes; \ 675 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 676 --(ip)->i_lfs->lfs_uinodes; \ 677 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 678 --(ip)->i_lfs->lfs_uinodes; \ 679 (ip)->i_flag &= ~(flags); \ 680 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 681 panic("lfs_uinodes < 0"); \ 682 } \ 683 } while (0) 684 685 #define LFS_ITIMES(ip, acc, mod, cre) \ 686 while ((ip)->i_flag & (IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY)) \ 687 lfs_itimes(ip, acc, mod, cre) 688 689 /* 690 * "struct vnode" associated definitions 691 */ 692 693 /* Heuristic emptiness measure */ 694 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 695 !(vp->v_type == VREG && (vp)->v_iflag & VI_ONWORKLST) &&\ 696 VTOI(vp)->i_lfs_nbtree == 0) 697 698 #define WRITEINPROG(vp) ((vp)->v_numoutput > 0 || \ 699 (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 700 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING)))) 701 702 703 /* 704 * On-disk and in-memory checkpoint segment usage structure. 705 */ 706 typedef struct segusage SEGUSE; 707 struct segusage { 708 u_int32_t su_nbytes; /* 0: number of live bytes */ 709 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 710 u_int16_t su_nsums; /* 8: number of summaries in segment */ 711 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 712 713 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 714 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 715 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 716 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 717 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 718 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 719 u_int32_t su_flags; /* 12: segment flags */ 720 u_int64_t su_lastmod; /* 16: last modified timestamp */ 721 }; 722 723 typedef struct segusage_v1 SEGUSE_V1; 724 struct segusage_v1 { 725 u_int32_t su_nbytes; /* 0: number of live bytes */ 726 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 727 u_int16_t su_nsums; /* 8: number of summaries in segment */ 728 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 729 u_int32_t su_flags; /* 12: segment flags */ 730 }; 731 732 #define SEGUPB(fs) (fs->lfs_sepb) 733 #define SEGTABSIZE_SU(fs) \ 734 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 735 736 #ifdef _KERNEL 737 # define SHARE_IFLOCK(F) \ 738 do { \ 739 rw_enter(&(F)->lfs_iflock, RW_READER); \ 740 } while(0) 741 # define UNSHARE_IFLOCK(F) \ 742 do { \ 743 rw_exit(&(F)->lfs_iflock); \ 744 } while(0) 745 #else /* ! _KERNEL */ 746 # define SHARE_IFLOCK(F) 747 # define UNSHARE_IFLOCK(F) 748 #endif /* ! _KERNEL */ 749 750 /* Read in the block with a specific segment usage entry from the ifile. */ 751 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 752 int _e; \ 753 SHARE_IFLOCK(F); \ 754 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 755 if ((_e = bread((F)->lfs_ivnode, \ 756 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 757 (F)->lfs_bsize, NOCRED, 0, &(BP))) != 0) \ 758 panic("lfs: ifile read: %d", _e); \ 759 if ((F)->lfs_version == 1) \ 760 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 761 ((IN) & ((F)->lfs_sepb - 1))); \ 762 else \ 763 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 764 UNSHARE_IFLOCK(F); \ 765 } while (0) 766 767 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 768 if ((SP)->su_nbytes == 0) \ 769 (SP)->su_flags |= SEGUSE_EMPTY; \ 770 else \ 771 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 772 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 773 LFS_BWRITE_LOG(BP); \ 774 } while (0) 775 776 /* 777 * On-disk file information. One per file with data blocks in the segment. 778 */ 779 typedef struct finfo FINFO; 780 struct finfo { 781 u_int32_t fi_nblocks; /* number of blocks */ 782 u_int32_t fi_version; /* version number */ 783 u_int32_t fi_ino; /* inode number */ 784 u_int32_t fi_lastlength; /* length of last block in array */ 785 int32_t fi_blocks[1]; /* array of logical block numbers */ 786 }; 787 /* sizeof FINFO except fi_blocks */ 788 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 789 790 /* 791 * Index file inode entries. 792 */ 793 typedef struct ifile IFILE; 794 struct ifile { 795 u_int32_t if_version; /* inode version number */ 796 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 797 int32_t if_daddr; /* inode disk address */ 798 #define LFS_ORPHAN_NEXTFREE (~(u_int32_t)0) /* indicate orphaned file */ 799 u_int32_t if_nextfree; /* next-unallocated inode */ 800 u_int32_t if_atime_sec; /* Last access time, seconds */ 801 u_int32_t if_atime_nsec; /* and nanoseconds */ 802 }; 803 804 typedef struct ifile_v1 IFILE_V1; 805 struct ifile_v1 { 806 u_int32_t if_version; /* inode version number */ 807 int32_t if_daddr; /* inode disk address */ 808 u_int32_t if_nextfree; /* next-unallocated inode */ 809 #if LFS_ATIME_IFILE 810 struct timespec if_atime; /* Last access time */ 811 #endif 812 }; 813 814 /* 815 * LFSv1 compatibility code is not allowed to touch if_atime, since it 816 * may not be mapped! 817 */ 818 /* Read in the block with a specific inode from the ifile. */ 819 #define LFS_IENTRY(IP, F, IN, BP) do { \ 820 int _e; \ 821 SHARE_IFLOCK(F); \ 822 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 823 if ((_e = bread((F)->lfs_ivnode, \ 824 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 825 (F)->lfs_bsize, NOCRED, 0, &(BP))) != 0) \ 826 panic("lfs: ifile ino %d read %d", (int)(IN), _e); \ 827 if ((F)->lfs_version == 1) \ 828 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 829 (IN) % (F)->lfs_ifpb); \ 830 else \ 831 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 832 UNSHARE_IFLOCK(F); \ 833 } while (0) 834 835 /* 836 * Cleaner information structure. This resides in the ifile and is used 837 * to pass information from the kernel to the cleaner. 838 */ 839 typedef struct _cleanerinfo { 840 u_int32_t clean; /* number of clean segments */ 841 u_int32_t dirty; /* number of dirty segments */ 842 int32_t bfree; /* disk blocks free */ 843 int32_t avail; /* disk blocks available */ 844 u_int32_t free_head; /* head of the inode free list */ 845 u_int32_t free_tail; /* tail of the inode free list */ 846 #define LFS_CLEANER_MUST_CLEAN 0x01 847 u_int32_t flags; /* status word from the kernel */ 848 } CLEANERINFO; 849 850 #define CLEANSIZE_SU(fs) \ 851 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 852 853 /* Read in the block with the cleaner info from the ifile. */ 854 #define LFS_CLEANERINFO(CP, F, BP) do { \ 855 SHARE_IFLOCK(F); \ 856 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 857 if (bread((F)->lfs_ivnode, \ 858 (daddr_t)0, (F)->lfs_bsize, NOCRED, 0, &(BP))) \ 859 panic("lfs: ifile read"); \ 860 (CP) = (CLEANERINFO *)(BP)->b_data; \ 861 UNSHARE_IFLOCK(F); \ 862 } while (0) 863 864 /* 865 * Synchronize the Ifile cleaner info with current avail and bfree. 866 */ 867 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 868 mutex_enter(&lfs_lock); \ 869 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 870 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 871 (fs)->lfs_favail) { \ 872 (cip)->bfree = (fs)->lfs_bfree; \ 873 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 874 (fs)->lfs_favail; \ 875 if (((bp)->b_flags & B_GATHERED) == 0) { \ 876 (fs)->lfs_flags |= LFS_IFDIRTY; \ 877 } \ 878 mutex_exit(&lfs_lock); \ 879 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 880 } else { \ 881 mutex_exit(&lfs_lock); \ 882 brelse(bp, 0); \ 883 } \ 884 } while (0) 885 886 /* 887 * Get the head of the inode free list. 888 * Always called with the segment lock held. 889 */ 890 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 891 if ((FS)->lfs_version > 1) { \ 892 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 893 (FS)->lfs_freehd = (CIP)->free_head; \ 894 brelse(BP, 0); \ 895 } \ 896 *(FREEP) = (FS)->lfs_freehd; \ 897 } while (0) 898 899 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 900 (FS)->lfs_freehd = (VAL); \ 901 if ((FS)->lfs_version > 1) { \ 902 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 903 (CIP)->free_head = (VAL); \ 904 LFS_BWRITE_LOG(BP); \ 905 mutex_enter(&lfs_lock); \ 906 (FS)->lfs_flags |= LFS_IFDIRTY; \ 907 mutex_exit(&lfs_lock); \ 908 } \ 909 } while (0) 910 911 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 912 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 913 *(FREEP) = (CIP)->free_tail; \ 914 brelse(BP, 0); \ 915 } while (0) 916 917 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 918 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 919 (CIP)->free_tail = (VAL); \ 920 LFS_BWRITE_LOG(BP); \ 921 mutex_enter(&lfs_lock); \ 922 (FS)->lfs_flags |= LFS_IFDIRTY; \ 923 mutex_exit(&lfs_lock); \ 924 } while (0) 925 926 /* 927 * On-disk segment summary information 928 */ 929 typedef struct segsum_v1 SEGSUM_V1; 930 struct segsum_v1 { 931 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 932 u_int32_t ss_datasum; /* 4: check sum of data */ 933 u_int32_t ss_magic; /* 8: segment summary magic number */ 934 #define SS_MAGIC 0x061561 935 int32_t ss_next; /* 12: next segment */ 936 u_int32_t ss_create; /* 16: creation time stamp */ 937 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 938 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 939 940 #define SS_DIROP 0x01 /* segment begins a dirop */ 941 #define SS_CONT 0x02 /* more partials to finish this write*/ 942 #define SS_CLEAN 0x04 /* written by the cleaner */ 943 #define SS_RFW 0x08 /* written by the roll-forward agent */ 944 #define SS_RECLAIM 0x10 /* written by the roll-forward agent */ 945 u_int16_t ss_flags; /* 24: used for directory operations */ 946 u_int16_t ss_pad; /* 26: extra space */ 947 /* FINFO's and inode daddr's... */ 948 }; 949 950 typedef struct segsum SEGSUM; 951 struct segsum { 952 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 953 u_int32_t ss_datasum; /* 4: check sum of data */ 954 u_int32_t ss_magic; /* 8: segment summary magic number */ 955 int32_t ss_next; /* 12: next segment */ 956 u_int32_t ss_ident; /* 16: roll-forward fsid */ 957 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 958 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 959 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 960 u_int16_t ss_flags; /* 24: used for directory operations */ 961 u_int8_t ss_pad[2]; /* 26: extra space */ 962 u_int32_t ss_reclino; /* 28: inode being reclaimed */ 963 u_int64_t ss_serial; /* 32: serial number */ 964 u_int64_t ss_create; /* 40: time stamp */ 965 /* FINFO's and inode daddr's... */ 966 }; 967 968 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 969 970 971 /* 972 * On-disk super block. 973 */ 974 struct dlfs { 975 #define LFS_MAGIC 0x070162 976 u_int32_t dlfs_magic; /* 0: magic number */ 977 #define LFS_VERSION 2 978 u_int32_t dlfs_version; /* 4: version number */ 979 980 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 981 /* number of frags in fs (v2) */ 982 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 983 /* number of bytes per segment (v2) */ 984 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 985 u_int32_t dlfs_bsize; /* 20: file system block size */ 986 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 987 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 988 989 /* Checkpoint region. */ 990 u_int32_t dlfs_freehd; /* 32: start of the free list */ 991 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 992 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 993 int32_t dlfs_avail; /* 44: blocks available for writing */ 994 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 995 int32_t dlfs_idaddr; /* 52: inode file disk address */ 996 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 997 int32_t dlfs_lastseg; /* 60: address of last segment written */ 998 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 999 int32_t dlfs_curseg; /* 68: current segment being written */ 1000 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 1001 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 1002 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 1003 #define dlfs_otstamp dlfs_inopf 1004 1005 /* These are configuration parameters. */ 1006 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 1007 1008 /* These fields can be computed from the others. */ 1009 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 1010 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 1011 u_int32_t dlfs_inopb; /* 100: inodes per block */ 1012 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 1013 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 1014 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 1015 u_int32_t dlfs_nseg; /* 116: number of segments */ 1016 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 1017 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 1018 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 1019 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 1020 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 1021 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 1022 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 1023 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 1024 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 1025 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 1026 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 1027 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 1028 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 1029 1030 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 1031 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 1032 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 1033 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 1034 1035 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 1036 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 1037 #define LFS_PF_CLEAN 0x1 1038 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 1039 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 1040 u_int32_t dlfs_minfreeseg; /* 328: segments not counted in bfree */ 1041 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 1042 u_int64_t dlfs_serial; /* 336: serial number */ 1043 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 1044 int32_t dlfs_start; /* 348: start of segment 0 */ 1045 u_int64_t dlfs_tstamp; /* 352: time stamp */ 1046 #define LFS_44INODEFMT 0 1047 #define LFS_MAXINODEFMT 0 1048 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 1049 u_int32_t dlfs_interleave; /* 364: segment interleave */ 1050 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 1051 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 1052 u_int32_t dlfs_resvseg; /* 376: segments reserved for the cleaner */ 1053 int8_t dlfs_pad[128]; /* 380: round to 512 bytes */ 1054 /* Checksum -- last valid disk field. */ 1055 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 1056 }; 1057 1058 /* Type used for the inode bitmap */ 1059 typedef u_int32_t lfs_bm_t; 1060 1061 /* 1062 * Linked list of segments whose byte count needs updating following a 1063 * file truncation. 1064 */ 1065 struct segdelta { 1066 long segnum; 1067 size_t num; 1068 LIST_ENTRY(segdelta) list; 1069 }; 1070 1071 /* 1072 * In-memory super block. 1073 */ 1074 struct lfs { 1075 struct dlfs lfs_dlfs; /* on-disk parameters */ 1076 #define lfs_magic lfs_dlfs.dlfs_magic 1077 #define lfs_version lfs_dlfs.dlfs_version 1078 #define lfs_size lfs_dlfs.dlfs_size 1079 #define lfs_ssize lfs_dlfs.dlfs_ssize 1080 #define lfs_dsize lfs_dlfs.dlfs_dsize 1081 #define lfs_bsize lfs_dlfs.dlfs_bsize 1082 #define lfs_fsize lfs_dlfs.dlfs_fsize 1083 #define lfs_frag lfs_dlfs.dlfs_frag 1084 #define lfs_freehd lfs_dlfs.dlfs_freehd 1085 #define lfs_bfree lfs_dlfs.dlfs_bfree 1086 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 1087 #define lfs_avail lfs_dlfs.dlfs_avail 1088 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 1089 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 1090 #define lfs_ifile lfs_dlfs.dlfs_ifile 1091 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 1092 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 1093 #define lfs_curseg lfs_dlfs.dlfs_curseg 1094 #define lfs_offset lfs_dlfs.dlfs_offset 1095 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 1096 #define lfs_otstamp lfs_dlfs.dlfs_inopf 1097 #define lfs_inopf lfs_dlfs.dlfs_inopf 1098 #define lfs_minfree lfs_dlfs.dlfs_minfree 1099 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 1100 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 1101 #define lfs_inopb lfs_dlfs.dlfs_inopb 1102 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 1103 #define lfs_sepb lfs_dlfs.dlfs_sepb 1104 #define lfs_nindir lfs_dlfs.dlfs_nindir 1105 #define lfs_nseg lfs_dlfs.dlfs_nseg 1106 #define lfs_nspf lfs_dlfs.dlfs_nspf 1107 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 1108 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 1109 #define lfs_segmask lfs_dlfs.dlfs_segmask 1110 #define lfs_segshift lfs_dlfs.dlfs_segshift 1111 #define lfs_bmask lfs_dlfs.dlfs_bmask 1112 #define lfs_bshift lfs_dlfs.dlfs_bshift 1113 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 1114 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 1115 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 1116 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 1117 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 1118 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 1119 #define lfs_sushift lfs_dlfs.dlfs_sushift 1120 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 1121 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 1122 #define lfs_cksum lfs_dlfs.dlfs_cksum 1123 #define lfs_pflags lfs_dlfs.dlfs_pflags 1124 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 1125 #define lfs_nclean lfs_dlfs.dlfs_nclean 1126 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 1127 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 1128 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 1129 #define lfs_serial lfs_dlfs.dlfs_serial 1130 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 1131 #define lfs_start lfs_dlfs.dlfs_start 1132 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 1133 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 1134 #define lfs_interleave lfs_dlfs.dlfs_interleave 1135 #define lfs_ident lfs_dlfs.dlfs_ident 1136 #define lfs_resvseg lfs_dlfs.dlfs_resvseg 1137 1138 /* These fields are set at mount time and are meaningless on disk. */ 1139 struct segment *lfs_sp; /* current segment being written */ 1140 struct vnode *lfs_ivnode; /* vnode for the ifile */ 1141 u_int32_t lfs_seglock; /* single-thread the segment writer */ 1142 pid_t lfs_lockpid; /* pid of lock holder */ 1143 lwpid_t lfs_locklwp; /* lwp of lock holder */ 1144 u_int32_t lfs_iocount; /* number of ios pending */ 1145 u_int32_t lfs_writer; /* don't allow any dirops to start */ 1146 u_int32_t lfs_dirops; /* count of active directory ops */ 1147 u_int32_t lfs_dirvcount; /* count of VDIROP nodes in this fs */ 1148 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 1149 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 1150 int8_t lfs_fmod; /* super block modified flag */ 1151 int8_t lfs_ronly; /* mounted read-only flag */ 1152 #define LFS_NOTYET 0x01 1153 #define LFS_IFDIRTY 0x02 1154 #define LFS_WARNED 0x04 1155 #define LFS_UNDIROP 0x08 1156 int8_t lfs_flags; /* currently unused flag */ 1157 u_int16_t lfs_activesb; /* toggle between superblocks */ 1158 daddr_t lfs_sbactive; /* disk address of current sb write */ 1159 struct vnode *lfs_flushvp; /* vnode being flushed */ 1160 int lfs_flushvp_fakevref; /* fake vref count for flushvp */ 1161 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 1162 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 1163 size_t lfs_devbsize; /* Device block size */ 1164 size_t lfs_devbshift; /* Device block shift */ 1165 krwlock_t lfs_fraglock; 1166 krwlock_t lfs_iflock; /* Ifile lock */ 1167 kcondvar_t lfs_stopcv; /* Wrap lock */ 1168 struct lwp *lfs_stoplwp; 1169 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 1170 int lfs_nadirop; /* number of active dirop nodes */ 1171 long lfs_ravail; /* blocks pre-reserved for writing */ 1172 long lfs_favail; /* blocks pre-reserved for writing */ 1173 res_t *lfs_resblk; /* Reserved memory for pageout */ 1174 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 1175 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 1176 #define LFS_RESHASH_WIDTH 17 1177 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 1178 int lfs_pdflush; /* pagedaemon wants us to flush */ 1179 u_int32_t **lfs_suflags; /* Segment use flags */ 1180 #ifdef _KERNEL 1181 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 1182 struct pool lfs_bpppool; /* Pool for bpp */ 1183 struct pool lfs_segpool; /* Pool for struct segment */ 1184 #endif /* _KERNEL */ 1185 #define LFS_MAX_CLEANIND 64 1186 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 1187 int lfs_cleanind; /* Index into intervals */ 1188 int lfs_sleepers; /* # procs sleeping this fs */ 1189 int lfs_pages; /* dirty pages blaming this fs */ 1190 lfs_bm_t *lfs_ino_bitmap; /* Inuse inodes bitmap */ 1191 int lfs_nowrap; /* Suspend log wrap */ 1192 int lfs_wrappass; /* Allow first log wrap requester to pass */ 1193 int lfs_wrapstatus; /* Wrap status */ 1194 int lfs_reclino; /* Inode being reclaimed */ 1195 int lfs_startseg; /* Segment we started writing at */ 1196 LIST_HEAD(, segdelta) lfs_segdhd; /* List of pending trunc accounting events */ 1197 }; 1198 1199 /* NINDIR is the number of indirects in a file system block. */ 1200 #define NINDIR(fs) ((fs)->lfs_nindir) 1201 1202 /* INOPB is the number of inodes in a secondary storage block. */ 1203 #define INOPB(fs) ((fs)->lfs_inopb) 1204 /* INOPF is the number of inodes in a fragment. */ 1205 #define INOPF(fs) ((fs)->lfs_inopf) 1206 1207 #define blksize(fs, ip, lbn) \ 1208 (((lbn) >= ULFS_NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 1209 ? (fs)->lfs_bsize \ 1210 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 1211 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 1212 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 1213 ((int)((loc) & (fs)->lfs_ffmask)) 1214 1215 #if defined (_KERNEL) 1216 #define fsbtodb(fs, b) ((b) << ((fs)->lfs_ffshift - DEV_BSHIFT)) 1217 #define dbtofsb(fs, b) ((b) >> ((fs)->lfs_ffshift - DEV_BSHIFT)) 1218 #else 1219 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 1220 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 1221 #endif 1222 1223 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 1224 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 1225 1226 #define fsbtob(fs, b) ((b) << (fs)->lfs_ffshift) 1227 #define btofsb(fs, b) ((b) >> (fs)->lfs_ffshift) 1228 1229 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 1230 ((loc) >> (fs)->lfs_ffshift) 1231 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 1232 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 1233 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 1234 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 1235 #define fragstoblks(fs, frags)/* calculates (frags / fs->fs_frag) */ \ 1236 ((frags) >> (fs)->lfs_fbshift) 1237 #define blkstofrags(fs, blks) /* calculates (blks * fs->fs_frag) */ \ 1238 ((blks) << (fs)->lfs_fbshift) 1239 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 1240 ((fsb) & ((fs)->lfs_frag - 1)) 1241 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 1242 ((fsb) &~ ((fs)->lfs_frag - 1)) 1243 #define dblksize(fs, dp, lbn) \ 1244 (((lbn) >= ULFS_NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 1245 ? (fs)->lfs_bsize \ 1246 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 1247 1248 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 1249 lblktosize((fs), (fs)->lfs_ssize) : \ 1250 (fs)->lfs_ssize) 1251 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 1252 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 1253 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 1254 #define dtosn(fs, daddr) /* block address to segment number */ \ 1255 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 1256 #define sntod(fs, sn) /* segment number to disk address */ \ 1257 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 1258 1259 /* 1260 * Structures used by lfs_bmapv and lfs_markv to communicate information 1261 * about inodes and data blocks. 1262 */ 1263 typedef struct block_info { 1264 u_int32_t bi_inode; /* inode # */ 1265 int32_t bi_lbn; /* logical block w/in file */ 1266 int32_t bi_daddr; /* disk address of block */ 1267 u_int64_t bi_segcreate; /* origin segment create time */ 1268 int bi_version; /* file version number */ 1269 void *bi_bp; /* data buffer */ 1270 int bi_size; /* size of the block (if fragment) */ 1271 } BLOCK_INFO; 1272 1273 /* Compatibility for 1.5 binaries */ 1274 typedef struct block_info_15 { 1275 u_int32_t bi_inode; /* inode # */ 1276 int32_t bi_lbn; /* logical block w/in file */ 1277 int32_t bi_daddr; /* disk address of block */ 1278 u_int32_t bi_segcreate; /* origin segment create time */ 1279 int bi_version; /* file version number */ 1280 void *bi_bp; /* data buffer */ 1281 int bi_size; /* size of the block (if fragment) */ 1282 } BLOCK_INFO_15; 1283 1284 /* In-memory description of a segment about to be written. */ 1285 struct segment { 1286 struct lfs *fs; /* file system pointer */ 1287 struct buf **bpp; /* pointer to buffer array */ 1288 struct buf **cbpp; /* pointer to next available bp */ 1289 struct buf **start_bpp; /* pointer to first bp in this set */ 1290 struct buf *ibp; /* buffer pointer to inode page */ 1291 struct ulfs1_dinode *idp; /* pointer to ifile dinode */ 1292 struct finfo *fip; /* current fileinfo pointer */ 1293 struct vnode *vp; /* vnode being gathered */ 1294 void *segsum; /* segment summary info */ 1295 u_int32_t ninodes; /* number of inodes in this segment */ 1296 int32_t seg_bytes_left; /* bytes left in segment */ 1297 int32_t sum_bytes_left; /* bytes left in summary block */ 1298 u_int32_t seg_number; /* number of this segment */ 1299 int32_t *start_lbp; /* beginning lbn for this set */ 1300 1301 #define SEGM_CKP 0x0001 /* doing a checkpoint */ 1302 #define SEGM_CLEAN 0x0002 /* cleaner call; don't sort */ 1303 #define SEGM_SYNC 0x0004 /* wait for segment */ 1304 #define SEGM_PROT 0x0008 /* don't inactivate at segunlock */ 1305 #define SEGM_PAGEDAEMON 0x0010 /* pagedaemon called us */ 1306 #define SEGM_WRITERD 0x0020 /* LFS writed called us */ 1307 #define SEGM_FORCE_CKP 0x0040 /* Force checkpoint right away */ 1308 #define SEGM_RECLAIM 0x0080 /* Writing to reclaim vnode */ 1309 #define SEGM_SINGLE 0x0100 /* Opportunistic writevnodes */ 1310 u_int16_t seg_flags; /* run-time flags for this segment */ 1311 u_int32_t seg_iocount; /* number of ios pending */ 1312 int ndupino; /* number of duplicate inodes */ 1313 }; 1314 1315 #ifdef _KERNEL 1316 struct lfs_cluster { 1317 size_t bufsize; /* Size of kept data */ 1318 struct buf **bpp; /* Array of kept buffers */ 1319 int bufcount; /* Number of kept buffers */ 1320 #define LFS_CL_MALLOC 0x00000001 1321 #define LFS_CL_SHIFT 0x00000002 1322 #define LFS_CL_SYNC 0x00000004 1323 u_int32_t flags; /* Flags */ 1324 struct lfs *fs; /* LFS that this belongs to */ 1325 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 1326 }; 1327 1328 /* 1329 * Splay tree containing block numbers allocated through lfs_balloc. 1330 */ 1331 struct lbnentry { 1332 SPLAY_ENTRY(lbnentry) entry; 1333 daddr_t lbn; 1334 }; 1335 #endif /* _KERNEL */ 1336 1337 /* 1338 * LFS inode extensions. 1339 */ 1340 struct lfs_inode_ext { 1341 off_t lfs_osize; /* size of file on disk */ 1342 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 1343 size_t lfs_fragsize[ULFS_NDADDR]; /* size of on-disk direct blocks */ 1344 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 1345 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 1346 #define LFSI_NO_GOP_WRITE 0x01 1347 #define LFSI_DELETED 0x02 1348 #define LFSI_WRAPBLOCK 0x04 1349 #define LFSI_WRAPWAIT 0x08 1350 #define LFSI_BMAP 0x10 1351 u_int32_t lfs_iflags; /* Inode flags */ 1352 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 1353 #ifdef _KERNEL 1354 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 1355 int lfs_nbtree; /* Size of tree */ 1356 LIST_HEAD(, segdelta) lfs_segdhd; 1357 #endif 1358 int16_t lfs_odnlink; /* on-disk nlink count for cleaner */ 1359 }; 1360 #define i_lfs_osize inode_ext.lfs->lfs_osize 1361 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 1362 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 1363 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 1364 #define i_lfs_pchain inode_ext.lfs->lfs_pchain 1365 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 1366 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 1367 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 1368 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 1369 #define i_lfs_segdhd inode_ext.lfs->lfs_segdhd 1370 #define i_lfs_odnlink inode_ext.lfs->lfs_odnlink 1371 1372 /* 1373 * Macros for determining free space on the disk, with the variable metadata 1374 * of segment summaries and inode blocks taken into account. 1375 */ 1376 /* 1377 * Estimate number of clean blocks not available for writing because 1378 * they will contain metadata or overhead. This is calculated as 1379 * 1380 * E = ((C * M / D) * D + (0) * (T - D)) / T 1381 * or more simply 1382 * E = (C * M) / T 1383 * 1384 * where 1385 * C is the clean space, 1386 * D is the dirty space, 1387 * M is the dirty metadata, and 1388 * T = C + D is the total space on disk. 1389 * 1390 * This approximates the old formula of E = C * M / D when D is close to T, 1391 * but avoids falsely reporting "disk full" when the sample size (D) is small. 1392 */ 1393 #define LFS_EST_CMETA(F) (int32_t)(( \ 1394 ((F)->lfs_dmeta * (int64_t)(F)->lfs_nclean) / \ 1395 ((F)->lfs_nseg))) 1396 1397 /* Estimate total size of the disk not including metadata */ 1398 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 1399 1400 /* Estimate number of blocks actually available for writing */ 1401 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) ? \ 1402 (F)->lfs_bfree - LFS_EST_CMETA(F) : 0) 1403 1404 /* Amount of non-meta space not available to mortal man */ 1405 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1406 (u_int64_t)(F)->lfs_minfree) / \ 1407 100) 1408 1409 /* Can credential C write BB blocks */ 1410 #define ISSPACE(F, BB, C) \ 1411 ((((C) == NOCRED || kauth_cred_geteuid(C) == 0) && \ 1412 LFS_EST_BFREE(F) >= (BB)) || \ 1413 (kauth_cred_geteuid(C) != 0 && IS_FREESPACE(F, BB))) 1414 1415 /* Can an ordinary user write BB blocks */ 1416 #define IS_FREESPACE(F, BB) \ 1417 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1418 1419 /* 1420 * The minimum number of blocks to create a new inode. This is: 1421 * directory direct block (1) + ULFS_NIADDR indirect blocks + inode block (1) + 1422 * ifile direct block (1) + ULFS_NIADDR indirect blocks = 3 + 2 * ULFS_NIADDR blocks. 1423 */ 1424 #define LFS_NRESERVE(F) (btofsb((F), (2 * ULFS_NIADDR + 3) << (F)->lfs_bshift)) 1425 1426 /* Statistics Counters */ 1427 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1428 u_int segsused; 1429 u_int psegwrites; 1430 u_int psyncwrites; 1431 u_int pcleanwrites; 1432 u_int blocktot; 1433 u_int cleanblocks; 1434 u_int ncheckpoints; 1435 u_int nwrites; 1436 u_int nsync_writes; 1437 u_int wait_exceeded; 1438 u_int write_exceeded; 1439 u_int flush_invoked; 1440 u_int vflush_invoked; 1441 u_int clean_inlocked; 1442 u_int clean_vnlocked; 1443 u_int segs_reclaimed; 1444 }; 1445 #ifdef _KERNEL 1446 extern struct lfs_stats lfs_stats; 1447 #endif 1448 1449 /* Fcntls to take the place of the lfs syscalls */ 1450 struct lfs_fcntl_markv { 1451 BLOCK_INFO *blkiov; /* blocks to relocate */ 1452 int blkcnt; /* number of blocks */ 1453 }; 1454 1455 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 14, struct timeval) 1456 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 15, struct timeval) 1457 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1458 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1459 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1460 1461 struct lfs_fhandle { 1462 char space[28]; /* FHANDLE_SIZE_COMPAT (but used from userland too) */ 1463 }; 1464 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1465 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1466 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1467 #define LFCNWRAPSTOP _FCNR_FSPRIV('L', 9, int) 1468 #define LFCNWRAPGO _FCNR_FSPRIV('L', 10, int) 1469 #define LFCNIFILEFH _FCNW_FSPRIV('L', 11, struct lfs_fhandle) 1470 #define LFCNWRAPPASS _FCNR_FSPRIV('L', 12, int) 1471 # define LFS_WRAP_GOING 0x0 1472 # define LFS_WRAP_WAITING 0x1 1473 #define LFCNWRAPSTATUS _FCNW_FSPRIV('L', 13, int) 1474 1475 /* 1476 * Compat. Defined for kernel only. Userland always uses 1477 * "the one true version". 1478 */ 1479 #ifdef _KERNEL 1480 #include <compat/sys/time_types.h> 1481 1482 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval50) 1483 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval50) 1484 #define LFCNIFILEFH_COMPAT _FCNW_FSPRIV('L', 5, struct lfs_fhandle) 1485 #define LFCNIFILEFH_COMPAT2 _FCN_FSPRIV(F_FSOUT, 'L', 11, 32) 1486 #define LFCNWRAPSTOP_COMPAT _FCNO_FSPRIV('L', 9) 1487 #define LFCNWRAPGO_COMPAT _FCNO_FSPRIV('L', 10) 1488 #define LFCNSEGWAITALL_COMPAT_50 _FCNR_FSPRIV('L', 0, struct timeval50) 1489 #define LFCNSEGWAIT_COMPAT_50 _FCNR_FSPRIV('L', 1, struct timeval50) 1490 #endif 1491 1492 #ifdef _KERNEL 1493 /* XXX MP */ 1494 #define LFS_SEGLOCK_HELD(fs) \ 1495 ((fs)->lfs_seglock != 0 && \ 1496 (fs)->lfs_lockpid == curproc->p_pid && \ 1497 (fs)->lfs_locklwp == curlwp->l_lid) 1498 #endif /* _KERNEL */ 1499 1500 /* Debug segment lock */ 1501 #ifdef notyet 1502 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1503 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1504 # define ASSERT_DUNNO_SEGLOCK(fs) 1505 # define ASSERT_MAYBE_SEGLOCK(fs) 1506 #else /* !notyet */ 1507 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1508 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1509 LFS_SEGLOCK_HELD(fs))) 1510 # define ASSERT_SEGLOCK(fs) do { \ 1511 if (!LFS_SEGLOCK_HELD(fs)) { \ 1512 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1513 } \ 1514 } while(0) 1515 # define ASSERT_NO_SEGLOCK(fs) do { \ 1516 if (LFS_SEGLOCK_HELD(fs)) { \ 1517 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1518 } \ 1519 } while(0) 1520 # define ASSERT_MAYBE_SEGLOCK(x) 1521 #endif /* !notyet */ 1522 1523 /* 1524 * Arguments to mount LFS filesystems 1525 */ 1526 struct ulfs_args { 1527 char *fspec; /* block special device to mount */ 1528 }; 1529 1530 __BEGIN_DECLS 1531 void lfs_itimes(struct inode *, const struct timespec *, 1532 const struct timespec *, const struct timespec *); 1533 __END_DECLS 1534 1535 #endif /* !_UFS_LFS_LFS_H_ */ 1536
Indexes created Tue Sep 30 17:09:57 GMT 2025