1 /* $NetBSD: lfs.h,v 1.42 2002/12/01 00:12:10 matt Exp $ */ 2 3 /*- 4 * Copyright (c) 1999, 2000 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Konrad E. Schroder <perseant@hhhh.org>. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. All advertising materials mentioning features or use of this software 19 * must display the following acknowledgement: 20 * This product includes software developed by the NetBSD 21 * Foundation, Inc. and its contributors. 22 * 4. Neither the name of The NetBSD Foundation nor the names of its 23 * contributors may be used to endorse or promote products derived 24 * from this software without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 27 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 28 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 29 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 30 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 31 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 32 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 33 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 34 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 35 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 36 * POSSIBILITY OF SUCH DAMAGE. 37 */ 38 /*- 39 * Copyright (c) 1991, 1993 40 * The Regents of the University of California. All rights reserved. 41 * 42 * Redistribution and use in source and binary forms, with or without 43 * modification, are permitted provided that the following conditions 44 * are met: 45 * 1. Redistributions of source code must retain the above copyright 46 * notice, this list of conditions and the following disclaimer. 47 * 2. Redistributions in binary form must reproduce the above copyright 48 * notice, this list of conditions and the following disclaimer in the 49 * documentation and/or other materials provided with the distribution. 50 * 3. All advertising materials mentioning features or use of this software 51 * must display the following acknowledgement: 52 * This product includes software developed by the University of 53 * California, Berkeley and its contributors. 54 * 4. Neither the name of the University nor the names of its contributors 55 * may be used to endorse or promote products derived from this software 56 * without specific prior written permission. 57 * 58 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 59 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 60 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 61 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 62 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 63 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 64 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 65 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 66 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 67 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 68 * SUCH DAMAGE. 69 * 70 * @(#)lfs.h 8.9 (Berkeley) 5/8/95 71 */ 72 73 #ifndef _UFS_LFS_LFS_H_ 74 #define _UFS_LFS_LFS_H_ 75 76 /* 77 * Compile-time options for LFS. 78 */ 79 #define LFS_IFIND_RETRIES 16 80 #define LFS_EAGAIN_FAIL /* markv fail with EAGAIN if ino is locked */ 81 #define LFS_DEBUG_RFW /* print roll-forward debugging info */ 82 #define LFS_NO_PAGEMOVE /* Use malloc/copy to write clusters */ 83 #define LFS_AGGRESSIVE_SEGLOCK 84 #define LFS_LOGLENGTH 1024 85 86 /* #define DEBUG_LFS */ /* Intensive debugging of LFS subsystem */ 87 88 #ifdef LFS_NO_PAGEMOVE 89 # define LFS_MALLOC_SUMMARY 90 #endif 91 92 /* 93 * Parameters and generic definitions 94 */ 95 #define BW_CLEAN 1 96 #define MIN_FREE_SEGS 2 97 #define LFS_MAX_ACTIVE 10 98 #define LFS_MAXDIROP (desiredvnodes >> 2) 99 #ifndef LFS_ATIME_IFILE 100 # define LFS_ATIME_IFILE 0 101 #endif 102 103 /* 104 * #define WRITE_THRESHHOLD ((nbuf >> 1) - 10) 105 * #define WAIT_THRESHHOLD (nbuf - (nbuf >> 2) - 10) 106 */ 107 #define LFS_MAX_BUFS ((nbuf >> 2) - 10) 108 #define LFS_WAIT_BUFS ((nbuf >> 1) - (nbuf >> 3) - 10) 109 /* These are new ... is LFS taking up too much memory in its buffers? */ 110 #define LFS_MAX_BYTES (((bufpages >> 2) - 10) * NBPG) 111 #define LFS_WAIT_BYTES (((bufpages >> 1) - (bufpages >> 3) - 10) * NBPG) 112 #define LFS_BUFWAIT 2 113 114 #define LFS_LOCK_BUF(bp) do { \ 115 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \ 116 ++locked_queue_count; \ 117 locked_queue_bytes += bp->b_bufsize; \ 118 } \ 119 (bp)->b_flags |= B_LOCKED; \ 120 } while (0) 121 122 #define LFS_UNLOCK_BUF(bp) do { \ 123 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \ 124 --locked_queue_count; \ 125 locked_queue_bytes -= bp->b_bufsize; \ 126 if (locked_queue_count < LFS_WAIT_BUFS && \ 127 locked_queue_bytes < LFS_WAIT_BYTES) \ 128 wakeup(&locked_queue_count); \ 129 } \ 130 (bp)->b_flags &= ~B_LOCKED; \ 131 } while (0) 132 133 #ifdef DEBUG_LOCKED_LIST 134 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 135 int _s; \ 136 extern int locked_queue_count; \ 137 extern long locked_queue_bytes; \ 138 _s = splbio(); \ 139 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 140 splx(_s); \ 141 wakeup(&locked_queue_count); \ 142 } while (0) 143 #else 144 # define LFS_DEBUG_COUNTLOCKED(m) 145 #endif 146 147 /* For convenience */ 148 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_ACCESSED|IN_CLEANING) 149 150 #define LFS_SET_UINO(ip, flags) do { \ 151 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 152 ++(ip)->i_lfs->lfs_uinodes; \ 153 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 154 ++(ip)->i_lfs->lfs_uinodes; \ 155 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 156 ++(ip)->i_lfs->lfs_uinodes; \ 157 (ip)->i_flag |= (flags); \ 158 } while (0) 159 160 #define LFS_CLR_UINO(ip, flags) do { \ 161 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 162 --(ip)->i_lfs->lfs_uinodes; \ 163 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 164 --(ip)->i_lfs->lfs_uinodes; \ 165 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 166 --(ip)->i_lfs->lfs_uinodes; \ 167 (ip)->i_flag &= ~(flags); \ 168 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 169 panic("lfs_uinodes < 0"); \ 170 } \ 171 } while (0) 172 173 #ifdef DEBUG 174 struct lfs_log_entry { 175 char *op; 176 char *file; 177 int line; 178 ufs_daddr_t block; 179 unsigned long flags; 180 }; 181 extern int lfs_lognum; 182 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 183 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 184 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags) do { \ 185 int _s; \ 186 \ 187 _s = splbio(); \ 188 lfs_log[lfs_lognum].op = theop; \ 189 lfs_log[lfs_lognum].file = thefile; \ 190 lfs_log[lfs_lognum].line = (theline); \ 191 lfs_log[lfs_lognum].block = (lbn); \ 192 lfs_log[lfs_lognum].flags = (theflags); \ 193 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 194 splx(_s); \ 195 } while (0) 196 197 # define LFS_BCLEAN_LOG(fs, bp) do { \ 198 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 199 LFS_ENTER_LOG("clear", __FILE__, __LINE__, bp->b_lblkno, bp->b_flags); \ 200 } while (0) 201 #else 202 # define LFS_BCLEAN_LOG(fs, bp) 203 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 204 #endif 205 206 #define LFS_ITIMES(ip, acc, mod, cre) do { \ 207 struct lfs *_fs = (ip)->i_lfs; \ 208 \ 209 if ((ip)->i_flag & IN_ACCESS) { \ 210 (ip)->i_ffs_atime = (acc)->tv_sec; \ 211 (ip)->i_ffs_atimensec = (acc)->tv_nsec; \ 212 if ((ip)->i_lfs->lfs_version > 1) { \ 213 struct buf *ibp; \ 214 IFILE *ifp; \ 215 \ 216 LFS_IENTRY(ifp, ip->i_lfs, ip->i_number, ibp); \ 217 ifp->if_atime_sec = (acc)->tv_sec; \ 218 ifp->if_atime_nsec = (acc)->tv_nsec; \ 219 LFS_BWRITE_LOG(ibp); \ 220 _fs->lfs_flags |= LFS_IFDIRTY; \ 221 } else { \ 222 LFS_SET_UINO(ip, IN_ACCESSED); \ 223 } \ 224 } \ 225 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) { \ 226 if ((ip)->i_flag & IN_UPDATE) { \ 227 (ip)->i_ffs_mtime = (mod)->tv_sec; \ 228 (ip)->i_ffs_mtimensec = (mod)->tv_nsec; \ 229 (ip)->i_modrev++; \ 230 } \ 231 if ((ip)->i_flag & IN_CHANGE) { \ 232 (ip)->i_ffs_ctime = (cre)->tv_sec; \ 233 (ip)->i_ffs_ctimensec = (cre)->tv_nsec; \ 234 } \ 235 LFS_SET_UINO(ip, IN_MODIFIED); \ 236 } \ 237 (ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE); \ 238 } while (0) 239 240 #define WRITEINPROG(vp) (vp->v_dirtyblkhd.lh_first && !(VTOI(vp)->i_flag & \ 241 (IN_MODIFIED | IN_ACCESSED | IN_CLEANING))) 242 243 /* Here begins the berkeley code */ 244 245 #define LFS_LABELPAD 8192 /* LFS label size */ 246 #define LFS_SBPAD 8192 /* LFS superblock size */ 247 248 /* On-disk and in-memory checkpoint segment usage structure. */ 249 typedef struct segusage SEGUSE; 250 struct segusage { 251 u_int32_t su_nbytes; /* 0: number of live bytes */ 252 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 253 u_int16_t su_nsums; /* 8: number of summaries in segment */ 254 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 255 256 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 257 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 258 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 259 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 260 u_int32_t su_flags; /* 12: segment flags */ 261 u_int64_t su_lastmod; /* 16: last modified timestamp */ 262 }; 263 264 typedef struct segusage_v1 SEGUSE_V1; 265 struct segusage_v1 { 266 u_int32_t su_nbytes; /* 0: number of live bytes */ 267 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 268 u_int16_t su_nsums; /* 8: number of summaries in segment */ 269 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 270 u_int32_t su_flags; /* 12: segment flags */ 271 }; 272 273 #define SEGUPB(fs) (fs->lfs_sepb) 274 #define SEGTABSIZE_SU(fs) \ 275 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 276 277 /* On-disk file information. One per file with data blocks in the segment. */ 278 typedef struct finfo FINFO; 279 struct finfo { 280 u_int32_t fi_nblocks; /* number of blocks */ 281 u_int32_t fi_version; /* version number */ 282 u_int32_t fi_ino; /* inode number */ 283 u_int32_t fi_lastlength; /* length of last block in array */ 284 ufs_daddr_t fi_blocks[1]; /* array of logical block numbers */ 285 }; 286 287 288 /* On-disk super block. */ 289 struct dlfs { 290 #define LFS_MAGIC 0x070162 291 u_int32_t dlfs_magic; /* 0: magic number */ 292 #define LFS_VERSION 2 293 u_int32_t dlfs_version; /* 4: version number */ 294 295 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 296 /* number of frags in fs (v2) */ 297 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 298 /* number of bytes per segment (v2) */ 299 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 300 u_int32_t dlfs_bsize; /* 20: file system block size */ 301 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 302 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 303 304 /* Checkpoint region. */ 305 u_int32_t dlfs_free; /* 32: start of the free list */ 306 u_int32_t dlfs_bfree; /* 36: number of free disk blocks */ 307 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 308 int32_t dlfs_avail; /* 44: blocks available for writing */ 309 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 310 ufs_daddr_t dlfs_idaddr; /* 52: inode file disk address */ 311 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 312 ufs_daddr_t dlfs_lastseg; /* 60: address of last segment written */ 313 ufs_daddr_t dlfs_nextseg; /* 64: address of next segment to write */ 314 ufs_daddr_t dlfs_curseg; /* 68: current segment being written */ 315 ufs_daddr_t dlfs_offset; /* 72: offset in curseg for next partial */ 316 ufs_daddr_t dlfs_lastpseg; /* 76: address of last partial written */ 317 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 318 #define dlfs_otstamp dlfs_inopf 319 320 /* These are configuration parameters. */ 321 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 322 323 /* These fields can be computed from the others. */ 324 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 325 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 326 u_int32_t dlfs_inopb; /* 100: inodes per block */ 327 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 328 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 329 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 330 u_int32_t dlfs_nseg; /* 116: number of segments */ 331 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 332 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 333 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 334 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 335 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 336 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 337 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 338 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 339 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 340 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 341 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 342 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 343 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 344 345 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 346 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 347 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 348 ufs_daddr_t dlfs_sboffs[LFS_MAXNUMSB]; 349 350 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 351 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 352 #define LFS_PF_CLEAN 0x1 353 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 354 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 355 u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */ 356 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 357 u_int64_t dlfs_serial; /* 336: serial number */ 358 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 359 ufs_daddr_t dlfs_start; /* 348: start of segment 0 */ 360 u_int64_t dlfs_tstamp; /* 352: time stamp */ 361 #define LFS_44INODEFMT 0 362 #define LFS_MAXINODEFMT 0 363 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 364 u_int32_t dlfs_interleave; /* 364: segment interleave */ 365 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 366 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 367 int8_t dlfs_pad[132]; /* 376: round to 512 bytes */ 368 /* Checksum -- last valid disk field. */ 369 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 370 }; 371 372 /* Maximum number of io's we can have pending at once */ 373 #define LFS_THROTTLE 32 /* XXX should be better paramtrized - ? */ 374 375 /* In-memory super block. */ 376 struct lfs { 377 struct dlfs lfs_dlfs; /* on-disk parameters */ 378 #define lfs_magic lfs_dlfs.dlfs_magic 379 #define lfs_version lfs_dlfs.dlfs_version 380 #define lfs_size lfs_dlfs.dlfs_size 381 #define lfs_ssize lfs_dlfs.dlfs_ssize 382 #define lfs_dsize lfs_dlfs.dlfs_dsize 383 #define lfs_bsize lfs_dlfs.dlfs_bsize 384 #define lfs_fsize lfs_dlfs.dlfs_fsize 385 #define lfs_frag lfs_dlfs.dlfs_frag 386 #define lfs_free lfs_dlfs.dlfs_free 387 #define lfs_bfree lfs_dlfs.dlfs_bfree 388 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 389 #define lfs_avail lfs_dlfs.dlfs_avail 390 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 391 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 392 #define lfs_ifile lfs_dlfs.dlfs_ifile 393 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 394 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 395 #define lfs_curseg lfs_dlfs.dlfs_curseg 396 #define lfs_offset lfs_dlfs.dlfs_offset 397 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 398 #define lfs_otstamp lfs_dlfs.dlfs_inopf 399 #define lfs_inopf lfs_dlfs.dlfs_inopf 400 #define lfs_minfree lfs_dlfs.dlfs_minfree 401 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 402 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 403 #define lfs_inopb lfs_dlfs.dlfs_inopb 404 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 405 #define lfs_sepb lfs_dlfs.dlfs_sepb 406 #define lfs_nindir lfs_dlfs.dlfs_nindir 407 #define lfs_nseg lfs_dlfs.dlfs_nseg 408 #define lfs_nspf lfs_dlfs.dlfs_nspf 409 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 410 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 411 #define lfs_segmask lfs_dlfs.dlfs_segmask 412 #define lfs_segshift lfs_dlfs.dlfs_segshift 413 #define lfs_bmask lfs_dlfs.dlfs_bmask 414 #define lfs_bshift lfs_dlfs.dlfs_bshift 415 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 416 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 417 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 418 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 419 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 420 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 421 #define lfs_sushift lfs_dlfs.dlfs_sushift 422 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 423 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 424 #define lfs_cksum lfs_dlfs.dlfs_cksum 425 #define lfs_pflags lfs_dlfs.dlfs_pflags 426 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 427 #define lfs_nclean lfs_dlfs.dlfs_nclean 428 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 429 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 430 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 431 #define lfs_serial lfs_dlfs.dlfs_serial 432 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 433 #define lfs_start lfs_dlfs.dlfs_start 434 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 435 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 436 #define lfs_interleave lfs_dlfs.dlfs_interleave 437 #define lfs_ident lfs_dlfs.dlfs_ident 438 439 /* These fields are set at mount time and are meaningless on disk. */ 440 struct segment *lfs_sp; /* current segment being written */ 441 struct vnode *lfs_ivnode; /* vnode for the ifile */ 442 u_int32_t lfs_seglock; /* single-thread the segment writer */ 443 pid_t lfs_lockpid; /* pid of lock holder */ 444 u_int32_t lfs_iocount; /* number of ios pending */ 445 u_int32_t lfs_writer; /* don't allow any dirops to start */ 446 u_int32_t lfs_dirops; /* count of active directory ops */ 447 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 448 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 449 int8_t lfs_fmod; /* super block modified flag */ 450 int8_t lfs_ronly; /* mounted read-only flag */ 451 #define LFS_NOTYET 0x01 452 #define LFS_IFDIRTY 0x02 453 #define LFS_WARNED 0x04 454 int8_t lfs_flags; /* currently unused flag */ 455 u_int16_t lfs_activesb; /* toggle between superblocks */ 456 #ifdef LFS_TRACK_IOS 457 daddr_t lfs_pending[LFS_THROTTLE]; /* daddrs of pending writes */ 458 #endif /* LFS_TRACK_IOS */ 459 daddr_t lfs_sbactive; /* disk address of in-progress sb write */ 460 struct vnode *lfs_flushvp; /* vnode being flushed */ 461 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 462 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 463 size_t lfs_devbsize; /* Device block size */ 464 size_t lfs_devbshift; /* Device block shift */ 465 struct lock lfs_freelock; 466 struct lock lfs_fraglock; 467 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 468 int lfs_nadirop; /* number of active dirop nodes */ 469 long lfs_ravail; /* blocks pre-reserved for writing */ 470 }; 471 472 /* 473 * Inode 0: out-of-band inode number 474 * Inode 1: IFILE inode number 475 * Inode 2: root inode 476 * Inode 3: lost+found inode number 477 */ 478 #define LFS_UNUSED_INUM 0 /* out of band inode number */ 479 #define LFS_IFILE_INUM 1 /* IFILE inode number */ 480 #define LOSTFOUNDINO 3 /* lost+found inode number */ 481 #define LFS_FIRST_INUM 4 /* first free inode number */ 482 483 /* Address calculations for metadata located in the inode */ 484 #define S_INDIR(fs) -NDADDR 485 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 486 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 487 488 /* Unassigned disk addresses. */ 489 #define UNASSIGNED -1 490 #define UNWRITTEN -2 491 492 /* Unused logical block number */ 493 #define LFS_UNUSED_LBN -1 494 495 typedef struct ifile IFILE; 496 struct ifile { 497 u_int32_t if_version; /* inode version number */ 498 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 499 ufs_daddr_t if_daddr; /* inode disk address */ 500 ino_t if_nextfree; /* next-unallocated inode */ 501 /* XXX - when inode format changes, this changes too */ 502 u_int32_t if_atime_sec; /* Last access time, seconds */ 503 u_int32_t if_atime_nsec; /* and nanoseconds */ 504 }; 505 506 typedef struct ifile_v1 IFILE_V1; 507 struct ifile_v1 { 508 u_int32_t if_version; /* inode version number */ 509 ufs_daddr_t if_daddr; /* inode disk address */ 510 ino_t if_nextfree; /* next-unallocated inode */ 511 #if LFS_ATIME_IFILE 512 struct timespec if_atime; /* Last access time */ 513 #endif 514 }; 515 516 /* 517 * Cleaner information structure. This resides in the ifile and is used 518 * to pass information between the cleaner and the kernel. 519 */ 520 typedef struct _cleanerinfo { 521 u_int32_t clean; /* number of clean segments */ 522 u_int32_t dirty; /* number of dirty segments */ 523 u_int32_t bfree; /* disk blocks free */ 524 int32_t avail; /* disk blocks available */ 525 u_int32_t free_head; /* head of the inode free list */ 526 u_int32_t free_tail; /* tail of the inode free list */ 527 } CLEANERINFO; 528 529 #define CLEANSIZE_SU(fs) \ 530 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 531 532 /* 533 * All summary blocks are the same size, so we can always read a summary 534 * block easily from a segment. 535 */ 536 #define LFS_V1_SUMMARY_SIZE 512 537 #define LFS_DFL_SUMMARY_SIZE 512 538 539 /* On-disk segment summary information */ 540 typedef struct segsum_v1 SEGSUM_V1; 541 struct segsum_v1 { 542 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 543 u_int32_t ss_datasum; /* 4: check sum of data */ 544 u_int32_t ss_magic; /* 8: segment summary magic number */ 545 #define SS_MAGIC 0x061561 546 ufs_daddr_t ss_next; /* 12: next segment */ 547 u_int32_t ss_create; /* 16: creation time stamp */ 548 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 549 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 550 551 #define SS_DIROP 0x01 /* segment begins a dirop */ 552 #define SS_CONT 0x02 /* more partials to finish this write*/ 553 u_int16_t ss_flags; /* 24: used for directory operations */ 554 u_int16_t ss_pad; /* 26: extra space */ 555 /* FINFO's and inode daddr's... */ 556 }; 557 558 typedef struct segsum SEGSUM; 559 struct segsum { 560 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 561 u_int32_t ss_datasum; /* 4: check sum of data */ 562 u_int32_t ss_magic; /* 8: segment summary magic number */ 563 ufs_daddr_t ss_next; /* 12: next segment */ 564 u_int32_t ss_ident; /* 16: roll-forward fsid */ 565 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 566 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 567 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 568 u_int16_t ss_flags; /* 24: used for directory operations */ 569 u_int8_t ss_pad[6]; /* 26: extra space */ 570 u_int64_t ss_serial; /* 32: serial number */ 571 u_int64_t ss_create; /* 40: time stamp */ 572 /* FINFO's and inode daddr's... */ 573 }; 574 575 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 576 577 /* NINDIR is the number of indirects in a file system block. */ 578 #define NINDIR(fs) ((fs)->lfs_nindir) 579 580 /* INOPB is the number of inodes in a secondary storage block. */ 581 #define INOPB(fs) ((fs)->lfs_inopb) 582 /* INOPF is the number of inodes in a fragment. */ 583 #define INOPF(fs) ((fs)->lfs_inopf) 584 585 #define blksize(fs, ip, lbn) \ 586 (((lbn) >= NDADDR || (ip)->i_ffs_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 587 ? (fs)->lfs_bsize \ 588 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs_size)))) 589 #define blkoff(fs, loc) ((int)(loc) & (fs)->lfs_bmask) 590 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 591 ((int)((loc) & (fs)->lfs_ffmask)) 592 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 593 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 594 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 595 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 596 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 597 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 598 /* Same as above, but named like dbtob(), btodb() */ 599 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 600 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 601 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 602 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 603 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 604 (fs)->lfs_fsbtodb)) 605 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 606 (fs)->lfs_fsbtodb)) 607 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 608 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 609 ((loc) >> (fs)->lfs_ffshift) 610 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 611 ((int)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 612 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 613 ((int)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 614 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 615 ((frags) >> (fs)->lfs_fbshift) 616 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 617 ((blks) << (fs)->lfs_fbshift) 618 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 619 ((fsb) & ((fs)->lfs_frag - 1)) 620 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 621 ((fsb) &~ ((fs)->lfs_frag - 1)) 622 #define dblksize(fs, dip, lbn) \ 623 (((lbn) >= NDADDR || (dip)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 624 ? (fs)->lfs_bsize \ 625 : (fragroundup(fs, blkoff(fs, (dip)->di_size)))) 626 627 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 628 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 629 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 630 #define dtosn(fs, daddr) /* block address to segment number */ \ 631 (((daddr) - (fs)->lfs_start) / segtod((fs), 1)) 632 #define sntod(fs, sn) /* segment number to disk address */ \ 633 ((ufs_daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 634 635 /* Read in the block with the cleaner info from the ifile. */ 636 #define LFS_CLEANERINFO(CP, F, BP) do { \ 637 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 638 if (bread((F)->lfs_ivnode, \ 639 (ufs_daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 640 panic("lfs: ifile read"); \ 641 (CP) = (CLEANERINFO *)(BP)->b_data; \ 642 } while(0) 643 644 /* Synchronize the Ifile cleaner info with current avail and bfree */ 645 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 646 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 647 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail) { \ 648 (cip)->bfree = (fs)->lfs_bfree; \ 649 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail; \ 650 if (((bp)->b_flags & B_GATHERED) == 0) \ 651 (fs)->lfs_flags |= LFS_IFDIRTY; \ 652 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 653 } else \ 654 brelse(bp); \ 655 } while (0) 656 657 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 658 if ((FS)->lfs_version > 1) { \ 659 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 660 (FS)->lfs_free = (CIP)->free_head; \ 661 brelse(BP); \ 662 } \ 663 *(FREEP) = (FS)->lfs_free; \ 664 } while (0) 665 666 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 667 (FS)->lfs_free = (VAL); \ 668 if ((FS)->lfs_version > 1) { \ 669 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 670 (CIP)->free_head = (VAL); \ 671 LFS_BWRITE_LOG(BP); \ 672 (FS)->lfs_flags |= LFS_IFDIRTY; \ 673 } \ 674 } while (0) 675 676 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 677 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 678 *(FREEP) = (CIP)->free_tail; \ 679 brelse(BP); \ 680 } while (0) 681 682 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 683 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 684 (CIP)->free_tail = (VAL); \ 685 LFS_BWRITE_LOG(BP); \ 686 (FS)->lfs_flags |= LFS_IFDIRTY; \ 687 } while (0) 688 689 /* 690 * XXX - v1 compatibility code is not allowed to touch if_atime, since it 691 * may not be mapped! 692 */ 693 /* Read in the block with a specific inode from the ifile. */ 694 #define LFS_IENTRY(IP, F, IN, BP) do { \ 695 int _e; \ 696 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 697 if ((_e = bread((F)->lfs_ivnode, \ 698 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 699 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 700 panic("lfs: ifile read %d", _e); \ 701 if ((F)->lfs_version == 1) \ 702 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + (IN) % (F)->lfs_ifpb); \ 703 else \ 704 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 705 } while(0) 706 707 /* Read in the block with a specific segment usage entry from the ifile. */ 708 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 709 int _e; \ 710 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 711 if ((_e = bread((F)->lfs_ivnode, \ 712 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 713 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 714 panic("lfs: ifile read: %d", _e); \ 715 if ((F)->lfs_version == 1) \ 716 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 717 ((IN) & ((F)->lfs_sepb - 1))); \ 718 else \ 719 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 720 } while(0) 721 722 /* Determine if a buffer belongs to the ifile */ 723 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 724 725 /* 726 * Structures used by lfs_bmapv and lfs_markv to communicate information 727 * about inodes and data blocks. 728 */ 729 typedef struct block_info { 730 ino_t bi_inode; /* inode # */ 731 ufs_daddr_t bi_lbn; /* logical block w/in file */ 732 ufs_daddr_t bi_daddr; /* disk address of block */ 733 u_int64_t bi_segcreate; /* origin segment create time */ 734 int bi_version; /* file version number */ 735 void *bi_bp; /* data buffer */ 736 int bi_size; /* size of the block (if fragment) */ 737 } BLOCK_INFO; 738 739 /* Compatibility for 1.5 binaries */ 740 typedef struct block_info_15 { 741 ino_t bi_inode; /* inode # */ 742 ufs_daddr_t bi_lbn; /* logical block w/in file */ 743 ufs_daddr_t bi_daddr; /* disk address of block */ 744 u_int32_t bi_segcreate; /* origin segment create time */ 745 int bi_version; /* file version number */ 746 void *bi_bp; /* data buffer */ 747 int bi_size; /* size of the block (if fragment) */ 748 } BLOCK_INFO_15; 749 750 /* In-memory description of a segment about to be written. */ 751 struct segment { 752 struct lfs *fs; /* file system pointer */ 753 struct buf **bpp; /* pointer to buffer array */ 754 struct buf **cbpp; /* pointer to next available bp */ 755 struct buf **start_bpp; /* pointer to first bp in this set */ 756 struct buf *ibp; /* buffer pointer to inode page */ 757 struct dinode *idp; /* pointer to ifile dinode */ 758 struct finfo *fip; /* current fileinfo pointer */ 759 struct vnode *vp; /* vnode being gathered */ 760 void *segsum; /* segment summary info */ 761 u_int32_t ninodes; /* number of inodes in this segment */ 762 u_int32_t seg_bytes_left; /* bytes left in segment */ 763 u_int32_t sum_bytes_left; /* bytes left in summary block */ 764 u_int32_t seg_number; /* number of this segment */ 765 ufs_daddr_t *start_lbp; /* beginning lbn for this set */ 766 767 #define SEGM_CKP 0x01 /* doing a checkpoint */ 768 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 769 #define SEGM_SYNC 0x04 /* wait for segment */ 770 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 771 u_int16_t seg_flags; /* run-time flags for this segment */ 772 u_int32_t seg_iocount; /* number of ios pending */ 773 int ndupino; /* number of duplicate inodes */ 774 }; 775 776 struct lfs_cluster { 777 struct buf **bpp; /* Array of kept buffers */ 778 int bufcount; /* Number of kept buffers */ 779 size_t bufsize; /* Size of kept data */ 780 #define LFS_CL_MALLOC 0x00000001 781 #define LFS_CL_SHIFT 0x00000002 782 #define LFS_CL_SYNC 0x00000004 783 u_int32_t flags; /* Flags */ 784 struct lfs *fs; /* LFS that this belongs to */ 785 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 786 void *saveaddr; /* Original contents of saveaddr */ 787 char *olddata; /* Original b_data, if LFS_CL_MALLOC */ 788 }; 789 790 /* 791 * Macros for determining free space on the disk, with the variable metadata 792 * of segment summaries and inode blocks taken into account. 793 */ 794 /* Estimate number of clean blocks not available for writing */ 795 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 796 (int64_t)(F)->lfs_nclean) / \ 797 ((F)->lfs_nseg - (F)->lfs_nclean))) 798 799 /* Estimate total size of the disk not including metadata */ 800 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 801 802 /* Estimate number of blocks actually available for writing */ 803 #define LFS_EST_BFREE(F) ((F)->lfs_bfree - LFS_EST_CMETA(F) - (F)->lfs_dmeta) 804 805 /* Amount of non-meta space not available to mortal man */ 806 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 807 (u_int64_t)(F)->lfs_minfree) / \ 808 100) 809 810 /* Can credential C write BB blocks */ 811 #define ISSPACE(F, BB, C) \ 812 ((((C) == NOCRED || (C)->cr_uid == 0) && \ 813 LFS_EST_BFREE(F) >= (BB)) || \ 814 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB))) 815 816 /* Can an ordinary user write BB blocks */ 817 #define IS_FREESPACE(F, BB) \ 818 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 819 820 /* Statistics Counters */ 821 struct lfs_stats { 822 u_int segsused; 823 u_int psegwrites; 824 u_int psyncwrites; 825 u_int pcleanwrites; 826 u_int blocktot; 827 u_int cleanblocks; 828 u_int ncheckpoints; 829 u_int nwrites; 830 u_int nsync_writes; 831 u_int wait_exceeded; 832 u_int write_exceeded; 833 u_int flush_invoked; 834 u_int vflush_invoked; 835 }; 836 #ifdef _KERNEL 837 extern struct lfs_stats lfs_stats; 838 #endif 839 840 #endif /* !_UFS_LFS_LFS_H_ */ 841