lfs.h revision 1.92
1 /* $NetBSD: lfs.h,v 1.92 2005/08/23 08:05:13 christos Exp $ */ 2 3 /*- 4 * Copyright (c) 1999, 2000, 2001, 2002, 2003 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 (at) 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. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)lfs.h 8.9 (Berkeley) 5/8/95 67 */ 68 69 #ifndef _UFS_LFS_LFS_H_ 70 #define _UFS_LFS_LFS_H_ 71 72 /* 73 * Compile-time options for LFS. 74 */ 75 #define LFS_IFIND_RETRIES 16 76 #define LFS_LOGLENGTH 1024 /* size of debugging log */ 77 #define LFS_MAX_ACTIVE 10 /* Dirty segments before ckp forced */ 78 79 /* 80 * Fixed filesystem layout parameters 81 */ 82 #define LFS_LABELPAD 8192 /* LFS label size */ 83 #define LFS_SBPAD 8192 /* LFS superblock size */ 84 85 #define LFS_UNUSED_INUM 0 /* 0: out of band inode number */ 86 #define LFS_IFILE_INUM 1 /* 1: IFILE inode number */ 87 /* 2: Root inode number */ 88 #define LOSTFOUNDINO 3 /* 3: lost+found inode number */ 89 #define LFS_FIRST_INUM 4 /* 4: first free inode number */ 90 91 #define LFS_V1_SUMMARY_SIZE 512 /* V1 fixed summary size */ 92 #define LFS_DFL_SUMMARY_SIZE 512 /* Default summary size */ 93 94 #define LFS_MAX_DADDR 0x7fffffff /* Highest addressable fsb */ 95 96 #define LFS_MAXNAMLEN 255 /* maximum name length in a dir */ 97 98 /* Adjustable filesystem parameters */ 99 #define MIN_FREE_SEGS 2 100 #ifndef LFS_ATIME_IFILE 101 # define LFS_ATIME_IFILE 0 /* Store atime info in ifile (optional in LFSv1) */ 102 #endif 103 #define LFS_MARKV_MAXBLKCNT 65536 /* Max block count for lfs_markv() */ 104 105 /* Misc. definitions */ 106 #define BW_CLEAN 1 /* Flag for lfs_bwrite_ext() */ 107 #define PG_DELWRI PG_PAGER1 /* Local def for delayed pageout */ 108 109 /* Resource limits */ 110 #define LFS_MAX_BUFS ((nbuf >> 2) - 10) 111 #define LFS_WAIT_BUFS ((nbuf >> 1) - (nbuf >> 3) - 10) 112 #define LFS_INVERSE_MAX_BUFS(n) (((n) + 10) << 2) 113 #define LFS_MAX_BYTES ((bufmem_lowater >> 2) - 10 * PAGE_SIZE) 114 #define LFS_INVERSE_MAX_BYTES(n) (((n) + 10 * PAGE_SIZE) << 2) 115 #define LFS_WAIT_BYTES ((bufmem_lowater >> 1) - (bufmem_lowater >> 3) - 10 * PAGE_SIZE) 116 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3)) 117 #define LFS_MAX_PAGES \ 118 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemin) >> 8) 119 #define LFS_WAIT_PAGES \ 120 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemax) >> 8) 121 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */ 122 123 /* 124 * Reserved blocks for lfs_malloc 125 */ 126 127 /* Structure to keep reserved blocks */ 128 typedef struct lfs_res_blk { 129 void *p; 130 LIST_ENTRY(lfs_res_blk) res; 131 int size; 132 char inuse; 133 } res_t; 134 135 /* Types for lfs_newbuf and lfs_malloc */ 136 #define LFS_NB_UNKNOWN -1 137 #define LFS_NB_SUMMARY 0 138 #define LFS_NB_SBLOCK 1 139 #define LFS_NB_IBLOCK 2 140 #define LFS_NB_CLUSTER 3 141 #define LFS_NB_CLEAN 4 142 #define LFS_NB_BLKIOV 5 143 #define LFS_NB_COUNT 6 /* always last */ 144 145 /* Number of reserved memory blocks of each type */ 146 #define LFS_N_SUMMARIES 2 147 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 148 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 149 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 150 #define LFS_N_CLEAN 0 151 #define LFS_N_BLKIOV 1 152 153 /* Total count of "large" (non-pool) types */ 154 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 155 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV) 156 157 /* Counts for pool types */ 158 #define LFS_N_CL LFS_N_CLUSTERS 159 #define LFS_N_BPP 2 160 #define LFS_N_SEG 2 161 162 /* 163 * "struct buf" associated definitions 164 */ 165 166 /* Unassigned disk addresses. */ 167 #define UNASSIGNED -1 168 #define UNWRITTEN -2 169 170 /* Unused logical block number */ 171 #define LFS_UNUSED_LBN -1 172 173 /* Determine if a buffer belongs to the ifile */ 174 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 175 176 # define LFS_LOCK_BUF(bp) do { \ 177 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \ 178 simple_lock(&lfs_subsys_lock); \ 179 ++locked_queue_count; \ 180 locked_queue_bytes += bp->b_bufsize; \ 181 simple_unlock(&lfs_subsys_lock); \ 182 } \ 183 (bp)->b_flags |= B_LOCKED; \ 184 } while (0) 185 186 # define LFS_UNLOCK_BUF(bp) do { \ 187 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \ 188 simple_lock(&lfs_subsys_lock); \ 189 --locked_queue_count; \ 190 locked_queue_bytes -= bp->b_bufsize; \ 191 if (locked_queue_count < LFS_WAIT_BUFS && \ 192 locked_queue_bytes < LFS_WAIT_BYTES) \ 193 wakeup(&locked_queue_count); \ 194 simple_unlock(&lfs_subsys_lock); \ 195 } \ 196 (bp)->b_flags &= ~B_LOCKED; \ 197 } while (0) 198 199 #ifdef _KERNEL 200 201 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 202 203 # define LFS_IS_MALLOC_BUF(bp) (((bp)->b_flags & B_CALL) && \ 204 (bp)->b_iodone == lfs_callback) 205 206 # ifdef DEBUG 207 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 208 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 209 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 210 wakeup(&locked_queue_count); \ 211 } \ 212 } while (0) 213 # else 214 # define LFS_DEBUG_COUNTLOCKED(m) 215 # endif 216 217 /* log for debugging writes to the Ifile */ 218 # ifdef DEBUG 219 struct lfs_log_entry { 220 const char *op; 221 const char *file; 222 int pid; 223 int line; 224 daddr_t block; 225 unsigned long flags; 226 }; 227 extern int lfs_lognum; 228 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 229 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 230 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 231 int _s; \ 232 \ 233 simple_lock(&lfs_subsys_lock); \ 234 _s = splbio(); \ 235 lfs_log[lfs_lognum].op = theop; \ 236 lfs_log[lfs_lognum].file = thefile; \ 237 lfs_log[lfs_lognum].line = (theline); \ 238 lfs_log[lfs_lognum].pid = (thepid); \ 239 lfs_log[lfs_lognum].block = (lbn); \ 240 lfs_log[lfs_lognum].flags = (theflags); \ 241 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 242 splx(_s); \ 243 simple_unlock(&lfs_subsys_lock); \ 244 } while (0) 245 246 # define LFS_BCLEAN_LOG(fs, bp) do { \ 247 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 248 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 249 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 250 } while (0) 251 252 /* Must match list in lfs_vfsops.c ! */ 253 # define DLOG_RF 0 /* roll forward */ 254 # define DLOG_ALLOC 1 /* inode alloc */ 255 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 256 # define DLOG_FLUSH 3 /* flush */ 257 # define DLOG_LLIST 4 /* locked list accounting */ 258 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 259 # define DLOG_VNODE 6 /* vflush/writevnodes */ 260 # define DLOG_SEG 7 /* segwrite */ 261 # define DLOG_SU 8 /* seguse accounting */ 262 # define DLOG_CLEAN 9 /* cleaner routines */ 263 # define DLOG_MOUNT 10 /* mount/unmount */ 264 # define DLOG_PAGE 11 /* putpages/gop_write */ 265 # define DLOG_DIROP 12 /* dirop accounting */ 266 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 267 # define DLOG_MAX 14 /* The terminator */ 268 # define DLOG(a) lfs_debug_log a 269 # else /* ! DEBUG */ 270 # define LFS_BCLEAN_LOG(fs, bp) 271 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 272 # define DLOG(a) 273 # endif /* ! DEBUG */ 274 #else /* ! _KERNEL */ 275 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 276 #endif /* _KERNEL */ 277 278 #ifdef _KERNEL 279 /* Filehandle structure for exported LFSes */ 280 struct lfid { 281 struct ufid lfid_ufid; 282 #define lfid_len lfid_ufid.ufid_len 283 #define lfid_ino lfid_ufid.ufid_ino 284 #define lfid_gen lfid_ufid.ufid_gen 285 uint32_t lfid_ident; 286 }; 287 #endif /* _KERNEL */ 288 289 /* 290 * "struct inode" associated definitions 291 */ 292 293 /* Address calculations for metadata located in the inode */ 294 #define S_INDIR(fs) -NDADDR 295 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 296 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 297 298 /* For convenience */ 299 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 300 301 #define LFS_SET_UINO(ip, flags) do { \ 302 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 303 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 304 ++(ip)->i_lfs->lfs_uinodes; \ 305 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 306 ++(ip)->i_lfs->lfs_uinodes; \ 307 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 308 ++(ip)->i_lfs->lfs_uinodes; \ 309 (ip)->i_flag |= (flags); \ 310 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 311 } while (0) 312 313 #define LFS_CLR_UINO(ip, flags) do { \ 314 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 315 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 316 --(ip)->i_lfs->lfs_uinodes; \ 317 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 318 --(ip)->i_lfs->lfs_uinodes; \ 319 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 320 --(ip)->i_lfs->lfs_uinodes; \ 321 (ip)->i_flag &= ~(flags); \ 322 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 323 panic("lfs_uinodes < 0"); \ 324 } \ 325 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 326 } while (0) 327 328 #define LFS_ITIMES(ip, acc, mod, cre) do { \ 329 struct lfs *_fs = (ip)->i_lfs; \ 330 \ 331 if ((ip)->i_flag & IN_ACCESS) { \ 332 (ip)->i_ffs1_atime = (acc)->tv_sec; \ 333 (ip)->i_ffs1_atimensec = (acc)->tv_nsec; \ 334 if ((ip)->i_lfs->lfs_version > 1) { \ 335 struct buf *_ibp; \ 336 IFILE *_ifp; \ 337 \ 338 LFS_IENTRY(_ifp, ip->i_lfs, ip->i_number, _ibp); \ 339 _ifp->if_atime_sec = (acc)->tv_sec; \ 340 _ifp->if_atime_nsec = (acc)->tv_nsec; \ 341 LFS_BWRITE_LOG(_ibp); \ 342 simple_lock(&_fs->lfs_interlock); \ 343 _fs->lfs_flags |= LFS_IFDIRTY; \ 344 simple_unlock(&_fs->lfs_interlock); \ 345 } else { \ 346 LFS_SET_UINO(ip, IN_ACCESSED); \ 347 } \ 348 } \ 349 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFY)) { \ 350 if ((ip)->i_flag & (IN_UPDATE | IN_MODIFY)) { \ 351 (ip)->i_ffs1_mtime = (mod)->tv_sec; \ 352 (ip)->i_ffs1_mtimensec = (mod)->tv_nsec; \ 353 (ip)->i_modrev++; \ 354 } \ 355 if ((ip)->i_flag & (IN_CHANGE | IN_MODIFY)) { \ 356 (ip)->i_ffs1_ctime = (cre)->tv_sec; \ 357 (ip)->i_ffs1_ctimensec = (cre)->tv_nsec; \ 358 } \ 359 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) \ 360 LFS_SET_UINO(ip, IN_MODIFIED); \ 361 if ((ip)->i_flag & IN_MODIFY) \ 362 LFS_SET_UINO(ip, IN_ACCESSED); \ 363 } \ 364 (ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY);\ 365 } while (0) 366 367 /* 368 * "struct vnode" associated definitions 369 */ 370 371 /* Heuristic emptiness measure */ 372 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 373 !((vp)->v_flag & VONWORKLST) && \ 374 VTOI(vp)->i_lfs_nbtree == 0) 375 376 /* XXX Shouldn't we use v_numoutput instead? */ 377 #define WRITEINPROG(vp) (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 378 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING))) 379 380 381 /* 382 * On-disk and in-memory checkpoint segment usage structure. 383 */ 384 typedef struct segusage SEGUSE; 385 struct segusage { 386 u_int32_t su_nbytes; /* 0: number of live bytes */ 387 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 388 u_int16_t su_nsums; /* 8: number of summaries in segment */ 389 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 390 391 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 392 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 393 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 394 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 395 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 396 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 397 u_int32_t su_flags; /* 12: segment flags */ 398 u_int64_t su_lastmod; /* 16: last modified timestamp */ 399 }; 400 401 typedef struct segusage_v1 SEGUSE_V1; 402 struct segusage_v1 { 403 u_int32_t su_nbytes; /* 0: number of live bytes */ 404 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 405 u_int16_t su_nsums; /* 8: number of summaries in segment */ 406 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 407 u_int32_t su_flags; /* 12: segment flags */ 408 }; 409 410 #define SEGUPB(fs) (fs->lfs_sepb) 411 #define SEGTABSIZE_SU(fs) \ 412 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 413 414 #ifdef _KERNEL 415 # define SHARE_IFLOCK(F) \ 416 do { \ 417 simple_lock(&(F)->lfs_interlock); \ 418 lockmgr(&(F)->lfs_iflock, LK_SHARED, &(F)->lfs_interlock); \ 419 simple_unlock(&(F)->lfs_interlock); \ 420 } while(0) 421 # define UNSHARE_IFLOCK(F) \ 422 do { \ 423 simple_lock(&(F)->lfs_interlock); \ 424 lockmgr(&(F)->lfs_iflock, LK_RELEASE, &(F)->lfs_interlock); \ 425 simple_unlock(&(F)->lfs_interlock); \ 426 } while(0) 427 #else /* ! _KERNEL */ 428 # define SHARE_IFLOCK(F) 429 # define UNSHARE_IFLOCK(F) 430 #endif /* ! _KERNEL */ 431 432 /* Read in the block with a specific segment usage entry from the ifile. */ 433 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 434 int _e; \ 435 SHARE_IFLOCK(F); \ 436 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 437 if ((_e = bread((F)->lfs_ivnode, \ 438 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 439 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 440 panic("lfs: ifile read: %d", _e); \ 441 if ((F)->lfs_version == 1) \ 442 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 443 ((IN) & ((F)->lfs_sepb - 1))); \ 444 else \ 445 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 446 UNSHARE_IFLOCK(F); \ 447 } while (0) 448 449 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 450 if ((SP)->su_nbytes == 0) \ 451 (SP)->su_flags |= SEGUSE_EMPTY; \ 452 else \ 453 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 454 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 455 LFS_BWRITE_LOG(BP); \ 456 } while (0) 457 458 /* 459 * On-disk file information. One per file with data blocks in the segment. 460 */ 461 typedef struct finfo FINFO; 462 struct finfo { 463 u_int32_t fi_nblocks; /* number of blocks */ 464 u_int32_t fi_version; /* version number */ 465 u_int32_t fi_ino; /* inode number */ 466 u_int32_t fi_lastlength; /* length of last block in array */ 467 int32_t fi_blocks[1]; /* array of logical block numbers */ 468 }; 469 /* sizeof FINFO except fi_blocks */ 470 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 471 472 /* 473 * Index file inode entries. 474 */ 475 typedef struct ifile IFILE; 476 struct ifile { 477 u_int32_t if_version; /* inode version number */ 478 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 479 int32_t if_daddr; /* inode disk address */ 480 u_int32_t if_nextfree; /* next-unallocated inode */ 481 u_int32_t if_atime_sec; /* Last access time, seconds */ 482 u_int32_t if_atime_nsec; /* and nanoseconds */ 483 }; 484 485 typedef struct ifile_v1 IFILE_V1; 486 struct ifile_v1 { 487 u_int32_t if_version; /* inode version number */ 488 int32_t if_daddr; /* inode disk address */ 489 u_int32_t if_nextfree; /* next-unallocated inode */ 490 #if LFS_ATIME_IFILE 491 struct timespec if_atime; /* Last access time */ 492 #endif 493 }; 494 495 /* 496 * LFSv1 compatibility code is not allowed to touch if_atime, since it 497 * may not be mapped! 498 */ 499 /* Read in the block with a specific inode from the ifile. */ 500 #define LFS_IENTRY(IP, F, IN, BP) do { \ 501 int _e; \ 502 SHARE_IFLOCK(F); \ 503 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 504 if ((_e = bread((F)->lfs_ivnode, \ 505 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 506 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 507 panic("lfs: ifile read %d", _e); \ 508 if ((F)->lfs_version == 1) \ 509 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 510 (IN) % (F)->lfs_ifpb); \ 511 else \ 512 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 513 UNSHARE_IFLOCK(F); \ 514 } while (0) 515 516 /* 517 * Cleaner information structure. This resides in the ifile and is used 518 * to pass information from the kernel to the cleaner. 519 */ 520 typedef struct _cleanerinfo { 521 u_int32_t clean; /* number of clean segments */ 522 u_int32_t dirty; /* number of dirty segments */ 523 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 /* Read in the block with the cleaner info from the ifile. */ 533 #define LFS_CLEANERINFO(CP, F, BP) do { \ 534 SHARE_IFLOCK(F); \ 535 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 536 if (bread((F)->lfs_ivnode, \ 537 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 538 panic("lfs: ifile read"); \ 539 (CP) = (CLEANERINFO *)(BP)->b_data; \ 540 UNSHARE_IFLOCK(F); \ 541 } while (0) 542 543 /* 544 * Synchronize the Ifile cleaner info with current avail and bfree. 545 */ 546 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 547 simple_lock(&(fs)->lfs_interlock); \ 548 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 549 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 550 (fs)->lfs_favail) { \ 551 (cip)->bfree = (fs)->lfs_bfree; \ 552 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 553 (fs)->lfs_favail; \ 554 if (((bp)->b_flags & B_GATHERED) == 0) { \ 555 (fs)->lfs_flags |= LFS_IFDIRTY; \ 556 } \ 557 simple_unlock(&(fs)->lfs_interlock); \ 558 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 559 } else { \ 560 simple_unlock(&(fs)->lfs_interlock); \ 561 brelse(bp); \ 562 } \ 563 } while (0) 564 565 /* 566 * Get the head of the inode free list. 567 * Always caled with the segment lock held. 568 */ 569 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 570 if ((FS)->lfs_version > 1) { \ 571 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 572 (FS)->lfs_freehd = (CIP)->free_head; \ 573 brelse(BP); \ 574 } \ 575 *(FREEP) = (FS)->lfs_freehd; \ 576 } while (0) 577 578 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 579 (FS)->lfs_freehd = (VAL); \ 580 if ((FS)->lfs_version > 1) { \ 581 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 582 (CIP)->free_head = (VAL); \ 583 LFS_BWRITE_LOG(BP); \ 584 simple_lock(&fs->lfs_interlock); \ 585 (FS)->lfs_flags |= LFS_IFDIRTY; \ 586 simple_unlock(&fs->lfs_interlock); \ 587 } \ 588 } while (0) 589 590 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 591 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 592 *(FREEP) = (CIP)->free_tail; \ 593 brelse(BP); \ 594 } while (0) 595 596 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 597 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 598 (CIP)->free_tail = (VAL); \ 599 LFS_BWRITE_LOG(BP); \ 600 simple_lock(&fs->lfs_interlock); \ 601 (FS)->lfs_flags |= LFS_IFDIRTY; \ 602 simple_unlock(&fs->lfs_interlock); \ 603 } while (0) 604 605 /* 606 * On-disk segment summary information 607 */ 608 typedef struct segsum_v1 SEGSUM_V1; 609 struct segsum_v1 { 610 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 611 u_int32_t ss_datasum; /* 4: check sum of data */ 612 u_int32_t ss_magic; /* 8: segment summary magic number */ 613 #define SS_MAGIC 0x061561 614 int32_t ss_next; /* 12: next segment */ 615 u_int32_t ss_create; /* 16: creation time stamp */ 616 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 617 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 618 619 #define SS_DIROP 0x01 /* segment begins a dirop */ 620 #define SS_CONT 0x02 /* more partials to finish this write*/ 621 u_int16_t ss_flags; /* 24: used for directory operations */ 622 u_int16_t ss_pad; /* 26: extra space */ 623 /* FINFO's and inode daddr's... */ 624 }; 625 626 typedef struct segsum SEGSUM; 627 struct segsum { 628 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 629 u_int32_t ss_datasum; /* 4: check sum of data */ 630 u_int32_t ss_magic; /* 8: segment summary magic number */ 631 int32_t ss_next; /* 12: next segment */ 632 u_int32_t ss_ident; /* 16: roll-forward fsid */ 633 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 634 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 635 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 636 u_int16_t ss_flags; /* 24: used for directory operations */ 637 u_int8_t ss_pad[6]; /* 26: extra space */ 638 u_int64_t ss_serial; /* 32: serial number */ 639 u_int64_t ss_create; /* 40: time stamp */ 640 /* FINFO's and inode daddr's... */ 641 }; 642 643 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 644 645 646 /* 647 * On-disk super block. 648 */ 649 struct dlfs { 650 #define LFS_MAGIC 0x070162 651 u_int32_t dlfs_magic; /* 0: magic number */ 652 #define LFS_VERSION 2 653 u_int32_t dlfs_version; /* 4: version number */ 654 655 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 656 /* number of frags in fs (v2) */ 657 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 658 /* number of bytes per segment (v2) */ 659 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 660 u_int32_t dlfs_bsize; /* 20: file system block size */ 661 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 662 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 663 664 /* Checkpoint region. */ 665 u_int32_t dlfs_freehd; /* 32: start of the free list */ 666 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 667 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 668 int32_t dlfs_avail; /* 44: blocks available for writing */ 669 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 670 int32_t dlfs_idaddr; /* 52: inode file disk address */ 671 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 672 int32_t dlfs_lastseg; /* 60: address of last segment written */ 673 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 674 int32_t dlfs_curseg; /* 68: current segment being written */ 675 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 676 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 677 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 678 #define dlfs_otstamp dlfs_inopf 679 680 /* These are configuration parameters. */ 681 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 682 683 /* These fields can be computed from the others. */ 684 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 685 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 686 u_int32_t dlfs_inopb; /* 100: inodes per block */ 687 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 688 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 689 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 690 u_int32_t dlfs_nseg; /* 116: number of segments */ 691 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 692 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 693 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 694 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 695 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 696 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 697 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 698 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 699 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 700 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 701 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 702 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 703 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 704 705 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 706 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 707 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 708 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 709 710 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 711 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 712 #define LFS_PF_CLEAN 0x1 713 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 714 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 715 u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */ 716 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 717 u_int64_t dlfs_serial; /* 336: serial number */ 718 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 719 int32_t dlfs_start; /* 348: start of segment 0 */ 720 u_int64_t dlfs_tstamp; /* 352: time stamp */ 721 #define LFS_44INODEFMT 0 722 #define LFS_MAXINODEFMT 0 723 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 724 u_int32_t dlfs_interleave; /* 364: segment interleave */ 725 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 726 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 727 int8_t dlfs_pad[132]; /* 376: round to 512 bytes */ 728 /* Checksum -- last valid disk field. */ 729 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 730 }; 731 732 /* 733 * In-memory super block. 734 */ 735 struct lfs { 736 struct dlfs lfs_dlfs; /* on-disk parameters */ 737 #define lfs_magic lfs_dlfs.dlfs_magic 738 #define lfs_version lfs_dlfs.dlfs_version 739 #define lfs_size lfs_dlfs.dlfs_size 740 #define lfs_ssize lfs_dlfs.dlfs_ssize 741 #define lfs_dsize lfs_dlfs.dlfs_dsize 742 #define lfs_bsize lfs_dlfs.dlfs_bsize 743 #define lfs_fsize lfs_dlfs.dlfs_fsize 744 #define lfs_frag lfs_dlfs.dlfs_frag 745 #define lfs_freehd lfs_dlfs.dlfs_freehd 746 #define lfs_bfree lfs_dlfs.dlfs_bfree 747 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 748 #define lfs_avail lfs_dlfs.dlfs_avail 749 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 750 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 751 #define lfs_ifile lfs_dlfs.dlfs_ifile 752 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 753 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 754 #define lfs_curseg lfs_dlfs.dlfs_curseg 755 #define lfs_offset lfs_dlfs.dlfs_offset 756 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 757 #define lfs_otstamp lfs_dlfs.dlfs_inopf 758 #define lfs_inopf lfs_dlfs.dlfs_inopf 759 #define lfs_minfree lfs_dlfs.dlfs_minfree 760 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 761 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 762 #define lfs_inopb lfs_dlfs.dlfs_inopb 763 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 764 #define lfs_sepb lfs_dlfs.dlfs_sepb 765 #define lfs_nindir lfs_dlfs.dlfs_nindir 766 #define lfs_nseg lfs_dlfs.dlfs_nseg 767 #define lfs_nspf lfs_dlfs.dlfs_nspf 768 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 769 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 770 #define lfs_segmask lfs_dlfs.dlfs_segmask 771 #define lfs_segshift lfs_dlfs.dlfs_segshift 772 #define lfs_bmask lfs_dlfs.dlfs_bmask 773 #define lfs_bshift lfs_dlfs.dlfs_bshift 774 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 775 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 776 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 777 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 778 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 779 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 780 #define lfs_sushift lfs_dlfs.dlfs_sushift 781 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 782 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 783 #define lfs_cksum lfs_dlfs.dlfs_cksum 784 #define lfs_pflags lfs_dlfs.dlfs_pflags 785 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 786 #define lfs_nclean lfs_dlfs.dlfs_nclean 787 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 788 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 789 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 790 #define lfs_serial lfs_dlfs.dlfs_serial 791 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 792 #define lfs_start lfs_dlfs.dlfs_start 793 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 794 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 795 #define lfs_interleave lfs_dlfs.dlfs_interleave 796 #define lfs_ident lfs_dlfs.dlfs_ident 797 798 /* These fields are set at mount time and are meaningless on disk. */ 799 struct segment *lfs_sp; /* current segment being written */ 800 struct vnode *lfs_ivnode; /* vnode for the ifile */ 801 u_int32_t lfs_seglock; /* single-thread the segment writer */ 802 pid_t lfs_lockpid; /* pid of lock holder */ 803 u_int32_t lfs_iocount; /* number of ios pending */ 804 u_int32_t lfs_writer; /* don't allow any dirops to start */ 805 u_int32_t lfs_dirops; /* count of active directory ops */ 806 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 807 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 808 int8_t lfs_fmod; /* super block modified flag */ 809 int8_t lfs_ronly; /* mounted read-only flag */ 810 #define LFS_NOTYET 0x01 811 #define LFS_IFDIRTY 0x02 812 #define LFS_WARNED 0x04 813 #define LFS_UNDIROP 0x08 814 int8_t lfs_flags; /* currently unused flag */ 815 u_int16_t lfs_activesb; /* toggle between superblocks */ 816 daddr_t lfs_sbactive; /* disk address of current sb write */ 817 struct vnode *lfs_flushvp; /* vnode being flushed */ 818 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 819 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 820 size_t lfs_devbsize; /* Device block size */ 821 size_t lfs_devbshift; /* Device block shift */ 822 struct lock lfs_fraglock; 823 struct lock lfs_iflock; /* Ifile lock */ 824 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 825 int lfs_nadirop; /* number of active dirop nodes */ 826 long lfs_ravail; /* blocks pre-reserved for writing */ 827 long lfs_favail; /* blocks pre-reserved for writing */ 828 res_t *lfs_resblk; /* Reserved memory for pageout */ 829 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 830 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 831 #define LFS_RESHASH_WIDTH 17 832 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 833 int lfs_pdflush; /* pagedaemon wants us to flush */ 834 u_int32_t **lfs_suflags; /* Segment use flags */ 835 #ifdef _KERNEL 836 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 837 struct pool lfs_bpppool; /* Pool for bpp */ 838 struct pool lfs_segpool; /* Pool for struct segment */ 839 #endif /* _KERNEL */ 840 #define LFS_MAX_CLEANIND 64 841 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 842 int lfs_cleanind; /* Index into intervals */ 843 struct simplelock lfs_interlock; /* lock for lfs_seglock */ 844 int lfs_sleepers; /* # procs sleeping this fs */ 845 int lfs_pages; /* dirty pages blaming this fs */ 846 }; 847 848 /* NINDIR is the number of indirects in a file system block. */ 849 #define NINDIR(fs) ((fs)->lfs_nindir) 850 851 /* INOPB is the number of inodes in a secondary storage block. */ 852 #define INOPB(fs) ((fs)->lfs_inopb) 853 /* INOPF is the number of inodes in a fragment. */ 854 #define INOPF(fs) ((fs)->lfs_inopf) 855 856 #define blksize(fs, ip, lbn) \ 857 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 858 ? (fs)->lfs_bsize \ 859 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 860 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 861 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 862 ((int)((loc) & (fs)->lfs_ffmask)) 863 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 864 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 865 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 866 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 867 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 868 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 869 /* Same as above, but named like dbtob(), btodb() */ 870 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 871 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 872 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 873 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 874 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 875 (fs)->lfs_fsbtodb)) 876 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 877 (fs)->lfs_fsbtodb)) 878 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 879 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 880 ((loc) >> (fs)->lfs_ffshift) 881 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 882 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 883 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 884 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 885 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 886 ((frags) >> (fs)->lfs_fbshift) 887 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 888 ((blks) << (fs)->lfs_fbshift) 889 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 890 ((fsb) & ((fs)->lfs_frag - 1)) 891 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 892 ((fsb) &~ ((fs)->lfs_frag - 1)) 893 #define dblksize(fs, dp, lbn) \ 894 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 895 ? (fs)->lfs_bsize \ 896 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 897 898 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 899 lblktosize((fs), (fs)->lfs_ssize) : \ 900 (fs)->lfs_ssize) 901 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 902 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 903 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 904 #define dtosn(fs, daddr) /* block address to segment number */ \ 905 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 906 #define sntod(fs, sn) /* segment number to disk address */ \ 907 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 908 909 /* 910 * Structures used by lfs_bmapv and lfs_markv to communicate information 911 * about inodes and data blocks. 912 */ 913 typedef struct block_info { 914 u_int32_t bi_inode; /* inode # */ 915 int32_t bi_lbn; /* logical block w/in file */ 916 int32_t bi_daddr; /* disk address of block */ 917 u_int64_t bi_segcreate; /* origin segment create time */ 918 int bi_version; /* file version number */ 919 void *bi_bp; /* data buffer */ 920 int bi_size; /* size of the block (if fragment) */ 921 } BLOCK_INFO; 922 923 /* Compatibility for 1.5 binaries */ 924 typedef struct block_info_15 { 925 u_int32_t bi_inode; /* inode # */ 926 int32_t bi_lbn; /* logical block w/in file */ 927 int32_t bi_daddr; /* disk address of block */ 928 u_int32_t bi_segcreate; /* origin segment create time */ 929 int bi_version; /* file version number */ 930 void *bi_bp; /* data buffer */ 931 int bi_size; /* size of the block (if fragment) */ 932 } BLOCK_INFO_15; 933 934 /* In-memory description of a segment about to be written. */ 935 struct segment { 936 struct lfs *fs; /* file system pointer */ 937 struct buf **bpp; /* pointer to buffer array */ 938 struct buf **cbpp; /* pointer to next available bp */ 939 struct buf **start_bpp; /* pointer to first bp in this set */ 940 struct buf *ibp; /* buffer pointer to inode page */ 941 struct ufs1_dinode *idp; /* pointer to ifile dinode */ 942 struct finfo *fip; /* current fileinfo pointer */ 943 struct vnode *vp; /* vnode being gathered */ 944 void *segsum; /* segment summary info */ 945 u_int32_t ninodes; /* number of inodes in this segment */ 946 int32_t seg_bytes_left; /* bytes left in segment */ 947 int32_t sum_bytes_left; /* bytes left in summary block */ 948 u_int32_t seg_number; /* number of this segment */ 949 int32_t *start_lbp; /* beginning lbn for this set */ 950 951 #define SEGM_CKP 0x01 /* doing a checkpoint */ 952 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 953 #define SEGM_SYNC 0x04 /* wait for segment */ 954 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 955 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */ 956 #define SEGM_WRITERD 0x20 /* LFS writed called us */ 957 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */ 958 u_int16_t seg_flags; /* run-time flags for this segment */ 959 u_int32_t seg_iocount; /* number of ios pending */ 960 int ndupino; /* number of duplicate inodes */ 961 }; 962 963 #ifdef _KERNEL 964 struct lfs_cluster { 965 size_t bufsize; /* Size of kept data */ 966 struct buf **bpp; /* Array of kept buffers */ 967 int bufcount; /* Number of kept buffers */ 968 #define LFS_CL_MALLOC 0x00000001 969 #define LFS_CL_SHIFT 0x00000002 970 #define LFS_CL_SYNC 0x00000004 971 u_int32_t flags; /* Flags */ 972 struct lfs *fs; /* LFS that this belongs to */ 973 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 974 }; 975 976 /* 977 * Splay tree containing block numbers allocated through lfs_balloc. 978 */ 979 struct lbnentry { 980 SPLAY_ENTRY(lbnentry) entry; 981 daddr_t lbn; 982 }; 983 #endif /* _KERNEL */ 984 985 /* 986 * LFS inode extensions. 987 */ 988 struct lfs_inode_ext { 989 off_t lfs_osize; /* size of file on disk */ 990 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 991 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */ 992 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 993 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 994 #define LFSI_NO_GOP_WRITE 0x01 995 u_int32_t lfs_iflags; /* Inode flags */ 996 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 997 #ifdef _KERNEL 998 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 999 int lfs_nbtree; /* Size of tree */ 1000 #endif 1001 }; 1002 #define i_lfs_osize inode_ext.lfs->lfs_osize 1003 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 1004 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 1005 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 1006 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 1007 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 1008 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 1009 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 1010 1011 /* 1012 * Macros for determining free space on the disk, with the variable metadata 1013 * of segment summaries and inode blocks taken into account. 1014 */ 1015 /* Estimate number of clean blocks not available for writing */ 1016 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 1017 (int64_t)(F)->lfs_nclean) / \ 1018 ((F)->lfs_nseg - (F)->lfs_nclean))) 1019 1020 /* Estimate total size of the disk not including metadata */ 1021 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 1022 1023 /* Estimate number of blocks actually available for writing */ 1024 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) + (F)->lfs_dmeta ? (F)->lfs_bfree - LFS_EST_CMETA(F) - (F)->lfs_dmeta : 0) 1025 1026 /* Amount of non-meta space not available to mortal man */ 1027 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1028 (u_int64_t)(F)->lfs_minfree) / \ 1029 100) 1030 1031 /* Can credential C write BB blocks */ 1032 #define ISSPACE(F, BB, C) \ 1033 ((((C) == NOCRED || (C)->cr_uid == 0) && \ 1034 LFS_EST_BFREE(F) >= (BB)) || \ 1035 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB))) 1036 1037 /* Can an ordinary user write BB blocks */ 1038 #define IS_FREESPACE(F, BB) \ 1039 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1040 1041 /* 1042 * The minimum number of blocks to create a new inode. This is: 1043 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1044 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1045 */ 1046 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1047 1048 /* Statistics Counters */ 1049 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1050 u_int segsused; 1051 u_int psegwrites; 1052 u_int psyncwrites; 1053 u_int pcleanwrites; 1054 u_int blocktot; 1055 u_int cleanblocks; 1056 u_int ncheckpoints; 1057 u_int nwrites; 1058 u_int nsync_writes; 1059 u_int wait_exceeded; 1060 u_int write_exceeded; 1061 u_int flush_invoked; 1062 u_int vflush_invoked; 1063 u_int clean_inlocked; 1064 u_int clean_vnlocked; 1065 u_int segs_reclaimed; 1066 }; 1067 #ifdef _KERNEL 1068 extern struct lfs_stats lfs_stats; 1069 #endif 1070 1071 /* Fcntls to take the place of the lfs syscalls */ 1072 struct lfs_fcntl_markv { 1073 BLOCK_INFO *blkiov; /* blocks to relocate */ 1074 int blkcnt; /* number of blocks */ 1075 }; 1076 1077 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1078 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1079 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1080 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1081 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1082 #define LFCNIFILEFH _FCNW_FSPRIV('L', 5, struct fhandle) 1083 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1084 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1085 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1086 /* Compat for NetBSD 2.x bug */ 1087 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1088 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1089 1090 #ifdef _KERNEL 1091 /* XXX MP */ 1092 #define LFS_SEGLOCK_HELD(fs) \ 1093 ((fs)->lfs_seglock != 0 && (fs)->lfs_lockpid == curproc->p_pid) 1094 #endif /* _KERNEL */ 1095 1096 /* Debug segment lock */ 1097 #ifdef notyet 1098 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1099 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1100 # define ASSERT_DUNNO_SEGLOCK(fs) 1101 # define ASSERT_MAYBE_SEGLOCK(fs) 1102 #else /* !notyet */ 1103 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1104 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1105 LFS_SEGLOCK_HELD(fs))) 1106 # define ASSERT_SEGLOCK(fs) do { \ 1107 if (!LFS_SEGLOCK_HELD(fs)) { \ 1108 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1109 } \ 1110 } while(0) 1111 # define ASSERT_NO_SEGLOCK(fs) do { \ 1112 if (LFS_SEGLOCK_HELD(fs)) { \ 1113 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1114 } \ 1115 } while(0) 1116 # define ASSERT_MAYBE_SEGLOCK(x) 1117 #endif /* !notyet */ 1118 1119 #endif /* !_UFS_LFS_LFS_H_ */ 1120
Indexes created Mon Sep 29 21:09:56 GMT 2025