lfs.h revision 1.88
1 /* $NetBSD: lfs.h,v 1.88 2005/05/29 21:25:24 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 /* Adjustable filesystem parameters */ 97 #define MIN_FREE_SEGS 2 98 #ifndef LFS_ATIME_IFILE 99 # define LFS_ATIME_IFILE 0 /* Store atime info in ifile (optional in LFSv1) */ 100 #endif 101 #define LFS_MARKV_MAXBLKCNT 65536 /* Max block count for lfs_markv() */ 102 103 /* Misc. definitions */ 104 #define BW_CLEAN 1 /* Flag for lfs_bwrite_ext() */ 105 #define PG_DELWRI PG_PAGER1 /* Local def for delayed pageout */ 106 107 /* Resource limits */ 108 #define LFS_MAX_BUFS ((nbuf >> 2) - 10) 109 #define LFS_WAIT_BUFS ((nbuf >> 1) - (nbuf >> 3) - 10) 110 #define LFS_INVERSE_MAX_BUFS(n) (((n) + 10) << 2) 111 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 112 #define LFS_MAX_BYTES ((bufmem_lowater >> 2) - 10 * PAGE_SIZE) 113 #define LFS_INVERSE_MAX_BYTES(n) (((n) + 10 * PAGE_SIZE) << 2) 114 #define LFS_WAIT_BYTES ((bufmem_lowater >> 1) - (bufmem_lowater >> 3) - 10 * PAGE_SIZE) 115 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3)) 116 #define LFS_MAX_PAGES \ 117 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemin) >> 8) 118 #define LFS_WAIT_PAGES \ 119 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemax) >> 8) 120 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */ 121 122 /* 123 * Reserved blocks for lfs_malloc 124 */ 125 126 /* Structure to keep reserved blocks */ 127 typedef struct lfs_res_blk { 128 void *p; 129 LIST_ENTRY(lfs_res_blk) res; 130 int size; 131 char inuse; 132 } res_t; 133 134 /* Types for lfs_newbuf and lfs_malloc */ 135 #define LFS_NB_UNKNOWN -1 136 #define LFS_NB_SUMMARY 0 137 #define LFS_NB_SBLOCK 1 138 #define LFS_NB_IBLOCK 2 139 #define LFS_NB_CLUSTER 3 140 #define LFS_NB_CLEAN 4 141 #define LFS_NB_BLKIOV 5 142 #define LFS_NB_COUNT 6 /* always last */ 143 144 /* Number of reserved memory blocks of each type */ 145 #define LFS_N_SUMMARIES 2 146 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 147 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 148 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 149 #define LFS_N_CLEAN 0 150 #define LFS_N_BLKIOV 1 151 152 /* Total count of "large" (non-pool) types */ 153 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 154 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV) 155 156 /* Counts for pool types */ 157 #define LFS_N_CL LFS_N_CLUSTERS 158 #define LFS_N_BPP 2 159 #define LFS_N_SEG 2 160 161 /* 162 * "struct buf" associated definitions 163 */ 164 165 /* Unassigned disk addresses. */ 166 #define UNASSIGNED -1 167 #define UNWRITTEN -2 168 169 /* Unused logical block number */ 170 #define LFS_UNUSED_LBN -1 171 172 /* Determine if a buffer belongs to the ifile */ 173 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 174 175 # define LFS_LOCK_BUF(bp) do { \ 176 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \ 177 simple_lock(&lfs_subsys_lock); \ 178 ++locked_queue_count; \ 179 locked_queue_bytes += bp->b_bufsize; \ 180 simple_unlock(&lfs_subsys_lock); \ 181 } \ 182 (bp)->b_flags |= B_LOCKED; \ 183 } while (0) 184 185 # define LFS_UNLOCK_BUF(bp) do { \ 186 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \ 187 simple_lock(&lfs_subsys_lock); \ 188 --locked_queue_count; \ 189 locked_queue_bytes -= bp->b_bufsize; \ 190 if (locked_queue_count < LFS_WAIT_BUFS && \ 191 locked_queue_bytes < LFS_WAIT_BYTES) \ 192 wakeup(&locked_queue_count); \ 193 simple_unlock(&lfs_subsys_lock); \ 194 } \ 195 (bp)->b_flags &= ~B_LOCKED; \ 196 } while (0) 197 198 #ifdef _KERNEL 199 # define LFS_IS_MALLOC_BUF(bp) (((bp)->b_flags & B_CALL) && \ 200 (bp)->b_iodone == lfs_callback) 201 202 # ifdef DEBUG 203 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 204 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 205 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 206 wakeup(&locked_queue_count); \ 207 } \ 208 } while (0) 209 # else 210 # define LFS_DEBUG_COUNTLOCKED(m) 211 # endif 212 213 /* log for debugging writes to the Ifile */ 214 # ifdef DEBUG 215 struct lfs_log_entry { 216 const char *op; 217 const char *file; 218 int pid; 219 int line; 220 daddr_t block; 221 unsigned long flags; 222 }; 223 extern int lfs_lognum; 224 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 225 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 226 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 227 int _s; \ 228 \ 229 simple_lock(&lfs_subsys_lock); \ 230 _s = splbio(); \ 231 lfs_log[lfs_lognum].op = theop; \ 232 lfs_log[lfs_lognum].file = thefile; \ 233 lfs_log[lfs_lognum].line = (theline); \ 234 lfs_log[lfs_lognum].pid = (thepid); \ 235 lfs_log[lfs_lognum].block = (lbn); \ 236 lfs_log[lfs_lognum].flags = (theflags); \ 237 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 238 splx(_s); \ 239 simple_unlock(&lfs_subsys_lock); \ 240 } while (0) 241 242 # define LFS_BCLEAN_LOG(fs, bp) do { \ 243 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 244 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 245 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 246 } while (0) 247 248 /* Must match list in lfs_vfsops.c ! */ 249 # define DLOG_RF 0 /* roll forward */ 250 # define DLOG_ALLOC 1 /* inode alloc */ 251 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 252 # define DLOG_FLUSH 3 /* flush */ 253 # define DLOG_LLIST 4 /* locked list accounting */ 254 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 255 # define DLOG_VNODE 6 /* vflush/writevnodes */ 256 # define DLOG_SEG 7 /* segwrite */ 257 # define DLOG_SU 8 /* seguse accounting */ 258 # define DLOG_CLEAN 9 /* cleaner routines */ 259 # define DLOG_MOUNT 10 /* mount/unmount */ 260 # define DLOG_PAGE 11 /* putpages/gop_write */ 261 # define DLOG_DIROP 12 /* dirop accounting */ 262 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 263 # define DLOG_MAX 14 /* The terminator */ 264 # define DLOG(a) lfs_debug_log a 265 # else /* ! DEBUG */ 266 # define LFS_BCLEAN_LOG(fs, bp) 267 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 268 # define DLOG(a) 269 # endif /* ! DEBUG */ 270 #else /* ! _KERNEL */ 271 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 272 #endif /* _KERNEL */ 273 274 #ifdef _KERNEL 275 /* Filehandle structure for exported LFSes */ 276 struct lfid { 277 struct ufid lfid_ufid; 278 #define lfid_len lfid_ufid.ufid_len 279 #define lfid_ino lfid_ufid.ufid_ino 280 #define lfid_gen lfid_ufid.ufid_gen 281 uint32_t lfid_ident; 282 }; 283 #endif /* _KERNEL */ 284 285 /* 286 * "struct inode" associated definitions 287 */ 288 289 /* Address calculations for metadata located in the inode */ 290 #define S_INDIR(fs) -NDADDR 291 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 292 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 293 294 /* For convenience */ 295 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 296 297 #define LFS_SET_UINO(ip, flags) do { \ 298 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 299 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 300 ++(ip)->i_lfs->lfs_uinodes; \ 301 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 302 ++(ip)->i_lfs->lfs_uinodes; \ 303 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 304 ++(ip)->i_lfs->lfs_uinodes; \ 305 (ip)->i_flag |= (flags); \ 306 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 307 } while (0) 308 309 #define LFS_CLR_UINO(ip, flags) do { \ 310 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 311 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 312 --(ip)->i_lfs->lfs_uinodes; \ 313 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 314 --(ip)->i_lfs->lfs_uinodes; \ 315 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 316 --(ip)->i_lfs->lfs_uinodes; \ 317 (ip)->i_flag &= ~(flags); \ 318 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 319 panic("lfs_uinodes < 0"); \ 320 } \ 321 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 322 } while (0) 323 324 #define LFS_ITIMES(ip, acc, mod, cre) do { \ 325 struct lfs *_fs = (ip)->i_lfs; \ 326 \ 327 if ((ip)->i_flag & IN_ACCESS) { \ 328 (ip)->i_ffs1_atime = (acc)->tv_sec; \ 329 (ip)->i_ffs1_atimensec = (acc)->tv_nsec; \ 330 if ((ip)->i_lfs->lfs_version > 1) { \ 331 struct buf *_ibp; \ 332 IFILE *_ifp; \ 333 \ 334 LFS_IENTRY(_ifp, ip->i_lfs, ip->i_number, _ibp); \ 335 _ifp->if_atime_sec = (acc)->tv_sec; \ 336 _ifp->if_atime_nsec = (acc)->tv_nsec; \ 337 LFS_BWRITE_LOG(_ibp); \ 338 simple_lock(&_fs->lfs_interlock); \ 339 _fs->lfs_flags |= LFS_IFDIRTY; \ 340 simple_unlock(&_fs->lfs_interlock); \ 341 } else { \ 342 LFS_SET_UINO(ip, IN_ACCESSED); \ 343 } \ 344 } \ 345 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFY)) { \ 346 if ((ip)->i_flag & (IN_UPDATE | IN_MODIFY)) { \ 347 (ip)->i_ffs1_mtime = (mod)->tv_sec; \ 348 (ip)->i_ffs1_mtimensec = (mod)->tv_nsec; \ 349 (ip)->i_modrev++; \ 350 } \ 351 if ((ip)->i_flag & (IN_CHANGE | IN_MODIFY)) { \ 352 (ip)->i_ffs1_ctime = (cre)->tv_sec; \ 353 (ip)->i_ffs1_ctimensec = (cre)->tv_nsec; \ 354 } \ 355 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) \ 356 LFS_SET_UINO(ip, IN_MODIFIED); \ 357 if ((ip)->i_flag & IN_MODIFY) \ 358 LFS_SET_UINO(ip, IN_ACCESSED); \ 359 } \ 360 (ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY);\ 361 } while (0) 362 363 /* 364 * "struct vnode" associated definitions 365 */ 366 367 /* Heuristic emptiness measure */ 368 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 369 !((vp)->v_flag & VONWORKLST) && \ 370 VTOI(vp)->i_lfs_nbtree == 0) 371 372 /* XXX Shouldn't we use v_numoutput instead? */ 373 #define WRITEINPROG(vp) (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 374 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING))) 375 376 377 /* 378 * On-disk and in-memory checkpoint segment usage structure. 379 */ 380 typedef struct segusage SEGUSE; 381 struct segusage { 382 u_int32_t su_nbytes; /* 0: number of live bytes */ 383 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 384 u_int16_t su_nsums; /* 8: number of summaries in segment */ 385 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 386 387 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 388 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 389 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 390 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 391 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 392 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 393 u_int32_t su_flags; /* 12: segment flags */ 394 u_int64_t su_lastmod; /* 16: last modified timestamp */ 395 }; 396 397 typedef struct segusage_v1 SEGUSE_V1; 398 struct segusage_v1 { 399 u_int32_t su_nbytes; /* 0: number of live bytes */ 400 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 401 u_int16_t su_nsums; /* 8: number of summaries in segment */ 402 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 403 u_int32_t su_flags; /* 12: segment flags */ 404 }; 405 406 #define SEGUPB(fs) (fs->lfs_sepb) 407 #define SEGTABSIZE_SU(fs) \ 408 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 409 410 #ifdef _KERNEL 411 # define SHARE_IFLOCK(F) \ 412 do { \ 413 simple_lock(&(F)->lfs_interlock); \ 414 lockmgr(&(F)->lfs_iflock, LK_SHARED, &(F)->lfs_interlock); \ 415 simple_unlock(&(F)->lfs_interlock); \ 416 } while(0) 417 # define UNSHARE_IFLOCK(F) \ 418 do { \ 419 simple_lock(&(F)->lfs_interlock); \ 420 lockmgr(&(F)->lfs_iflock, LK_RELEASE, &(F)->lfs_interlock); \ 421 simple_unlock(&(F)->lfs_interlock); \ 422 } while(0) 423 #else /* ! _KERNEL */ 424 # define SHARE_IFLOCK(F) 425 # define UNSHARE_IFLOCK(F) 426 #endif /* ! _KERNEL */ 427 428 /* Read in the block with a specific segment usage entry from the ifile. */ 429 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 430 int _e; \ 431 SHARE_IFLOCK(F); \ 432 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 433 if ((_e = bread((F)->lfs_ivnode, \ 434 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 435 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 436 panic("lfs: ifile read: %d", _e); \ 437 if ((F)->lfs_version == 1) \ 438 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 439 ((IN) & ((F)->lfs_sepb - 1))); \ 440 else \ 441 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 442 UNSHARE_IFLOCK(F); \ 443 } while (0) 444 445 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 446 if ((SP)->su_nbytes == 0) \ 447 (SP)->su_flags |= SEGUSE_EMPTY; \ 448 else \ 449 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 450 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 451 LFS_BWRITE_LOG(BP); \ 452 } while (0) 453 454 /* 455 * On-disk file information. One per file with data blocks in the segment. 456 */ 457 typedef struct finfo FINFO; 458 struct finfo { 459 u_int32_t fi_nblocks; /* number of blocks */ 460 u_int32_t fi_version; /* version number */ 461 u_int32_t fi_ino; /* inode number */ 462 u_int32_t fi_lastlength; /* length of last block in array */ 463 int32_t fi_blocks[1]; /* array of logical block numbers */ 464 }; 465 /* sizeof FINFO except fi_blocks */ 466 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 467 468 /* 469 * Index file inode entries. 470 */ 471 typedef struct ifile IFILE; 472 struct ifile { 473 u_int32_t if_version; /* inode version number */ 474 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 475 int32_t if_daddr; /* inode disk address */ 476 ino_t if_nextfree; /* next-unallocated inode */ 477 u_int32_t if_atime_sec; /* Last access time, seconds */ 478 u_int32_t if_atime_nsec; /* and nanoseconds */ 479 }; 480 481 typedef struct ifile_v1 IFILE_V1; 482 struct ifile_v1 { 483 u_int32_t if_version; /* inode version number */ 484 int32_t if_daddr; /* inode disk address */ 485 ino_t if_nextfree; /* next-unallocated inode */ 486 #if LFS_ATIME_IFILE 487 struct timespec if_atime; /* Last access time */ 488 #endif 489 }; 490 491 /* 492 * LFSv1 compatibility code is not allowed to touch if_atime, since it 493 * may not be mapped! 494 */ 495 /* Read in the block with a specific inode from the ifile. */ 496 #define LFS_IENTRY(IP, F, IN, BP) do { \ 497 int _e; \ 498 SHARE_IFLOCK(F); \ 499 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 500 if ((_e = bread((F)->lfs_ivnode, \ 501 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 502 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 503 panic("lfs: ifile read %d", _e); \ 504 if ((F)->lfs_version == 1) \ 505 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 506 (IN) % (F)->lfs_ifpb); \ 507 else \ 508 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 509 UNSHARE_IFLOCK(F); \ 510 } while (0) 511 512 /* 513 * Cleaner information structure. This resides in the ifile and is used 514 * to pass information from the kernel to the cleaner. 515 */ 516 typedef struct _cleanerinfo { 517 u_int32_t clean; /* number of clean segments */ 518 u_int32_t dirty; /* number of dirty segments */ 519 int32_t bfree; /* disk blocks free */ 520 int32_t avail; /* disk blocks available */ 521 u_int32_t free_head; /* head of the inode free list */ 522 u_int32_t free_tail; /* tail of the inode free list */ 523 } CLEANERINFO; 524 525 #define CLEANSIZE_SU(fs) \ 526 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 527 528 /* Read in the block with the cleaner info from the ifile. */ 529 #define LFS_CLEANERINFO(CP, F, BP) do { \ 530 SHARE_IFLOCK(F); \ 531 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 532 if (bread((F)->lfs_ivnode, \ 533 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 534 panic("lfs: ifile read"); \ 535 (CP) = (CLEANERINFO *)(BP)->b_data; \ 536 UNSHARE_IFLOCK(F); \ 537 } while (0) 538 539 /* 540 * Synchronize the Ifile cleaner info with current avail and bfree. 541 */ 542 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 543 simple_lock(&(fs)->lfs_interlock); \ 544 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 545 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 546 (fs)->lfs_favail) { \ 547 (cip)->bfree = (fs)->lfs_bfree; \ 548 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 549 (fs)->lfs_favail; \ 550 if (((bp)->b_flags & B_GATHERED) == 0) { \ 551 (fs)->lfs_flags |= LFS_IFDIRTY; \ 552 } \ 553 simple_unlock(&(fs)->lfs_interlock); \ 554 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 555 } else { \ 556 simple_unlock(&(fs)->lfs_interlock); \ 557 brelse(bp); \ 558 } \ 559 } while (0) 560 561 /* 562 * Get the head of the inode free list. 563 * Always caled with the segment lock held. 564 */ 565 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 566 if ((FS)->lfs_version > 1) { \ 567 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 568 (FS)->lfs_freehd = (CIP)->free_head; \ 569 brelse(BP); \ 570 } \ 571 *(FREEP) = (FS)->lfs_freehd; \ 572 } while (0) 573 574 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 575 (FS)->lfs_freehd = (VAL); \ 576 if ((FS)->lfs_version > 1) { \ 577 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 578 (CIP)->free_head = (VAL); \ 579 LFS_BWRITE_LOG(BP); \ 580 simple_lock(&fs->lfs_interlock); \ 581 (FS)->lfs_flags |= LFS_IFDIRTY; \ 582 simple_unlock(&fs->lfs_interlock); \ 583 } \ 584 } while (0) 585 586 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 587 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 588 *(FREEP) = (CIP)->free_tail; \ 589 brelse(BP); \ 590 } while (0) 591 592 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 593 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 594 (CIP)->free_tail = (VAL); \ 595 LFS_BWRITE_LOG(BP); \ 596 simple_lock(&fs->lfs_interlock); \ 597 (FS)->lfs_flags |= LFS_IFDIRTY; \ 598 simple_unlock(&fs->lfs_interlock); \ 599 } while (0) 600 601 /* 602 * On-disk segment summary information 603 */ 604 typedef struct segsum_v1 SEGSUM_V1; 605 struct segsum_v1 { 606 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 607 u_int32_t ss_datasum; /* 4: check sum of data */ 608 u_int32_t ss_magic; /* 8: segment summary magic number */ 609 #define SS_MAGIC 0x061561 610 int32_t ss_next; /* 12: next segment */ 611 u_int32_t ss_create; /* 16: creation time stamp */ 612 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 613 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 614 615 #define SS_DIROP 0x01 /* segment begins a dirop */ 616 #define SS_CONT 0x02 /* more partials to finish this write*/ 617 u_int16_t ss_flags; /* 24: used for directory operations */ 618 u_int16_t ss_pad; /* 26: extra space */ 619 /* FINFO's and inode daddr's... */ 620 }; 621 622 typedef struct segsum SEGSUM; 623 struct segsum { 624 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 625 u_int32_t ss_datasum; /* 4: check sum of data */ 626 u_int32_t ss_magic; /* 8: segment summary magic number */ 627 int32_t ss_next; /* 12: next segment */ 628 u_int32_t ss_ident; /* 16: roll-forward fsid */ 629 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 630 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 631 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 632 u_int16_t ss_flags; /* 24: used for directory operations */ 633 u_int8_t ss_pad[6]; /* 26: extra space */ 634 u_int64_t ss_serial; /* 32: serial number */ 635 u_int64_t ss_create; /* 40: time stamp */ 636 /* FINFO's and inode daddr's... */ 637 }; 638 639 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 640 641 642 /* 643 * On-disk super block. 644 */ 645 struct dlfs { 646 #define LFS_MAGIC 0x070162 647 u_int32_t dlfs_magic; /* 0: magic number */ 648 #define LFS_VERSION 2 649 u_int32_t dlfs_version; /* 4: version number */ 650 651 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 652 /* number of frags in fs (v2) */ 653 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 654 /* number of bytes per segment (v2) */ 655 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 656 u_int32_t dlfs_bsize; /* 20: file system block size */ 657 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 658 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 659 660 /* Checkpoint region. */ 661 u_int32_t dlfs_freehd; /* 32: start of the free list */ 662 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 663 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 664 int32_t dlfs_avail; /* 44: blocks available for writing */ 665 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 666 int32_t dlfs_idaddr; /* 52: inode file disk address */ 667 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 668 int32_t dlfs_lastseg; /* 60: address of last segment written */ 669 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 670 int32_t dlfs_curseg; /* 68: current segment being written */ 671 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 672 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 673 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 674 #define dlfs_otstamp dlfs_inopf 675 676 /* These are configuration parameters. */ 677 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 678 679 /* These fields can be computed from the others. */ 680 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 681 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 682 u_int32_t dlfs_inopb; /* 100: inodes per block */ 683 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 684 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 685 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 686 u_int32_t dlfs_nseg; /* 116: number of segments */ 687 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 688 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 689 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 690 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 691 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 692 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 693 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 694 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 695 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 696 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 697 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 698 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 699 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 700 701 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 702 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 703 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 704 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 705 706 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 707 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 708 #define LFS_PF_CLEAN 0x1 709 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 710 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 711 u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */ 712 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 713 u_int64_t dlfs_serial; /* 336: serial number */ 714 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 715 int32_t dlfs_start; /* 348: start of segment 0 */ 716 u_int64_t dlfs_tstamp; /* 352: time stamp */ 717 #define LFS_44INODEFMT 0 718 #define LFS_MAXINODEFMT 0 719 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 720 u_int32_t dlfs_interleave; /* 364: segment interleave */ 721 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 722 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 723 int8_t dlfs_pad[132]; /* 376: round to 512 bytes */ 724 /* Checksum -- last valid disk field. */ 725 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 726 }; 727 728 /* 729 * In-memory super block. 730 */ 731 struct lfs { 732 struct dlfs lfs_dlfs; /* on-disk parameters */ 733 #define lfs_magic lfs_dlfs.dlfs_magic 734 #define lfs_version lfs_dlfs.dlfs_version 735 #define lfs_size lfs_dlfs.dlfs_size 736 #define lfs_ssize lfs_dlfs.dlfs_ssize 737 #define lfs_dsize lfs_dlfs.dlfs_dsize 738 #define lfs_bsize lfs_dlfs.dlfs_bsize 739 #define lfs_fsize lfs_dlfs.dlfs_fsize 740 #define lfs_frag lfs_dlfs.dlfs_frag 741 #define lfs_freehd lfs_dlfs.dlfs_freehd 742 #define lfs_bfree lfs_dlfs.dlfs_bfree 743 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 744 #define lfs_avail lfs_dlfs.dlfs_avail 745 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 746 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 747 #define lfs_ifile lfs_dlfs.dlfs_ifile 748 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 749 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 750 #define lfs_curseg lfs_dlfs.dlfs_curseg 751 #define lfs_offset lfs_dlfs.dlfs_offset 752 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 753 #define lfs_otstamp lfs_dlfs.dlfs_inopf 754 #define lfs_inopf lfs_dlfs.dlfs_inopf 755 #define lfs_minfree lfs_dlfs.dlfs_minfree 756 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 757 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 758 #define lfs_inopb lfs_dlfs.dlfs_inopb 759 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 760 #define lfs_sepb lfs_dlfs.dlfs_sepb 761 #define lfs_nindir lfs_dlfs.dlfs_nindir 762 #define lfs_nseg lfs_dlfs.dlfs_nseg 763 #define lfs_nspf lfs_dlfs.dlfs_nspf 764 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 765 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 766 #define lfs_segmask lfs_dlfs.dlfs_segmask 767 #define lfs_segshift lfs_dlfs.dlfs_segshift 768 #define lfs_bmask lfs_dlfs.dlfs_bmask 769 #define lfs_bshift lfs_dlfs.dlfs_bshift 770 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 771 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 772 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 773 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 774 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 775 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 776 #define lfs_sushift lfs_dlfs.dlfs_sushift 777 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 778 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 779 #define lfs_cksum lfs_dlfs.dlfs_cksum 780 #define lfs_pflags lfs_dlfs.dlfs_pflags 781 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 782 #define lfs_nclean lfs_dlfs.dlfs_nclean 783 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 784 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 785 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 786 #define lfs_serial lfs_dlfs.dlfs_serial 787 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 788 #define lfs_start lfs_dlfs.dlfs_start 789 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 790 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 791 #define lfs_interleave lfs_dlfs.dlfs_interleave 792 #define lfs_ident lfs_dlfs.dlfs_ident 793 794 /* These fields are set at mount time and are meaningless on disk. */ 795 struct segment *lfs_sp; /* current segment being written */ 796 struct vnode *lfs_ivnode; /* vnode for the ifile */ 797 u_int32_t lfs_seglock; /* single-thread the segment writer */ 798 pid_t lfs_lockpid; /* pid of lock holder */ 799 u_int32_t lfs_iocount; /* number of ios pending */ 800 u_int32_t lfs_writer; /* don't allow any dirops to start */ 801 u_int32_t lfs_dirops; /* count of active directory ops */ 802 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 803 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 804 int8_t lfs_fmod; /* super block modified flag */ 805 int8_t lfs_ronly; /* mounted read-only flag */ 806 #define LFS_NOTYET 0x01 807 #define LFS_IFDIRTY 0x02 808 #define LFS_WARNED 0x04 809 #define LFS_UNDIROP 0x08 810 int8_t lfs_flags; /* currently unused flag */ 811 u_int16_t lfs_activesb; /* toggle between superblocks */ 812 daddr_t lfs_sbactive; /* disk address of current sb write */ 813 struct vnode *lfs_flushvp; /* vnode being flushed */ 814 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 815 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 816 size_t lfs_devbsize; /* Device block size */ 817 size_t lfs_devbshift; /* Device block shift */ 818 struct lock lfs_fraglock; 819 struct lock lfs_iflock; /* Ifile lock */ 820 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 821 int lfs_nadirop; /* number of active dirop nodes */ 822 long lfs_ravail; /* blocks pre-reserved for writing */ 823 long lfs_favail; /* blocks pre-reserved for writing */ 824 res_t *lfs_resblk; /* Reserved memory for pageout */ 825 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 826 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 827 #define LFS_RESHASH_WIDTH 17 828 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 829 int lfs_pdflush; /* pagedaemon wants us to flush */ 830 u_int32_t **lfs_suflags; /* Segment use flags */ 831 #ifdef _KERNEL 832 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 833 struct pool lfs_bpppool; /* Pool for bpp */ 834 struct pool lfs_segpool; /* Pool for struct segment */ 835 #endif /* _KERNEL */ 836 #define LFS_MAX_CLEANIND 64 837 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 838 int lfs_cleanind; /* Index into intervals */ 839 struct simplelock lfs_interlock; /* lock for lfs_seglock */ 840 int lfs_sleepers; /* # procs sleeping this fs */ 841 int lfs_pages; /* dirty pages blaming this fs */ 842 }; 843 844 /* NINDIR is the number of indirects in a file system block. */ 845 #define NINDIR(fs) ((fs)->lfs_nindir) 846 847 /* INOPB is the number of inodes in a secondary storage block. */ 848 #define INOPB(fs) ((fs)->lfs_inopb) 849 /* INOPF is the number of inodes in a fragment. */ 850 #define INOPF(fs) ((fs)->lfs_inopf) 851 852 #define blksize(fs, ip, lbn) \ 853 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 854 ? (fs)->lfs_bsize \ 855 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 856 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 857 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 858 ((int)((loc) & (fs)->lfs_ffmask)) 859 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 860 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 861 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 862 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 863 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 864 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 865 /* Same as above, but named like dbtob(), btodb() */ 866 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 867 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 868 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 869 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 870 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 871 (fs)->lfs_fsbtodb)) 872 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 873 (fs)->lfs_fsbtodb)) 874 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 875 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 876 ((loc) >> (fs)->lfs_ffshift) 877 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 878 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 879 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 880 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 881 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 882 ((frags) >> (fs)->lfs_fbshift) 883 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 884 ((blks) << (fs)->lfs_fbshift) 885 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 886 ((fsb) & ((fs)->lfs_frag - 1)) 887 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 888 ((fsb) &~ ((fs)->lfs_frag - 1)) 889 #define dblksize(fs, dp, lbn) \ 890 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 891 ? (fs)->lfs_bsize \ 892 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 893 894 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 895 lblktosize((fs), (fs)->lfs_ssize) : \ 896 (fs)->lfs_ssize) 897 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 898 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 899 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 900 #define dtosn(fs, daddr) /* block address to segment number */ \ 901 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 902 #define sntod(fs, sn) /* segment number to disk address */ \ 903 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 904 905 /* 906 * Structures used by lfs_bmapv and lfs_markv to communicate information 907 * about inodes and data blocks. 908 */ 909 typedef struct block_info { 910 ino_t bi_inode; /* inode # */ 911 int32_t bi_lbn; /* logical block w/in file */ 912 int32_t bi_daddr; /* disk address of block */ 913 u_int64_t bi_segcreate; /* origin segment create time */ 914 int bi_version; /* file version number */ 915 void *bi_bp; /* data buffer */ 916 int bi_size; /* size of the block (if fragment) */ 917 } BLOCK_INFO; 918 919 /* Compatibility for 1.5 binaries */ 920 typedef struct block_info_15 { 921 ino_t bi_inode; /* inode # */ 922 int32_t bi_lbn; /* logical block w/in file */ 923 int32_t bi_daddr; /* disk address of block */ 924 u_int32_t bi_segcreate; /* origin segment create time */ 925 int bi_version; /* file version number */ 926 void *bi_bp; /* data buffer */ 927 int bi_size; /* size of the block (if fragment) */ 928 } BLOCK_INFO_15; 929 930 /* In-memory description of a segment about to be written. */ 931 struct segment { 932 struct lfs *fs; /* file system pointer */ 933 struct buf **bpp; /* pointer to buffer array */ 934 struct buf **cbpp; /* pointer to next available bp */ 935 struct buf **start_bpp; /* pointer to first bp in this set */ 936 struct buf *ibp; /* buffer pointer to inode page */ 937 struct ufs1_dinode *idp; /* pointer to ifile dinode */ 938 struct finfo *fip; /* current fileinfo pointer */ 939 struct vnode *vp; /* vnode being gathered */ 940 void *segsum; /* segment summary info */ 941 u_int32_t ninodes; /* number of inodes in this segment */ 942 int32_t seg_bytes_left; /* bytes left in segment */ 943 int32_t sum_bytes_left; /* bytes left in summary block */ 944 u_int32_t seg_number; /* number of this segment */ 945 int32_t *start_lbp; /* beginning lbn for this set */ 946 947 #define SEGM_CKP 0x01 /* doing a checkpoint */ 948 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 949 #define SEGM_SYNC 0x04 /* wait for segment */ 950 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 951 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */ 952 #define SEGM_WRITERD 0x20 /* LFS writed called us */ 953 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */ 954 u_int16_t seg_flags; /* run-time flags for this segment */ 955 u_int32_t seg_iocount; /* number of ios pending */ 956 int ndupino; /* number of duplicate inodes */ 957 }; 958 959 #ifdef _KERNEL 960 struct lfs_cluster { 961 size_t bufsize; /* Size of kept data */ 962 struct buf **bpp; /* Array of kept buffers */ 963 int bufcount; /* Number of kept buffers */ 964 #define LFS_CL_MALLOC 0x00000001 965 #define LFS_CL_SHIFT 0x00000002 966 #define LFS_CL_SYNC 0x00000004 967 u_int32_t flags; /* Flags */ 968 struct lfs *fs; /* LFS that this belongs to */ 969 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 970 }; 971 972 /* 973 * Splay tree containing block numbers allocated through lfs_balloc. 974 */ 975 struct lbnentry { 976 SPLAY_ENTRY(lbnentry) entry; 977 daddr_t lbn; 978 }; 979 #endif /* _KERNEL */ 980 981 /* 982 * LFS inode extensions. 983 */ 984 struct lfs_inode_ext { 985 off_t lfs_osize; /* size of file on disk */ 986 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 987 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */ 988 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 989 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 990 #define LFSI_NO_GOP_WRITE 0x01 991 u_int32_t lfs_iflags; /* Inode flags */ 992 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 993 #ifdef _KERNEL 994 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 995 int lfs_nbtree; /* Size of tree */ 996 #endif 997 }; 998 #define i_lfs_osize inode_ext.lfs->lfs_osize 999 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 1000 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 1001 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 1002 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 1003 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 1004 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 1005 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 1006 1007 /* 1008 * Macros for determining free space on the disk, with the variable metadata 1009 * of segment summaries and inode blocks taken into account. 1010 */ 1011 /* Estimate number of clean blocks not available for writing */ 1012 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 1013 (int64_t)(F)->lfs_nclean) / \ 1014 ((F)->lfs_nseg - (F)->lfs_nclean))) 1015 1016 /* Estimate total size of the disk not including metadata */ 1017 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 1018 1019 /* Estimate number of blocks actually available for writing */ 1020 #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) 1021 1022 /* Amount of non-meta space not available to mortal man */ 1023 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1024 (u_int64_t)(F)->lfs_minfree) / \ 1025 100) 1026 1027 /* Can credential C write BB blocks */ 1028 #define ISSPACE(F, BB, C) \ 1029 ((((C) == NOCRED || (C)->cr_uid == 0) && \ 1030 LFS_EST_BFREE(F) >= (BB)) || \ 1031 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB))) 1032 1033 /* Can an ordinary user write BB blocks */ 1034 #define IS_FREESPACE(F, BB) \ 1035 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1036 1037 /* 1038 * The minimum number of blocks to create a new inode. This is: 1039 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1040 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1041 */ 1042 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1043 1044 /* Statistics Counters */ 1045 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1046 u_int segsused; 1047 u_int psegwrites; 1048 u_int psyncwrites; 1049 u_int pcleanwrites; 1050 u_int blocktot; 1051 u_int cleanblocks; 1052 u_int ncheckpoints; 1053 u_int nwrites; 1054 u_int nsync_writes; 1055 u_int wait_exceeded; 1056 u_int write_exceeded; 1057 u_int flush_invoked; 1058 u_int vflush_invoked; 1059 u_int clean_inlocked; 1060 u_int clean_vnlocked; 1061 u_int segs_reclaimed; 1062 }; 1063 #ifdef _KERNEL 1064 extern struct lfs_stats lfs_stats; 1065 #endif 1066 1067 /* Fcntls to take the place of the lfs syscalls */ 1068 struct lfs_fcntl_markv { 1069 BLOCK_INFO *blkiov; /* blocks to relocate */ 1070 int blkcnt; /* number of blocks */ 1071 }; 1072 1073 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1074 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1075 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1076 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1077 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1078 #define LFCNIFILEFH _FCNW_FSPRIV('L', 5, struct fhandle) 1079 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1080 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1081 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1082 /* Compat for NetBSD 2.x bug */ 1083 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1084 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1085 1086 #ifdef _KERNEL 1087 /* XXX MP */ 1088 #define LFS_SEGLOCK_HELD(fs) \ 1089 ((fs)->lfs_seglock != 0 && (fs)->lfs_lockpid == curproc->p_pid) 1090 #endif /* _KERNEL */ 1091 1092 /* Debug segment lock */ 1093 #ifdef notyet 1094 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1095 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1096 # define ASSERT_DUNNO_SEGLOCK(fs) 1097 # define ASSERT_MAYBE_SEGLOCK(fs) 1098 #else /* !notyet */ 1099 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1100 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1101 LFS_SEGLOCK_HELD(fs))) 1102 # define ASSERT_SEGLOCK(fs) do { \ 1103 if (!LFS_SEGLOCK_HELD(fs)) { \ 1104 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1105 } \ 1106 } while(0) 1107 # define ASSERT_NO_SEGLOCK(fs) do { \ 1108 if (LFS_SEGLOCK_HELD(fs)) { \ 1109 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1110 } \ 1111 } while(0) 1112 # define ASSERT_MAYBE_SEGLOCK(x) 1113 #endif /* !notyet */ 1114 1115 #endif /* !_UFS_LFS_LFS_H_ */ 1116
Indexes created Tue Sep 30 06:09:59 GMT 2025