lfs.h revision 1.78
1 /* $NetBSD: lfs.h,v 1.78 2005/04/01 21:59:46 perseant 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_COUNT 5 /* always last */ 142 143 /* Number of reserved memory blocks of each type */ 144 #define LFS_N_SUMMARIES 2 145 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 146 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 147 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 148 #define LFS_N_CLEAN 0 149 150 /* Total count of "large" (non-pool) types */ 151 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 152 LFS_N_CLUSTERS + LFS_N_CLEAN) 153 154 /* Counts for pool types */ 155 #define LFS_N_CL LFS_N_CLUSTERS 156 #define LFS_N_BPP 2 157 #define LFS_N_SEG 2 158 159 /* 160 * "struct buf" associated definitions 161 */ 162 163 /* Unassigned disk addresses. */ 164 #define UNASSIGNED -1 165 #define UNWRITTEN -2 166 167 /* Unused logical block number */ 168 #define LFS_UNUSED_LBN -1 169 170 /* Determine if a buffer belongs to the ifile */ 171 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 172 173 # define LFS_LOCK_BUF(bp) do { \ 174 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \ 175 simple_lock(&lfs_subsys_lock); \ 176 ++locked_queue_count; \ 177 locked_queue_bytes += bp->b_bufsize; \ 178 simple_unlock(&lfs_subsys_lock); \ 179 } \ 180 (bp)->b_flags |= B_LOCKED; \ 181 } while (0) 182 183 # define LFS_UNLOCK_BUF(bp) do { \ 184 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \ 185 simple_lock(&lfs_subsys_lock); \ 186 --locked_queue_count; \ 187 locked_queue_bytes -= bp->b_bufsize; \ 188 if (locked_queue_count < LFS_WAIT_BUFS && \ 189 locked_queue_bytes < LFS_WAIT_BYTES) \ 190 wakeup(&locked_queue_count); \ 191 simple_unlock(&lfs_subsys_lock); \ 192 } \ 193 (bp)->b_flags &= ~B_LOCKED; \ 194 } while (0) 195 196 #ifdef _KERNEL 197 # define LFS_IS_MALLOC_BUF(bp) (((bp)->b_flags & B_CALL) && \ 198 (bp)->b_iodone == lfs_callback) 199 200 # ifdef DEBUG 201 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 202 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 203 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 204 wakeup(&locked_queue_count); \ 205 } \ 206 } while (0) 207 # else 208 # define LFS_DEBUG_COUNTLOCKED(m) 209 # endif 210 211 /* log for debugging writes to the Ifile */ 212 # ifdef DEBUG 213 struct lfs_log_entry { 214 char *op; 215 char *file; 216 int pid; 217 int line; 218 daddr_t block; 219 unsigned long flags; 220 }; 221 extern int lfs_lognum; 222 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 223 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 224 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 225 int _s; \ 226 \ 227 simple_lock(&lfs_subsys_lock); \ 228 _s = splbio(); \ 229 lfs_log[lfs_lognum].op = theop; \ 230 lfs_log[lfs_lognum].file = thefile; \ 231 lfs_log[lfs_lognum].line = (theline); \ 232 lfs_log[lfs_lognum].pid = (thepid); \ 233 lfs_log[lfs_lognum].block = (lbn); \ 234 lfs_log[lfs_lognum].flags = (theflags); \ 235 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 236 splx(_s); \ 237 simple_unlock(&lfs_subsys_lock); \ 238 } while (0) 239 240 # define LFS_BCLEAN_LOG(fs, bp) do { \ 241 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 242 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 243 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 244 } while (0) 245 246 /* Must match list in lfs_vfsops.c ! */ 247 # define DLOG_RF 0 /* roll forward */ 248 # define DLOG_ALLOC 1 /* inode alloc */ 249 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 250 # define DLOG_FLUSH 3 /* flush */ 251 # define DLOG_LLIST 4 /* locked list accounting */ 252 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 253 # define DLOG_VNODE 6 /* vflush/writevnodes */ 254 # define DLOG_SEG 7 /* segwrite */ 255 # define DLOG_SU 8 /* seguse accounting */ 256 # define DLOG_CLEAN 9 /* cleaner routines */ 257 # define DLOG_MOUNT 10 /* mount/unmount */ 258 # define DLOG_PAGE 11 /* putpages/gop_write */ 259 # define DLOG_DIROP 12 /* dirop accounting */ 260 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 261 # define DLOG_MAX 14 /* The terminator */ 262 # define DLOG(a) lfs_debug_log a 263 # else /* ! DEBUG */ 264 # define LFS_BCLEAN_LOG(fs, bp) 265 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 266 # define DLOG(a) 267 # endif /* ! DEBUG */ 268 #else /* ! _KERNEL */ 269 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 270 #endif /* _KERNEL */ 271 272 #ifdef _KERNEL 273 /* Filehandle structure for exported LFSes */ 274 struct lfid { 275 struct ufid lfid_ufid; 276 #define lfid_len lfid_ufid.ufid_len 277 #define lfid_ino lfid_ufid.ufid_ino 278 #define lfid_gen lfid_ufid.ufid_gen 279 uint32_t lfid_ident; 280 }; 281 #endif /* _KERNEL */ 282 283 /* 284 * "struct inode" associated definitions 285 */ 286 287 /* Address calculations for metadata located in the inode */ 288 #define S_INDIR(fs) -NDADDR 289 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 290 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 291 292 /* For convenience */ 293 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 294 295 #define LFS_SET_UINO(ip, flags) do { \ 296 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 297 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 298 ++(ip)->i_lfs->lfs_uinodes; \ 299 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 300 ++(ip)->i_lfs->lfs_uinodes; \ 301 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 302 ++(ip)->i_lfs->lfs_uinodes; \ 303 (ip)->i_flag |= (flags); \ 304 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 305 } while (0) 306 307 #define LFS_CLR_UINO(ip, flags) do { \ 308 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 309 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 310 --(ip)->i_lfs->lfs_uinodes; \ 311 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 312 --(ip)->i_lfs->lfs_uinodes; \ 313 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 314 --(ip)->i_lfs->lfs_uinodes; \ 315 (ip)->i_flag &= ~(flags); \ 316 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 317 panic("lfs_uinodes < 0"); \ 318 } \ 319 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 320 } while (0) 321 322 #define LFS_ITIMES(ip, acc, mod, cre) do { \ 323 struct lfs *_fs = (ip)->i_lfs; \ 324 \ 325 if ((ip)->i_flag & IN_ACCESS) { \ 326 (ip)->i_ffs1_atime = (acc)->tv_sec; \ 327 (ip)->i_ffs1_atimensec = (acc)->tv_nsec; \ 328 if ((ip)->i_lfs->lfs_version > 1) { \ 329 struct buf *_ibp; \ 330 IFILE *_ifp; \ 331 \ 332 LFS_IENTRY(_ifp, ip->i_lfs, ip->i_number, _ibp); \ 333 _ifp->if_atime_sec = (acc)->tv_sec; \ 334 _ifp->if_atime_nsec = (acc)->tv_nsec; \ 335 LFS_BWRITE_LOG(_ibp); \ 336 simple_lock(&_fs->lfs_interlock); \ 337 _fs->lfs_flags |= LFS_IFDIRTY; \ 338 simple_unlock(&_fs->lfs_interlock); \ 339 } else { \ 340 LFS_SET_UINO(ip, IN_ACCESSED); \ 341 } \ 342 } \ 343 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFY)) { \ 344 if ((ip)->i_flag & (IN_UPDATE | IN_MODIFY)) { \ 345 (ip)->i_ffs1_mtime = (mod)->tv_sec; \ 346 (ip)->i_ffs1_mtimensec = (mod)->tv_nsec; \ 347 (ip)->i_modrev++; \ 348 } \ 349 if ((ip)->i_flag & (IN_CHANGE | IN_MODIFY)) { \ 350 (ip)->i_ffs1_ctime = (cre)->tv_sec; \ 351 (ip)->i_ffs1_ctimensec = (cre)->tv_nsec; \ 352 } \ 353 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) \ 354 LFS_SET_UINO(ip, IN_MODIFIED); \ 355 if ((ip)->i_flag & IN_MODIFY) \ 356 LFS_SET_UINO(ip, IN_ACCESSED); \ 357 } \ 358 (ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY);\ 359 } while (0) 360 361 /* 362 * "struct vnode" associated definitions 363 */ 364 365 /* Heuristic emptiness measure */ 366 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 367 !((vp)->v_flag & VONWORKLST)) 368 369 /* XXX Shouldn't we use v_numoutput instead? */ 370 #define WRITEINPROG(vp) (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 371 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING))) 372 373 374 /* 375 * On-disk and in-memory checkpoint segment usage structure. 376 */ 377 typedef struct segusage SEGUSE; 378 struct segusage { 379 u_int32_t su_nbytes; /* 0: number of live bytes */ 380 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 381 u_int16_t su_nsums; /* 8: number of summaries in segment */ 382 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 383 384 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 385 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 386 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 387 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 388 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 389 u_int32_t su_flags; /* 12: segment flags */ 390 u_int64_t su_lastmod; /* 16: last modified timestamp */ 391 }; 392 393 typedef struct segusage_v1 SEGUSE_V1; 394 struct segusage_v1 { 395 u_int32_t su_nbytes; /* 0: number of live bytes */ 396 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 397 u_int16_t su_nsums; /* 8: number of summaries in segment */ 398 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 399 u_int32_t su_flags; /* 12: segment flags */ 400 }; 401 402 #define SEGUPB(fs) (fs->lfs_sepb) 403 #define SEGTABSIZE_SU(fs) \ 404 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 405 406 /* Read in the block with a specific segment usage entry from the ifile. */ 407 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 408 int _e; \ 409 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 410 if ((_e = bread((F)->lfs_ivnode, \ 411 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 412 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 413 panic("lfs: ifile read: %d", _e); \ 414 if ((F)->lfs_version == 1) \ 415 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 416 ((IN) & ((F)->lfs_sepb - 1))); \ 417 else \ 418 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 419 } while (0) 420 421 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 422 if ((SP)->su_nbytes == 0) \ 423 (SP)->su_flags |= SEGUSE_EMPTY; \ 424 else \ 425 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 426 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 427 LFS_BWRITE_LOG(BP); \ 428 } while (0) 429 430 /* 431 * On-disk file information. One per file with data blocks in the segment. 432 */ 433 typedef struct finfo FINFO; 434 struct finfo { 435 u_int32_t fi_nblocks; /* number of blocks */ 436 u_int32_t fi_version; /* version number */ 437 u_int32_t fi_ino; /* inode number */ 438 u_int32_t fi_lastlength; /* length of last block in array */ 439 int32_t fi_blocks[1]; /* array of logical block numbers */ 440 }; 441 /* sizeof FINFO except fi_blocks */ 442 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 443 444 /* 445 * Index file inode entries. 446 */ 447 typedef struct ifile IFILE; 448 struct ifile { 449 u_int32_t if_version; /* inode version number */ 450 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 451 int32_t if_daddr; /* inode disk address */ 452 ino_t if_nextfree; /* next-unallocated inode */ 453 u_int32_t if_atime_sec; /* Last access time, seconds */ 454 u_int32_t if_atime_nsec; /* and nanoseconds */ 455 }; 456 457 typedef struct ifile_v1 IFILE_V1; 458 struct ifile_v1 { 459 u_int32_t if_version; /* inode version number */ 460 int32_t if_daddr; /* inode disk address */ 461 ino_t if_nextfree; /* next-unallocated inode */ 462 #if LFS_ATIME_IFILE 463 struct timespec if_atime; /* Last access time */ 464 #endif 465 }; 466 467 /* 468 * LFSv1 compatibility code is not allowed to touch if_atime, since it 469 * may not be mapped! 470 */ 471 /* Read in the block with a specific inode from the ifile. */ 472 #define LFS_IENTRY(IP, F, IN, BP) do { \ 473 int _e; \ 474 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 475 if ((_e = bread((F)->lfs_ivnode, \ 476 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 477 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 478 panic("lfs: ifile read %d", _e); \ 479 if ((F)->lfs_version == 1) \ 480 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 481 (IN) % (F)->lfs_ifpb); \ 482 else \ 483 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 484 } while (0) 485 486 /* 487 * Cleaner information structure. This resides in the ifile and is used 488 * to pass information from the kernel to the cleaner. 489 */ 490 typedef struct _cleanerinfo { 491 u_int32_t clean; /* number of clean segments */ 492 u_int32_t dirty; /* number of dirty segments */ 493 int32_t bfree; /* disk blocks free */ 494 int32_t avail; /* disk blocks available */ 495 u_int32_t free_head; /* head of the inode free list */ 496 u_int32_t free_tail; /* tail of the inode free list */ 497 } CLEANERINFO; 498 499 #define CLEANSIZE_SU(fs) \ 500 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 501 502 /* Read in the block with the cleaner info from the ifile. */ 503 #define LFS_CLEANERINFO(CP, F, BP) do { \ 504 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 505 if (bread((F)->lfs_ivnode, \ 506 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 507 panic("lfs: ifile read"); \ 508 (CP) = (CLEANERINFO *)(BP)->b_data; \ 509 } while (0) 510 511 /* 512 * Synchronize the Ifile cleaner info with current avail and bfree. 513 */ 514 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 515 simple_lock(&(fs)->lfs_interlock); \ 516 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 517 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 518 (fs)->lfs_favail) { \ 519 (cip)->bfree = (fs)->lfs_bfree; \ 520 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 521 (fs)->lfs_favail; \ 522 if (((bp)->b_flags & B_GATHERED) == 0) { \ 523 (fs)->lfs_flags |= LFS_IFDIRTY; \ 524 } \ 525 simple_unlock(&(fs)->lfs_interlock); \ 526 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 527 } else { \ 528 simple_unlock(&(fs)->lfs_interlock); \ 529 brelse(bp); \ 530 } \ 531 } while (0) 532 533 /* 534 * Get the head of the inode free list. 535 * Always caled with the segment lock held. 536 */ 537 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 538 if ((FS)->lfs_version > 1) { \ 539 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 540 (FS)->lfs_freehd = (CIP)->free_head; \ 541 brelse(BP); \ 542 } \ 543 *(FREEP) = (FS)->lfs_freehd; \ 544 } while (0) 545 546 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 547 (FS)->lfs_freehd = (VAL); \ 548 if ((FS)->lfs_version > 1) { \ 549 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 550 (CIP)->free_head = (VAL); \ 551 LFS_BWRITE_LOG(BP); \ 552 simple_lock(&fs->lfs_interlock); \ 553 (FS)->lfs_flags |= LFS_IFDIRTY; \ 554 simple_unlock(&fs->lfs_interlock); \ 555 } \ 556 } while (0) 557 558 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 559 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 560 *(FREEP) = (CIP)->free_tail; \ 561 brelse(BP); \ 562 } while (0) 563 564 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 565 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 566 (CIP)->free_tail = (VAL); \ 567 LFS_BWRITE_LOG(BP); \ 568 simple_lock(&fs->lfs_interlock); \ 569 (FS)->lfs_flags |= LFS_IFDIRTY; \ 570 simple_unlock(&fs->lfs_interlock); \ 571 } while (0) 572 573 /* 574 * On-disk segment summary information 575 */ 576 typedef struct segsum_v1 SEGSUM_V1; 577 struct segsum_v1 { 578 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 579 u_int32_t ss_datasum; /* 4: check sum of data */ 580 u_int32_t ss_magic; /* 8: segment summary magic number */ 581 #define SS_MAGIC 0x061561 582 int32_t ss_next; /* 12: next segment */ 583 u_int32_t ss_create; /* 16: creation time stamp */ 584 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 585 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 586 587 #define SS_DIROP 0x01 /* segment begins a dirop */ 588 #define SS_CONT 0x02 /* more partials to finish this write*/ 589 u_int16_t ss_flags; /* 24: used for directory operations */ 590 u_int16_t ss_pad; /* 26: extra space */ 591 /* FINFO's and inode daddr's... */ 592 }; 593 594 typedef struct segsum SEGSUM; 595 struct segsum { 596 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 597 u_int32_t ss_datasum; /* 4: check sum of data */ 598 u_int32_t ss_magic; /* 8: segment summary magic number */ 599 int32_t ss_next; /* 12: next segment */ 600 u_int32_t ss_ident; /* 16: roll-forward fsid */ 601 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 602 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 603 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 604 u_int16_t ss_flags; /* 24: used for directory operations */ 605 u_int8_t ss_pad[6]; /* 26: extra space */ 606 u_int64_t ss_serial; /* 32: serial number */ 607 u_int64_t ss_create; /* 40: time stamp */ 608 /* FINFO's and inode daddr's... */ 609 }; 610 611 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 612 613 614 /* 615 * On-disk super block. 616 */ 617 struct dlfs { 618 #define LFS_MAGIC 0x070162 619 u_int32_t dlfs_magic; /* 0: magic number */ 620 #define LFS_VERSION 2 621 u_int32_t dlfs_version; /* 4: version number */ 622 623 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 624 /* number of frags in fs (v2) */ 625 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 626 /* number of bytes per segment (v2) */ 627 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 628 u_int32_t dlfs_bsize; /* 20: file system block size */ 629 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 630 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 631 632 /* Checkpoint region. */ 633 u_int32_t dlfs_freehd; /* 32: start of the free list */ 634 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 635 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 636 int32_t dlfs_avail; /* 44: blocks available for writing */ 637 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 638 int32_t dlfs_idaddr; /* 52: inode file disk address */ 639 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 640 int32_t dlfs_lastseg; /* 60: address of last segment written */ 641 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 642 int32_t dlfs_curseg; /* 68: current segment being written */ 643 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 644 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 645 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 646 #define dlfs_otstamp dlfs_inopf 647 648 /* These are configuration parameters. */ 649 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 650 651 /* These fields can be computed from the others. */ 652 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 653 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 654 u_int32_t dlfs_inopb; /* 100: inodes per block */ 655 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 656 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 657 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 658 u_int32_t dlfs_nseg; /* 116: number of segments */ 659 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 660 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 661 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 662 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 663 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 664 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 665 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 666 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 667 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 668 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 669 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 670 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 671 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 672 673 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 674 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 675 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 676 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 677 678 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 679 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 680 #define LFS_PF_CLEAN 0x1 681 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 682 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 683 u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */ 684 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 685 u_int64_t dlfs_serial; /* 336: serial number */ 686 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 687 int32_t dlfs_start; /* 348: start of segment 0 */ 688 u_int64_t dlfs_tstamp; /* 352: time stamp */ 689 #define LFS_44INODEFMT 0 690 #define LFS_MAXINODEFMT 0 691 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 692 u_int32_t dlfs_interleave; /* 364: segment interleave */ 693 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 694 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 695 int8_t dlfs_pad[132]; /* 376: round to 512 bytes */ 696 /* Checksum -- last valid disk field. */ 697 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 698 }; 699 700 /* 701 * In-memory super block. 702 */ 703 struct lfs { 704 struct dlfs lfs_dlfs; /* on-disk parameters */ 705 #define lfs_magic lfs_dlfs.dlfs_magic 706 #define lfs_version lfs_dlfs.dlfs_version 707 #define lfs_size lfs_dlfs.dlfs_size 708 #define lfs_ssize lfs_dlfs.dlfs_ssize 709 #define lfs_dsize lfs_dlfs.dlfs_dsize 710 #define lfs_bsize lfs_dlfs.dlfs_bsize 711 #define lfs_fsize lfs_dlfs.dlfs_fsize 712 #define lfs_frag lfs_dlfs.dlfs_frag 713 #define lfs_freehd lfs_dlfs.dlfs_freehd 714 #define lfs_bfree lfs_dlfs.dlfs_bfree 715 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 716 #define lfs_avail lfs_dlfs.dlfs_avail 717 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 718 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 719 #define lfs_ifile lfs_dlfs.dlfs_ifile 720 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 721 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 722 #define lfs_curseg lfs_dlfs.dlfs_curseg 723 #define lfs_offset lfs_dlfs.dlfs_offset 724 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 725 #define lfs_otstamp lfs_dlfs.dlfs_inopf 726 #define lfs_inopf lfs_dlfs.dlfs_inopf 727 #define lfs_minfree lfs_dlfs.dlfs_minfree 728 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 729 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 730 #define lfs_inopb lfs_dlfs.dlfs_inopb 731 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 732 #define lfs_sepb lfs_dlfs.dlfs_sepb 733 #define lfs_nindir lfs_dlfs.dlfs_nindir 734 #define lfs_nseg lfs_dlfs.dlfs_nseg 735 #define lfs_nspf lfs_dlfs.dlfs_nspf 736 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 737 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 738 #define lfs_segmask lfs_dlfs.dlfs_segmask 739 #define lfs_segshift lfs_dlfs.dlfs_segshift 740 #define lfs_bmask lfs_dlfs.dlfs_bmask 741 #define lfs_bshift lfs_dlfs.dlfs_bshift 742 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 743 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 744 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 745 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 746 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 747 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 748 #define lfs_sushift lfs_dlfs.dlfs_sushift 749 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 750 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 751 #define lfs_cksum lfs_dlfs.dlfs_cksum 752 #define lfs_pflags lfs_dlfs.dlfs_pflags 753 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 754 #define lfs_nclean lfs_dlfs.dlfs_nclean 755 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 756 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 757 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 758 #define lfs_serial lfs_dlfs.dlfs_serial 759 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 760 #define lfs_start lfs_dlfs.dlfs_start 761 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 762 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 763 #define lfs_interleave lfs_dlfs.dlfs_interleave 764 #define lfs_ident lfs_dlfs.dlfs_ident 765 766 /* These fields are set at mount time and are meaningless on disk. */ 767 struct segment *lfs_sp; /* current segment being written */ 768 struct vnode *lfs_ivnode; /* vnode for the ifile */ 769 u_int32_t lfs_seglock; /* single-thread the segment writer */ 770 pid_t lfs_lockpid; /* pid of lock holder */ 771 u_int32_t lfs_iocount; /* number of ios pending */ 772 u_int32_t lfs_writer; /* don't allow any dirops to start */ 773 u_int32_t lfs_dirops; /* count of active directory ops */ 774 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 775 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 776 int8_t lfs_fmod; /* super block modified flag */ 777 int8_t lfs_ronly; /* mounted read-only flag */ 778 #define LFS_NOTYET 0x01 779 #define LFS_IFDIRTY 0x02 780 #define LFS_WARNED 0x04 781 #define LFS_UNDIROP 0x08 782 int8_t lfs_flags; /* currently unused flag */ 783 u_int16_t lfs_activesb; /* toggle between superblocks */ 784 daddr_t lfs_sbactive; /* disk address of current sb write */ 785 struct vnode *lfs_flushvp; /* vnode being flushed */ 786 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 787 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 788 size_t lfs_devbsize; /* Device block size */ 789 size_t lfs_devbshift; /* Device block shift */ 790 struct lock lfs_fraglock; 791 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 792 int lfs_nadirop; /* number of active dirop nodes */ 793 long lfs_ravail; /* blocks pre-reserved for writing */ 794 long lfs_favail; /* blocks pre-reserved for writing */ 795 res_t *lfs_resblk; /* Reserved memory for pageout */ 796 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 797 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 798 #define LFS_RESHASH_WIDTH 17 799 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 800 int lfs_pdflush; /* pagedaemon wants us to flush */ 801 u_int32_t **lfs_suflags; /* Segment use flags */ 802 #ifdef _KERNEL 803 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 804 struct pool lfs_bpppool; /* Pool for bpp */ 805 struct pool lfs_segpool; /* Pool for struct segment */ 806 #endif /* _KERNEL */ 807 #define LFS_MAX_CLEANIND 64 808 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 809 int lfs_cleanind; /* Index into intervals */ 810 struct simplelock lfs_interlock; /* lock for lfs_seglock */ 811 int lfs_sleepers; /* # procs sleeping this fs */ 812 int lfs_pages; /* dirty pages blaming this fs */ 813 }; 814 815 /* NINDIR is the number of indirects in a file system block. */ 816 #define NINDIR(fs) ((fs)->lfs_nindir) 817 818 /* INOPB is the number of inodes in a secondary storage block. */ 819 #define INOPB(fs) ((fs)->lfs_inopb) 820 /* INOPF is the number of inodes in a fragment. */ 821 #define INOPF(fs) ((fs)->lfs_inopf) 822 823 #define blksize(fs, ip, lbn) \ 824 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 825 ? (fs)->lfs_bsize \ 826 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 827 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 828 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 829 ((int)((loc) & (fs)->lfs_ffmask)) 830 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 831 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 832 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 833 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 834 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 835 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 836 /* Same as above, but named like dbtob(), btodb() */ 837 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 838 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 839 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 840 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 841 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 842 (fs)->lfs_fsbtodb)) 843 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 844 (fs)->lfs_fsbtodb)) 845 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 846 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 847 ((loc) >> (fs)->lfs_ffshift) 848 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 849 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 850 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 851 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 852 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 853 ((frags) >> (fs)->lfs_fbshift) 854 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 855 ((blks) << (fs)->lfs_fbshift) 856 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 857 ((fsb) & ((fs)->lfs_frag - 1)) 858 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 859 ((fsb) &~ ((fs)->lfs_frag - 1)) 860 #define dblksize(fs, dp, lbn) \ 861 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 862 ? (fs)->lfs_bsize \ 863 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 864 865 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 866 lblktosize((fs), (fs)->lfs_ssize) : \ 867 (fs)->lfs_ssize) 868 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 869 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 870 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 871 #define dtosn(fs, daddr) /* block address to segment number */ \ 872 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 873 #define sntod(fs, sn) /* segment number to disk address */ \ 874 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 875 876 /* 877 * Structures used by lfs_bmapv and lfs_markv to communicate information 878 * about inodes and data blocks. 879 */ 880 typedef struct block_info { 881 ino_t bi_inode; /* inode # */ 882 int32_t bi_lbn; /* logical block w/in file */ 883 int32_t bi_daddr; /* disk address of block */ 884 u_int64_t bi_segcreate; /* origin segment create time */ 885 int bi_version; /* file version number */ 886 void *bi_bp; /* data buffer */ 887 int bi_size; /* size of the block (if fragment) */ 888 } BLOCK_INFO; 889 890 /* Compatibility for 1.5 binaries */ 891 typedef struct block_info_15 { 892 ino_t bi_inode; /* inode # */ 893 int32_t bi_lbn; /* logical block w/in file */ 894 int32_t bi_daddr; /* disk address of block */ 895 u_int32_t bi_segcreate; /* origin segment create time */ 896 int bi_version; /* file version number */ 897 void *bi_bp; /* data buffer */ 898 int bi_size; /* size of the block (if fragment) */ 899 } BLOCK_INFO_15; 900 901 /* In-memory description of a segment about to be written. */ 902 struct segment { 903 struct lfs *fs; /* file system pointer */ 904 struct buf **bpp; /* pointer to buffer array */ 905 struct buf **cbpp; /* pointer to next available bp */ 906 struct buf **start_bpp; /* pointer to first bp in this set */ 907 struct buf *ibp; /* buffer pointer to inode page */ 908 struct ufs1_dinode *idp; /* pointer to ifile dinode */ 909 struct finfo *fip; /* current fileinfo pointer */ 910 struct vnode *vp; /* vnode being gathered */ 911 void *segsum; /* segment summary info */ 912 u_int32_t ninodes; /* number of inodes in this segment */ 913 int32_t seg_bytes_left; /* bytes left in segment */ 914 int32_t sum_bytes_left; /* bytes left in summary block */ 915 u_int32_t seg_number; /* number of this segment */ 916 int32_t *start_lbp; /* beginning lbn for this set */ 917 918 #define SEGM_CKP 0x01 /* doing a checkpoint */ 919 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 920 #define SEGM_SYNC 0x04 /* wait for segment */ 921 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 922 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */ 923 #define SEGM_WRITERD 0x20 /* LFS writed called us */ 924 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */ 925 u_int16_t seg_flags; /* run-time flags for this segment */ 926 u_int32_t seg_iocount; /* number of ios pending */ 927 int ndupino; /* number of duplicate inodes */ 928 }; 929 930 #ifdef _KERNEL 931 struct lfs_cluster { 932 size_t bufsize; /* Size of kept data */ 933 struct buf **bpp; /* Array of kept buffers */ 934 int bufcount; /* Number of kept buffers */ 935 #define LFS_CL_MALLOC 0x00000001 936 #define LFS_CL_SHIFT 0x00000002 937 #define LFS_CL_SYNC 0x00000004 938 u_int32_t flags; /* Flags */ 939 struct lfs *fs; /* LFS that this belongs to */ 940 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 941 }; 942 #endif /* _KERNEL */ 943 944 /* 945 * List containing block numbers allocated through lfs_balloc. 946 */ 947 struct lbnentry { 948 LIST_ENTRY(lbnentry) entry; 949 daddr_t lbn; 950 }; 951 952 /* 953 * LFS inode extensions. 954 */ 955 struct lfs_inode_ext { 956 off_t lfs_osize; /* size of file on disk */ 957 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 958 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */ 959 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 960 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 961 /* Blocks allocated for write */ 962 #define LFS_BLIST_HASH_WIDTH 17 963 LIST_HEAD(, lbnentry) lfs_blist[LFS_BLIST_HASH_WIDTH]; 964 #define LFSI_NO_GOP_WRITE 0x01 965 u_int32_t lfs_iflags; /* Inode flags */ 966 }; 967 #define i_lfs_osize inode_ext.lfs->lfs_osize 968 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 969 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 970 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 971 #define i_lfs_blist inode_ext.lfs->lfs_blist 972 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 973 974 /* 975 * Macros for determining free space on the disk, with the variable metadata 976 * of segment summaries and inode blocks taken into account. 977 */ 978 /* Estimate number of clean blocks not available for writing */ 979 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 980 (int64_t)(F)->lfs_nclean) / \ 981 ((F)->lfs_nseg - (F)->lfs_nclean))) 982 983 /* Estimate total size of the disk not including metadata */ 984 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 985 986 /* Estimate number of blocks actually available for writing */ 987 #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) 988 989 /* Amount of non-meta space not available to mortal man */ 990 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 991 (u_int64_t)(F)->lfs_minfree) / \ 992 100) 993 994 /* Can credential C write BB blocks */ 995 #define ISSPACE(F, BB, C) \ 996 ((((C) == NOCRED || (C)->cr_uid == 0) && \ 997 LFS_EST_BFREE(F) >= (BB)) || \ 998 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB))) 999 1000 /* Can an ordinary user write BB blocks */ 1001 #define IS_FREESPACE(F, BB) \ 1002 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1003 1004 /* 1005 * The minimum number of blocks to create a new inode. This is: 1006 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1007 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1008 */ 1009 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1010 1011 /* Statistics Counters */ 1012 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1013 u_int segsused; 1014 u_int psegwrites; 1015 u_int psyncwrites; 1016 u_int pcleanwrites; 1017 u_int blocktot; 1018 u_int cleanblocks; 1019 u_int ncheckpoints; 1020 u_int nwrites; 1021 u_int nsync_writes; 1022 u_int wait_exceeded; 1023 u_int write_exceeded; 1024 u_int flush_invoked; 1025 u_int vflush_invoked; 1026 u_int clean_inlocked; 1027 u_int clean_vnlocked; 1028 }; 1029 #ifdef _KERNEL 1030 extern struct lfs_stats lfs_stats; 1031 #endif 1032 1033 /* Fcntls to take the place of the lfs syscalls */ 1034 struct lfs_fcntl_markv { 1035 BLOCK_INFO *blkiov; /* blocks to relocate */ 1036 int blkcnt; /* number of blocks */ 1037 }; 1038 1039 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1040 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1041 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1042 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1043 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1044 #define LFCNIFILEFH _FCNW_FSPRIV('L', 5, struct fhandle) 1045 /* Compat for NetBSD 2.x error */ 1046 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1047 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1048 1049 #ifdef _KERNEL 1050 /* XXX MP */ 1051 #define LFS_SEGLOCK_HELD(fs) \ 1052 ((fs)->lfs_seglock != 0 && (fs)->lfs_lockpid == curproc->p_pid) 1053 #endif /* _KERNEL */ 1054 1055 /* Debug segment lock */ 1056 #ifdef notyet 1057 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1058 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1059 # define ASSERT_DUNNO_SEGLOCK(fs) 1060 # define ASSERT_MAYBE_SEGLOCK(fs) 1061 #else /* !notyet */ 1062 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1063 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1064 LFS_SEGLOCK_HELD(fs))) 1065 # define ASSERT_SEGLOCK(fs) do { \ 1066 if (!LFS_SEGLOCK_HELD(fs)) { \ 1067 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1068 } \ 1069 } while(0) 1070 # define ASSERT_NO_SEGLOCK(fs) do { \ 1071 if (LFS_SEGLOCK_HELD(fs)) { \ 1072 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1073 } \ 1074 } while(0) 1075 # define ASSERT_MAYBE_SEGLOCK(x) 1076 #endif /* !notyet */ 1077 1078 #endif /* !_UFS_LFS_LFS_H_ */ 1079
Indexes created Mon Sep 29 21:09:56 GMT 2025