lfs.h revision 1.103
1 /* $NetBSD: lfs.h,v 1.103 2006/04/17 20:02:34 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 #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 /* How starved can we be before we start holding back page writes */ 124 #define LFS_STARVED_FOR_SEGS(fs) ((fs)->lfs_nclean < (fs)->lfs_minfreeseg / 2 + 1) 125 126 /* 127 * Reserved blocks for lfs_malloc 128 */ 129 130 /* Structure to keep reserved blocks */ 131 typedef struct lfs_res_blk { 132 void *p; 133 LIST_ENTRY(lfs_res_blk) res; 134 int size; 135 char inuse; 136 } res_t; 137 138 /* Types for lfs_newbuf and lfs_malloc */ 139 #define LFS_NB_UNKNOWN -1 140 #define LFS_NB_SUMMARY 0 141 #define LFS_NB_SBLOCK 1 142 #define LFS_NB_IBLOCK 2 143 #define LFS_NB_CLUSTER 3 144 #define LFS_NB_CLEAN 4 145 #define LFS_NB_BLKIOV 5 146 #define LFS_NB_COUNT 6 /* always last */ 147 148 /* Number of reserved memory blocks of each type */ 149 #define LFS_N_SUMMARIES 2 150 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 151 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 152 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 153 #define LFS_N_CLEAN 0 154 #define LFS_N_BLKIOV 1 155 156 /* Total count of "large" (non-pool) types */ 157 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 158 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV) 159 160 /* Counts for pool types */ 161 #define LFS_N_CL LFS_N_CLUSTERS 162 #define LFS_N_BPP 2 163 #define LFS_N_SEG 2 164 165 /* 166 * "struct buf" associated definitions 167 */ 168 169 /* Unassigned disk addresses. */ 170 #define UNASSIGNED -1 171 #define UNWRITTEN -2 172 173 /* Unused logical block number */ 174 #define LFS_UNUSED_LBN -1 175 176 /* Determine if a buffer belongs to the ifile */ 177 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 178 179 # define LFS_LOCK_BUF(bp) do { \ 180 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \ 181 simple_lock(&lfs_subsys_lock); \ 182 ++locked_queue_count; \ 183 locked_queue_bytes += bp->b_bufsize; \ 184 simple_unlock(&lfs_subsys_lock); \ 185 } \ 186 (bp)->b_flags |= B_LOCKED; \ 187 } while (0) 188 189 # define LFS_UNLOCK_BUF(bp) do { \ 190 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \ 191 simple_lock(&lfs_subsys_lock); \ 192 --locked_queue_count; \ 193 locked_queue_bytes -= bp->b_bufsize; \ 194 if (locked_queue_count < LFS_WAIT_BUFS && \ 195 locked_queue_bytes < LFS_WAIT_BYTES) \ 196 wakeup(&locked_queue_count); \ 197 simple_unlock(&lfs_subsys_lock); \ 198 } \ 199 (bp)->b_flags &= ~B_LOCKED; \ 200 } while (0) 201 202 #ifdef _KERNEL 203 204 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 205 206 # define LFS_IS_MALLOC_BUF(bp) (((bp)->b_flags & B_CALL) && \ 207 (bp)->b_iodone == lfs_callback) 208 209 # ifdef DEBUG 210 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 211 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 212 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 213 wakeup(&locked_queue_count); \ 214 } \ 215 } while (0) 216 # else 217 # define LFS_DEBUG_COUNTLOCKED(m) 218 # endif 219 220 /* log for debugging writes to the Ifile */ 221 # ifdef DEBUG 222 struct lfs_log_entry { 223 const char *op; 224 const char *file; 225 int pid; 226 int line; 227 daddr_t block; 228 unsigned long flags; 229 }; 230 extern int lfs_lognum; 231 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 232 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 233 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 234 int _s; \ 235 \ 236 simple_lock(&lfs_subsys_lock); \ 237 _s = splbio(); \ 238 lfs_log[lfs_lognum].op = theop; \ 239 lfs_log[lfs_lognum].file = thefile; \ 240 lfs_log[lfs_lognum].line = (theline); \ 241 lfs_log[lfs_lognum].pid = (thepid); \ 242 lfs_log[lfs_lognum].block = (lbn); \ 243 lfs_log[lfs_lognum].flags = (theflags); \ 244 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 245 splx(_s); \ 246 simple_unlock(&lfs_subsys_lock); \ 247 } while (0) 248 249 # define LFS_BCLEAN_LOG(fs, bp) do { \ 250 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 251 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 252 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 253 } while (0) 254 255 /* Must match list in lfs_vfsops.c ! */ 256 # define DLOG_RF 0 /* roll forward */ 257 # define DLOG_ALLOC 1 /* inode alloc */ 258 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 259 # define DLOG_FLUSH 3 /* flush */ 260 # define DLOG_LLIST 4 /* locked list accounting */ 261 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 262 # define DLOG_VNODE 6 /* vflush/writevnodes */ 263 # define DLOG_SEG 7 /* segwrite */ 264 # define DLOG_SU 8 /* seguse accounting */ 265 # define DLOG_CLEAN 9 /* cleaner routines */ 266 # define DLOG_MOUNT 10 /* mount/unmount */ 267 # define DLOG_PAGE 11 /* putpages/gop_write */ 268 # define DLOG_DIROP 12 /* dirop accounting */ 269 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 270 # define DLOG_MAX 14 /* The terminator */ 271 # define DLOG(a) lfs_debug_log a 272 # else /* ! DEBUG */ 273 # define LFS_BCLEAN_LOG(fs, bp) 274 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 275 # define DLOG(a) 276 # endif /* ! DEBUG */ 277 #else /* ! _KERNEL */ 278 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 279 #endif /* _KERNEL */ 280 281 #ifdef _KERNEL 282 /* Filehandle structure for exported LFSes */ 283 struct lfid { 284 struct ufid lfid_ufid; 285 #define lfid_len lfid_ufid.ufid_len 286 #define lfid_ino lfid_ufid.ufid_ino 287 #define lfid_gen lfid_ufid.ufid_gen 288 uint32_t lfid_ident; 289 }; 290 #endif /* _KERNEL */ 291 292 /* 293 * "struct inode" associated definitions 294 */ 295 296 /* Address calculations for metadata located in the inode */ 297 #define S_INDIR(fs) -NDADDR 298 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 299 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 300 301 /* For convenience */ 302 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 303 304 #define LFS_SET_UINO(ip, flags) do { \ 305 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 306 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 307 ++(ip)->i_lfs->lfs_uinodes; \ 308 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 309 ++(ip)->i_lfs->lfs_uinodes; \ 310 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 311 ++(ip)->i_lfs->lfs_uinodes; \ 312 (ip)->i_flag |= (flags); \ 313 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 314 } while (0) 315 316 #define LFS_CLR_UINO(ip, flags) do { \ 317 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 318 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 319 --(ip)->i_lfs->lfs_uinodes; \ 320 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 321 --(ip)->i_lfs->lfs_uinodes; \ 322 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 323 --(ip)->i_lfs->lfs_uinodes; \ 324 (ip)->i_flag &= ~(flags); \ 325 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 326 panic("lfs_uinodes < 0"); \ 327 } \ 328 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 329 } while (0) 330 331 #define LFS_ITIMES(ip, acc, mod, cre) \ 332 while ((ip)->i_flag & (IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY)) \ 333 lfs_itimes(ip, acc, mod, cre) 334 335 /* 336 * "struct vnode" associated definitions 337 */ 338 339 /* Heuristic emptiness measure */ 340 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 341 !((vp)->v_flag & VONWORKLST) && \ 342 VTOI(vp)->i_lfs_nbtree == 0) 343 344 /* XXX Shouldn't we use v_numoutput instead? */ 345 #define WRITEINPROG(vp) (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 346 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING))) 347 348 349 /* 350 * On-disk and in-memory checkpoint segment usage structure. 351 */ 352 typedef struct segusage SEGUSE; 353 struct segusage { 354 u_int32_t su_nbytes; /* 0: number of live bytes */ 355 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 356 u_int16_t su_nsums; /* 8: number of summaries in segment */ 357 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 358 359 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 360 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 361 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 362 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 363 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 364 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 365 u_int32_t su_flags; /* 12: segment flags */ 366 u_int64_t su_lastmod; /* 16: last modified timestamp */ 367 }; 368 369 typedef struct segusage_v1 SEGUSE_V1; 370 struct segusage_v1 { 371 u_int32_t su_nbytes; /* 0: number of live bytes */ 372 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 373 u_int16_t su_nsums; /* 8: number of summaries in segment */ 374 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 375 u_int32_t su_flags; /* 12: segment flags */ 376 }; 377 378 #define SEGUPB(fs) (fs->lfs_sepb) 379 #define SEGTABSIZE_SU(fs) \ 380 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 381 382 #ifdef _KERNEL 383 # define SHARE_IFLOCK(F) \ 384 do { \ 385 simple_lock(&(F)->lfs_interlock); \ 386 lockmgr(&(F)->lfs_iflock, LK_SHARED, &(F)->lfs_interlock); \ 387 simple_unlock(&(F)->lfs_interlock); \ 388 } while(0) 389 # define UNSHARE_IFLOCK(F) \ 390 do { \ 391 simple_lock(&(F)->lfs_interlock); \ 392 lockmgr(&(F)->lfs_iflock, LK_RELEASE, &(F)->lfs_interlock); \ 393 simple_unlock(&(F)->lfs_interlock); \ 394 } while(0) 395 #else /* ! _KERNEL */ 396 # define SHARE_IFLOCK(F) 397 # define UNSHARE_IFLOCK(F) 398 #endif /* ! _KERNEL */ 399 400 /* Read in the block with a specific segment usage entry from the ifile. */ 401 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 402 int _e; \ 403 SHARE_IFLOCK(F); \ 404 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 405 if ((_e = bread((F)->lfs_ivnode, \ 406 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 407 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 408 panic("lfs: ifile read: %d", _e); \ 409 if ((F)->lfs_version == 1) \ 410 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 411 ((IN) & ((F)->lfs_sepb - 1))); \ 412 else \ 413 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 414 UNSHARE_IFLOCK(F); \ 415 } while (0) 416 417 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 418 if ((SP)->su_nbytes == 0) \ 419 (SP)->su_flags |= SEGUSE_EMPTY; \ 420 else \ 421 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 422 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 423 LFS_BWRITE_LOG(BP); \ 424 } while (0) 425 426 /* 427 * On-disk file information. One per file with data blocks in the segment. 428 */ 429 typedef struct finfo FINFO; 430 struct finfo { 431 u_int32_t fi_nblocks; /* number of blocks */ 432 u_int32_t fi_version; /* version number */ 433 u_int32_t fi_ino; /* inode number */ 434 u_int32_t fi_lastlength; /* length of last block in array */ 435 int32_t fi_blocks[1]; /* array of logical block numbers */ 436 }; 437 /* sizeof FINFO except fi_blocks */ 438 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 439 440 /* 441 * Index file inode entries. 442 */ 443 typedef struct ifile IFILE; 444 struct ifile { 445 u_int32_t if_version; /* inode version number */ 446 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 447 int32_t if_daddr; /* inode disk address */ 448 u_int32_t if_nextfree; /* next-unallocated inode */ 449 u_int32_t if_atime_sec; /* Last access time, seconds */ 450 u_int32_t if_atime_nsec; /* and nanoseconds */ 451 }; 452 453 typedef struct ifile_v1 IFILE_V1; 454 struct ifile_v1 { 455 u_int32_t if_version; /* inode version number */ 456 int32_t if_daddr; /* inode disk address */ 457 u_int32_t if_nextfree; /* next-unallocated inode */ 458 #if LFS_ATIME_IFILE 459 struct timespec if_atime; /* Last access time */ 460 #endif 461 }; 462 463 /* 464 * LFSv1 compatibility code is not allowed to touch if_atime, since it 465 * may not be mapped! 466 */ 467 /* Read in the block with a specific inode from the ifile. */ 468 #define LFS_IENTRY(IP, F, IN, BP) do { \ 469 int _e; \ 470 SHARE_IFLOCK(F); \ 471 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 472 if ((_e = bread((F)->lfs_ivnode, \ 473 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 474 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 475 panic("lfs: ifile ino %d read %d", (int)(IN), _e); \ 476 if ((F)->lfs_version == 1) \ 477 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 478 (IN) % (F)->lfs_ifpb); \ 479 else \ 480 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 481 UNSHARE_IFLOCK(F); \ 482 } while (0) 483 484 /* 485 * Cleaner information structure. This resides in the ifile and is used 486 * to pass information from the kernel to the cleaner. 487 */ 488 typedef struct _cleanerinfo { 489 u_int32_t clean; /* number of clean segments */ 490 u_int32_t dirty; /* number of dirty segments */ 491 int32_t bfree; /* disk blocks free */ 492 int32_t avail; /* disk blocks available */ 493 u_int32_t free_head; /* head of the inode free list */ 494 u_int32_t free_tail; /* tail of the inode free list */ 495 } CLEANERINFO; 496 497 #define CLEANSIZE_SU(fs) \ 498 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 499 500 /* Read in the block with the cleaner info from the ifile. */ 501 #define LFS_CLEANERINFO(CP, F, BP) do { \ 502 SHARE_IFLOCK(F); \ 503 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 504 if (bread((F)->lfs_ivnode, \ 505 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 506 panic("lfs: ifile read"); \ 507 (CP) = (CLEANERINFO *)(BP)->b_data; \ 508 UNSHARE_IFLOCK(F); \ 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 /* Type used for the inode bitmap */ 701 typedef u_int32_t lfs_bm_t; 702 703 /* 704 * In-memory super block. 705 */ 706 struct lfs { 707 struct dlfs lfs_dlfs; /* on-disk parameters */ 708 #define lfs_magic lfs_dlfs.dlfs_magic 709 #define lfs_version lfs_dlfs.dlfs_version 710 #define lfs_size lfs_dlfs.dlfs_size 711 #define lfs_ssize lfs_dlfs.dlfs_ssize 712 #define lfs_dsize lfs_dlfs.dlfs_dsize 713 #define lfs_bsize lfs_dlfs.dlfs_bsize 714 #define lfs_fsize lfs_dlfs.dlfs_fsize 715 #define lfs_frag lfs_dlfs.dlfs_frag 716 #define lfs_freehd lfs_dlfs.dlfs_freehd 717 #define lfs_bfree lfs_dlfs.dlfs_bfree 718 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 719 #define lfs_avail lfs_dlfs.dlfs_avail 720 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 721 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 722 #define lfs_ifile lfs_dlfs.dlfs_ifile 723 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 724 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 725 #define lfs_curseg lfs_dlfs.dlfs_curseg 726 #define lfs_offset lfs_dlfs.dlfs_offset 727 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 728 #define lfs_otstamp lfs_dlfs.dlfs_inopf 729 #define lfs_inopf lfs_dlfs.dlfs_inopf 730 #define lfs_minfree lfs_dlfs.dlfs_minfree 731 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 732 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 733 #define lfs_inopb lfs_dlfs.dlfs_inopb 734 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 735 #define lfs_sepb lfs_dlfs.dlfs_sepb 736 #define lfs_nindir lfs_dlfs.dlfs_nindir 737 #define lfs_nseg lfs_dlfs.dlfs_nseg 738 #define lfs_nspf lfs_dlfs.dlfs_nspf 739 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 740 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 741 #define lfs_segmask lfs_dlfs.dlfs_segmask 742 #define lfs_segshift lfs_dlfs.dlfs_segshift 743 #define lfs_bmask lfs_dlfs.dlfs_bmask 744 #define lfs_bshift lfs_dlfs.dlfs_bshift 745 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 746 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 747 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 748 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 749 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 750 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 751 #define lfs_sushift lfs_dlfs.dlfs_sushift 752 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 753 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 754 #define lfs_cksum lfs_dlfs.dlfs_cksum 755 #define lfs_pflags lfs_dlfs.dlfs_pflags 756 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 757 #define lfs_nclean lfs_dlfs.dlfs_nclean 758 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 759 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 760 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 761 #define lfs_serial lfs_dlfs.dlfs_serial 762 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 763 #define lfs_start lfs_dlfs.dlfs_start 764 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 765 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 766 #define lfs_interleave lfs_dlfs.dlfs_interleave 767 #define lfs_ident lfs_dlfs.dlfs_ident 768 769 /* These fields are set at mount time and are meaningless on disk. */ 770 struct segment *lfs_sp; /* current segment being written */ 771 struct vnode *lfs_ivnode; /* vnode for the ifile */ 772 u_int32_t lfs_seglock; /* single-thread the segment writer */ 773 pid_t lfs_lockpid; /* pid of lock holder */ 774 lwpid_t lfs_locklwp; /* lwp of lock holder */ 775 u_int32_t lfs_iocount; /* number of ios pending */ 776 u_int32_t lfs_writer; /* don't allow any dirops to start */ 777 u_int32_t lfs_dirops; /* count of active directory ops */ 778 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 779 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 780 int8_t lfs_fmod; /* super block modified flag */ 781 int8_t lfs_ronly; /* mounted read-only flag */ 782 #define LFS_NOTYET 0x01 783 #define LFS_IFDIRTY 0x02 784 #define LFS_WARNED 0x04 785 #define LFS_UNDIROP 0x08 786 int8_t lfs_flags; /* currently unused flag */ 787 u_int16_t lfs_activesb; /* toggle between superblocks */ 788 daddr_t lfs_sbactive; /* disk address of current sb write */ 789 struct vnode *lfs_flushvp; /* vnode being flushed */ 790 int lfs_flushvp_fakevref; /* fake vref count for flushvp */ 791 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 792 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 793 size_t lfs_devbsize; /* Device block size */ 794 size_t lfs_devbshift; /* Device block shift */ 795 struct lock lfs_fraglock; 796 struct lock lfs_iflock; /* Ifile lock */ 797 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 798 int lfs_nadirop; /* number of active dirop nodes */ 799 long lfs_ravail; /* blocks pre-reserved for writing */ 800 long lfs_favail; /* blocks pre-reserved for writing */ 801 res_t *lfs_resblk; /* Reserved memory for pageout */ 802 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 803 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 804 #define LFS_RESHASH_WIDTH 17 805 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 806 int lfs_pdflush; /* pagedaemon wants us to flush */ 807 u_int32_t **lfs_suflags; /* Segment use flags */ 808 #ifdef _KERNEL 809 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 810 struct pool lfs_bpppool; /* Pool for bpp */ 811 struct pool lfs_segpool; /* Pool for struct segment */ 812 #endif /* _KERNEL */ 813 #define LFS_MAX_CLEANIND 64 814 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 815 int lfs_cleanind; /* Index into intervals */ 816 struct simplelock lfs_interlock; /* lock for lfs_seglock */ 817 int lfs_sleepers; /* # procs sleeping this fs */ 818 int lfs_pages; /* dirty pages blaming this fs */ 819 lfs_bm_t *lfs_ino_bitmap; /* Inuse inodes bitmap */ 820 int lfs_nowrap; /* Suspend log wrap */ 821 }; 822 823 /* NINDIR is the number of indirects in a file system block. */ 824 #define NINDIR(fs) ((fs)->lfs_nindir) 825 826 /* INOPB is the number of inodes in a secondary storage block. */ 827 #define INOPB(fs) ((fs)->lfs_inopb) 828 /* INOPF is the number of inodes in a fragment. */ 829 #define INOPF(fs) ((fs)->lfs_inopf) 830 831 #define blksize(fs, ip, lbn) \ 832 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 833 ? (fs)->lfs_bsize \ 834 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 835 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 836 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 837 ((int)((loc) & (fs)->lfs_ffmask)) 838 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 839 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 840 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 841 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 842 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 843 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 844 /* Same as above, but named like dbtob(), btodb() */ 845 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 846 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 847 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 848 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 849 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 850 (fs)->lfs_fsbtodb)) 851 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 852 (fs)->lfs_fsbtodb)) 853 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 854 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 855 ((loc) >> (fs)->lfs_ffshift) 856 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 857 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 858 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 859 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 860 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 861 ((frags) >> (fs)->lfs_fbshift) 862 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 863 ((blks) << (fs)->lfs_fbshift) 864 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 865 ((fsb) & ((fs)->lfs_frag - 1)) 866 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 867 ((fsb) &~ ((fs)->lfs_frag - 1)) 868 #define dblksize(fs, dp, lbn) \ 869 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 870 ? (fs)->lfs_bsize \ 871 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 872 873 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 874 lblktosize((fs), (fs)->lfs_ssize) : \ 875 (fs)->lfs_ssize) 876 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 877 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 878 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 879 #define dtosn(fs, daddr) /* block address to segment number */ \ 880 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 881 #define sntod(fs, sn) /* segment number to disk address */ \ 882 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 883 884 /* 885 * Structures used by lfs_bmapv and lfs_markv to communicate information 886 * about inodes and data blocks. 887 */ 888 typedef struct block_info { 889 u_int32_t bi_inode; /* inode # */ 890 int32_t bi_lbn; /* logical block w/in file */ 891 int32_t bi_daddr; /* disk address of block */ 892 u_int64_t bi_segcreate; /* origin segment create time */ 893 int bi_version; /* file version number */ 894 void *bi_bp; /* data buffer */ 895 int bi_size; /* size of the block (if fragment) */ 896 } BLOCK_INFO; 897 898 /* Compatibility for 1.5 binaries */ 899 typedef struct block_info_15 { 900 u_int32_t bi_inode; /* inode # */ 901 int32_t bi_lbn; /* logical block w/in file */ 902 int32_t bi_daddr; /* disk address of block */ 903 u_int32_t bi_segcreate; /* origin segment create time */ 904 int bi_version; /* file version number */ 905 void *bi_bp; /* data buffer */ 906 int bi_size; /* size of the block (if fragment) */ 907 } BLOCK_INFO_15; 908 909 /* In-memory description of a segment about to be written. */ 910 struct segment { 911 struct lfs *fs; /* file system pointer */ 912 struct buf **bpp; /* pointer to buffer array */ 913 struct buf **cbpp; /* pointer to next available bp */ 914 struct buf **start_bpp; /* pointer to first bp in this set */ 915 struct buf *ibp; /* buffer pointer to inode page */ 916 struct ufs1_dinode *idp; /* pointer to ifile dinode */ 917 struct finfo *fip; /* current fileinfo pointer */ 918 struct vnode *vp; /* vnode being gathered */ 919 void *segsum; /* segment summary info */ 920 u_int32_t ninodes; /* number of inodes in this segment */ 921 int32_t seg_bytes_left; /* bytes left in segment */ 922 int32_t sum_bytes_left; /* bytes left in summary block */ 923 u_int32_t seg_number; /* number of this segment */ 924 int32_t *start_lbp; /* beginning lbn for this set */ 925 926 #define SEGM_CKP 0x01 /* doing a checkpoint */ 927 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 928 #define SEGM_SYNC 0x04 /* wait for segment */ 929 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 930 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */ 931 #define SEGM_WRITERD 0x20 /* LFS writed called us */ 932 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */ 933 u_int16_t seg_flags; /* run-time flags for this segment */ 934 u_int32_t seg_iocount; /* number of ios pending */ 935 int ndupino; /* number of duplicate inodes */ 936 }; 937 938 #ifdef _KERNEL 939 struct lfs_cluster { 940 size_t bufsize; /* Size of kept data */ 941 struct buf **bpp; /* Array of kept buffers */ 942 int bufcount; /* Number of kept buffers */ 943 #define LFS_CL_MALLOC 0x00000001 944 #define LFS_CL_SHIFT 0x00000002 945 #define LFS_CL_SYNC 0x00000004 946 u_int32_t flags; /* Flags */ 947 struct lfs *fs; /* LFS that this belongs to */ 948 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 949 }; 950 951 /* 952 * Splay tree containing block numbers allocated through lfs_balloc. 953 */ 954 struct lbnentry { 955 SPLAY_ENTRY(lbnentry) entry; 956 daddr_t lbn; 957 }; 958 #endif /* _KERNEL */ 959 960 /* 961 * LFS inode extensions. 962 */ 963 struct lfs_inode_ext { 964 off_t lfs_osize; /* size of file on disk */ 965 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 966 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */ 967 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 968 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 969 #define LFSI_NO_GOP_WRITE 0x01 970 u_int32_t lfs_iflags; /* Inode flags */ 971 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 972 #ifdef _KERNEL 973 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 974 int lfs_nbtree; /* Size of tree */ 975 #endif 976 }; 977 #define i_lfs_osize inode_ext.lfs->lfs_osize 978 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 979 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 980 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 981 #define i_lfs_pchain inode_ext.lfs->lfs_pchain 982 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 983 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 984 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 985 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 986 987 /* 988 * Macros for determining free space on the disk, with the variable metadata 989 * of segment summaries and inode blocks taken into account. 990 */ 991 /* Estimate number of clean blocks not available for writing */ 992 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 993 (int64_t)(F)->lfs_nclean) / \ 994 ((F)->lfs_nseg - (F)->lfs_nclean))) 995 996 /* Estimate total size of the disk not including metadata */ 997 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 998 999 /* Estimate number of blocks actually available for writing */ 1000 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) ? \ 1001 (F)->lfs_bfree - LFS_EST_CMETA(F) : 0) 1002 1003 /* Amount of non-meta space not available to mortal man */ 1004 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1005 (u_int64_t)(F)->lfs_minfree) / \ 1006 100) 1007 1008 /* Can credential C write BB blocks */ 1009 #define ISSPACE(F, BB, C) \ 1010 ((((C) == NOCRED || (C)->cr_uid == 0) && \ 1011 LFS_EST_BFREE(F) >= (BB)) || \ 1012 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB))) 1013 1014 /* Can an ordinary user write BB blocks */ 1015 #define IS_FREESPACE(F, BB) \ 1016 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1017 1018 /* 1019 * The minimum number of blocks to create a new inode. This is: 1020 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1021 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1022 */ 1023 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1024 1025 /* Statistics Counters */ 1026 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1027 u_int segsused; 1028 u_int psegwrites; 1029 u_int psyncwrites; 1030 u_int pcleanwrites; 1031 u_int blocktot; 1032 u_int cleanblocks; 1033 u_int ncheckpoints; 1034 u_int nwrites; 1035 u_int nsync_writes; 1036 u_int wait_exceeded; 1037 u_int write_exceeded; 1038 u_int flush_invoked; 1039 u_int vflush_invoked; 1040 u_int clean_inlocked; 1041 u_int clean_vnlocked; 1042 u_int segs_reclaimed; 1043 }; 1044 #ifdef _KERNEL 1045 extern struct lfs_stats lfs_stats; 1046 #endif 1047 1048 /* Fcntls to take the place of the lfs syscalls */ 1049 struct lfs_fcntl_markv { 1050 BLOCK_INFO *blkiov; /* blocks to relocate */ 1051 int blkcnt; /* number of blocks */ 1052 }; 1053 1054 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1055 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1056 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1057 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1058 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1059 #define LFCNIFILEFH _FCNW_FSPRIV('L', 5, struct fhandle) 1060 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1061 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1062 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1063 #define LFCNWRAPSTOP _FCNO_FSPRIV('L', 9) 1064 #define LFCNWRAPGO _FCNO_FSPRIV('L', 10) 1065 /* Compat for NetBSD 2.x bug */ 1066 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1067 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1068 1069 #ifdef _KERNEL 1070 /* XXX MP */ 1071 #define LFS_SEGLOCK_HELD(fs) \ 1072 ((fs)->lfs_seglock != 0 && \ 1073 (fs)->lfs_lockpid == curproc->p_pid && \ 1074 (fs)->lfs_locklwp == curlwp->l_lid) 1075 #endif /* _KERNEL */ 1076 1077 /* Debug segment lock */ 1078 #ifdef notyet 1079 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1080 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1081 # define ASSERT_DUNNO_SEGLOCK(fs) 1082 # define ASSERT_MAYBE_SEGLOCK(fs) 1083 #else /* !notyet */ 1084 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1085 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1086 LFS_SEGLOCK_HELD(fs))) 1087 # define ASSERT_SEGLOCK(fs) do { \ 1088 if (!LFS_SEGLOCK_HELD(fs)) { \ 1089 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1090 } \ 1091 } while(0) 1092 # define ASSERT_NO_SEGLOCK(fs) do { \ 1093 if (LFS_SEGLOCK_HELD(fs)) { \ 1094 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1095 } \ 1096 } while(0) 1097 # define ASSERT_MAYBE_SEGLOCK(x) 1098 #endif /* !notyet */ 1099 1100 __BEGIN_DECLS 1101 void lfs_itimes(struct inode *, const struct timespec *, 1102 const struct timespec *, const struct timespec *); 1103 __END_DECLS 1104 1105 #endif /* !_UFS_LFS_LFS_H_ */ 1106
Indexes created Mon Sep 29 15:09:53 GMT 2025