lfs.h revision 1.112
1 /* $NetBSD: lfs.h,v 1.112 2006/07/31 16:34:44 martin 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 20 100 #define MIN_RESV_SEGS 15 101 #ifndef LFS_ATIME_IFILE 102 # define LFS_ATIME_IFILE 0 /* Store atime info in ifile (optional in LFSv1) */ 103 #endif 104 #define LFS_MARKV_MAXBLKCNT 65536 /* Max block count for lfs_markv() */ 105 106 /* Misc. definitions */ 107 #define BW_CLEAN 1 /* Flag for lfs_bwrite_ext() */ 108 #define PG_DELWRI PG_PAGER1 /* Local def for delayed pageout */ 109 110 /* Resource limits */ 111 #define LFS_MAX_BUFS ((nbuf >> 2) - 10) 112 #define LFS_WAIT_BUFS ((nbuf >> 1) - (nbuf >> 3) - 10) 113 #define LFS_INVERSE_MAX_BUFS(n) (((n) + 10) << 2) 114 #define LFS_MAX_BYTES ((bufmem_lowater >> 2) - 10 * PAGE_SIZE) 115 #define LFS_INVERSE_MAX_BYTES(n) (((n) + 10 * PAGE_SIZE) << 2) 116 #define LFS_WAIT_BYTES ((bufmem_lowater >> 1) - (bufmem_lowater >> 3) - 10 * PAGE_SIZE) 117 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3)) 118 #define SIZEOF_DIROP(fs) (2 * ((fs)->lfs_bsize + DINODE1_SIZE)) 119 #define LFS_MAX_FSDIROP(fs) \ 120 ((fs)->lfs_nclean <= (fs)->lfs_resvseg ? 0 : \ 121 (((fs)->lfs_nclean - (fs)->lfs_resvseg) * (fs)->lfs_ssize) / \ 122 (2 * SIZEOF_DIROP(fs))) 123 #define LFS_MAX_PAGES \ 124 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemin) >> 8) 125 #define LFS_WAIT_PAGES \ 126 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemax) >> 8) 127 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */ 128 129 /* How starved can we be before we start holding back page writes */ 130 #define LFS_STARVED_FOR_SEGS(fs) ((fs)->lfs_nclean < (fs)->lfs_resvseg) 131 132 /* 133 * Reserved blocks for lfs_malloc 134 */ 135 136 /* Structure to keep reserved blocks */ 137 typedef struct lfs_res_blk { 138 void *p; 139 LIST_ENTRY(lfs_res_blk) res; 140 int size; 141 char inuse; 142 } res_t; 143 144 /* Types for lfs_newbuf and lfs_malloc */ 145 #define LFS_NB_UNKNOWN -1 146 #define LFS_NB_SUMMARY 0 147 #define LFS_NB_SBLOCK 1 148 #define LFS_NB_IBLOCK 2 149 #define LFS_NB_CLUSTER 3 150 #define LFS_NB_CLEAN 4 151 #define LFS_NB_BLKIOV 5 152 #define LFS_NB_COUNT 6 /* always last */ 153 154 /* Number of reserved memory blocks of each type */ 155 #define LFS_N_SUMMARIES 2 156 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 157 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 158 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 159 #define LFS_N_CLEAN 0 160 #define LFS_N_BLKIOV 1 161 162 /* Total count of "large" (non-pool) types */ 163 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 164 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV) 165 166 /* Counts for pool types */ 167 #define LFS_N_CL LFS_N_CLUSTERS 168 #define LFS_N_BPP 2 169 #define LFS_N_SEG 2 170 171 /* 172 * "struct buf" associated definitions 173 */ 174 175 /* Unassigned disk addresses. */ 176 #define UNASSIGNED -1 177 #define UNWRITTEN -2 178 179 /* Unused logical block number */ 180 #define LFS_UNUSED_LBN -1 181 182 /* Determine if a buffer belongs to the ifile */ 183 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 184 185 # define LFS_LOCK_BUF(bp) do { \ 186 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \ 187 simple_lock(&lfs_subsys_lock); \ 188 ++locked_queue_count; \ 189 locked_queue_bytes += bp->b_bufsize; \ 190 simple_unlock(&lfs_subsys_lock); \ 191 } \ 192 (bp)->b_flags |= B_LOCKED; \ 193 } while (0) 194 195 # define LFS_UNLOCK_BUF(bp) do { \ 196 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \ 197 simple_lock(&lfs_subsys_lock); \ 198 --locked_queue_count; \ 199 locked_queue_bytes -= bp->b_bufsize; \ 200 if (locked_queue_count < LFS_WAIT_BUFS && \ 201 locked_queue_bytes < LFS_WAIT_BYTES) \ 202 wakeup(&locked_queue_count); \ 203 simple_unlock(&lfs_subsys_lock); \ 204 } \ 205 (bp)->b_flags &= ~B_LOCKED; \ 206 } while (0) 207 208 #ifdef _KERNEL 209 210 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 211 212 # define LFS_IS_MALLOC_BUF(bp) (((bp)->b_flags & B_CALL) && \ 213 (bp)->b_iodone == lfs_callback) 214 215 # ifdef DEBUG 216 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 217 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 218 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 219 wakeup(&locked_queue_count); \ 220 } \ 221 } while (0) 222 # else 223 # define LFS_DEBUG_COUNTLOCKED(m) 224 # endif 225 226 /* log for debugging writes to the Ifile */ 227 # ifdef DEBUG 228 struct lfs_log_entry { 229 const char *op; 230 const char *file; 231 int pid; 232 int line; 233 daddr_t block; 234 unsigned long flags; 235 }; 236 extern int lfs_lognum; 237 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 238 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 239 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 240 int _s; \ 241 \ 242 simple_lock(&lfs_subsys_lock); \ 243 _s = splbio(); \ 244 lfs_log[lfs_lognum].op = theop; \ 245 lfs_log[lfs_lognum].file = thefile; \ 246 lfs_log[lfs_lognum].line = (theline); \ 247 lfs_log[lfs_lognum].pid = (thepid); \ 248 lfs_log[lfs_lognum].block = (lbn); \ 249 lfs_log[lfs_lognum].flags = (theflags); \ 250 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 251 splx(_s); \ 252 simple_unlock(&lfs_subsys_lock); \ 253 } while (0) 254 255 # define LFS_BCLEAN_LOG(fs, bp) do { \ 256 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 257 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 258 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 259 } while (0) 260 261 /* Must match list in lfs_vfsops.c ! */ 262 # define DLOG_RF 0 /* roll forward */ 263 # define DLOG_ALLOC 1 /* inode alloc */ 264 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 265 # define DLOG_FLUSH 3 /* flush */ 266 # define DLOG_LLIST 4 /* locked list accounting */ 267 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 268 # define DLOG_VNODE 6 /* vflush/writevnodes */ 269 # define DLOG_SEG 7 /* segwrite */ 270 # define DLOG_SU 8 /* seguse accounting */ 271 # define DLOG_CLEAN 9 /* cleaner routines */ 272 # define DLOG_MOUNT 10 /* mount/unmount */ 273 # define DLOG_PAGE 11 /* putpages/gop_write */ 274 # define DLOG_DIROP 12 /* dirop accounting */ 275 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 276 # define DLOG_MAX 14 /* The terminator */ 277 # define DLOG(a) lfs_debug_log a 278 # else /* ! DEBUG */ 279 # define LFS_BCLEAN_LOG(fs, bp) 280 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 281 # define DLOG(a) 282 # endif /* ! DEBUG */ 283 #else /* ! _KERNEL */ 284 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 285 #endif /* _KERNEL */ 286 287 #ifdef _KERNEL 288 /* Filehandle structure for exported LFSes */ 289 struct lfid { 290 struct ufid lfid_ufid; 291 #define lfid_len lfid_ufid.ufid_len 292 #define lfid_ino lfid_ufid.ufid_ino 293 #define lfid_gen lfid_ufid.ufid_gen 294 uint32_t lfid_ident; 295 }; 296 #endif /* _KERNEL */ 297 298 /* 299 * "struct inode" associated definitions 300 */ 301 302 /* Address calculations for metadata located in the inode */ 303 #define S_INDIR(fs) -NDADDR 304 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 305 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 306 307 /* For convenience */ 308 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 309 310 #define LFS_SET_UINO(ip, flags) do { \ 311 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 312 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 313 ++(ip)->i_lfs->lfs_uinodes; \ 314 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 315 ++(ip)->i_lfs->lfs_uinodes; \ 316 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 317 ++(ip)->i_lfs->lfs_uinodes; \ 318 (ip)->i_flag |= (flags); \ 319 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 320 } while (0) 321 322 #define LFS_CLR_UINO(ip, flags) do { \ 323 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 324 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 325 --(ip)->i_lfs->lfs_uinodes; \ 326 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 327 --(ip)->i_lfs->lfs_uinodes; \ 328 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 329 --(ip)->i_lfs->lfs_uinodes; \ 330 (ip)->i_flag &= ~(flags); \ 331 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 332 panic("lfs_uinodes < 0"); \ 333 } \ 334 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 335 } while (0) 336 337 #define LFS_ITIMES(ip, acc, mod, cre) \ 338 while ((ip)->i_flag & (IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY)) \ 339 lfs_itimes(ip, acc, mod, cre) 340 341 /* 342 * "struct vnode" associated definitions 343 */ 344 345 /* Heuristic emptiness measure */ 346 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 347 !((vp)->v_flag & VONWORKLST) && \ 348 VTOI(vp)->i_lfs_nbtree == 0) 349 350 #define WRITEINPROG(vp) ((vp)->v_numoutput > 0 || \ 351 (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 352 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING)))) 353 354 355 /* 356 * On-disk and in-memory checkpoint segment usage structure. 357 */ 358 typedef struct segusage SEGUSE; 359 struct segusage { 360 u_int32_t su_nbytes; /* 0: number of live bytes */ 361 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 362 u_int16_t su_nsums; /* 8: number of summaries in segment */ 363 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 364 365 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 366 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 367 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 368 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 369 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 370 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 371 u_int32_t su_flags; /* 12: segment flags */ 372 u_int64_t su_lastmod; /* 16: last modified timestamp */ 373 }; 374 375 typedef struct segusage_v1 SEGUSE_V1; 376 struct segusage_v1 { 377 u_int32_t su_nbytes; /* 0: number of live bytes */ 378 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 379 u_int16_t su_nsums; /* 8: number of summaries in segment */ 380 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 381 u_int32_t su_flags; /* 12: segment flags */ 382 }; 383 384 #define SEGUPB(fs) (fs->lfs_sepb) 385 #define SEGTABSIZE_SU(fs) \ 386 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 387 388 #ifdef _KERNEL 389 # define SHARE_IFLOCK(F) \ 390 do { \ 391 simple_lock(&(F)->lfs_interlock); \ 392 lockmgr(&(F)->lfs_iflock, LK_SHARED, &(F)->lfs_interlock); \ 393 simple_unlock(&(F)->lfs_interlock); \ 394 } while(0) 395 # define UNSHARE_IFLOCK(F) \ 396 do { \ 397 simple_lock(&(F)->lfs_interlock); \ 398 lockmgr(&(F)->lfs_iflock, LK_RELEASE, &(F)->lfs_interlock); \ 399 simple_unlock(&(F)->lfs_interlock); \ 400 } while(0) 401 #else /* ! _KERNEL */ 402 # define SHARE_IFLOCK(F) 403 # define UNSHARE_IFLOCK(F) 404 #endif /* ! _KERNEL */ 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 SHARE_IFLOCK(F); \ 410 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 411 if ((_e = bread((F)->lfs_ivnode, \ 412 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 413 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 414 panic("lfs: ifile read: %d", _e); \ 415 if ((F)->lfs_version == 1) \ 416 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 417 ((IN) & ((F)->lfs_sepb - 1))); \ 418 else \ 419 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 420 UNSHARE_IFLOCK(F); \ 421 } while (0) 422 423 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 424 if ((SP)->su_nbytes == 0) \ 425 (SP)->su_flags |= SEGUSE_EMPTY; \ 426 else \ 427 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 428 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 429 LFS_BWRITE_LOG(BP); \ 430 } while (0) 431 432 /* 433 * On-disk file information. One per file with data blocks in the segment. 434 */ 435 typedef struct finfo FINFO; 436 struct finfo { 437 u_int32_t fi_nblocks; /* number of blocks */ 438 u_int32_t fi_version; /* version number */ 439 u_int32_t fi_ino; /* inode number */ 440 u_int32_t fi_lastlength; /* length of last block in array */ 441 int32_t fi_blocks[1]; /* array of logical block numbers */ 442 }; 443 /* sizeof FINFO except fi_blocks */ 444 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 445 446 /* 447 * Index file inode entries. 448 */ 449 typedef struct ifile IFILE; 450 struct ifile { 451 u_int32_t if_version; /* inode version number */ 452 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 453 int32_t if_daddr; /* inode disk address */ 454 u_int32_t if_nextfree; /* next-unallocated inode */ 455 u_int32_t if_atime_sec; /* Last access time, seconds */ 456 u_int32_t if_atime_nsec; /* and nanoseconds */ 457 }; 458 459 typedef struct ifile_v1 IFILE_V1; 460 struct ifile_v1 { 461 u_int32_t if_version; /* inode version number */ 462 int32_t if_daddr; /* inode disk address */ 463 u_int32_t if_nextfree; /* next-unallocated inode */ 464 #if LFS_ATIME_IFILE 465 struct timespec if_atime; /* Last access time */ 466 #endif 467 }; 468 469 /* 470 * LFSv1 compatibility code is not allowed to touch if_atime, since it 471 * may not be mapped! 472 */ 473 /* Read in the block with a specific inode from the ifile. */ 474 #define LFS_IENTRY(IP, F, IN, BP) do { \ 475 int _e; \ 476 SHARE_IFLOCK(F); \ 477 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 478 if ((_e = bread((F)->lfs_ivnode, \ 479 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 480 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 481 panic("lfs: ifile ino %d read %d", (int)(IN), _e); \ 482 if ((F)->lfs_version == 1) \ 483 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 484 (IN) % (F)->lfs_ifpb); \ 485 else \ 486 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 487 UNSHARE_IFLOCK(F); \ 488 } while (0) 489 490 /* 491 * Cleaner information structure. This resides in the ifile and is used 492 * to pass information from the kernel to the cleaner. 493 */ 494 typedef struct _cleanerinfo { 495 u_int32_t clean; /* number of clean segments */ 496 u_int32_t dirty; /* number of dirty segments */ 497 int32_t bfree; /* disk blocks free */ 498 int32_t avail; /* disk blocks available */ 499 u_int32_t free_head; /* head of the inode free list */ 500 u_int32_t free_tail; /* tail of the inode free list */ 501 } CLEANERINFO; 502 503 #define CLEANSIZE_SU(fs) \ 504 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 505 506 /* Read in the block with the cleaner info from the ifile. */ 507 #define LFS_CLEANERINFO(CP, F, BP) do { \ 508 SHARE_IFLOCK(F); \ 509 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 510 if (bread((F)->lfs_ivnode, \ 511 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 512 panic("lfs: ifile read"); \ 513 (CP) = (CLEANERINFO *)(BP)->b_data; \ 514 UNSHARE_IFLOCK(F); \ 515 } while (0) 516 517 /* 518 * Synchronize the Ifile cleaner info with current avail and bfree. 519 */ 520 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 521 simple_lock(&(fs)->lfs_interlock); \ 522 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 523 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 524 (fs)->lfs_favail) { \ 525 (cip)->bfree = (fs)->lfs_bfree; \ 526 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 527 (fs)->lfs_favail; \ 528 if (((bp)->b_flags & B_GATHERED) == 0) { \ 529 (fs)->lfs_flags |= LFS_IFDIRTY; \ 530 } \ 531 simple_unlock(&(fs)->lfs_interlock); \ 532 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 533 } else { \ 534 simple_unlock(&(fs)->lfs_interlock); \ 535 brelse(bp); \ 536 } \ 537 } while (0) 538 539 /* 540 * Get the head of the inode free list. 541 * Always caled with the segment lock held. 542 */ 543 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 544 if ((FS)->lfs_version > 1) { \ 545 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 546 (FS)->lfs_freehd = (CIP)->free_head; \ 547 brelse(BP); \ 548 } \ 549 *(FREEP) = (FS)->lfs_freehd; \ 550 } while (0) 551 552 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 553 (FS)->lfs_freehd = (VAL); \ 554 if ((FS)->lfs_version > 1) { \ 555 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 556 (CIP)->free_head = (VAL); \ 557 LFS_BWRITE_LOG(BP); \ 558 simple_lock(&fs->lfs_interlock); \ 559 (FS)->lfs_flags |= LFS_IFDIRTY; \ 560 simple_unlock(&fs->lfs_interlock); \ 561 } \ 562 } while (0) 563 564 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 565 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 566 *(FREEP) = (CIP)->free_tail; \ 567 brelse(BP); \ 568 } while (0) 569 570 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 571 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 572 (CIP)->free_tail = (VAL); \ 573 LFS_BWRITE_LOG(BP); \ 574 simple_lock(&fs->lfs_interlock); \ 575 (FS)->lfs_flags |= LFS_IFDIRTY; \ 576 simple_unlock(&fs->lfs_interlock); \ 577 } while (0) 578 579 /* 580 * On-disk segment summary information 581 */ 582 typedef struct segsum_v1 SEGSUM_V1; 583 struct segsum_v1 { 584 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 585 u_int32_t ss_datasum; /* 4: check sum of data */ 586 u_int32_t ss_magic; /* 8: segment summary magic number */ 587 #define SS_MAGIC 0x061561 588 int32_t ss_next; /* 12: next segment */ 589 u_int32_t ss_create; /* 16: creation time stamp */ 590 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 591 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 592 593 #define SS_DIROP 0x01 /* segment begins a dirop */ 594 #define SS_CONT 0x02 /* more partials to finish this write*/ 595 #define SS_CLEAN 0x04 /* written by the cleaner */ 596 u_int16_t ss_flags; /* 24: used for directory operations */ 597 u_int16_t ss_pad; /* 26: extra space */ 598 /* FINFO's and inode daddr's... */ 599 }; 600 601 typedef struct segsum SEGSUM; 602 struct segsum { 603 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 604 u_int32_t ss_datasum; /* 4: check sum of data */ 605 u_int32_t ss_magic; /* 8: segment summary magic number */ 606 int32_t ss_next; /* 12: next segment */ 607 u_int32_t ss_ident; /* 16: roll-forward fsid */ 608 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 609 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 610 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 611 u_int16_t ss_flags; /* 24: used for directory operations */ 612 u_int8_t ss_pad[6]; /* 26: extra space */ 613 u_int64_t ss_serial; /* 32: serial number */ 614 u_int64_t ss_create; /* 40: time stamp */ 615 /* FINFO's and inode daddr's... */ 616 }; 617 618 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 619 620 621 /* 622 * On-disk super block. 623 */ 624 struct dlfs { 625 #define LFS_MAGIC 0x070162 626 u_int32_t dlfs_magic; /* 0: magic number */ 627 #define LFS_VERSION 2 628 u_int32_t dlfs_version; /* 4: version number */ 629 630 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 631 /* number of frags in fs (v2) */ 632 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 633 /* number of bytes per segment (v2) */ 634 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 635 u_int32_t dlfs_bsize; /* 20: file system block size */ 636 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 637 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 638 639 /* Checkpoint region. */ 640 u_int32_t dlfs_freehd; /* 32: start of the free list */ 641 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 642 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 643 int32_t dlfs_avail; /* 44: blocks available for writing */ 644 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 645 int32_t dlfs_idaddr; /* 52: inode file disk address */ 646 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 647 int32_t dlfs_lastseg; /* 60: address of last segment written */ 648 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 649 int32_t dlfs_curseg; /* 68: current segment being written */ 650 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 651 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 652 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 653 #define dlfs_otstamp dlfs_inopf 654 655 /* These are configuration parameters. */ 656 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 657 658 /* These fields can be computed from the others. */ 659 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 660 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 661 u_int32_t dlfs_inopb; /* 100: inodes per block */ 662 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 663 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 664 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 665 u_int32_t dlfs_nseg; /* 116: number of segments */ 666 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 667 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 668 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 669 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 670 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 671 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 672 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 673 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 674 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 675 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 676 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 677 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 678 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 679 680 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 681 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 682 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 683 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 684 685 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 686 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 687 #define LFS_PF_CLEAN 0x1 688 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 689 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 690 u_int32_t dlfs_minfreeseg; /* 328: segments not counted in bfree */ 691 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 692 u_int64_t dlfs_serial; /* 336: serial number */ 693 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 694 int32_t dlfs_start; /* 348: start of segment 0 */ 695 u_int64_t dlfs_tstamp; /* 352: time stamp */ 696 #define LFS_44INODEFMT 0 697 #define LFS_MAXINODEFMT 0 698 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 699 u_int32_t dlfs_interleave; /* 364: segment interleave */ 700 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 701 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 702 u_int32_t dlfs_resvseg; /* 376: segments reserved for the cleaner */ 703 int8_t dlfs_pad[128]; /* 380: round to 512 bytes */ 704 /* Checksum -- last valid disk field. */ 705 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 706 }; 707 708 /* Type used for the inode bitmap */ 709 typedef u_int32_t lfs_bm_t; 710 711 /* 712 * Linked list of segments whose byte count needs updating following a 713 * file truncation. 714 */ 715 struct segdelta { 716 long segnum; 717 size_t num; 718 LIST_ENTRY(segdelta) list; 719 }; 720 721 /* 722 * In-memory super block. 723 */ 724 struct lfs { 725 struct dlfs lfs_dlfs; /* on-disk parameters */ 726 #define lfs_magic lfs_dlfs.dlfs_magic 727 #define lfs_version lfs_dlfs.dlfs_version 728 #define lfs_size lfs_dlfs.dlfs_size 729 #define lfs_ssize lfs_dlfs.dlfs_ssize 730 #define lfs_dsize lfs_dlfs.dlfs_dsize 731 #define lfs_bsize lfs_dlfs.dlfs_bsize 732 #define lfs_fsize lfs_dlfs.dlfs_fsize 733 #define lfs_frag lfs_dlfs.dlfs_frag 734 #define lfs_freehd lfs_dlfs.dlfs_freehd 735 #define lfs_bfree lfs_dlfs.dlfs_bfree 736 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 737 #define lfs_avail lfs_dlfs.dlfs_avail 738 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 739 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 740 #define lfs_ifile lfs_dlfs.dlfs_ifile 741 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 742 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 743 #define lfs_curseg lfs_dlfs.dlfs_curseg 744 #define lfs_offset lfs_dlfs.dlfs_offset 745 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 746 #define lfs_otstamp lfs_dlfs.dlfs_inopf 747 #define lfs_inopf lfs_dlfs.dlfs_inopf 748 #define lfs_minfree lfs_dlfs.dlfs_minfree 749 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 750 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 751 #define lfs_inopb lfs_dlfs.dlfs_inopb 752 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 753 #define lfs_sepb lfs_dlfs.dlfs_sepb 754 #define lfs_nindir lfs_dlfs.dlfs_nindir 755 #define lfs_nseg lfs_dlfs.dlfs_nseg 756 #define lfs_nspf lfs_dlfs.dlfs_nspf 757 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 758 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 759 #define lfs_segmask lfs_dlfs.dlfs_segmask 760 #define lfs_segshift lfs_dlfs.dlfs_segshift 761 #define lfs_bmask lfs_dlfs.dlfs_bmask 762 #define lfs_bshift lfs_dlfs.dlfs_bshift 763 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 764 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 765 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 766 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 767 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 768 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 769 #define lfs_sushift lfs_dlfs.dlfs_sushift 770 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 771 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 772 #define lfs_cksum lfs_dlfs.dlfs_cksum 773 #define lfs_pflags lfs_dlfs.dlfs_pflags 774 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 775 #define lfs_nclean lfs_dlfs.dlfs_nclean 776 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 777 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 778 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 779 #define lfs_serial lfs_dlfs.dlfs_serial 780 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 781 #define lfs_start lfs_dlfs.dlfs_start 782 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 783 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 784 #define lfs_interleave lfs_dlfs.dlfs_interleave 785 #define lfs_ident lfs_dlfs.dlfs_ident 786 #define lfs_resvseg lfs_dlfs.dlfs_resvseg 787 788 /* These fields are set at mount time and are meaningless on disk. */ 789 struct segment *lfs_sp; /* current segment being written */ 790 struct vnode *lfs_ivnode; /* vnode for the ifile */ 791 u_int32_t lfs_seglock; /* single-thread the segment writer */ 792 pid_t lfs_lockpid; /* pid of lock holder */ 793 lwpid_t lfs_locklwp; /* lwp of lock holder */ 794 u_int32_t lfs_iocount; /* number of ios pending */ 795 u_int32_t lfs_writer; /* don't allow any dirops to start */ 796 u_int32_t lfs_dirops; /* count of active directory ops */ 797 u_int32_t lfs_dirvcount; /* count of VDIROP nodes in this fs */ 798 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 799 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 800 int8_t lfs_fmod; /* super block modified flag */ 801 int8_t lfs_ronly; /* mounted read-only flag */ 802 #define LFS_NOTYET 0x01 803 #define LFS_IFDIRTY 0x02 804 #define LFS_WARNED 0x04 805 #define LFS_UNDIROP 0x08 806 int8_t lfs_flags; /* currently unused flag */ 807 u_int16_t lfs_activesb; /* toggle between superblocks */ 808 daddr_t lfs_sbactive; /* disk address of current sb write */ 809 struct vnode *lfs_flushvp; /* vnode being flushed */ 810 int lfs_flushvp_fakevref; /* fake vref count for flushvp */ 811 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 812 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 813 size_t lfs_devbsize; /* Device block size */ 814 size_t lfs_devbshift; /* Device block shift */ 815 struct lock lfs_fraglock; 816 struct lock lfs_iflock; /* Ifile lock */ 817 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 818 int lfs_nadirop; /* number of active dirop nodes */ 819 long lfs_ravail; /* blocks pre-reserved for writing */ 820 long lfs_favail; /* blocks pre-reserved for writing */ 821 res_t *lfs_resblk; /* Reserved memory for pageout */ 822 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 823 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 824 #define LFS_RESHASH_WIDTH 17 825 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 826 int lfs_pdflush; /* pagedaemon wants us to flush */ 827 u_int32_t **lfs_suflags; /* Segment use flags */ 828 #ifdef _KERNEL 829 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 830 struct pool lfs_bpppool; /* Pool for bpp */ 831 struct pool lfs_segpool; /* Pool for struct segment */ 832 #endif /* _KERNEL */ 833 #define LFS_MAX_CLEANIND 64 834 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 835 int lfs_cleanind; /* Index into intervals */ 836 struct simplelock lfs_interlock; /* lock for lfs_seglock */ 837 int lfs_sleepers; /* # procs sleeping this fs */ 838 int lfs_pages; /* dirty pages blaming this fs */ 839 lfs_bm_t *lfs_ino_bitmap; /* Inuse inodes bitmap */ 840 int lfs_nowrap; /* Suspend log wrap */ 841 LIST_HEAD(, segdelta) lfs_segdhd; /* List of pending trunc accounting events */ 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 u_int32_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 u_int32_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 #define LFSI_DELETED 0x02 992 #define LFSI_WRAPBLOCK 0x04 993 u_int32_t lfs_iflags; /* Inode flags */ 994 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 995 #ifdef _KERNEL 996 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 997 int lfs_nbtree; /* Size of tree */ 998 LIST_HEAD(, segdelta) lfs_segdhd; 999 #endif 1000 }; 1001 #define i_lfs_osize inode_ext.lfs->lfs_osize 1002 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 1003 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 1004 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 1005 #define i_lfs_pchain inode_ext.lfs->lfs_pchain 1006 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 1007 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 1008 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 1009 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 1010 #define i_lfs_segdhd inode_ext.lfs->lfs_segdhd 1011 1012 /* 1013 * Macros for determining free space on the disk, with the variable metadata 1014 * of segment summaries and inode blocks taken into account. 1015 */ 1016 /* Estimate number of clean blocks not available for writing */ 1017 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 1018 (int64_t)(F)->lfs_nclean) / \ 1019 ((F)->lfs_nseg - (F)->lfs_nclean))) 1020 1021 /* Estimate total size of the disk not including metadata */ 1022 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 1023 1024 /* Estimate number of blocks actually available for writing */ 1025 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) ? \ 1026 (F)->lfs_bfree - LFS_EST_CMETA(F) : 0) 1027 1028 /* Amount of non-meta space not available to mortal man */ 1029 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1030 (u_int64_t)(F)->lfs_minfree) / \ 1031 100) 1032 1033 /* Can credential C write BB blocks */ 1034 #define ISSPACE(F, BB, C) \ 1035 ((((C) == NOCRED || kauth_cred_geteuid(C) == 0) && \ 1036 LFS_EST_BFREE(F) >= (BB)) || \ 1037 (kauth_cred_geteuid(C) != 0 && IS_FREESPACE(F, BB))) 1038 1039 /* Can an ordinary user write BB blocks */ 1040 #define IS_FREESPACE(F, BB) \ 1041 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1042 1043 /* 1044 * The minimum number of blocks to create a new inode. This is: 1045 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1046 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1047 */ 1048 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1049 1050 /* Statistics Counters */ 1051 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1052 u_int segsused; 1053 u_int psegwrites; 1054 u_int psyncwrites; 1055 u_int pcleanwrites; 1056 u_int blocktot; 1057 u_int cleanblocks; 1058 u_int ncheckpoints; 1059 u_int nwrites; 1060 u_int nsync_writes; 1061 u_int wait_exceeded; 1062 u_int write_exceeded; 1063 u_int flush_invoked; 1064 u_int vflush_invoked; 1065 u_int clean_inlocked; 1066 u_int clean_vnlocked; 1067 u_int segs_reclaimed; 1068 }; 1069 #ifdef _KERNEL 1070 extern struct lfs_stats lfs_stats; 1071 #endif 1072 1073 /* Fcntls to take the place of the lfs syscalls */ 1074 struct lfs_fcntl_markv { 1075 BLOCK_INFO *blkiov; /* blocks to relocate */ 1076 int blkcnt; /* number of blocks */ 1077 }; 1078 1079 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1080 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1081 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1082 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1083 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1084 1085 struct lfs_fhandle { 1086 char space[32]; 1087 }; 1088 struct lfs_compat_30_fhandle { 1089 char space[24]; 1090 }; 1091 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1092 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1093 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1094 #define LFCNWRAPSTOP _FCNR_FSPRIV('L', 9, int) 1095 #define LFCNWRAPGO _FCNR_FSPRIV('L', 10, int) 1096 #define LFCNIFILEFH _FCNW_FSPRIV('L', 11, struct lfs_fhandle) 1097 /* Compat */ 1098 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1099 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1100 #define LFCNIFILEFH_COMPAT _FCNW_FSPRIV('L', 5, struct lfs_compat_30_fhandle) 1101 #define LFCNWRAPSTOP_COMPAT _FCNO_FSPRIV('L', 9) 1102 #define LFCNWRAPGO_COMPAT _FCNO_FSPRIV('L', 10) 1103 1104 #ifdef _KERNEL 1105 /* XXX MP */ 1106 #define LFS_SEGLOCK_HELD(fs) \ 1107 ((fs)->lfs_seglock != 0 && \ 1108 (fs)->lfs_lockpid == curproc->p_pid && \ 1109 (fs)->lfs_locklwp == curlwp->l_lid) 1110 #endif /* _KERNEL */ 1111 1112 /* Debug segment lock */ 1113 #ifdef notyet 1114 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1115 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1116 # define ASSERT_DUNNO_SEGLOCK(fs) 1117 # define ASSERT_MAYBE_SEGLOCK(fs) 1118 #else /* !notyet */ 1119 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1120 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1121 LFS_SEGLOCK_HELD(fs))) 1122 # define ASSERT_SEGLOCK(fs) do { \ 1123 if (!LFS_SEGLOCK_HELD(fs)) { \ 1124 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1125 } \ 1126 } while(0) 1127 # define ASSERT_NO_SEGLOCK(fs) do { \ 1128 if (LFS_SEGLOCK_HELD(fs)) { \ 1129 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1130 } \ 1131 } while(0) 1132 # define ASSERT_MAYBE_SEGLOCK(x) 1133 #endif /* !notyet */ 1134 1135 __BEGIN_DECLS 1136 void lfs_itimes(struct inode *, const struct timespec *, 1137 const struct timespec *, const struct timespec *); 1138 __END_DECLS 1139 1140 #endif /* !_UFS_LFS_LFS_H_ */ 1141
Indexes created Sun Sep 28 12:09:53 GMT 2025