lfs.h revision 1.116
1 /* $NetBSD: lfs.h,v 1.116 2006/09/15 18:50:49 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 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_RESOURCE(x, u) (((x) >> 2) - 10 * (u)) 112 #define LFS_WAIT_RESOURCE(x, u) (((x) >> 1) - ((x) >> 3) - 10 * (u)) 113 #define LFS_INVERSE_MAX_RESOURCE(x, u) (((x) + 10 * (u)) << 2) 114 #define LFS_MAX_BUFS LFS_MAX_RESOURCE(nbuf, 1) 115 #define LFS_WAIT_BUFS LFS_WAIT_RESOURCE(nbuf, 1) 116 #define LFS_INVERSE_MAX_BUFS(n) LFS_INVERSE_MAX_RESOURCE(n, 1) 117 #define LFS_MAX_BYTES LFS_MAX_RESOURCE(bufmem_lowater, PAGE_SIZE) 118 #define LFS_INVERSE_MAX_BYTES(n) LFS_INVERSE_MAX_RESOURCE(n, PAGE_SIZE) 119 #define LFS_WAIT_BYTES LFS_WAIT_RESOURCE(bufmem_lowater, PAGE_SIZE) 120 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3)) 121 #define SIZEOF_DIROP(fs) (2 * ((fs)->lfs_bsize + DINODE1_SIZE)) 122 #define LFS_MAX_FSDIROP(fs) \ 123 ((fs)->lfs_nclean <= (fs)->lfs_resvseg ? 0 : \ 124 (((fs)->lfs_nclean - (fs)->lfs_resvseg) * (fs)->lfs_ssize) / \ 125 (2 * SIZEOF_DIROP(fs))) 126 #define LFS_MAX_PAGES lfs_max_pages() 127 #define LFS_WAIT_PAGES lfs_wait_pages() 128 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */ 129 130 #ifdef _KERNEL 131 int lfs_wait_pages(void); 132 int lfs_max_pages(void); 133 #endif /* _KERNEL */ 134 135 /* How starved can we be before we start holding back page writes */ 136 #define LFS_STARVED_FOR_SEGS(fs) ((fs)->lfs_nclean < (fs)->lfs_resvseg) 137 138 /* 139 * Reserved blocks for lfs_malloc 140 */ 141 142 /* Structure to keep reserved blocks */ 143 typedef struct lfs_res_blk { 144 void *p; 145 LIST_ENTRY(lfs_res_blk) res; 146 int size; 147 char inuse; 148 } res_t; 149 150 /* Types for lfs_newbuf and lfs_malloc */ 151 #define LFS_NB_UNKNOWN -1 152 #define LFS_NB_SUMMARY 0 153 #define LFS_NB_SBLOCK 1 154 #define LFS_NB_IBLOCK 2 155 #define LFS_NB_CLUSTER 3 156 #define LFS_NB_CLEAN 4 157 #define LFS_NB_BLKIOV 5 158 #define LFS_NB_COUNT 6 /* always last */ 159 160 /* Number of reserved memory blocks of each type */ 161 #define LFS_N_SUMMARIES 2 162 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 163 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 164 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 165 #define LFS_N_CLEAN 0 166 #define LFS_N_BLKIOV 1 167 168 /* Total count of "large" (non-pool) types */ 169 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 170 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV) 171 172 /* Counts for pool types */ 173 #define LFS_N_CL LFS_N_CLUSTERS 174 #define LFS_N_BPP 2 175 #define LFS_N_SEG 2 176 177 /* 178 * "struct buf" associated definitions 179 */ 180 181 /* Unassigned disk addresses. */ 182 #define UNASSIGNED -1 183 #define UNWRITTEN -2 184 185 /* Unused logical block number */ 186 #define LFS_UNUSED_LBN -1 187 188 /* Determine if a buffer belongs to the ifile */ 189 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 190 191 # define LFS_LOCK_BUF(bp) do { \ 192 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \ 193 simple_lock(&lfs_subsys_lock); \ 194 ++locked_queue_count; \ 195 locked_queue_bytes += bp->b_bufsize; \ 196 simple_unlock(&lfs_subsys_lock); \ 197 } \ 198 (bp)->b_flags |= B_LOCKED; \ 199 } while (0) 200 201 # define LFS_UNLOCK_BUF(bp) do { \ 202 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \ 203 simple_lock(&lfs_subsys_lock); \ 204 --locked_queue_count; \ 205 locked_queue_bytes -= bp->b_bufsize; \ 206 if (locked_queue_count < LFS_WAIT_BUFS && \ 207 locked_queue_bytes < LFS_WAIT_BYTES) \ 208 wakeup(&locked_queue_count); \ 209 simple_unlock(&lfs_subsys_lock); \ 210 } \ 211 (bp)->b_flags &= ~B_LOCKED; \ 212 } while (0) 213 214 #ifdef _KERNEL 215 216 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 217 218 # define LFS_IS_MALLOC_BUF(bp) (((bp)->b_flags & B_CALL) && \ 219 (bp)->b_iodone == lfs_callback) 220 221 # ifdef DEBUG 222 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 223 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 224 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 225 wakeup(&locked_queue_count); \ 226 } \ 227 } while (0) 228 # else 229 # define LFS_DEBUG_COUNTLOCKED(m) 230 # endif 231 232 /* log for debugging writes to the Ifile */ 233 # ifdef DEBUG 234 struct lfs_log_entry { 235 const char *op; 236 const char *file; 237 int pid; 238 int line; 239 daddr_t block; 240 unsigned long flags; 241 }; 242 extern int lfs_lognum; 243 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 244 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 245 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 246 int _s; \ 247 \ 248 simple_lock(&lfs_subsys_lock); \ 249 _s = splbio(); \ 250 lfs_log[lfs_lognum].op = theop; \ 251 lfs_log[lfs_lognum].file = thefile; \ 252 lfs_log[lfs_lognum].line = (theline); \ 253 lfs_log[lfs_lognum].pid = (thepid); \ 254 lfs_log[lfs_lognum].block = (lbn); \ 255 lfs_log[lfs_lognum].flags = (theflags); \ 256 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 257 splx(_s); \ 258 simple_unlock(&lfs_subsys_lock); \ 259 } while (0) 260 261 # define LFS_BCLEAN_LOG(fs, bp) do { \ 262 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 263 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 264 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 265 } while (0) 266 267 /* Must match list in lfs_vfsops.c ! */ 268 # define DLOG_RF 0 /* roll forward */ 269 # define DLOG_ALLOC 1 /* inode alloc */ 270 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 271 # define DLOG_FLUSH 3 /* flush */ 272 # define DLOG_LLIST 4 /* locked list accounting */ 273 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 274 # define DLOG_VNODE 6 /* vflush/writevnodes */ 275 # define DLOG_SEG 7 /* segwrite */ 276 # define DLOG_SU 8 /* seguse accounting */ 277 # define DLOG_CLEAN 9 /* cleaner routines */ 278 # define DLOG_MOUNT 10 /* mount/unmount */ 279 # define DLOG_PAGE 11 /* putpages/gop_write */ 280 # define DLOG_DIROP 12 /* dirop accounting */ 281 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 282 # define DLOG_MAX 14 /* The terminator */ 283 # define DLOG(a) lfs_debug_log a 284 # else /* ! DEBUG */ 285 # define LFS_BCLEAN_LOG(fs, bp) 286 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 287 # define DLOG(a) 288 # endif /* ! DEBUG */ 289 #else /* ! _KERNEL */ 290 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 291 #endif /* _KERNEL */ 292 293 #ifdef _KERNEL 294 /* Filehandle structure for exported LFSes */ 295 struct lfid { 296 struct ufid lfid_ufid; 297 #define lfid_len lfid_ufid.ufid_len 298 #define lfid_ino lfid_ufid.ufid_ino 299 #define lfid_gen lfid_ufid.ufid_gen 300 uint32_t lfid_ident; 301 }; 302 #endif /* _KERNEL */ 303 304 /* 305 * "struct inode" associated definitions 306 */ 307 308 /* Address calculations for metadata located in the inode */ 309 #define S_INDIR(fs) -NDADDR 310 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 311 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 312 313 /* For convenience */ 314 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 315 316 #define LFS_SET_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 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 326 } while (0) 327 328 #define LFS_CLR_UINO(ip, flags) do { \ 329 simple_lock(&(ip)->i_lfs->lfs_interlock); \ 330 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 331 --(ip)->i_lfs->lfs_uinodes; \ 332 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 333 --(ip)->i_lfs->lfs_uinodes; \ 334 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 335 --(ip)->i_lfs->lfs_uinodes; \ 336 (ip)->i_flag &= ~(flags); \ 337 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 338 panic("lfs_uinodes < 0"); \ 339 } \ 340 simple_unlock(&(ip)->i_lfs->lfs_interlock); \ 341 } while (0) 342 343 #define LFS_ITIMES(ip, acc, mod, cre) \ 344 while ((ip)->i_flag & (IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY)) \ 345 lfs_itimes(ip, acc, mod, cre) 346 347 /* 348 * "struct vnode" associated definitions 349 */ 350 351 /* Heuristic emptiness measure */ 352 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 353 !(vp->v_type == VREG && (vp)->v_flag & VONWORKLST) &&\ 354 VTOI(vp)->i_lfs_nbtree == 0) 355 356 #define WRITEINPROG(vp) ((vp)->v_numoutput > 0 || \ 357 (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 358 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING)))) 359 360 361 /* 362 * On-disk and in-memory checkpoint segment usage structure. 363 */ 364 typedef struct segusage SEGUSE; 365 struct segusage { 366 u_int32_t su_nbytes; /* 0: number of live bytes */ 367 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 368 u_int16_t su_nsums; /* 8: number of summaries in segment */ 369 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 370 371 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 372 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 373 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 374 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 375 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 376 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 377 u_int32_t su_flags; /* 12: segment flags */ 378 u_int64_t su_lastmod; /* 16: last modified timestamp */ 379 }; 380 381 typedef struct segusage_v1 SEGUSE_V1; 382 struct segusage_v1 { 383 u_int32_t su_nbytes; /* 0: number of live bytes */ 384 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 385 u_int16_t su_nsums; /* 8: number of summaries in segment */ 386 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 387 u_int32_t su_flags; /* 12: segment flags */ 388 }; 389 390 #define SEGUPB(fs) (fs->lfs_sepb) 391 #define SEGTABSIZE_SU(fs) \ 392 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 393 394 #ifdef _KERNEL 395 # define SHARE_IFLOCK(F) \ 396 do { \ 397 simple_lock(&(F)->lfs_interlock); \ 398 lockmgr(&(F)->lfs_iflock, LK_SHARED, &(F)->lfs_interlock); \ 399 simple_unlock(&(F)->lfs_interlock); \ 400 } while(0) 401 # define UNSHARE_IFLOCK(F) \ 402 do { \ 403 simple_lock(&(F)->lfs_interlock); \ 404 lockmgr(&(F)->lfs_iflock, LK_RELEASE, &(F)->lfs_interlock); \ 405 simple_unlock(&(F)->lfs_interlock); \ 406 } while(0) 407 #else /* ! _KERNEL */ 408 # define SHARE_IFLOCK(F) 409 # define UNSHARE_IFLOCK(F) 410 #endif /* ! _KERNEL */ 411 412 /* Read in the block with a specific segment usage entry from the ifile. */ 413 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 414 int _e; \ 415 SHARE_IFLOCK(F); \ 416 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 417 if ((_e = bread((F)->lfs_ivnode, \ 418 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 419 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 420 panic("lfs: ifile read: %d", _e); \ 421 if ((F)->lfs_version == 1) \ 422 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 423 ((IN) & ((F)->lfs_sepb - 1))); \ 424 else \ 425 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 426 UNSHARE_IFLOCK(F); \ 427 } while (0) 428 429 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 430 if ((SP)->su_nbytes == 0) \ 431 (SP)->su_flags |= SEGUSE_EMPTY; \ 432 else \ 433 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 434 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 435 LFS_BWRITE_LOG(BP); \ 436 } while (0) 437 438 /* 439 * On-disk file information. One per file with data blocks in the segment. 440 */ 441 typedef struct finfo FINFO; 442 struct finfo { 443 u_int32_t fi_nblocks; /* number of blocks */ 444 u_int32_t fi_version; /* version number */ 445 u_int32_t fi_ino; /* inode number */ 446 u_int32_t fi_lastlength; /* length of last block in array */ 447 int32_t fi_blocks[1]; /* array of logical block numbers */ 448 }; 449 /* sizeof FINFO except fi_blocks */ 450 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 451 452 /* 453 * Index file inode entries. 454 */ 455 typedef struct ifile IFILE; 456 struct ifile { 457 u_int32_t if_version; /* inode version number */ 458 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 459 int32_t if_daddr; /* inode disk address */ 460 #define LFS_ORPHAN_NEXTFREE (~(u_int32_t)0) /* indicate orphaned file */ 461 u_int32_t if_nextfree; /* next-unallocated inode */ 462 u_int32_t if_atime_sec; /* Last access time, seconds */ 463 u_int32_t if_atime_nsec; /* and nanoseconds */ 464 }; 465 466 typedef struct ifile_v1 IFILE_V1; 467 struct ifile_v1 { 468 u_int32_t if_version; /* inode version number */ 469 int32_t if_daddr; /* inode disk address */ 470 u_int32_t if_nextfree; /* next-unallocated inode */ 471 #if LFS_ATIME_IFILE 472 struct timespec if_atime; /* Last access time */ 473 #endif 474 }; 475 476 /* 477 * LFSv1 compatibility code is not allowed to touch if_atime, since it 478 * may not be mapped! 479 */ 480 /* Read in the block with a specific inode from the ifile. */ 481 #define LFS_IENTRY(IP, F, IN, BP) do { \ 482 int _e; \ 483 SHARE_IFLOCK(F); \ 484 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 485 if ((_e = bread((F)->lfs_ivnode, \ 486 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 487 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 488 panic("lfs: ifile ino %d read %d", (int)(IN), _e); \ 489 if ((F)->lfs_version == 1) \ 490 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 491 (IN) % (F)->lfs_ifpb); \ 492 else \ 493 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 494 UNSHARE_IFLOCK(F); \ 495 } while (0) 496 497 /* 498 * Cleaner information structure. This resides in the ifile and is used 499 * to pass information from the kernel to the cleaner. 500 */ 501 typedef struct _cleanerinfo { 502 u_int32_t clean; /* number of clean segments */ 503 u_int32_t dirty; /* number of dirty segments */ 504 int32_t bfree; /* disk blocks free */ 505 int32_t avail; /* disk blocks available */ 506 u_int32_t free_head; /* head of the inode free list */ 507 u_int32_t free_tail; /* tail of the inode free list */ 508 #define LFS_CLEANER_MUST_CLEAN 0x01 509 u_int32_t flags; /* status word from the kernel */ 510 } CLEANERINFO; 511 512 #define CLEANSIZE_SU(fs) \ 513 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 514 515 /* Read in the block with the cleaner info from the ifile. */ 516 #define LFS_CLEANERINFO(CP, F, BP) do { \ 517 SHARE_IFLOCK(F); \ 518 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 519 if (bread((F)->lfs_ivnode, \ 520 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 521 panic("lfs: ifile read"); \ 522 (CP) = (CLEANERINFO *)(BP)->b_data; \ 523 UNSHARE_IFLOCK(F); \ 524 } while (0) 525 526 /* 527 * Synchronize the Ifile cleaner info with current avail and bfree. 528 */ 529 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 530 simple_lock(&(fs)->lfs_interlock); \ 531 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 532 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 533 (fs)->lfs_favail) { \ 534 (cip)->bfree = (fs)->lfs_bfree; \ 535 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 536 (fs)->lfs_favail; \ 537 if (((bp)->b_flags & B_GATHERED) == 0) { \ 538 (fs)->lfs_flags |= LFS_IFDIRTY; \ 539 } \ 540 simple_unlock(&(fs)->lfs_interlock); \ 541 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 542 } else { \ 543 simple_unlock(&(fs)->lfs_interlock); \ 544 brelse(bp); \ 545 } \ 546 } while (0) 547 548 /* 549 * Get the head of the inode free list. 550 * Always caled with the segment lock held. 551 */ 552 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 553 if ((FS)->lfs_version > 1) { \ 554 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 555 (FS)->lfs_freehd = (CIP)->free_head; \ 556 brelse(BP); \ 557 } \ 558 *(FREEP) = (FS)->lfs_freehd; \ 559 } while (0) 560 561 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 562 (FS)->lfs_freehd = (VAL); \ 563 if ((FS)->lfs_version > 1) { \ 564 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 565 (CIP)->free_head = (VAL); \ 566 LFS_BWRITE_LOG(BP); \ 567 simple_lock(&fs->lfs_interlock); \ 568 (FS)->lfs_flags |= LFS_IFDIRTY; \ 569 simple_unlock(&fs->lfs_interlock); \ 570 } \ 571 } while (0) 572 573 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 574 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 575 *(FREEP) = (CIP)->free_tail; \ 576 brelse(BP); \ 577 } while (0) 578 579 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 580 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 581 (CIP)->free_tail = (VAL); \ 582 LFS_BWRITE_LOG(BP); \ 583 simple_lock(&fs->lfs_interlock); \ 584 (FS)->lfs_flags |= LFS_IFDIRTY; \ 585 simple_unlock(&fs->lfs_interlock); \ 586 } while (0) 587 588 /* 589 * On-disk segment summary information 590 */ 591 typedef struct segsum_v1 SEGSUM_V1; 592 struct segsum_v1 { 593 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 594 u_int32_t ss_datasum; /* 4: check sum of data */ 595 u_int32_t ss_magic; /* 8: segment summary magic number */ 596 #define SS_MAGIC 0x061561 597 int32_t ss_next; /* 12: next segment */ 598 u_int32_t ss_create; /* 16: creation time stamp */ 599 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 600 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 601 602 #define SS_DIROP 0x01 /* segment begins a dirop */ 603 #define SS_CONT 0x02 /* more partials to finish this write*/ 604 #define SS_CLEAN 0x04 /* written by the cleaner */ 605 #define SS_RFW 0x08 /* written by the roll-forward agent */ 606 u_int16_t ss_flags; /* 24: used for directory operations */ 607 u_int16_t ss_pad; /* 26: extra space */ 608 /* FINFO's and inode daddr's... */ 609 }; 610 611 typedef struct segsum SEGSUM; 612 struct segsum { 613 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 614 u_int32_t ss_datasum; /* 4: check sum of data */ 615 u_int32_t ss_magic; /* 8: segment summary magic number */ 616 int32_t ss_next; /* 12: next segment */ 617 u_int32_t ss_ident; /* 16: roll-forward fsid */ 618 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 619 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 620 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 621 u_int16_t ss_flags; /* 24: used for directory operations */ 622 u_int8_t ss_pad[6]; /* 26: extra space */ 623 u_int64_t ss_serial; /* 32: serial number */ 624 u_int64_t ss_create; /* 40: time stamp */ 625 /* FINFO's and inode daddr's... */ 626 }; 627 628 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 629 630 631 /* 632 * On-disk super block. 633 */ 634 struct dlfs { 635 #define LFS_MAGIC 0x070162 636 u_int32_t dlfs_magic; /* 0: magic number */ 637 #define LFS_VERSION 2 638 u_int32_t dlfs_version; /* 4: version number */ 639 640 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 641 /* number of frags in fs (v2) */ 642 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 643 /* number of bytes per segment (v2) */ 644 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 645 u_int32_t dlfs_bsize; /* 20: file system block size */ 646 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 647 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 648 649 /* Checkpoint region. */ 650 u_int32_t dlfs_freehd; /* 32: start of the free list */ 651 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 652 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 653 int32_t dlfs_avail; /* 44: blocks available for writing */ 654 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 655 int32_t dlfs_idaddr; /* 52: inode file disk address */ 656 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 657 int32_t dlfs_lastseg; /* 60: address of last segment written */ 658 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 659 int32_t dlfs_curseg; /* 68: current segment being written */ 660 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 661 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 662 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 663 #define dlfs_otstamp dlfs_inopf 664 665 /* These are configuration parameters. */ 666 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 667 668 /* These fields can be computed from the others. */ 669 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 670 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 671 u_int32_t dlfs_inopb; /* 100: inodes per block */ 672 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 673 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 674 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 675 u_int32_t dlfs_nseg; /* 116: number of segments */ 676 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 677 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 678 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 679 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 680 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 681 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 682 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 683 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 684 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 685 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 686 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 687 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 688 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 689 690 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 691 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 692 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 693 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 694 695 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 696 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 697 #define LFS_PF_CLEAN 0x1 698 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 699 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 700 u_int32_t dlfs_minfreeseg; /* 328: segments not counted in bfree */ 701 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 702 u_int64_t dlfs_serial; /* 336: serial number */ 703 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 704 int32_t dlfs_start; /* 348: start of segment 0 */ 705 u_int64_t dlfs_tstamp; /* 352: time stamp */ 706 #define LFS_44INODEFMT 0 707 #define LFS_MAXINODEFMT 0 708 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 709 u_int32_t dlfs_interleave; /* 364: segment interleave */ 710 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 711 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 712 u_int32_t dlfs_resvseg; /* 376: segments reserved for the cleaner */ 713 int8_t dlfs_pad[128]; /* 380: round to 512 bytes */ 714 /* Checksum -- last valid disk field. */ 715 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 716 }; 717 718 /* Type used for the inode bitmap */ 719 typedef u_int32_t lfs_bm_t; 720 721 /* 722 * Linked list of segments whose byte count needs updating following a 723 * file truncation. 724 */ 725 struct segdelta { 726 long segnum; 727 size_t num; 728 LIST_ENTRY(segdelta) list; 729 }; 730 731 /* 732 * In-memory super block. 733 */ 734 struct lfs { 735 struct dlfs lfs_dlfs; /* on-disk parameters */ 736 #define lfs_magic lfs_dlfs.dlfs_magic 737 #define lfs_version lfs_dlfs.dlfs_version 738 #define lfs_size lfs_dlfs.dlfs_size 739 #define lfs_ssize lfs_dlfs.dlfs_ssize 740 #define lfs_dsize lfs_dlfs.dlfs_dsize 741 #define lfs_bsize lfs_dlfs.dlfs_bsize 742 #define lfs_fsize lfs_dlfs.dlfs_fsize 743 #define lfs_frag lfs_dlfs.dlfs_frag 744 #define lfs_freehd lfs_dlfs.dlfs_freehd 745 #define lfs_bfree lfs_dlfs.dlfs_bfree 746 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 747 #define lfs_avail lfs_dlfs.dlfs_avail 748 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 749 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 750 #define lfs_ifile lfs_dlfs.dlfs_ifile 751 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 752 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 753 #define lfs_curseg lfs_dlfs.dlfs_curseg 754 #define lfs_offset lfs_dlfs.dlfs_offset 755 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 756 #define lfs_otstamp lfs_dlfs.dlfs_inopf 757 #define lfs_inopf lfs_dlfs.dlfs_inopf 758 #define lfs_minfree lfs_dlfs.dlfs_minfree 759 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 760 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 761 #define lfs_inopb lfs_dlfs.dlfs_inopb 762 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 763 #define lfs_sepb lfs_dlfs.dlfs_sepb 764 #define lfs_nindir lfs_dlfs.dlfs_nindir 765 #define lfs_nseg lfs_dlfs.dlfs_nseg 766 #define lfs_nspf lfs_dlfs.dlfs_nspf 767 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 768 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 769 #define lfs_segmask lfs_dlfs.dlfs_segmask 770 #define lfs_segshift lfs_dlfs.dlfs_segshift 771 #define lfs_bmask lfs_dlfs.dlfs_bmask 772 #define lfs_bshift lfs_dlfs.dlfs_bshift 773 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 774 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 775 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 776 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 777 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 778 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 779 #define lfs_sushift lfs_dlfs.dlfs_sushift 780 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 781 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 782 #define lfs_cksum lfs_dlfs.dlfs_cksum 783 #define lfs_pflags lfs_dlfs.dlfs_pflags 784 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 785 #define lfs_nclean lfs_dlfs.dlfs_nclean 786 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 787 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 788 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 789 #define lfs_serial lfs_dlfs.dlfs_serial 790 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 791 #define lfs_start lfs_dlfs.dlfs_start 792 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 793 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 794 #define lfs_interleave lfs_dlfs.dlfs_interleave 795 #define lfs_ident lfs_dlfs.dlfs_ident 796 #define lfs_resvseg lfs_dlfs.dlfs_resvseg 797 798 /* These fields are set at mount time and are meaningless on disk. */ 799 struct segment *lfs_sp; /* current segment being written */ 800 struct vnode *lfs_ivnode; /* vnode for the ifile */ 801 u_int32_t lfs_seglock; /* single-thread the segment writer */ 802 pid_t lfs_lockpid; /* pid of lock holder */ 803 lwpid_t lfs_locklwp; /* lwp of lock holder */ 804 u_int32_t lfs_iocount; /* number of ios pending */ 805 u_int32_t lfs_writer; /* don't allow any dirops to start */ 806 u_int32_t lfs_dirops; /* count of active directory ops */ 807 u_int32_t lfs_dirvcount; /* count of VDIROP nodes in this fs */ 808 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 809 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 810 int8_t lfs_fmod; /* super block modified flag */ 811 int8_t lfs_ronly; /* mounted read-only flag */ 812 #define LFS_NOTYET 0x01 813 #define LFS_IFDIRTY 0x02 814 #define LFS_WARNED 0x04 815 #define LFS_UNDIROP 0x08 816 int8_t lfs_flags; /* currently unused flag */ 817 u_int16_t lfs_activesb; /* toggle between superblocks */ 818 daddr_t lfs_sbactive; /* disk address of current sb write */ 819 struct vnode *lfs_flushvp; /* vnode being flushed */ 820 int lfs_flushvp_fakevref; /* fake vref count for flushvp */ 821 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 822 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 823 size_t lfs_devbsize; /* Device block size */ 824 size_t lfs_devbshift; /* Device block shift */ 825 struct lock lfs_fraglock; 826 struct lock lfs_iflock; /* Ifile lock */ 827 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 828 int lfs_nadirop; /* number of active dirop nodes */ 829 long lfs_ravail; /* blocks pre-reserved for writing */ 830 long lfs_favail; /* blocks pre-reserved for writing */ 831 res_t *lfs_resblk; /* Reserved memory for pageout */ 832 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 833 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 834 #define LFS_RESHASH_WIDTH 17 835 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 836 int lfs_pdflush; /* pagedaemon wants us to flush */ 837 u_int32_t **lfs_suflags; /* Segment use flags */ 838 #ifdef _KERNEL 839 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 840 struct pool lfs_bpppool; /* Pool for bpp */ 841 struct pool lfs_segpool; /* Pool for struct segment */ 842 #endif /* _KERNEL */ 843 #define LFS_MAX_CLEANIND 64 844 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 845 int lfs_cleanind; /* Index into intervals */ 846 struct simplelock lfs_interlock; /* lock for lfs_seglock */ 847 int lfs_sleepers; /* # procs sleeping this fs */ 848 int lfs_pages; /* dirty pages blaming this fs */ 849 lfs_bm_t *lfs_ino_bitmap; /* Inuse inodes bitmap */ 850 int lfs_nowrap; /* Suspend log wrap */ 851 int lfs_wrappass; /* Allow first log wrap requester to pass */ 852 int lfs_wrapstatus; /* Wrap status */ 853 LIST_HEAD(, segdelta) lfs_segdhd; /* List of pending trunc accounting events */ 854 }; 855 856 /* NINDIR is the number of indirects in a file system block. */ 857 #define NINDIR(fs) ((fs)->lfs_nindir) 858 859 /* INOPB is the number of inodes in a secondary storage block. */ 860 #define INOPB(fs) ((fs)->lfs_inopb) 861 /* INOPF is the number of inodes in a fragment. */ 862 #define INOPF(fs) ((fs)->lfs_inopf) 863 864 #define blksize(fs, ip, lbn) \ 865 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 866 ? (fs)->lfs_bsize \ 867 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 868 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 869 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 870 ((int)((loc) & (fs)->lfs_ffmask)) 871 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 872 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 873 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 874 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 875 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 876 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 877 /* Same as above, but named like dbtob(), btodb() */ 878 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 879 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 880 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 881 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 882 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 883 (fs)->lfs_fsbtodb)) 884 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 885 (fs)->lfs_fsbtodb)) 886 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 887 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 888 ((loc) >> (fs)->lfs_ffshift) 889 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 890 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 891 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 892 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 893 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 894 ((frags) >> (fs)->lfs_fbshift) 895 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 896 ((blks) << (fs)->lfs_fbshift) 897 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 898 ((fsb) & ((fs)->lfs_frag - 1)) 899 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 900 ((fsb) &~ ((fs)->lfs_frag - 1)) 901 #define dblksize(fs, dp, lbn) \ 902 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 903 ? (fs)->lfs_bsize \ 904 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 905 906 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 907 lblktosize((fs), (fs)->lfs_ssize) : \ 908 (fs)->lfs_ssize) 909 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 910 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 911 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 912 #define dtosn(fs, daddr) /* block address to segment number */ \ 913 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 914 #define sntod(fs, sn) /* segment number to disk address */ \ 915 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 916 917 /* 918 * Structures used by lfs_bmapv and lfs_markv to communicate information 919 * about inodes and data blocks. 920 */ 921 typedef struct block_info { 922 u_int32_t bi_inode; /* inode # */ 923 int32_t bi_lbn; /* logical block w/in file */ 924 int32_t bi_daddr; /* disk address of block */ 925 u_int64_t bi_segcreate; /* origin segment create time */ 926 int bi_version; /* file version number */ 927 void *bi_bp; /* data buffer */ 928 int bi_size; /* size of the block (if fragment) */ 929 } BLOCK_INFO; 930 931 /* Compatibility for 1.5 binaries */ 932 typedef struct block_info_15 { 933 u_int32_t bi_inode; /* inode # */ 934 int32_t bi_lbn; /* logical block w/in file */ 935 int32_t bi_daddr; /* disk address of block */ 936 u_int32_t bi_segcreate; /* origin segment create time */ 937 int bi_version; /* file version number */ 938 void *bi_bp; /* data buffer */ 939 int bi_size; /* size of the block (if fragment) */ 940 } BLOCK_INFO_15; 941 942 /* In-memory description of a segment about to be written. */ 943 struct segment { 944 struct lfs *fs; /* file system pointer */ 945 struct buf **bpp; /* pointer to buffer array */ 946 struct buf **cbpp; /* pointer to next available bp */ 947 struct buf **start_bpp; /* pointer to first bp in this set */ 948 struct buf *ibp; /* buffer pointer to inode page */ 949 struct ufs1_dinode *idp; /* pointer to ifile dinode */ 950 struct finfo *fip; /* current fileinfo pointer */ 951 struct vnode *vp; /* vnode being gathered */ 952 void *segsum; /* segment summary info */ 953 u_int32_t ninodes; /* number of inodes in this segment */ 954 int32_t seg_bytes_left; /* bytes left in segment */ 955 int32_t sum_bytes_left; /* bytes left in summary block */ 956 u_int32_t seg_number; /* number of this segment */ 957 int32_t *start_lbp; /* beginning lbn for this set */ 958 959 #define SEGM_CKP 0x01 /* doing a checkpoint */ 960 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 961 #define SEGM_SYNC 0x04 /* wait for segment */ 962 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 963 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */ 964 #define SEGM_WRITERD 0x20 /* LFS writed called us */ 965 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */ 966 u_int16_t seg_flags; /* run-time flags for this segment */ 967 u_int32_t seg_iocount; /* number of ios pending */ 968 int ndupino; /* number of duplicate inodes */ 969 }; 970 971 #ifdef _KERNEL 972 struct lfs_cluster { 973 size_t bufsize; /* Size of kept data */ 974 struct buf **bpp; /* Array of kept buffers */ 975 int bufcount; /* Number of kept buffers */ 976 #define LFS_CL_MALLOC 0x00000001 977 #define LFS_CL_SHIFT 0x00000002 978 #define LFS_CL_SYNC 0x00000004 979 u_int32_t flags; /* Flags */ 980 struct lfs *fs; /* LFS that this belongs to */ 981 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 982 }; 983 984 /* 985 * Splay tree containing block numbers allocated through lfs_balloc. 986 */ 987 struct lbnentry { 988 SPLAY_ENTRY(lbnentry) entry; 989 daddr_t lbn; 990 }; 991 #endif /* _KERNEL */ 992 993 /* 994 * LFS inode extensions. 995 */ 996 struct lfs_inode_ext { 997 off_t lfs_osize; /* size of file on disk */ 998 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 999 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */ 1000 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 1001 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 1002 #define LFSI_NO_GOP_WRITE 0x01 1003 #define LFSI_DELETED 0x02 1004 #define LFSI_WRAPBLOCK 0x04 1005 #define LFSI_WRAPWAIT 0x08 1006 u_int32_t lfs_iflags; /* Inode flags */ 1007 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 1008 #ifdef _KERNEL 1009 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 1010 int lfs_nbtree; /* Size of tree */ 1011 LIST_HEAD(, segdelta) lfs_segdhd; 1012 #endif 1013 int16_t lfs_odnlink; /* on-disk nlink count for cleaner */ 1014 }; 1015 #define i_lfs_osize inode_ext.lfs->lfs_osize 1016 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 1017 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 1018 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 1019 #define i_lfs_pchain inode_ext.lfs->lfs_pchain 1020 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 1021 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 1022 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 1023 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 1024 #define i_lfs_segdhd inode_ext.lfs->lfs_segdhd 1025 #define i_lfs_odnlink inode_ext.lfs->lfs_odnlink 1026 1027 /* 1028 * Macros for determining free space on the disk, with the variable metadata 1029 * of segment summaries and inode blocks taken into account. 1030 */ 1031 /* Estimate number of clean blocks not available for writing */ 1032 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 1033 (int64_t)(F)->lfs_nclean) / \ 1034 ((F)->lfs_nseg - (F)->lfs_nclean))) 1035 1036 /* Estimate total size of the disk not including metadata */ 1037 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 1038 1039 /* Estimate number of blocks actually available for writing */ 1040 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) ? \ 1041 (F)->lfs_bfree - LFS_EST_CMETA(F) : 0) 1042 1043 /* Amount of non-meta space not available to mortal man */ 1044 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1045 (u_int64_t)(F)->lfs_minfree) / \ 1046 100) 1047 1048 /* Can credential C write BB blocks */ 1049 #define ISSPACE(F, BB, C) \ 1050 ((((C) == NOCRED || kauth_cred_geteuid(C) == 0) && \ 1051 LFS_EST_BFREE(F) >= (BB)) || \ 1052 (kauth_cred_geteuid(C) != 0 && IS_FREESPACE(F, BB))) 1053 1054 /* Can an ordinary user write BB blocks */ 1055 #define IS_FREESPACE(F, BB) \ 1056 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1057 1058 /* 1059 * The minimum number of blocks to create a new inode. This is: 1060 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1061 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1062 */ 1063 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1064 1065 /* Statistics Counters */ 1066 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1067 u_int segsused; 1068 u_int psegwrites; 1069 u_int psyncwrites; 1070 u_int pcleanwrites; 1071 u_int blocktot; 1072 u_int cleanblocks; 1073 u_int ncheckpoints; 1074 u_int nwrites; 1075 u_int nsync_writes; 1076 u_int wait_exceeded; 1077 u_int write_exceeded; 1078 u_int flush_invoked; 1079 u_int vflush_invoked; 1080 u_int clean_inlocked; 1081 u_int clean_vnlocked; 1082 u_int segs_reclaimed; 1083 }; 1084 #ifdef _KERNEL 1085 extern struct lfs_stats lfs_stats; 1086 #endif 1087 1088 /* Fcntls to take the place of the lfs syscalls */ 1089 struct lfs_fcntl_markv { 1090 BLOCK_INFO *blkiov; /* blocks to relocate */ 1091 int blkcnt; /* number of blocks */ 1092 }; 1093 1094 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1095 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1096 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1097 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1098 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1099 1100 struct lfs_fhandle { 1101 char space[28]; /* FHANDLE_SIZE_COMPAT (but used from userland too) */ 1102 }; 1103 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1104 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1105 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1106 #define LFCNWRAPSTOP _FCNR_FSPRIV('L', 9, int) 1107 #define LFCNWRAPGO _FCNR_FSPRIV('L', 10, int) 1108 #define LFCNIFILEFH _FCNW_FSPRIV('L', 11, struct lfs_fhandle) 1109 #define LFCNWRAPPASS _FCNR_FSPRIV('L', 12, int) 1110 # define LFS_WRAP_GOING 0x0 1111 # define LFS_WRAP_WAITING 0x1 1112 #define LFCNWRAPSTATUS _FCNW_FSPRIV('L', 13, int) 1113 /* Compat */ 1114 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1115 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1116 #define LFCNIFILEFH_COMPAT _FCNW_FSPRIV('L', 5, struct lfs_fhandle) 1117 #define LFCNIFILEFH_COMPAT2 _FCN_FSPRIV(F_FSOUT, 'L', 11, 32) 1118 #define LFCNWRAPSTOP_COMPAT _FCNO_FSPRIV('L', 9) 1119 #define LFCNWRAPGO_COMPAT _FCNO_FSPRIV('L', 10) 1120 1121 #ifdef _KERNEL 1122 /* XXX MP */ 1123 #define LFS_SEGLOCK_HELD(fs) \ 1124 ((fs)->lfs_seglock != 0 && \ 1125 (fs)->lfs_lockpid == curproc->p_pid && \ 1126 (fs)->lfs_locklwp == curlwp->l_lid) 1127 #endif /* _KERNEL */ 1128 1129 /* Debug segment lock */ 1130 #ifdef notyet 1131 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1132 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1133 # define ASSERT_DUNNO_SEGLOCK(fs) 1134 # define ASSERT_MAYBE_SEGLOCK(fs) 1135 #else /* !notyet */ 1136 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1137 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1138 LFS_SEGLOCK_HELD(fs))) 1139 # define ASSERT_SEGLOCK(fs) do { \ 1140 if (!LFS_SEGLOCK_HELD(fs)) { \ 1141 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1142 } \ 1143 } while(0) 1144 # define ASSERT_NO_SEGLOCK(fs) do { \ 1145 if (LFS_SEGLOCK_HELD(fs)) { \ 1146 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1147 } \ 1148 } while(0) 1149 # define ASSERT_MAYBE_SEGLOCK(x) 1150 #endif /* !notyet */ 1151 1152 __BEGIN_DECLS 1153 void lfs_itimes(struct inode *, const struct timespec *, 1154 const struct timespec *, const struct timespec *); 1155 __END_DECLS 1156 1157 #endif /* !_UFS_LFS_LFS_H_ */ 1158
Indexes created Tue Sep 30 17:09:57 GMT 2025