lfs.h revision 1.124
1 /* $NetBSD: lfs.h,v 1.124 2008/01/03 19:28:50 ad 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 #include <sys/rwlock.h> 73 #include <sys/mutex.h> 74 #include <sys/queue.h> 75 #include <sys/condvar.h> 76 77 /* 78 * Compile-time options for LFS. 79 */ 80 #define LFS_IFIND_RETRIES 16 81 #define LFS_LOGLENGTH 1024 /* size of debugging log */ 82 #define LFS_MAX_ACTIVE 10 /* Dirty segments before ckp forced */ 83 84 /* 85 * Fixed filesystem layout parameters 86 */ 87 #define LFS_LABELPAD 8192 /* LFS label size */ 88 #define LFS_SBPAD 8192 /* LFS superblock size */ 89 90 #define LFS_UNUSED_INUM 0 /* 0: out of band inode number */ 91 #define LFS_IFILE_INUM 1 /* 1: IFILE inode number */ 92 /* 2: Root inode number */ 93 #define LOSTFOUNDINO 3 /* 3: lost+found inode number */ 94 #define LFS_FIRST_INUM 4 /* 4: first free inode number */ 95 96 #define LFS_V1_SUMMARY_SIZE 512 /* V1 fixed summary size */ 97 #define LFS_DFL_SUMMARY_SIZE 512 /* Default summary size */ 98 99 #define LFS_MAX_DADDR 0x7fffffff /* Highest addressable fsb */ 100 101 #define LFS_MAXNAMLEN 255 /* maximum name length in a dir */ 102 103 /* Adjustable filesystem parameters */ 104 #define MIN_FREE_SEGS 20 105 #define MIN_RESV_SEGS 15 106 #ifndef LFS_ATIME_IFILE 107 # define LFS_ATIME_IFILE 0 /* Store atime info in ifile (optional in LFSv1) */ 108 #endif 109 #define LFS_MARKV_MAXBLKCNT 65536 /* Max block count for lfs_markv() */ 110 111 /* Misc. definitions */ 112 #define BW_CLEAN 1 /* Flag for lfs_bwrite_ext() */ 113 #define PG_DELWRI PG_PAGER1 /* Local def for delayed pageout */ 114 115 /* Resource limits */ 116 #define LFS_MAX_RESOURCE(x, u) (((x) >> 2) - 10 * (u)) 117 #define LFS_WAIT_RESOURCE(x, u) (((x) >> 1) - ((x) >> 3) - 10 * (u)) 118 #define LFS_INVERSE_MAX_RESOURCE(x, u) (((x) + 10 * (u)) << 2) 119 #define LFS_MAX_BUFS LFS_MAX_RESOURCE(nbuf, 1) 120 #define LFS_WAIT_BUFS LFS_WAIT_RESOURCE(nbuf, 1) 121 #define LFS_INVERSE_MAX_BUFS(n) LFS_INVERSE_MAX_RESOURCE(n, 1) 122 #define LFS_MAX_BYTES LFS_MAX_RESOURCE(bufmem_lowater, PAGE_SIZE) 123 #define LFS_INVERSE_MAX_BYTES(n) LFS_INVERSE_MAX_RESOURCE(n, PAGE_SIZE) 124 #define LFS_WAIT_BYTES LFS_WAIT_RESOURCE(bufmem_lowater, PAGE_SIZE) 125 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3)) 126 #define SIZEOF_DIROP(fs) (2 * ((fs)->lfs_bsize + DINODE1_SIZE)) 127 #define LFS_MAX_FSDIROP(fs) \ 128 ((fs)->lfs_nclean <= (fs)->lfs_resvseg ? 0 : \ 129 (((fs)->lfs_nclean - (fs)->lfs_resvseg) * (fs)->lfs_ssize) / \ 130 (2 * SIZEOF_DIROP(fs))) 131 #define LFS_MAX_PAGES lfs_max_pages() 132 #define LFS_WAIT_PAGES lfs_wait_pages() 133 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */ 134 135 #ifdef _KERNEL 136 int lfs_wait_pages(void); 137 int lfs_max_pages(void); 138 #endif /* _KERNEL */ 139 140 /* How starved can we be before we start holding back page writes */ 141 #define LFS_STARVED_FOR_SEGS(fs) ((fs)->lfs_nclean < (fs)->lfs_resvseg) 142 143 /* 144 * Reserved blocks for lfs_malloc 145 */ 146 147 /* Structure to keep reserved blocks */ 148 typedef struct lfs_res_blk { 149 void *p; 150 LIST_ENTRY(lfs_res_blk) res; 151 int size; 152 char inuse; 153 } res_t; 154 155 /* Types for lfs_newbuf and lfs_malloc */ 156 #define LFS_NB_UNKNOWN -1 157 #define LFS_NB_SUMMARY 0 158 #define LFS_NB_SBLOCK 1 159 #define LFS_NB_IBLOCK 2 160 #define LFS_NB_CLUSTER 3 161 #define LFS_NB_CLEAN 4 162 #define LFS_NB_BLKIOV 5 163 #define LFS_NB_COUNT 6 /* always last */ 164 165 /* Number of reserved memory blocks of each type */ 166 #define LFS_N_SUMMARIES 2 167 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */ 168 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */ 169 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */ 170 #define LFS_N_CLEAN 0 171 #define LFS_N_BLKIOV 1 172 173 /* Total count of "large" (non-pool) types */ 174 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \ 175 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV) 176 177 /* Counts for pool types */ 178 #define LFS_N_CL LFS_N_CLUSTERS 179 #define LFS_N_BPP 2 180 #define LFS_N_SEG 2 181 182 /* 183 * "struct buf" associated definitions 184 */ 185 186 /* Unassigned disk addresses. */ 187 #define UNASSIGNED -1 188 #define UNWRITTEN -2 189 190 /* Unused logical block number */ 191 #define LFS_UNUSED_LBN -1 192 193 /* Determine if a buffer belongs to the ifile */ 194 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM) 195 196 # define LFS_LOCK_BUF(bp) do { \ 197 KASSERT(mutex_owned(&bufcache_lock)); \ 198 if (((bp)->b_cflags & BC_LOCKED) == 0 && bp->b_iodone == NULL) {\ 199 mutex_enter(&lfs_lock); \ 200 ++locked_queue_count; \ 201 locked_queue_bytes += bp->b_bufsize; \ 202 mutex_exit(&lfs_lock); \ 203 } \ 204 (bp)->b_cflags |= BC_LOCKED; \ 205 } while (0) 206 207 # define LFS_UNLOCK_BUF(bp) do { \ 208 KASSERT(mutex_owned(&bufcache_lock)); \ 209 if (((bp)->b_cflags & BC_LOCKED) != 0 && bp->b_iodone == NULL) {\ 210 mutex_enter(&lfs_lock); \ 211 --locked_queue_count; \ 212 locked_queue_bytes -= bp->b_bufsize; \ 213 if (locked_queue_count < LFS_WAIT_BUFS && \ 214 locked_queue_bytes < LFS_WAIT_BYTES) \ 215 wakeup(&locked_queue_count); \ 216 mutex_exit(&lfs_lock); \ 217 } \ 218 (bp)->b_cflags &= ~BC_LOCKED; \ 219 } while (0) 220 221 #ifdef _KERNEL 222 223 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 224 225 # define LFS_IS_MALLOC_BUF(bp) ((bp)->b_iodone == lfs_callback) 226 227 # ifdef DEBUG 228 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 229 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 230 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 231 wakeup(&locked_queue_count); \ 232 } \ 233 } while (0) 234 # else 235 # define LFS_DEBUG_COUNTLOCKED(m) 236 # endif 237 238 /* log for debugging writes to the Ifile */ 239 # ifdef DEBUG 240 struct lfs_log_entry { 241 const char *op; 242 const char *file; 243 int pid; 244 int line; 245 daddr_t block; 246 unsigned long flags; 247 }; 248 extern int lfs_lognum; 249 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 250 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 251 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 252 int _s; \ 253 \ 254 mutex_enter(&lfs_lock); \ 255 _s = splbio(); \ 256 lfs_log[lfs_lognum].op = theop; \ 257 lfs_log[lfs_lognum].file = thefile; \ 258 lfs_log[lfs_lognum].line = (theline); \ 259 lfs_log[lfs_lognum].pid = (thepid); \ 260 lfs_log[lfs_lognum].block = (lbn); \ 261 lfs_log[lfs_lognum].flags = (theflags); \ 262 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 263 splx(_s); \ 264 mutex_exit(&lfs_lock); \ 265 } while (0) 266 267 # define LFS_BCLEAN_LOG(fs, bp) do { \ 268 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 269 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 270 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 271 } while (0) 272 273 /* Must match list in lfs_vfsops.c ! */ 274 # define DLOG_RF 0 /* roll forward */ 275 # define DLOG_ALLOC 1 /* inode alloc */ 276 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 277 # define DLOG_FLUSH 3 /* flush */ 278 # define DLOG_LLIST 4 /* locked list accounting */ 279 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 280 # define DLOG_VNODE 6 /* vflush/writevnodes */ 281 # define DLOG_SEG 7 /* segwrite */ 282 # define DLOG_SU 8 /* seguse accounting */ 283 # define DLOG_CLEAN 9 /* cleaner routines */ 284 # define DLOG_MOUNT 10 /* mount/unmount */ 285 # define DLOG_PAGE 11 /* putpages/gop_write */ 286 # define DLOG_DIROP 12 /* dirop accounting */ 287 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 288 # define DLOG_MAX 14 /* The terminator */ 289 # define DLOG(a) lfs_debug_log a 290 # else /* ! DEBUG */ 291 # define LFS_BCLEAN_LOG(fs, bp) 292 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 293 # define DLOG(a) 294 # endif /* ! DEBUG */ 295 #else /* ! _KERNEL */ 296 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 297 #endif /* _KERNEL */ 298 299 #ifdef _KERNEL 300 /* Filehandle structure for exported LFSes */ 301 struct lfid { 302 struct ufid lfid_ufid; 303 #define lfid_len lfid_ufid.ufid_len 304 #define lfid_ino lfid_ufid.ufid_ino 305 #define lfid_gen lfid_ufid.ufid_gen 306 uint32_t lfid_ident; 307 }; 308 #endif /* _KERNEL */ 309 310 /* 311 * "struct inode" associated definitions 312 */ 313 314 /* Address calculations for metadata located in the inode */ 315 #define S_INDIR(fs) -NDADDR 316 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 317 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 318 319 /* For convenience */ 320 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 321 322 #define LFS_SET_UINO(ip, flags) do { \ 323 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 324 ++(ip)->i_lfs->lfs_uinodes; \ 325 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 326 ++(ip)->i_lfs->lfs_uinodes; \ 327 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 328 ++(ip)->i_lfs->lfs_uinodes; \ 329 (ip)->i_flag |= (flags); \ 330 } while (0) 331 332 #define LFS_CLR_UINO(ip, flags) do { \ 333 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 334 --(ip)->i_lfs->lfs_uinodes; \ 335 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 336 --(ip)->i_lfs->lfs_uinodes; \ 337 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 338 --(ip)->i_lfs->lfs_uinodes; \ 339 (ip)->i_flag &= ~(flags); \ 340 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 341 panic("lfs_uinodes < 0"); \ 342 } \ 343 } while (0) 344 345 #define LFS_ITIMES(ip, acc, mod, cre) \ 346 while ((ip)->i_flag & (IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY)) \ 347 lfs_itimes(ip, acc, mod, cre) 348 349 /* 350 * "struct vnode" associated definitions 351 */ 352 353 /* Heuristic emptiness measure */ 354 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 355 !(vp->v_type == VREG && (vp)->v_iflag & VI_ONWORKLST) &&\ 356 VTOI(vp)->i_lfs_nbtree == 0) 357 358 #define WRITEINPROG(vp) ((vp)->v_numoutput > 0 || \ 359 (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \ 360 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING)))) 361 362 363 /* 364 * On-disk and in-memory checkpoint segment usage structure. 365 */ 366 typedef struct segusage SEGUSE; 367 struct segusage { 368 u_int32_t su_nbytes; /* 0: number of live bytes */ 369 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */ 370 u_int16_t su_nsums; /* 8: number of summaries in segment */ 371 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 372 373 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */ 374 #define SEGUSE_DIRTY 0x02 /* segment has data in it */ 375 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */ 376 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */ 377 #define SEGUSE_EMPTY 0x10 /* segment is empty */ 378 #define SEGUSE_INVAL 0x20 /* segment is invalid */ 379 u_int32_t su_flags; /* 12: segment flags */ 380 u_int64_t su_lastmod; /* 16: last modified timestamp */ 381 }; 382 383 typedef struct segusage_v1 SEGUSE_V1; 384 struct segusage_v1 { 385 u_int32_t su_nbytes; /* 0: number of live bytes */ 386 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */ 387 u_int16_t su_nsums; /* 8: number of summaries in segment */ 388 u_int16_t su_ninos; /* 10: number of inode blocks in seg */ 389 u_int32_t su_flags; /* 12: segment flags */ 390 }; 391 392 #define SEGUPB(fs) (fs->lfs_sepb) 393 #define SEGTABSIZE_SU(fs) \ 394 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb) 395 396 #ifdef _KERNEL 397 # define SHARE_IFLOCK(F) \ 398 do { \ 399 rw_enter(&(F)->lfs_iflock, RW_READER); \ 400 } while(0) 401 # define UNSHARE_IFLOCK(F) \ 402 do { \ 403 rw_exit(&(F)->lfs_iflock); \ 404 } while(0) 405 #else /* ! _KERNEL */ 406 # define SHARE_IFLOCK(F) 407 # define UNSHARE_IFLOCK(F) 408 #endif /* ! _KERNEL */ 409 410 /* Read in the block with a specific segment usage entry from the ifile. */ 411 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 412 int _e; \ 413 SHARE_IFLOCK(F); \ 414 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 415 if ((_e = bread((F)->lfs_ivnode, \ 416 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 417 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 418 panic("lfs: ifile read: %d", _e); \ 419 if ((F)->lfs_version == 1) \ 420 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 421 ((IN) & ((F)->lfs_sepb - 1))); \ 422 else \ 423 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 424 UNSHARE_IFLOCK(F); \ 425 } while (0) 426 427 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 428 if ((SP)->su_nbytes == 0) \ 429 (SP)->su_flags |= SEGUSE_EMPTY; \ 430 else \ 431 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 432 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 433 LFS_BWRITE_LOG(BP); \ 434 } while (0) 435 436 /* 437 * On-disk file information. One per file with data blocks in the segment. 438 */ 439 typedef struct finfo FINFO; 440 struct finfo { 441 u_int32_t fi_nblocks; /* number of blocks */ 442 u_int32_t fi_version; /* version number */ 443 u_int32_t fi_ino; /* inode number */ 444 u_int32_t fi_lastlength; /* length of last block in array */ 445 int32_t fi_blocks[1]; /* array of logical block numbers */ 446 }; 447 /* sizeof FINFO except fi_blocks */ 448 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 449 450 /* 451 * Index file inode entries. 452 */ 453 typedef struct ifile IFILE; 454 struct ifile { 455 u_int32_t if_version; /* inode version number */ 456 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 457 int32_t if_daddr; /* inode disk address */ 458 #define LFS_ORPHAN_NEXTFREE (~(u_int32_t)0) /* indicate orphaned file */ 459 u_int32_t if_nextfree; /* next-unallocated inode */ 460 u_int32_t if_atime_sec; /* Last access time, seconds */ 461 u_int32_t if_atime_nsec; /* and nanoseconds */ 462 }; 463 464 typedef struct ifile_v1 IFILE_V1; 465 struct ifile_v1 { 466 u_int32_t if_version; /* inode version number */ 467 int32_t if_daddr; /* inode disk address */ 468 u_int32_t if_nextfree; /* next-unallocated inode */ 469 #if LFS_ATIME_IFILE 470 struct timespec if_atime; /* Last access time */ 471 #endif 472 }; 473 474 /* 475 * LFSv1 compatibility code is not allowed to touch if_atime, since it 476 * may not be mapped! 477 */ 478 /* Read in the block with a specific inode from the ifile. */ 479 #define LFS_IENTRY(IP, F, IN, BP) do { \ 480 int _e; \ 481 SHARE_IFLOCK(F); \ 482 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 483 if ((_e = bread((F)->lfs_ivnode, \ 484 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 485 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 486 panic("lfs: ifile ino %d read %d", (int)(IN), _e); \ 487 if ((F)->lfs_version == 1) \ 488 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 489 (IN) % (F)->lfs_ifpb); \ 490 else \ 491 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 492 UNSHARE_IFLOCK(F); \ 493 } while (0) 494 495 /* 496 * Cleaner information structure. This resides in the ifile and is used 497 * to pass information from the kernel to the cleaner. 498 */ 499 typedef struct _cleanerinfo { 500 u_int32_t clean; /* number of clean segments */ 501 u_int32_t dirty; /* number of dirty segments */ 502 int32_t bfree; /* disk blocks free */ 503 int32_t avail; /* disk blocks available */ 504 u_int32_t free_head; /* head of the inode free list */ 505 u_int32_t free_tail; /* tail of the inode free list */ 506 #define LFS_CLEANER_MUST_CLEAN 0x01 507 u_int32_t flags; /* status word from the kernel */ 508 } CLEANERINFO; 509 510 #define CLEANSIZE_SU(fs) \ 511 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 512 513 /* Read in the block with the cleaner info from the ifile. */ 514 #define LFS_CLEANERINFO(CP, F, BP) do { \ 515 SHARE_IFLOCK(F); \ 516 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 517 if (bread((F)->lfs_ivnode, \ 518 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 519 panic("lfs: ifile read"); \ 520 (CP) = (CLEANERINFO *)(BP)->b_data; \ 521 UNSHARE_IFLOCK(F); \ 522 } while (0) 523 524 /* 525 * Synchronize the Ifile cleaner info with current avail and bfree. 526 */ 527 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 528 mutex_enter(&lfs_lock); \ 529 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 530 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 531 (fs)->lfs_favail) { \ 532 (cip)->bfree = (fs)->lfs_bfree; \ 533 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 534 (fs)->lfs_favail; \ 535 if (((bp)->b_flags & B_GATHERED) == 0) { \ 536 (fs)->lfs_flags |= LFS_IFDIRTY; \ 537 } \ 538 mutex_exit(&lfs_lock); \ 539 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 540 } else { \ 541 mutex_exit(&lfs_lock); \ 542 brelse(bp, 0); \ 543 } \ 544 } while (0) 545 546 /* 547 * Get the head of the inode free list. 548 * Always caled with the segment lock held. 549 */ 550 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 551 if ((FS)->lfs_version > 1) { \ 552 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 553 (FS)->lfs_freehd = (CIP)->free_head; \ 554 brelse(BP, 0); \ 555 } \ 556 *(FREEP) = (FS)->lfs_freehd; \ 557 } while (0) 558 559 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 560 (FS)->lfs_freehd = (VAL); \ 561 if ((FS)->lfs_version > 1) { \ 562 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 563 (CIP)->free_head = (VAL); \ 564 LFS_BWRITE_LOG(BP); \ 565 mutex_enter(&lfs_lock); \ 566 (FS)->lfs_flags |= LFS_IFDIRTY; \ 567 mutex_exit(&lfs_lock); \ 568 } \ 569 } while (0) 570 571 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 572 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 573 *(FREEP) = (CIP)->free_tail; \ 574 brelse(BP, 0); \ 575 } while (0) 576 577 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 578 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 579 (CIP)->free_tail = (VAL); \ 580 LFS_BWRITE_LOG(BP); \ 581 mutex_enter(&lfs_lock); \ 582 (FS)->lfs_flags |= LFS_IFDIRTY; \ 583 mutex_exit(&lfs_lock); \ 584 } while (0) 585 586 /* 587 * On-disk segment summary information 588 */ 589 typedef struct segsum_v1 SEGSUM_V1; 590 struct segsum_v1 { 591 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 592 u_int32_t ss_datasum; /* 4: check sum of data */ 593 u_int32_t ss_magic; /* 8: segment summary magic number */ 594 #define SS_MAGIC 0x061561 595 int32_t ss_next; /* 12: next segment */ 596 u_int32_t ss_create; /* 16: creation time stamp */ 597 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 598 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 599 600 #define SS_DIROP 0x01 /* segment begins a dirop */ 601 #define SS_CONT 0x02 /* more partials to finish this write*/ 602 #define SS_CLEAN 0x04 /* written by the cleaner */ 603 #define SS_RFW 0x08 /* written by the roll-forward agent */ 604 u_int16_t ss_flags; /* 24: used for directory operations */ 605 u_int16_t ss_pad; /* 26: extra space */ 606 /* FINFO's and inode daddr's... */ 607 }; 608 609 typedef struct segsum SEGSUM; 610 struct segsum { 611 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 612 u_int32_t ss_datasum; /* 4: check sum of data */ 613 u_int32_t ss_magic; /* 8: segment summary magic number */ 614 int32_t ss_next; /* 12: next segment */ 615 u_int32_t ss_ident; /* 16: roll-forward fsid */ 616 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 617 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 618 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 619 u_int16_t ss_flags; /* 24: used for directory operations */ 620 u_int8_t ss_pad[6]; /* 26: extra space */ 621 u_int64_t ss_serial; /* 32: serial number */ 622 u_int64_t ss_create; /* 40: time stamp */ 623 /* FINFO's and inode daddr's... */ 624 }; 625 626 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 627 628 629 /* 630 * On-disk super block. 631 */ 632 struct dlfs { 633 #define LFS_MAGIC 0x070162 634 u_int32_t dlfs_magic; /* 0: magic number */ 635 #define LFS_VERSION 2 636 u_int32_t dlfs_version; /* 4: version number */ 637 638 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 639 /* number of frags in fs (v2) */ 640 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 641 /* number of bytes per segment (v2) */ 642 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 643 u_int32_t dlfs_bsize; /* 20: file system block size */ 644 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 645 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 646 647 /* Checkpoint region. */ 648 u_int32_t dlfs_freehd; /* 32: start of the free list */ 649 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 650 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 651 int32_t dlfs_avail; /* 44: blocks available for writing */ 652 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 653 int32_t dlfs_idaddr; /* 52: inode file disk address */ 654 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 655 int32_t dlfs_lastseg; /* 60: address of last segment written */ 656 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 657 int32_t dlfs_curseg; /* 68: current segment being written */ 658 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 659 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 660 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 661 #define dlfs_otstamp dlfs_inopf 662 663 /* These are configuration parameters. */ 664 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 665 666 /* These fields can be computed from the others. */ 667 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 668 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 669 u_int32_t dlfs_inopb; /* 100: inodes per block */ 670 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 671 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 672 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 673 u_int32_t dlfs_nseg; /* 116: number of segments */ 674 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 675 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 676 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 677 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 678 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 679 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 680 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 681 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 682 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 683 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 684 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 685 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 686 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 687 688 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 689 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 690 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 691 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 692 693 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 694 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 695 #define LFS_PF_CLEAN 0x1 696 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 697 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 698 u_int32_t dlfs_minfreeseg; /* 328: segments not counted in bfree */ 699 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 700 u_int64_t dlfs_serial; /* 336: serial number */ 701 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 702 int32_t dlfs_start; /* 348: start of segment 0 */ 703 u_int64_t dlfs_tstamp; /* 352: time stamp */ 704 #define LFS_44INODEFMT 0 705 #define LFS_MAXINODEFMT 0 706 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 707 u_int32_t dlfs_interleave; /* 364: segment interleave */ 708 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 709 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 710 u_int32_t dlfs_resvseg; /* 376: segments reserved for the cleaner */ 711 int8_t dlfs_pad[128]; /* 380: round to 512 bytes */ 712 /* Checksum -- last valid disk field. */ 713 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 714 }; 715 716 /* Type used for the inode bitmap */ 717 typedef u_int32_t lfs_bm_t; 718 719 /* 720 * Linked list of segments whose byte count needs updating following a 721 * file truncation. 722 */ 723 struct segdelta { 724 long segnum; 725 size_t num; 726 LIST_ENTRY(segdelta) list; 727 }; 728 729 /* 730 * In-memory super block. 731 */ 732 struct lfs { 733 struct dlfs lfs_dlfs; /* on-disk parameters */ 734 #define lfs_magic lfs_dlfs.dlfs_magic 735 #define lfs_version lfs_dlfs.dlfs_version 736 #define lfs_size lfs_dlfs.dlfs_size 737 #define lfs_ssize lfs_dlfs.dlfs_ssize 738 #define lfs_dsize lfs_dlfs.dlfs_dsize 739 #define lfs_bsize lfs_dlfs.dlfs_bsize 740 #define lfs_fsize lfs_dlfs.dlfs_fsize 741 #define lfs_frag lfs_dlfs.dlfs_frag 742 #define lfs_freehd lfs_dlfs.dlfs_freehd 743 #define lfs_bfree lfs_dlfs.dlfs_bfree 744 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 745 #define lfs_avail lfs_dlfs.dlfs_avail 746 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 747 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 748 #define lfs_ifile lfs_dlfs.dlfs_ifile 749 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 750 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 751 #define lfs_curseg lfs_dlfs.dlfs_curseg 752 #define lfs_offset lfs_dlfs.dlfs_offset 753 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 754 #define lfs_otstamp lfs_dlfs.dlfs_inopf 755 #define lfs_inopf lfs_dlfs.dlfs_inopf 756 #define lfs_minfree lfs_dlfs.dlfs_minfree 757 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 758 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 759 #define lfs_inopb lfs_dlfs.dlfs_inopb 760 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 761 #define lfs_sepb lfs_dlfs.dlfs_sepb 762 #define lfs_nindir lfs_dlfs.dlfs_nindir 763 #define lfs_nseg lfs_dlfs.dlfs_nseg 764 #define lfs_nspf lfs_dlfs.dlfs_nspf 765 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 766 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 767 #define lfs_segmask lfs_dlfs.dlfs_segmask 768 #define lfs_segshift lfs_dlfs.dlfs_segshift 769 #define lfs_bmask lfs_dlfs.dlfs_bmask 770 #define lfs_bshift lfs_dlfs.dlfs_bshift 771 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 772 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 773 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 774 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 775 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 776 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 777 #define lfs_sushift lfs_dlfs.dlfs_sushift 778 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 779 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 780 #define lfs_cksum lfs_dlfs.dlfs_cksum 781 #define lfs_pflags lfs_dlfs.dlfs_pflags 782 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 783 #define lfs_nclean lfs_dlfs.dlfs_nclean 784 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 785 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 786 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 787 #define lfs_serial lfs_dlfs.dlfs_serial 788 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 789 #define lfs_start lfs_dlfs.dlfs_start 790 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 791 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 792 #define lfs_interleave lfs_dlfs.dlfs_interleave 793 #define lfs_ident lfs_dlfs.dlfs_ident 794 #define lfs_resvseg lfs_dlfs.dlfs_resvseg 795 796 /* These fields are set at mount time and are meaningless on disk. */ 797 struct segment *lfs_sp; /* current segment being written */ 798 struct vnode *lfs_ivnode; /* vnode for the ifile */ 799 u_int32_t lfs_seglock; /* single-thread the segment writer */ 800 pid_t lfs_lockpid; /* pid of lock holder */ 801 lwpid_t lfs_locklwp; /* lwp of lock holder */ 802 u_int32_t lfs_iocount; /* number of ios pending */ 803 u_int32_t lfs_writer; /* don't allow any dirops to start */ 804 u_int32_t lfs_dirops; /* count of active directory ops */ 805 u_int32_t lfs_dirvcount; /* count of VDIROP nodes in this fs */ 806 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 807 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 808 int8_t lfs_fmod; /* super block modified flag */ 809 int8_t lfs_ronly; /* mounted read-only flag */ 810 #define LFS_NOTYET 0x01 811 #define LFS_IFDIRTY 0x02 812 #define LFS_WARNED 0x04 813 #define LFS_UNDIROP 0x08 814 int8_t lfs_flags; /* currently unused flag */ 815 u_int16_t lfs_activesb; /* toggle between superblocks */ 816 daddr_t lfs_sbactive; /* disk address of current sb write */ 817 struct vnode *lfs_flushvp; /* vnode being flushed */ 818 int lfs_flushvp_fakevref; /* fake vref count for flushvp */ 819 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 820 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 821 size_t lfs_devbsize; /* Device block size */ 822 size_t lfs_devbshift; /* Device block shift */ 823 krwlock_t lfs_fraglock; 824 krwlock_t lfs_iflock; /* Ifile lock */ 825 kcondvar_t lfs_stopcv; /* Wrap lock */ 826 struct lwp *lfs_stoplwp; 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 int lfs_sleepers; /* # procs sleeping this fs */ 847 int lfs_pages; /* dirty pages blaming this fs */ 848 lfs_bm_t *lfs_ino_bitmap; /* Inuse inodes bitmap */ 849 int lfs_nowrap; /* Suspend log wrap */ 850 int lfs_wrappass; /* Allow first log wrap requester to pass */ 851 int lfs_wrapstatus; /* Wrap status */ 852 LIST_HEAD(, segdelta) lfs_segdhd; /* List of pending trunc accounting events */ 853 }; 854 855 /* NINDIR is the number of indirects in a file system block. */ 856 #define NINDIR(fs) ((fs)->lfs_nindir) 857 858 /* INOPB is the number of inodes in a secondary storage block. */ 859 #define INOPB(fs) ((fs)->lfs_inopb) 860 /* INOPF is the number of inodes in a fragment. */ 861 #define INOPF(fs) ((fs)->lfs_inopf) 862 863 #define blksize(fs, ip, lbn) \ 864 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 865 ? (fs)->lfs_bsize \ 866 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 867 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 868 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 869 ((int)((loc) & (fs)->lfs_ffmask)) 870 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 871 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 872 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 873 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 874 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 875 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 876 /* Same as above, but named like dbtob(), btodb() */ 877 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 878 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 879 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 880 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 881 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 882 (fs)->lfs_fsbtodb)) 883 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 884 (fs)->lfs_fsbtodb)) 885 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 886 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 887 ((loc) >> (fs)->lfs_ffshift) 888 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 889 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 890 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 891 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 892 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 893 ((frags) >> (fs)->lfs_fbshift) 894 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 895 ((blks) << (fs)->lfs_fbshift) 896 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 897 ((fsb) & ((fs)->lfs_frag - 1)) 898 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 899 ((fsb) &~ ((fs)->lfs_frag - 1)) 900 #define dblksize(fs, dp, lbn) \ 901 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 902 ? (fs)->lfs_bsize \ 903 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 904 905 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 906 lblktosize((fs), (fs)->lfs_ssize) : \ 907 (fs)->lfs_ssize) 908 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 909 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 910 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 911 #define dtosn(fs, daddr) /* block address to segment number */ \ 912 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 913 #define sntod(fs, sn) /* segment number to disk address */ \ 914 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 915 916 /* 917 * Structures used by lfs_bmapv and lfs_markv to communicate information 918 * about inodes and data blocks. 919 */ 920 typedef struct block_info { 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_int64_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; 929 930 /* Compatibility for 1.5 binaries */ 931 typedef struct block_info_15 { 932 u_int32_t bi_inode; /* inode # */ 933 int32_t bi_lbn; /* logical block w/in file */ 934 int32_t bi_daddr; /* disk address of block */ 935 u_int32_t bi_segcreate; /* origin segment create time */ 936 int bi_version; /* file version number */ 937 void *bi_bp; /* data buffer */ 938 int bi_size; /* size of the block (if fragment) */ 939 } BLOCK_INFO_15; 940 941 /* In-memory description of a segment about to be written. */ 942 struct segment { 943 struct lfs *fs; /* file system pointer */ 944 struct buf **bpp; /* pointer to buffer array */ 945 struct buf **cbpp; /* pointer to next available bp */ 946 struct buf **start_bpp; /* pointer to first bp in this set */ 947 struct buf *ibp; /* buffer pointer to inode page */ 948 struct ufs1_dinode *idp; /* pointer to ifile dinode */ 949 struct finfo *fip; /* current fileinfo pointer */ 950 struct vnode *vp; /* vnode being gathered */ 951 void *segsum; /* segment summary info */ 952 u_int32_t ninodes; /* number of inodes in this segment */ 953 int32_t seg_bytes_left; /* bytes left in segment */ 954 int32_t sum_bytes_left; /* bytes left in summary block */ 955 u_int32_t seg_number; /* number of this segment */ 956 int32_t *start_lbp; /* beginning lbn for this set */ 957 958 #define SEGM_CKP 0x01 /* doing a checkpoint */ 959 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 960 #define SEGM_SYNC 0x04 /* wait for segment */ 961 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 962 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */ 963 #define SEGM_WRITERD 0x20 /* LFS writed called us */ 964 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */ 965 u_int16_t seg_flags; /* run-time flags for this segment */ 966 u_int32_t seg_iocount; /* number of ios pending */ 967 int ndupino; /* number of duplicate inodes */ 968 }; 969 970 #ifdef _KERNEL 971 struct lfs_cluster { 972 size_t bufsize; /* Size of kept data */ 973 struct buf **bpp; /* Array of kept buffers */ 974 int bufcount; /* Number of kept buffers */ 975 #define LFS_CL_MALLOC 0x00000001 976 #define LFS_CL_SHIFT 0x00000002 977 #define LFS_CL_SYNC 0x00000004 978 u_int32_t flags; /* Flags */ 979 struct lfs *fs; /* LFS that this belongs to */ 980 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 981 }; 982 983 /* 984 * Splay tree containing block numbers allocated through lfs_balloc. 985 */ 986 struct lbnentry { 987 SPLAY_ENTRY(lbnentry) entry; 988 daddr_t lbn; 989 }; 990 #endif /* _KERNEL */ 991 992 /* 993 * LFS inode extensions. 994 */ 995 struct lfs_inode_ext { 996 off_t lfs_osize; /* size of file on disk */ 997 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 998 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */ 999 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 1000 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 1001 #define LFSI_NO_GOP_WRITE 0x01 1002 #define LFSI_DELETED 0x02 1003 #define LFSI_WRAPBLOCK 0x04 1004 #define LFSI_WRAPWAIT 0x08 1005 u_int32_t lfs_iflags; /* Inode flags */ 1006 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 1007 #ifdef _KERNEL 1008 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 1009 int lfs_nbtree; /* Size of tree */ 1010 LIST_HEAD(, segdelta) lfs_segdhd; 1011 #endif 1012 int16_t lfs_odnlink; /* on-disk nlink count for cleaner */ 1013 }; 1014 #define i_lfs_osize inode_ext.lfs->lfs_osize 1015 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 1016 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 1017 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 1018 #define i_lfs_pchain inode_ext.lfs->lfs_pchain 1019 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 1020 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 1021 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 1022 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 1023 #define i_lfs_segdhd inode_ext.lfs->lfs_segdhd 1024 #define i_lfs_odnlink inode_ext.lfs->lfs_odnlink 1025 1026 /* 1027 * Macros for determining free space on the disk, with the variable metadata 1028 * of segment summaries and inode blocks taken into account. 1029 */ 1030 /* Estimate number of clean blocks not available for writing */ 1031 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 1032 (int64_t)(F)->lfs_nclean) / \ 1033 ((F)->lfs_nseg - (F)->lfs_nclean))) 1034 1035 /* Estimate total size of the disk not including metadata */ 1036 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 1037 1038 /* Estimate number of blocks actually available for writing */ 1039 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) ? \ 1040 (F)->lfs_bfree - LFS_EST_CMETA(F) : 0) 1041 1042 /* Amount of non-meta space not available to mortal man */ 1043 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1044 (u_int64_t)(F)->lfs_minfree) / \ 1045 100) 1046 1047 /* Can credential C write BB blocks */ 1048 #define ISSPACE(F, BB, C) \ 1049 ((((C) == NOCRED || kauth_cred_geteuid(C) == 0) && \ 1050 LFS_EST_BFREE(F) >= (BB)) || \ 1051 (kauth_cred_geteuid(C) != 0 && IS_FREESPACE(F, BB))) 1052 1053 /* Can an ordinary user write BB blocks */ 1054 #define IS_FREESPACE(F, BB) \ 1055 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1056 1057 /* 1058 * The minimum number of blocks to create a new inode. This is: 1059 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1060 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1061 */ 1062 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1063 1064 /* Statistics Counters */ 1065 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1066 u_int segsused; 1067 u_int psegwrites; 1068 u_int psyncwrites; 1069 u_int pcleanwrites; 1070 u_int blocktot; 1071 u_int cleanblocks; 1072 u_int ncheckpoints; 1073 u_int nwrites; 1074 u_int nsync_writes; 1075 u_int wait_exceeded; 1076 u_int write_exceeded; 1077 u_int flush_invoked; 1078 u_int vflush_invoked; 1079 u_int clean_inlocked; 1080 u_int clean_vnlocked; 1081 u_int segs_reclaimed; 1082 }; 1083 #ifdef _KERNEL 1084 extern struct lfs_stats lfs_stats; 1085 #endif 1086 1087 /* Fcntls to take the place of the lfs syscalls */ 1088 struct lfs_fcntl_markv { 1089 BLOCK_INFO *blkiov; /* blocks to relocate */ 1090 int blkcnt; /* number of blocks */ 1091 }; 1092 1093 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1094 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1095 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1096 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1097 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1098 1099 struct lfs_fhandle { 1100 char space[28]; /* FHANDLE_SIZE_COMPAT (but used from userland too) */ 1101 }; 1102 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1103 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1104 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1105 #define LFCNWRAPSTOP _FCNR_FSPRIV('L', 9, int) 1106 #define LFCNWRAPGO _FCNR_FSPRIV('L', 10, int) 1107 #define LFCNIFILEFH _FCNW_FSPRIV('L', 11, struct lfs_fhandle) 1108 #define LFCNWRAPPASS _FCNR_FSPRIV('L', 12, int) 1109 # define LFS_WRAP_GOING 0x0 1110 # define LFS_WRAP_WAITING 0x1 1111 #define LFCNWRAPSTATUS _FCNW_FSPRIV('L', 13, int) 1112 /* Compat */ 1113 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1114 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1115 #define LFCNIFILEFH_COMPAT _FCNW_FSPRIV('L', 5, struct lfs_fhandle) 1116 #define LFCNIFILEFH_COMPAT2 _FCN_FSPRIV(F_FSOUT, 'L', 11, 32) 1117 #define LFCNWRAPSTOP_COMPAT _FCNO_FSPRIV('L', 9) 1118 #define LFCNWRAPGO_COMPAT _FCNO_FSPRIV('L', 10) 1119 1120 #ifdef _KERNEL 1121 /* XXX MP */ 1122 #define LFS_SEGLOCK_HELD(fs) \ 1123 ((fs)->lfs_seglock != 0 && \ 1124 (fs)->lfs_lockpid == curproc->p_pid && \ 1125 (fs)->lfs_locklwp == curlwp->l_lid) 1126 #endif /* _KERNEL */ 1127 1128 /* Debug segment lock */ 1129 #ifdef notyet 1130 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1131 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1132 # define ASSERT_DUNNO_SEGLOCK(fs) 1133 # define ASSERT_MAYBE_SEGLOCK(fs) 1134 #else /* !notyet */ 1135 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1136 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1137 LFS_SEGLOCK_HELD(fs))) 1138 # define ASSERT_SEGLOCK(fs) do { \ 1139 if (!LFS_SEGLOCK_HELD(fs)) { \ 1140 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1141 } \ 1142 } while(0) 1143 # define ASSERT_NO_SEGLOCK(fs) do { \ 1144 if (LFS_SEGLOCK_HELD(fs)) { \ 1145 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1146 } \ 1147 } while(0) 1148 # define ASSERT_MAYBE_SEGLOCK(x) 1149 #endif /* !notyet */ 1150 1151 __BEGIN_DECLS 1152 void lfs_itimes(struct inode *, const struct timespec *, 1153 const struct timespec *, const struct timespec *); 1154 __END_DECLS 1155 1156 #endif /* !_UFS_LFS_LFS_H_ */ 1157
Indexes created Tue Sep 30 17:09:57 GMT 2025