lfs.h revision 1.125
1 /* $NetBSD: lfs.h,v 1.125 2008/02/15 13:30:56 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 if (((bp)->b_flags & B_LOCKED) == 0 && bp->b_iodone == NULL) { \ 198 mutex_enter(&lfs_lock); \ 199 ++locked_queue_count; \ 200 locked_queue_bytes += bp->b_bufsize; \ 201 mutex_exit(&lfs_lock); \ 202 } \ 203 (bp)->b_flags |= B_LOCKED; \ 204 } while (0) 205 206 # define LFS_UNLOCK_BUF(bp) do { \ 207 if (((bp)->b_flags & B_LOCKED) != 0 && bp->b_iodone == NULL) { \ 208 mutex_enter(&lfs_lock); \ 209 --locked_queue_count; \ 210 locked_queue_bytes -= bp->b_bufsize; \ 211 if (locked_queue_count < LFS_WAIT_BUFS && \ 212 locked_queue_bytes < LFS_WAIT_BYTES) \ 213 wakeup(&locked_queue_count); \ 214 mutex_exit(&lfs_lock); \ 215 } \ 216 (bp)->b_flags &= ~B_LOCKED; \ 217 } while (0) 218 219 #ifdef _KERNEL 220 221 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */ 222 223 # define LFS_IS_MALLOC_BUF(bp) ((bp)->b_iodone == lfs_callback) 224 225 # ifdef DEBUG 226 # define LFS_DEBUG_COUNTLOCKED(m) do { \ 227 if (lfs_debug_log_subsys[DLOG_LLIST]) { \ 228 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \ 229 wakeup(&locked_queue_count); \ 230 } \ 231 } while (0) 232 # else 233 # define LFS_DEBUG_COUNTLOCKED(m) 234 # endif 235 236 /* log for debugging writes to the Ifile */ 237 # ifdef DEBUG 238 struct lfs_log_entry { 239 const char *op; 240 const char *file; 241 int pid; 242 int line; 243 daddr_t block; 244 unsigned long flags; 245 }; 246 extern int lfs_lognum; 247 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH]; 248 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__) 249 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\ 250 int _s; \ 251 \ 252 mutex_enter(&lfs_lock); \ 253 _s = splbio(); \ 254 lfs_log[lfs_lognum].op = theop; \ 255 lfs_log[lfs_lognum].file = thefile; \ 256 lfs_log[lfs_lognum].line = (theline); \ 257 lfs_log[lfs_lognum].pid = (thepid); \ 258 lfs_log[lfs_lognum].block = (lbn); \ 259 lfs_log[lfs_lognum].flags = (theflags); \ 260 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \ 261 splx(_s); \ 262 mutex_exit(&lfs_lock); \ 263 } while (0) 264 265 # define LFS_BCLEAN_LOG(fs, bp) do { \ 266 if ((bp)->b_vp == (fs)->lfs_ivnode) \ 267 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \ 268 bp->b_lblkno, bp->b_flags, curproc->p_pid);\ 269 } while (0) 270 271 /* Must match list in lfs_vfsops.c ! */ 272 # define DLOG_RF 0 /* roll forward */ 273 # define DLOG_ALLOC 1 /* inode alloc */ 274 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */ 275 # define DLOG_FLUSH 3 /* flush */ 276 # define DLOG_LLIST 4 /* locked list accounting */ 277 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */ 278 # define DLOG_VNODE 6 /* vflush/writevnodes */ 279 # define DLOG_SEG 7 /* segwrite */ 280 # define DLOG_SU 8 /* seguse accounting */ 281 # define DLOG_CLEAN 9 /* cleaner routines */ 282 # define DLOG_MOUNT 10 /* mount/unmount */ 283 # define DLOG_PAGE 11 /* putpages/gop_write */ 284 # define DLOG_DIROP 12 /* dirop accounting */ 285 # define DLOG_MALLOC 13 /* lfs_malloc accounting */ 286 # define DLOG_MAX 14 /* The terminator */ 287 # define DLOG(a) lfs_debug_log a 288 # else /* ! DEBUG */ 289 # define LFS_BCLEAN_LOG(fs, bp) 290 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 291 # define DLOG(a) 292 # endif /* ! DEBUG */ 293 #else /* ! _KERNEL */ 294 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)) 295 #endif /* _KERNEL */ 296 297 #ifdef _KERNEL 298 /* Filehandle structure for exported LFSes */ 299 struct lfid { 300 struct ufid lfid_ufid; 301 #define lfid_len lfid_ufid.ufid_len 302 #define lfid_ino lfid_ufid.ufid_ino 303 #define lfid_gen lfid_ufid.ufid_gen 304 uint32_t lfid_ident; 305 }; 306 #endif /* _KERNEL */ 307 308 /* 309 * "struct inode" associated definitions 310 */ 311 312 /* Address calculations for metadata located in the inode */ 313 #define S_INDIR(fs) -NDADDR 314 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1) 315 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1) 316 317 /* For convenience */ 318 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING) 319 320 #define LFS_SET_UINO(ip, flags) do { \ 321 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \ 322 ++(ip)->i_lfs->lfs_uinodes; \ 323 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \ 324 ++(ip)->i_lfs->lfs_uinodes; \ 325 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \ 326 ++(ip)->i_lfs->lfs_uinodes; \ 327 (ip)->i_flag |= (flags); \ 328 } while (0) 329 330 #define LFS_CLR_UINO(ip, flags) do { \ 331 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \ 332 --(ip)->i_lfs->lfs_uinodes; \ 333 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \ 334 --(ip)->i_lfs->lfs_uinodes; \ 335 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \ 336 --(ip)->i_lfs->lfs_uinodes; \ 337 (ip)->i_flag &= ~(flags); \ 338 if ((ip)->i_lfs->lfs_uinodes < 0) { \ 339 panic("lfs_uinodes < 0"); \ 340 } \ 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_iflag & VI_ONWORKLST) &&\ 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 rw_enter(&(F)->lfs_iflock, RW_READER); \ 398 } while(0) 399 # define UNSHARE_IFLOCK(F) \ 400 do { \ 401 rw_exit(&(F)->lfs_iflock); \ 402 } while(0) 403 #else /* ! _KERNEL */ 404 # define SHARE_IFLOCK(F) 405 # define UNSHARE_IFLOCK(F) 406 #endif /* ! _KERNEL */ 407 408 /* Read in the block with a specific segment usage entry from the ifile. */ 409 #define LFS_SEGENTRY(SP, F, IN, BP) do { \ 410 int _e; \ 411 SHARE_IFLOCK(F); \ 412 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 413 if ((_e = bread((F)->lfs_ivnode, \ 414 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \ 415 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 416 panic("lfs: ifile read: %d", _e); \ 417 if ((F)->lfs_version == 1) \ 418 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \ 419 ((IN) & ((F)->lfs_sepb - 1))); \ 420 else \ 421 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \ 422 UNSHARE_IFLOCK(F); \ 423 } while (0) 424 425 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \ 426 if ((SP)->su_nbytes == 0) \ 427 (SP)->su_flags |= SEGUSE_EMPTY; \ 428 else \ 429 (SP)->su_flags &= ~SEGUSE_EMPTY; \ 430 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \ 431 LFS_BWRITE_LOG(BP); \ 432 } while (0) 433 434 /* 435 * On-disk file information. One per file with data blocks in the segment. 436 */ 437 typedef struct finfo FINFO; 438 struct finfo { 439 u_int32_t fi_nblocks; /* number of blocks */ 440 u_int32_t fi_version; /* version number */ 441 u_int32_t fi_ino; /* inode number */ 442 u_int32_t fi_lastlength; /* length of last block in array */ 443 int32_t fi_blocks[1]; /* array of logical block numbers */ 444 }; 445 /* sizeof FINFO except fi_blocks */ 446 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t)) 447 448 /* 449 * Index file inode entries. 450 */ 451 typedef struct ifile IFILE; 452 struct ifile { 453 u_int32_t if_version; /* inode version number */ 454 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */ 455 int32_t if_daddr; /* inode disk address */ 456 #define LFS_ORPHAN_NEXTFREE (~(u_int32_t)0) /* indicate orphaned file */ 457 u_int32_t if_nextfree; /* next-unallocated inode */ 458 u_int32_t if_atime_sec; /* Last access time, seconds */ 459 u_int32_t if_atime_nsec; /* and nanoseconds */ 460 }; 461 462 typedef struct ifile_v1 IFILE_V1; 463 struct ifile_v1 { 464 u_int32_t if_version; /* inode version number */ 465 int32_t if_daddr; /* inode disk address */ 466 u_int32_t if_nextfree; /* next-unallocated inode */ 467 #if LFS_ATIME_IFILE 468 struct timespec if_atime; /* Last access time */ 469 #endif 470 }; 471 472 /* 473 * LFSv1 compatibility code is not allowed to touch if_atime, since it 474 * may not be mapped! 475 */ 476 /* Read in the block with a specific inode from the ifile. */ 477 #define LFS_IENTRY(IP, F, IN, BP) do { \ 478 int _e; \ 479 SHARE_IFLOCK(F); \ 480 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 481 if ((_e = bread((F)->lfs_ivnode, \ 482 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \ 483 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \ 484 panic("lfs: ifile ino %d read %d", (int)(IN), _e); \ 485 if ((F)->lfs_version == 1) \ 486 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \ 487 (IN) % (F)->lfs_ifpb); \ 488 else \ 489 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \ 490 UNSHARE_IFLOCK(F); \ 491 } while (0) 492 493 /* 494 * Cleaner information structure. This resides in the ifile and is used 495 * to pass information from the kernel to the cleaner. 496 */ 497 typedef struct _cleanerinfo { 498 u_int32_t clean; /* number of clean segments */ 499 u_int32_t dirty; /* number of dirty segments */ 500 int32_t bfree; /* disk blocks free */ 501 int32_t avail; /* disk blocks available */ 502 u_int32_t free_head; /* head of the inode free list */ 503 u_int32_t free_tail; /* tail of the inode free list */ 504 #define LFS_CLEANER_MUST_CLEAN 0x01 505 u_int32_t flags; /* status word from the kernel */ 506 } CLEANERINFO; 507 508 #define CLEANSIZE_SU(fs) \ 509 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift) 510 511 /* Read in the block with the cleaner info from the ifile. */ 512 #define LFS_CLEANERINFO(CP, F, BP) do { \ 513 SHARE_IFLOCK(F); \ 514 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \ 515 if (bread((F)->lfs_ivnode, \ 516 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \ 517 panic("lfs: ifile read"); \ 518 (CP) = (CLEANERINFO *)(BP)->b_data; \ 519 UNSHARE_IFLOCK(F); \ 520 } while (0) 521 522 /* 523 * Synchronize the Ifile cleaner info with current avail and bfree. 524 */ 525 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \ 526 mutex_enter(&lfs_lock); \ 527 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \ 528 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \ 529 (fs)->lfs_favail) { \ 530 (cip)->bfree = (fs)->lfs_bfree; \ 531 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \ 532 (fs)->lfs_favail; \ 533 if (((bp)->b_flags & B_GATHERED) == 0) { \ 534 (fs)->lfs_flags |= LFS_IFDIRTY; \ 535 } \ 536 mutex_exit(&lfs_lock); \ 537 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \ 538 } else { \ 539 mutex_exit(&lfs_lock); \ 540 brelse(bp, 0); \ 541 } \ 542 } while (0) 543 544 /* 545 * Get the head of the inode free list. 546 * Always caled with the segment lock held. 547 */ 548 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \ 549 if ((FS)->lfs_version > 1) { \ 550 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 551 (FS)->lfs_freehd = (CIP)->free_head; \ 552 brelse(BP, 0); \ 553 } \ 554 *(FREEP) = (FS)->lfs_freehd; \ 555 } while (0) 556 557 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \ 558 (FS)->lfs_freehd = (VAL); \ 559 if ((FS)->lfs_version > 1) { \ 560 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 561 (CIP)->free_head = (VAL); \ 562 LFS_BWRITE_LOG(BP); \ 563 mutex_enter(&lfs_lock); \ 564 (FS)->lfs_flags |= LFS_IFDIRTY; \ 565 mutex_exit(&lfs_lock); \ 566 } \ 567 } while (0) 568 569 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \ 570 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 571 *(FREEP) = (CIP)->free_tail; \ 572 brelse(BP, 0); \ 573 } while (0) 574 575 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \ 576 LFS_CLEANERINFO((CIP), (FS), (BP)); \ 577 (CIP)->free_tail = (VAL); \ 578 LFS_BWRITE_LOG(BP); \ 579 mutex_enter(&lfs_lock); \ 580 (FS)->lfs_flags |= LFS_IFDIRTY; \ 581 mutex_exit(&lfs_lock); \ 582 } while (0) 583 584 /* 585 * On-disk segment summary information 586 */ 587 typedef struct segsum_v1 SEGSUM_V1; 588 struct segsum_v1 { 589 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 590 u_int32_t ss_datasum; /* 4: check sum of data */ 591 u_int32_t ss_magic; /* 8: segment summary magic number */ 592 #define SS_MAGIC 0x061561 593 int32_t ss_next; /* 12: next segment */ 594 u_int32_t ss_create; /* 16: creation time stamp */ 595 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 596 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 597 598 #define SS_DIROP 0x01 /* segment begins a dirop */ 599 #define SS_CONT 0x02 /* more partials to finish this write*/ 600 #define SS_CLEAN 0x04 /* written by the cleaner */ 601 #define SS_RFW 0x08 /* written by the roll-forward agent */ 602 u_int16_t ss_flags; /* 24: used for directory operations */ 603 u_int16_t ss_pad; /* 26: extra space */ 604 /* FINFO's and inode daddr's... */ 605 }; 606 607 typedef struct segsum SEGSUM; 608 struct segsum { 609 u_int32_t ss_sumsum; /* 0: check sum of summary block */ 610 u_int32_t ss_datasum; /* 4: check sum of data */ 611 u_int32_t ss_magic; /* 8: segment summary magic number */ 612 int32_t ss_next; /* 12: next segment */ 613 u_int32_t ss_ident; /* 16: roll-forward fsid */ 614 #define ss_ocreate ss_ident /* ident is where create was in v1 */ 615 u_int16_t ss_nfinfo; /* 20: number of file info structures */ 616 u_int16_t ss_ninos; /* 22: number of inodes in summary */ 617 u_int16_t ss_flags; /* 24: used for directory operations */ 618 u_int8_t ss_pad[6]; /* 26: extra space */ 619 u_int64_t ss_serial; /* 32: serial number */ 620 u_int64_t ss_create; /* 40: time stamp */ 621 /* FINFO's and inode daddr's... */ 622 }; 623 624 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM)) 625 626 627 /* 628 * On-disk super block. 629 */ 630 struct dlfs { 631 #define LFS_MAGIC 0x070162 632 u_int32_t dlfs_magic; /* 0: magic number */ 633 #define LFS_VERSION 2 634 u_int32_t dlfs_version; /* 4: version number */ 635 636 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */ 637 /* number of frags in fs (v2) */ 638 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */ 639 /* number of bytes per segment (v2) */ 640 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */ 641 u_int32_t dlfs_bsize; /* 20: file system block size */ 642 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */ 643 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */ 644 645 /* Checkpoint region. */ 646 u_int32_t dlfs_freehd; /* 32: start of the free list */ 647 int32_t dlfs_bfree; /* 36: number of free disk blocks */ 648 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */ 649 int32_t dlfs_avail; /* 44: blocks available for writing */ 650 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */ 651 int32_t dlfs_idaddr; /* 52: inode file disk address */ 652 u_int32_t dlfs_ifile; /* 56: inode file inode number */ 653 int32_t dlfs_lastseg; /* 60: address of last segment written */ 654 int32_t dlfs_nextseg; /* 64: address of next segment to write */ 655 int32_t dlfs_curseg; /* 68: current segment being written */ 656 int32_t dlfs_offset; /* 72: offset in curseg for next partial */ 657 int32_t dlfs_lastpseg; /* 76: address of last partial written */ 658 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */ 659 #define dlfs_otstamp dlfs_inopf 660 661 /* These are configuration parameters. */ 662 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */ 663 664 /* These fields can be computed from the others. */ 665 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */ 666 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */ 667 u_int32_t dlfs_inopb; /* 100: inodes per block */ 668 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */ 669 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */ 670 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */ 671 u_int32_t dlfs_nseg; /* 116: number of segments */ 672 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */ 673 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */ 674 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */ 675 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */ 676 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */ 677 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */ 678 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */ 679 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */ 680 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */ 681 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */ 682 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */ 683 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */ 684 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */ 685 686 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */ 687 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */ 688 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */ 689 int32_t dlfs_sboffs[LFS_MAXNUMSB]; 690 691 u_int32_t dlfs_nclean; /* 228: Number of clean segments */ 692 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */ 693 #define LFS_PF_CLEAN 0x1 694 u_int16_t dlfs_pflags; /* 322: file system persistent flags */ 695 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */ 696 u_int32_t dlfs_minfreeseg; /* 328: segments not counted in bfree */ 697 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */ 698 u_int64_t dlfs_serial; /* 336: serial number */ 699 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */ 700 int32_t dlfs_start; /* 348: start of segment 0 */ 701 u_int64_t dlfs_tstamp; /* 352: time stamp */ 702 #define LFS_44INODEFMT 0 703 #define LFS_MAXINODEFMT 0 704 u_int32_t dlfs_inodefmt; /* 360: inode format version */ 705 u_int32_t dlfs_interleave; /* 364: segment interleave */ 706 u_int32_t dlfs_ident; /* 368: per-fs identifier */ 707 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */ 708 u_int32_t dlfs_resvseg; /* 376: segments reserved for the cleaner */ 709 int8_t dlfs_pad[128]; /* 380: round to 512 bytes */ 710 /* Checksum -- last valid disk field. */ 711 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */ 712 }; 713 714 /* Type used for the inode bitmap */ 715 typedef u_int32_t lfs_bm_t; 716 717 /* 718 * Linked list of segments whose byte count needs updating following a 719 * file truncation. 720 */ 721 struct segdelta { 722 long segnum; 723 size_t num; 724 LIST_ENTRY(segdelta) list; 725 }; 726 727 /* 728 * In-memory super block. 729 */ 730 struct lfs { 731 struct dlfs lfs_dlfs; /* on-disk parameters */ 732 #define lfs_magic lfs_dlfs.dlfs_magic 733 #define lfs_version lfs_dlfs.dlfs_version 734 #define lfs_size lfs_dlfs.dlfs_size 735 #define lfs_ssize lfs_dlfs.dlfs_ssize 736 #define lfs_dsize lfs_dlfs.dlfs_dsize 737 #define lfs_bsize lfs_dlfs.dlfs_bsize 738 #define lfs_fsize lfs_dlfs.dlfs_fsize 739 #define lfs_frag lfs_dlfs.dlfs_frag 740 #define lfs_freehd lfs_dlfs.dlfs_freehd 741 #define lfs_bfree lfs_dlfs.dlfs_bfree 742 #define lfs_nfiles lfs_dlfs.dlfs_nfiles 743 #define lfs_avail lfs_dlfs.dlfs_avail 744 #define lfs_uinodes lfs_dlfs.dlfs_uinodes 745 #define lfs_idaddr lfs_dlfs.dlfs_idaddr 746 #define lfs_ifile lfs_dlfs.dlfs_ifile 747 #define lfs_lastseg lfs_dlfs.dlfs_lastseg 748 #define lfs_nextseg lfs_dlfs.dlfs_nextseg 749 #define lfs_curseg lfs_dlfs.dlfs_curseg 750 #define lfs_offset lfs_dlfs.dlfs_offset 751 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg 752 #define lfs_otstamp lfs_dlfs.dlfs_inopf 753 #define lfs_inopf lfs_dlfs.dlfs_inopf 754 #define lfs_minfree lfs_dlfs.dlfs_minfree 755 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize 756 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg 757 #define lfs_inopb lfs_dlfs.dlfs_inopb 758 #define lfs_ifpb lfs_dlfs.dlfs_ifpb 759 #define lfs_sepb lfs_dlfs.dlfs_sepb 760 #define lfs_nindir lfs_dlfs.dlfs_nindir 761 #define lfs_nseg lfs_dlfs.dlfs_nseg 762 #define lfs_nspf lfs_dlfs.dlfs_nspf 763 #define lfs_cleansz lfs_dlfs.dlfs_cleansz 764 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz 765 #define lfs_segmask lfs_dlfs.dlfs_segmask 766 #define lfs_segshift lfs_dlfs.dlfs_segshift 767 #define lfs_bmask lfs_dlfs.dlfs_bmask 768 #define lfs_bshift lfs_dlfs.dlfs_bshift 769 #define lfs_ffmask lfs_dlfs.dlfs_ffmask 770 #define lfs_ffshift lfs_dlfs.dlfs_ffshift 771 #define lfs_fbmask lfs_dlfs.dlfs_fbmask 772 #define lfs_fbshift lfs_dlfs.dlfs_fbshift 773 #define lfs_blktodb lfs_dlfs.dlfs_blktodb 774 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb 775 #define lfs_sushift lfs_dlfs.dlfs_sushift 776 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen 777 #define lfs_sboffs lfs_dlfs.dlfs_sboffs 778 #define lfs_cksum lfs_dlfs.dlfs_cksum 779 #define lfs_pflags lfs_dlfs.dlfs_pflags 780 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt 781 #define lfs_nclean lfs_dlfs.dlfs_nclean 782 #define lfs_dmeta lfs_dlfs.dlfs_dmeta 783 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg 784 #define lfs_sumsize lfs_dlfs.dlfs_sumsize 785 #define lfs_serial lfs_dlfs.dlfs_serial 786 #define lfs_ibsize lfs_dlfs.dlfs_ibsize 787 #define lfs_start lfs_dlfs.dlfs_start 788 #define lfs_tstamp lfs_dlfs.dlfs_tstamp 789 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt 790 #define lfs_interleave lfs_dlfs.dlfs_interleave 791 #define lfs_ident lfs_dlfs.dlfs_ident 792 #define lfs_resvseg lfs_dlfs.dlfs_resvseg 793 794 /* These fields are set at mount time and are meaningless on disk. */ 795 struct segment *lfs_sp; /* current segment being written */ 796 struct vnode *lfs_ivnode; /* vnode for the ifile */ 797 u_int32_t lfs_seglock; /* single-thread the segment writer */ 798 pid_t lfs_lockpid; /* pid of lock holder */ 799 lwpid_t lfs_locklwp; /* lwp of lock holder */ 800 u_int32_t lfs_iocount; /* number of ios pending */ 801 u_int32_t lfs_writer; /* don't allow any dirops to start */ 802 u_int32_t lfs_dirops; /* count of active directory ops */ 803 u_int32_t lfs_dirvcount; /* count of VDIROP nodes in this fs */ 804 u_int32_t lfs_doifile; /* Write ifile blocks on next write */ 805 u_int32_t lfs_nactive; /* Number of segments since last ckp */ 806 int8_t lfs_fmod; /* super block modified flag */ 807 int8_t lfs_ronly; /* mounted read-only flag */ 808 #define LFS_NOTYET 0x01 809 #define LFS_IFDIRTY 0x02 810 #define LFS_WARNED 0x04 811 #define LFS_UNDIROP 0x08 812 int8_t lfs_flags; /* currently unused flag */ 813 u_int16_t lfs_activesb; /* toggle between superblocks */ 814 daddr_t lfs_sbactive; /* disk address of current sb write */ 815 struct vnode *lfs_flushvp; /* vnode being flushed */ 816 int lfs_flushvp_fakevref; /* fake vref count for flushvp */ 817 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */ 818 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */ 819 size_t lfs_devbsize; /* Device block size */ 820 size_t lfs_devbshift; /* Device block shift */ 821 krwlock_t lfs_fraglock; 822 krwlock_t lfs_iflock; /* Ifile lock */ 823 kcondvar_t lfs_stopcv; /* Wrap lock */ 824 struct lwp *lfs_stoplwp; 825 pid_t lfs_rfpid; /* Process ID of roll-forward agent */ 826 int lfs_nadirop; /* number of active dirop nodes */ 827 long lfs_ravail; /* blocks pre-reserved for writing */ 828 long lfs_favail; /* blocks pre-reserved for writing */ 829 res_t *lfs_resblk; /* Reserved memory for pageout */ 830 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */ 831 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */ 832 #define LFS_RESHASH_WIDTH 17 833 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH]; 834 int lfs_pdflush; /* pagedaemon wants us to flush */ 835 u_int32_t **lfs_suflags; /* Segment use flags */ 836 #ifdef _KERNEL 837 struct pool lfs_clpool; /* Pool for struct lfs_cluster */ 838 struct pool lfs_bpppool; /* Pool for bpp */ 839 struct pool lfs_segpool; /* Pool for struct segment */ 840 #endif /* _KERNEL */ 841 #define LFS_MAX_CLEANIND 64 842 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */ 843 int lfs_cleanind; /* Index into intervals */ 844 int lfs_sleepers; /* # procs sleeping this fs */ 845 int lfs_pages; /* dirty pages blaming this fs */ 846 lfs_bm_t *lfs_ino_bitmap; /* Inuse inodes bitmap */ 847 int lfs_nowrap; /* Suspend log wrap */ 848 int lfs_wrappass; /* Allow first log wrap requester to pass */ 849 int lfs_wrapstatus; /* Wrap status */ 850 LIST_HEAD(, segdelta) lfs_segdhd; /* List of pending trunc accounting events */ 851 }; 852 853 /* NINDIR is the number of indirects in a file system block. */ 854 #define NINDIR(fs) ((fs)->lfs_nindir) 855 856 /* INOPB is the number of inodes in a secondary storage block. */ 857 #define INOPB(fs) ((fs)->lfs_inopb) 858 /* INOPF is the number of inodes in a fragment. */ 859 #define INOPF(fs) ((fs)->lfs_inopf) 860 861 #define blksize(fs, ip, lbn) \ 862 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \ 863 ? (fs)->lfs_bsize \ 864 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size)))) 865 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask)) 866 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \ 867 ((int)((loc) & (fs)->lfs_ffmask)) 868 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb) 869 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb) 870 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 871 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift)) 872 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift) 873 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift) 874 /* Same as above, but named like dbtob(), btodb() */ 875 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \ 876 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 877 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \ 878 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb)) 879 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 880 (fs)->lfs_fsbtodb)) 881 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \ 882 (fs)->lfs_fsbtodb)) 883 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift) 884 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \ 885 ((loc) >> (fs)->lfs_ffshift) 886 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \ 887 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask))) 888 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \ 889 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask))) 890 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \ 891 ((frags) >> (fs)->lfs_fbshift) 892 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \ 893 ((blks) << (fs)->lfs_fbshift) 894 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \ 895 ((fsb) & ((fs)->lfs_frag - 1)) 896 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \ 897 ((fsb) &~ ((fs)->lfs_frag - 1)) 898 #define dblksize(fs, dp, lbn) \ 899 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\ 900 ? (fs)->lfs_bsize \ 901 : (fragroundup(fs, blkoff(fs, (dp)->di_size)))) 902 903 #define segsize(fs) ((fs)->lfs_version == 1 ? \ 904 lblktosize((fs), (fs)->lfs_ssize) : \ 905 (fs)->lfs_ssize) 906 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \ 907 (fs)->lfs_ssize << (fs)->lfs_blktodb : \ 908 btofsb((fs), (fs)->lfs_ssize)) * (seg)) 909 #define dtosn(fs, daddr) /* block address to segment number */ \ 910 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1))) 911 #define sntod(fs, sn) /* segment number to disk address */ \ 912 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start)) 913 914 /* 915 * Structures used by lfs_bmapv and lfs_markv to communicate information 916 * about inodes and data blocks. 917 */ 918 typedef struct block_info { 919 u_int32_t bi_inode; /* inode # */ 920 int32_t bi_lbn; /* logical block w/in file */ 921 int32_t bi_daddr; /* disk address of block */ 922 u_int64_t bi_segcreate; /* origin segment create time */ 923 int bi_version; /* file version number */ 924 void *bi_bp; /* data buffer */ 925 int bi_size; /* size of the block (if fragment) */ 926 } BLOCK_INFO; 927 928 /* Compatibility for 1.5 binaries */ 929 typedef struct block_info_15 { 930 u_int32_t bi_inode; /* inode # */ 931 int32_t bi_lbn; /* logical block w/in file */ 932 int32_t bi_daddr; /* disk address of block */ 933 u_int32_t bi_segcreate; /* origin segment create time */ 934 int bi_version; /* file version number */ 935 void *bi_bp; /* data buffer */ 936 int bi_size; /* size of the block (if fragment) */ 937 } BLOCK_INFO_15; 938 939 /* In-memory description of a segment about to be written. */ 940 struct segment { 941 struct lfs *fs; /* file system pointer */ 942 struct buf **bpp; /* pointer to buffer array */ 943 struct buf **cbpp; /* pointer to next available bp */ 944 struct buf **start_bpp; /* pointer to first bp in this set */ 945 struct buf *ibp; /* buffer pointer to inode page */ 946 struct ufs1_dinode *idp; /* pointer to ifile dinode */ 947 struct finfo *fip; /* current fileinfo pointer */ 948 struct vnode *vp; /* vnode being gathered */ 949 void *segsum; /* segment summary info */ 950 u_int32_t ninodes; /* number of inodes in this segment */ 951 int32_t seg_bytes_left; /* bytes left in segment */ 952 int32_t sum_bytes_left; /* bytes left in summary block */ 953 u_int32_t seg_number; /* number of this segment */ 954 int32_t *start_lbp; /* beginning lbn for this set */ 955 956 #define SEGM_CKP 0x01 /* doing a checkpoint */ 957 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */ 958 #define SEGM_SYNC 0x04 /* wait for segment */ 959 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */ 960 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */ 961 #define SEGM_WRITERD 0x20 /* LFS writed called us */ 962 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */ 963 u_int16_t seg_flags; /* run-time flags for this segment */ 964 u_int32_t seg_iocount; /* number of ios pending */ 965 int ndupino; /* number of duplicate inodes */ 966 }; 967 968 #ifdef _KERNEL 969 struct lfs_cluster { 970 size_t bufsize; /* Size of kept data */ 971 struct buf **bpp; /* Array of kept buffers */ 972 int bufcount; /* Number of kept buffers */ 973 #define LFS_CL_MALLOC 0x00000001 974 #define LFS_CL_SHIFT 0x00000002 975 #define LFS_CL_SYNC 0x00000004 976 u_int32_t flags; /* Flags */ 977 struct lfs *fs; /* LFS that this belongs to */ 978 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */ 979 }; 980 981 /* 982 * Splay tree containing block numbers allocated through lfs_balloc. 983 */ 984 struct lbnentry { 985 SPLAY_ENTRY(lbnentry) entry; 986 daddr_t lbn; 987 }; 988 #endif /* _KERNEL */ 989 990 /* 991 * LFS inode extensions. 992 */ 993 struct lfs_inode_ext { 994 off_t lfs_osize; /* size of file on disk */ 995 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */ 996 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */ 997 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */ 998 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */ 999 #define LFSI_NO_GOP_WRITE 0x01 1000 #define LFSI_DELETED 0x02 1001 #define LFSI_WRAPBLOCK 0x04 1002 #define LFSI_WRAPWAIT 0x08 1003 u_int32_t lfs_iflags; /* Inode flags */ 1004 daddr_t lfs_hiblk; /* Highest lbn held by inode */ 1005 #ifdef _KERNEL 1006 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */ 1007 int lfs_nbtree; /* Size of tree */ 1008 LIST_HEAD(, segdelta) lfs_segdhd; 1009 #endif 1010 int16_t lfs_odnlink; /* on-disk nlink count for cleaner */ 1011 }; 1012 #define i_lfs_osize inode_ext.lfs->lfs_osize 1013 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks 1014 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize 1015 #define i_lfs_dchain inode_ext.lfs->lfs_dchain 1016 #define i_lfs_pchain inode_ext.lfs->lfs_pchain 1017 #define i_lfs_iflags inode_ext.lfs->lfs_iflags 1018 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk 1019 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree 1020 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree 1021 #define i_lfs_segdhd inode_ext.lfs->lfs_segdhd 1022 #define i_lfs_odnlink inode_ext.lfs->lfs_odnlink 1023 1024 /* 1025 * Macros for determining free space on the disk, with the variable metadata 1026 * of segment summaries and inode blocks taken into account. 1027 */ 1028 /* Estimate number of clean blocks not available for writing */ 1029 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \ 1030 (int64_t)(F)->lfs_nclean) / \ 1031 ((F)->lfs_nseg - (F)->lfs_nclean))) 1032 1033 /* Estimate total size of the disk not including metadata */ 1034 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F)) 1035 1036 /* Estimate number of blocks actually available for writing */ 1037 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) ? \ 1038 (F)->lfs_bfree - LFS_EST_CMETA(F) : 0) 1039 1040 /* Amount of non-meta space not available to mortal man */ 1041 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \ 1042 (u_int64_t)(F)->lfs_minfree) / \ 1043 100) 1044 1045 /* Can credential C write BB blocks */ 1046 #define ISSPACE(F, BB, C) \ 1047 ((((C) == NOCRED || kauth_cred_geteuid(C) == 0) && \ 1048 LFS_EST_BFREE(F) >= (BB)) || \ 1049 (kauth_cred_geteuid(C) != 0 && IS_FREESPACE(F, BB))) 1050 1051 /* Can an ordinary user write BB blocks */ 1052 #define IS_FREESPACE(F, BB) \ 1053 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F)) 1054 1055 /* 1056 * The minimum number of blocks to create a new inode. This is: 1057 * directory direct block (1) + NIADDR indirect blocks + inode block (1) + 1058 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks. 1059 */ 1060 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift)) 1061 1062 /* Statistics Counters */ 1063 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */ 1064 u_int segsused; 1065 u_int psegwrites; 1066 u_int psyncwrites; 1067 u_int pcleanwrites; 1068 u_int blocktot; 1069 u_int cleanblocks; 1070 u_int ncheckpoints; 1071 u_int nwrites; 1072 u_int nsync_writes; 1073 u_int wait_exceeded; 1074 u_int write_exceeded; 1075 u_int flush_invoked; 1076 u_int vflush_invoked; 1077 u_int clean_inlocked; 1078 u_int clean_vnlocked; 1079 u_int segs_reclaimed; 1080 }; 1081 #ifdef _KERNEL 1082 extern struct lfs_stats lfs_stats; 1083 #endif 1084 1085 /* Fcntls to take the place of the lfs syscalls */ 1086 struct lfs_fcntl_markv { 1087 BLOCK_INFO *blkiov; /* blocks to relocate */ 1088 int blkcnt; /* number of blocks */ 1089 }; 1090 1091 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval) 1092 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval) 1093 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv) 1094 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv) 1095 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4) 1096 1097 struct lfs_fhandle { 1098 char space[28]; /* FHANDLE_SIZE_COMPAT (but used from userland too) */ 1099 }; 1100 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int) 1101 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int) 1102 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int) 1103 #define LFCNWRAPSTOP _FCNR_FSPRIV('L', 9, int) 1104 #define LFCNWRAPGO _FCNR_FSPRIV('L', 10, int) 1105 #define LFCNIFILEFH _FCNW_FSPRIV('L', 11, struct lfs_fhandle) 1106 #define LFCNWRAPPASS _FCNR_FSPRIV('L', 12, int) 1107 # define LFS_WRAP_GOING 0x0 1108 # define LFS_WRAP_WAITING 0x1 1109 #define LFCNWRAPSTATUS _FCNW_FSPRIV('L', 13, int) 1110 /* Compat */ 1111 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval) 1112 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval) 1113 #define LFCNIFILEFH_COMPAT _FCNW_FSPRIV('L', 5, struct lfs_fhandle) 1114 #define LFCNIFILEFH_COMPAT2 _FCN_FSPRIV(F_FSOUT, 'L', 11, 32) 1115 #define LFCNWRAPSTOP_COMPAT _FCNO_FSPRIV('L', 9) 1116 #define LFCNWRAPGO_COMPAT _FCNO_FSPRIV('L', 10) 1117 1118 #ifdef _KERNEL 1119 /* XXX MP */ 1120 #define LFS_SEGLOCK_HELD(fs) \ 1121 ((fs)->lfs_seglock != 0 && \ 1122 (fs)->lfs_lockpid == curproc->p_pid && \ 1123 (fs)->lfs_locklwp == curlwp->l_lid) 1124 #endif /* _KERNEL */ 1125 1126 /* Debug segment lock */ 1127 #ifdef notyet 1128 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs)) 1129 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs)) 1130 # define ASSERT_DUNNO_SEGLOCK(fs) 1131 # define ASSERT_MAYBE_SEGLOCK(fs) 1132 #else /* !notyet */ 1133 # define ASSERT_DUNNO_SEGLOCK(fs) \ 1134 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \ 1135 LFS_SEGLOCK_HELD(fs))) 1136 # define ASSERT_SEGLOCK(fs) do { \ 1137 if (!LFS_SEGLOCK_HELD(fs)) { \ 1138 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \ 1139 } \ 1140 } while(0) 1141 # define ASSERT_NO_SEGLOCK(fs) do { \ 1142 if (LFS_SEGLOCK_HELD(fs)) { \ 1143 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \ 1144 } \ 1145 } while(0) 1146 # define ASSERT_MAYBE_SEGLOCK(x) 1147 #endif /* !notyet */ 1148 1149 __BEGIN_DECLS 1150 void lfs_itimes(struct inode *, const struct timespec *, 1151 const struct timespec *, const struct timespec *); 1152 __END_DECLS 1153 1154 #endif /* !_UFS_LFS_LFS_H_ */ 1155
Indexes created Sun Sep 28 16:09:52 GMT 2025