lfs.h revision 1.151 1 /* $NetBSD: lfs.h,v 1.151 2013/06/08 23:04:49 dholland Exp $ */
2
3 /* from NetBSD: dinode.h,v 1.22 2013/01/22 09:39:18 dholland Exp */
4 /* from NetBSD: dir.h,v 1.21 2009/07/22 04:49:19 dholland Exp */
5
6 /*-
7 * Copyright (c) 1999, 2000, 2001, 2002, 2003 The NetBSD Foundation, Inc.
8 * All rights reserved.
9 *
10 * This code is derived from software contributed to The NetBSD Foundation
11 * by Konrad E. Schroder <perseant (at) hhhh.org>.
12 *
13 * Redistribution and use in source and binary forms, with or without
14 * modification, are permitted provided that the following conditions
15 * are met:
16 * 1. Redistributions of source code must retain the above copyright
17 * notice, this list of conditions and the following disclaimer.
18 * 2. Redistributions in binary form must reproduce the above copyright
19 * notice, this list of conditions and the following disclaimer in the
20 * documentation and/or other materials provided with the distribution.
21 *
22 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
23 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
24 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
25 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
26 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
27 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
28 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
29 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
30 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
31 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
32 * POSSIBILITY OF SUCH DAMAGE.
33 */
34 /*-
35 * Copyright (c) 1991, 1993
36 * The Regents of the University of California. All rights reserved.
37 *
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
40 * are met:
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. Neither the name of the University nor the names of its contributors
47 * may be used to endorse or promote products derived from this software
48 * without specific prior written permission.
49 *
50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
53 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60 * SUCH DAMAGE.
61 *
62 * @(#)lfs.h 8.9 (Berkeley) 5/8/95
63 */
64 /*
65 * Copyright (c) 2002 Networks Associates Technology, Inc.
66 * All rights reserved.
67 *
68 * This software was developed for the FreeBSD Project by Marshall
69 * Kirk McKusick and Network Associates Laboratories, the Security
70 * Research Division of Network Associates, Inc. under DARPA/SPAWAR
71 * contract N66001-01-C-8035 ("CBOSS"), as part of the DARPA CHATS
72 * research program
73 *
74 * Copyright (c) 1982, 1989, 1993
75 * The Regents of the University of California. All rights reserved.
76 * (c) UNIX System Laboratories, Inc.
77 * All or some portions of this file are derived from material licensed
78 * to the University of California by American Telephone and Telegraph
79 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
80 * the permission of UNIX System Laboratories, Inc.
81 *
82 * Redistribution and use in source and binary forms, with or without
83 * modification, are permitted provided that the following conditions
84 * are met:
85 * 1. Redistributions of source code must retain the above copyright
86 * notice, this list of conditions and the following disclaimer.
87 * 2. Redistributions in binary form must reproduce the above copyright
88 * notice, this list of conditions and the following disclaimer in the
89 * documentation and/or other materials provided with the distribution.
90 * 3. Neither the name of the University nor the names of its contributors
91 * may be used to endorse or promote products derived from this software
92 * without specific prior written permission.
93 *
94 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
95 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
96 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
97 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
98 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
99 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
100 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
101 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
102 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
103 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
104 * SUCH DAMAGE.
105 *
106 * @(#)dinode.h 8.9 (Berkeley) 3/29/95
107 */
108 /*
109 * Copyright (c) 1982, 1986, 1989, 1993
110 * The Regents of the University of California. All rights reserved.
111 * (c) UNIX System Laboratories, Inc.
112 * All or some portions of this file are derived from material licensed
113 * to the University of California by American Telephone and Telegraph
114 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
115 * the permission of UNIX System Laboratories, Inc.
116 *
117 * Redistribution and use in source and binary forms, with or without
118 * modification, are permitted provided that the following conditions
119 * are met:
120 * 1. Redistributions of source code must retain the above copyright
121 * notice, this list of conditions and the following disclaimer.
122 * 2. Redistributions in binary form must reproduce the above copyright
123 * notice, this list of conditions and the following disclaimer in the
124 * documentation and/or other materials provided with the distribution.
125 * 3. Neither the name of the University nor the names of its contributors
126 * may be used to endorse or promote products derived from this software
127 * without specific prior written permission.
128 *
129 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
130 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
131 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
132 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
133 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
134 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
135 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
136 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
137 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
138 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
139 * SUCH DAMAGE.
140 *
141 * @(#)dir.h 8.5 (Berkeley) 4/27/95
142 */
143
144 /*
145 * NOTE: COORDINATE ON-DISK FORMAT CHANGES WITH THE FREEBSD PROJECT.
146 */
147
148 #ifndef _UFS_LFS_LFS_H_
149 #define _UFS_LFS_LFS_H_
150
151 #include <sys/rwlock.h>
152 #include <sys/mutex.h>
153 #include <sys/queue.h>
154 #include <sys/condvar.h>
155 #include <sys/mount.h>
156 #include <sys/pool.h>
157
158 /*
159 * Compile-time options for LFS.
160 */
161 #define LFS_IFIND_RETRIES 16
162 #define LFS_LOGLENGTH 1024 /* size of debugging log */
163 #define LFS_MAX_ACTIVE 10 /* Dirty segments before ckp forced */
164
165 /*
166 * Fixed filesystem layout parameters
167 */
168 #define LFS_LABELPAD 8192 /* LFS label size */
169 #define LFS_SBPAD 8192 /* LFS superblock size */
170
171 #define LFS_UNUSED_INUM 0 /* 0: out of band inode number */
172 #define LFS_IFILE_INUM 1 /* 1: IFILE inode number */
173 /* 2: Root inode number */
174 #define LFS_LOSTFOUNDINO 3 /* 3: lost+found inode number */
175 #define LFS_FIRST_INUM 4 /* 4: first free inode number */
176
177 /*
178 * The root inode is the root of the file system. Inode 0 can't be used for
179 * normal purposes and historically bad blocks were linked to inode 1, thus
180 * the root inode is 2. (Inode 1 is no longer used for this purpose, however
181 * numerous dump tapes make this assumption, so we are stuck with it).
182 */
183 #define ULFS_ROOTINO ((ino_t)2)
184
185 /*
186 * The Whiteout inode# is a dummy non-zero inode number which will
187 * never be allocated to a real file. It is used as a place holder
188 * in the directory entry which has been tagged as a LFS_DT_WHT entry.
189 * See the comments about ULFS_ROOTINO above.
190 */
191 #define ULFS_WINO ((ino_t)1)
192
193
194 #define LFS_V1_SUMMARY_SIZE 512 /* V1 fixed summary size */
195 #define LFS_DFL_SUMMARY_SIZE 512 /* Default summary size */
196
197 #define LFS_MAX_DADDR 0x7fffffff /* Highest addressable fsb */
198
199 #define LFS_MAXNAMLEN 255 /* maximum name length in a dir */
200
201 #define ULFS_NXADDR 2
202 #define ULFS_NDADDR 12 /* Direct addresses in inode. */
203 #define ULFS_NIADDR 3 /* Indirect addresses in inode. */
204
205 /*
206 * Adjustable filesystem parameters
207 */
208 #define MIN_FREE_SEGS 20
209 #define MIN_RESV_SEGS 15
210 #ifndef LFS_ATIME_IFILE
211 # define LFS_ATIME_IFILE 0 /* Store atime info in ifile (optional in LFSv1) */
212 #endif
213 #define LFS_MARKV_MAXBLKCNT 65536 /* Max block count for lfs_markv() */
214
215 /* Misc. definitions */
216 #define BW_CLEAN 1 /* Flag for lfs_bwrite_ext() */
217 #define PG_DELWRI PG_PAGER1 /* Local def for delayed pageout */
218
219 /* Resource limits */
220 #define LFS_MAX_RESOURCE(x, u) (((x) >> 2) - 10 * (u))
221 #define LFS_WAIT_RESOURCE(x, u) (((x) >> 1) - ((x) >> 3) - 10 * (u))
222 #define LFS_INVERSE_MAX_RESOURCE(x, u) (((x) + 10 * (u)) << 2)
223 #define LFS_MAX_BUFS LFS_MAX_RESOURCE(nbuf, 1)
224 #define LFS_WAIT_BUFS LFS_WAIT_RESOURCE(nbuf, 1)
225 #define LFS_INVERSE_MAX_BUFS(n) LFS_INVERSE_MAX_RESOURCE(n, 1)
226 #define LFS_MAX_BYTES LFS_MAX_RESOURCE(bufmem_lowater, PAGE_SIZE)
227 #define LFS_INVERSE_MAX_BYTES(n) LFS_INVERSE_MAX_RESOURCE(n, PAGE_SIZE)
228 #define LFS_WAIT_BYTES LFS_WAIT_RESOURCE(bufmem_lowater, PAGE_SIZE)
229 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3))
230 #define SIZEOF_DIROP(fs) (2 * ((fs)->lfs_bsize + LFS_DINODE1_SIZE))
231 #define LFS_MAX_FSDIROP(fs) \
232 ((fs)->lfs_nclean <= (fs)->lfs_resvseg ? 0 : \
233 (((fs)->lfs_nclean - (fs)->lfs_resvseg) * (fs)->lfs_ssize) / \
234 (2 * SIZEOF_DIROP(fs)))
235 #define LFS_MAX_PAGES lfs_max_pages()
236 #define LFS_WAIT_PAGES lfs_wait_pages()
237 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */
238
239 #ifdef _KERNEL
240 int lfs_wait_pages(void);
241 int lfs_max_pages(void);
242 #endif /* _KERNEL */
243
244 /* How starved can we be before we start holding back page writes */
245 #define LFS_STARVED_FOR_SEGS(fs) ((fs)->lfs_nclean < (fs)->lfs_resvseg)
246
247 /*
248 * Reserved blocks for lfs_malloc
249 */
250
251 /* Structure to keep reserved blocks */
252 typedef struct lfs_res_blk {
253 void *p;
254 LIST_ENTRY(lfs_res_blk) res;
255 int size;
256 char inuse;
257 } res_t;
258
259 /* Types for lfs_newbuf and lfs_malloc */
260 #define LFS_NB_UNKNOWN -1
261 #define LFS_NB_SUMMARY 0
262 #define LFS_NB_SBLOCK 1
263 #define LFS_NB_IBLOCK 2
264 #define LFS_NB_CLUSTER 3
265 #define LFS_NB_CLEAN 4
266 #define LFS_NB_BLKIOV 5
267 #define LFS_NB_COUNT 6 /* always last */
268
269 /* Number of reserved memory blocks of each type */
270 #define LFS_N_SUMMARIES 2
271 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */
272 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */
273 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */
274 #define LFS_N_CLEAN 0
275 #define LFS_N_BLKIOV 1
276
277 /* Total count of "large" (non-pool) types */
278 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \
279 LFS_N_CLUSTERS + LFS_N_CLEAN + LFS_N_BLKIOV)
280
281 /* Counts for pool types */
282 #define LFS_N_CL LFS_N_CLUSTERS
283 #define LFS_N_BPP 2
284 #define LFS_N_SEG 2
285
286 /*
287 * Directories
288 */
289
290 /*
291 * A directory consists of some number of blocks of LFS_DIRBLKSIZ
292 * bytes, where LFS_DIRBLKSIZ is chosen such that it can be transferred
293 * to disk in a single atomic operation (e.g. 512 bytes on most machines).
294 *
295 * Each LFS_DIRBLKSIZ byte block contains some number of directory entry
296 * structures, which are of variable length. Each directory entry has
297 * a struct lfs_direct at the front of it, containing its inode number,
298 * the length of the entry, and the length of the name contained in
299 * the entry. These are followed by the name padded to a 4 byte boundary.
300 * All names are guaranteed null terminated.
301 * The maximum length of a name in a directory is LFS_MAXNAMLEN.
302 *
303 * The macro DIRSIZ(fmt, dp) gives the amount of space required to represent
304 * a directory entry. Free space in a directory is represented by
305 * entries which have dp->d_reclen > DIRSIZ(fmt, dp). All LFS_DIRBLKSIZ bytes
306 * in a directory block are claimed by the directory entries. This
307 * usually results in the last entry in a directory having a large
308 * dp->d_reclen. When entries are deleted from a directory, the
309 * space is returned to the previous entry in the same directory
310 * block by increasing its dp->d_reclen. If the first entry of
311 * a directory block is free, then its dp->d_ino is set to 0.
312 * Entries other than the first in a directory do not normally have
313 * dp->d_ino set to 0.
314 */
315
316 /*
317 * Directory block size.
318 */
319 #undef LFS_DIRBLKSIZ
320 #define LFS_DIRBLKSIZ DEV_BSIZE
321
322 /*
323 * Convert between stat structure types and directory types.
324 */
325 #define LFS_IFTODT(mode) (((mode) & 0170000) >> 12)
326 #define LFS_DTTOIF(dirtype) ((dirtype) << 12)
327
328 /*
329 * The LFS_DIRSIZ macro gives the minimum record length which will hold
330 * the directory entry. This requires the amount of space in struct lfs_direct
331 * without the d_name field, plus enough space for the name with a terminating
332 * null byte (dp->d_namlen+1), rounded up to a 4 byte boundary.
333 */
334 #define LFS_DIRECTSIZ(namlen) \
335 ((sizeof(struct lfs_direct) - (LFS_MAXNAMLEN+1)) + (((namlen)+1 + 3) &~ 3))
336
337 #if (BYTE_ORDER == LITTLE_ENDIAN)
338 #define LFS_DIRSIZ(oldfmt, dp, needswap) \
339 (((oldfmt) && !(needswap)) ? \
340 LFS_DIRECTSIZ((dp)->d_type) : LFS_DIRECTSIZ((dp)->d_namlen))
341 #else
342 #define LFS_DIRSIZ(oldfmt, dp, needswap) \
343 (((oldfmt) && (needswap)) ? \
344 LFS_DIRECTSIZ((dp)->d_type) : LFS_DIRECTSIZ((dp)->d_namlen))
345 #endif
346
347 /* Constants for the first argument of LFS_DIRSIZ */
348 #define LFS_OLDDIRFMT 1
349 #define LFS_NEWDIRFMT 0
350
351 /*
352 * Theoretically, directories can be more than 2Gb in length; however, in
353 * practice this seems unlikely. So, we define the type doff_t as a 32-bit
354 * quantity to keep down the cost of doing lookup on a 32-bit machine.
355 */
356 #define doff_t int32_t
357 #define lfs_doff_t int32_t
358 #define MAXDIRSIZE (0x7fffffff)
359
360 #define LFS_MAXNAMLEN 255
361
362 /*
363 * File types for d_type
364 */
365 #define LFS_DT_UNKNOWN 0
366 #define LFS_DT_FIFO 1
367 #define LFS_DT_CHR 2
368 #define LFS_DT_DIR 4
369 #define LFS_DT_BLK 6
370 #define LFS_DT_REG 8
371 #define LFS_DT_LNK 10
372 #define LFS_DT_SOCK 12
373 #define LFS_DT_WHT 14
374
375 /*
376 * (See notes above)
377 */
378 #define d_ino d_fileno
379 struct lfs_direct {
380 u_int32_t d_fileno; /* inode number of entry */
381 u_int16_t d_reclen; /* length of this record */
382 u_int8_t d_type; /* file type, see below */
383 u_int8_t d_namlen; /* length of string in d_name */
384 char d_name[LFS_MAXNAMLEN + 1];/* name with length <= LFS_MAXNAMLEN */
385 };
386
387 /*
388 * Template for manipulating directories. Should use struct lfs_direct's,
389 * but the name field is LFS_MAXNAMLEN - 1, and this just won't do.
390 */
391 struct lfs_dirtemplate {
392 u_int32_t dot_ino;
393 int16_t dot_reclen;
394 u_int8_t dot_type;
395 u_int8_t dot_namlen;
396 char dot_name[4]; /* must be multiple of 4 */
397 u_int32_t dotdot_ino;
398 int16_t dotdot_reclen;
399 u_int8_t dotdot_type;
400 u_int8_t dotdot_namlen;
401 char dotdot_name[4]; /* ditto */
402 };
403
404 /*
405 * This is the old format of directories, sans type element.
406 */
407 struct lfs_odirtemplate {
408 u_int32_t dot_ino;
409 int16_t dot_reclen;
410 u_int16_t dot_namlen;
411 char dot_name[4]; /* must be multiple of 4 */
412 u_int32_t dotdot_ino;
413 int16_t dotdot_reclen;
414 u_int16_t dotdot_namlen;
415 char dotdot_name[4]; /* ditto */
416 };
417
418 /*
419 * Inodes
420 */
421
422 /*
423 * A dinode contains all the meta-data associated with a LFS file.
424 * This structure defines the on-disk format of a dinode. Since
425 * this structure describes an on-disk structure, all its fields
426 * are defined by types with precise widths.
427 */
428
429 struct ulfs1_dinode {
430 u_int16_t di_mode; /* 0: IFMT, permissions; see below. */
431 int16_t di_nlink; /* 2: File link count. */
432 u_int32_t di_inumber; /* 4: Inode number. */
433 u_int64_t di_size; /* 8: File byte count. */
434 int32_t di_atime; /* 16: Last access time. */
435 int32_t di_atimensec; /* 20: Last access time. */
436 int32_t di_mtime; /* 24: Last modified time. */
437 int32_t di_mtimensec; /* 28: Last modified time. */
438 int32_t di_ctime; /* 32: Last inode change time. */
439 int32_t di_ctimensec; /* 36: Last inode change time. */
440 int32_t di_db[ULFS_NDADDR]; /* 40: Direct disk blocks. */
441 int32_t di_ib[ULFS_NIADDR]; /* 88: Indirect disk blocks. */
442 u_int32_t di_flags; /* 100: Status flags (chflags). */
443 u_int32_t di_blocks; /* 104: Blocks actually held. */
444 int32_t di_gen; /* 108: Generation number. */
445 u_int32_t di_uid; /* 112: File owner. */
446 u_int32_t di_gid; /* 116: File group. */
447 u_int64_t di_modrev; /* 120: i_modrev for NFSv4 */
448 };
449
450 struct ulfs2_dinode {
451 u_int16_t di_mode; /* 0: IFMT, permissions; see below. */
452 int16_t di_nlink; /* 2: File link count. */
453 u_int32_t di_uid; /* 4: File owner. */
454 u_int32_t di_gid; /* 8: File group. */
455 u_int32_t di_blksize; /* 12: Inode blocksize. */
456 u_int64_t di_size; /* 16: File byte count. */
457 u_int64_t di_blocks; /* 24: Bytes actually held. */
458 int64_t di_atime; /* 32: Last access time. */
459 int64_t di_mtime; /* 40: Last modified time. */
460 int64_t di_ctime; /* 48: Last inode change time. */
461 int64_t di_birthtime; /* 56: Inode creation time. */
462 int32_t di_mtimensec; /* 64: Last modified time. */
463 int32_t di_atimensec; /* 68: Last access time. */
464 int32_t di_ctimensec; /* 72: Last inode change time. */
465 int32_t di_birthnsec; /* 76: Inode creation time. */
466 int32_t di_gen; /* 80: Generation number. */
467 u_int32_t di_kernflags; /* 84: Kernel flags. */
468 u_int32_t di_flags; /* 88: Status flags (chflags). */
469 int32_t di_extsize; /* 92: External attributes block. */
470 int64_t di_extb[ULFS_NXADDR];/* 96: External attributes block. */
471 int64_t di_db[ULFS_NDADDR]; /* 112: Direct disk blocks. */
472 int64_t di_ib[ULFS_NIADDR]; /* 208: Indirect disk blocks. */
473 u_int64_t di_modrev; /* 232: i_modrev for NFSv4 */
474 int64_t di_spare[2]; /* 240: Reserved; currently unused */
475 };
476
477 /*
478 * The di_db fields may be overlaid with other information for
479 * file types that do not have associated disk storage. Block
480 * and character devices overlay the first data block with their
481 * dev_t value. Short symbolic links place their path in the
482 * di_db area.
483 */
484 #define di_rdev di_db[0]
485
486 /* Size of the on-disk inode. */
487 #define LFS_DINODE1_SIZE (sizeof(struct ulfs1_dinode)) /* 128 */
488 #define LFS_DINODE2_SIZE (sizeof(struct ulfs2_dinode))
489
490 /* File types, found in the upper bits of di_mode. */
491 #define LFS_IFMT 0170000 /* Mask of file type. */
492 #define LFS_IFIFO 0010000 /* Named pipe (fifo). */
493 #define LFS_IFCHR 0020000 /* Character device. */
494 #define LFS_IFDIR 0040000 /* Directory file. */
495 #define LFS_IFBLK 0060000 /* Block device. */
496 #define LFS_IFREG 0100000 /* Regular file. */
497 #define LFS_IFLNK 0120000 /* Symbolic link. */
498 #define LFS_IFSOCK 0140000 /* UNIX domain socket. */
499 #define LFS_IFWHT 0160000 /* Whiteout. */
500
501 /*
502 * Maximum length of a symlink that can be stored within the inode.
503 */
504 #define ULFS1_MAXSYMLINKLEN ((ULFS_NDADDR + ULFS_NIADDR) * sizeof(int32_t))
505 #define ULFS2_MAXSYMLINKLEN ((ULFS_NDADDR + ULFS_NIADDR) * sizeof(int64_t))
506
507 #define ULFS_MAXSYMLINKLEN(ip) \
508 ((ip)->i_ump->um_fstype == ULFS1) ? \
509 ULFS1_MAXSYMLINKLEN : ULFS2_MAXSYMLINKLEN
510
511 /*
512 * "struct buf" associated definitions
513 */
514
515 /* Unassigned disk addresses. */
516 #define UNASSIGNED -1
517 #define UNWRITTEN -2
518
519 /* Unused logical block number */
520 #define LFS_UNUSED_LBN -1
521
522 /* Determine if a buffer belongs to the ifile */
523 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM)
524
525 # define LFS_LOCK_BUF(bp) do { \
526 if (((bp)->b_flags & B_LOCKED) == 0 && bp->b_iodone == NULL) { \
527 mutex_enter(&lfs_lock); \
528 ++locked_queue_count; \
529 locked_queue_bytes += bp->b_bufsize; \
530 mutex_exit(&lfs_lock); \
531 } \
532 (bp)->b_flags |= B_LOCKED; \
533 } while (0)
534
535 # define LFS_UNLOCK_BUF(bp) do { \
536 if (((bp)->b_flags & B_LOCKED) != 0 && bp->b_iodone == NULL) { \
537 mutex_enter(&lfs_lock); \
538 --locked_queue_count; \
539 locked_queue_bytes -= bp->b_bufsize; \
540 if (locked_queue_count < LFS_WAIT_BUFS && \
541 locked_queue_bytes < LFS_WAIT_BYTES) \
542 cv_broadcast(&locked_queue_cv); \
543 mutex_exit(&lfs_lock); \
544 } \
545 (bp)->b_flags &= ~B_LOCKED; \
546 } while (0)
547
548 #ifdef _KERNEL
549
550 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */
551
552 # define LFS_IS_MALLOC_BUF(bp) ((bp)->b_iodone == lfs_callback)
553
554 # ifdef DEBUG
555 # define LFS_DEBUG_COUNTLOCKED(m) do { \
556 if (lfs_debug_log_subsys[DLOG_LLIST]) { \
557 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \
558 cv_broadcast(&locked_queue_cv); \
559 } \
560 } while (0)
561 # else
562 # define LFS_DEBUG_COUNTLOCKED(m)
563 # endif
564
565 /* log for debugging writes to the Ifile */
566 # ifdef DEBUG
567 struct lfs_log_entry {
568 const char *op;
569 const char *file;
570 int pid;
571 int line;
572 daddr_t block;
573 unsigned long flags;
574 };
575 extern int lfs_lognum;
576 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH];
577 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__)
578 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags, thepid) do {\
579 int _s; \
580 \
581 mutex_enter(&lfs_lock); \
582 _s = splbio(); \
583 lfs_log[lfs_lognum].op = theop; \
584 lfs_log[lfs_lognum].file = thefile; \
585 lfs_log[lfs_lognum].line = (theline); \
586 lfs_log[lfs_lognum].pid = (thepid); \
587 lfs_log[lfs_lognum].block = (lbn); \
588 lfs_log[lfs_lognum].flags = (theflags); \
589 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \
590 splx(_s); \
591 mutex_exit(&lfs_lock); \
592 } while (0)
593
594 # define LFS_BCLEAN_LOG(fs, bp) do { \
595 if ((bp)->b_vp == (fs)->lfs_ivnode) \
596 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \
597 bp->b_lblkno, bp->b_flags, curproc->p_pid);\
598 } while (0)
599
600 /* Must match list in lfs_vfsops.c ! */
601 # define DLOG_RF 0 /* roll forward */
602 # define DLOG_ALLOC 1 /* inode alloc */
603 # define DLOG_AVAIL 2 /* lfs_{,r,f}avail */
604 # define DLOG_FLUSH 3 /* flush */
605 # define DLOG_LLIST 4 /* locked list accounting */
606 # define DLOG_WVNODE 5 /* vflush/writevnodes verbose */
607 # define DLOG_VNODE 6 /* vflush/writevnodes */
608 # define DLOG_SEG 7 /* segwrite */
609 # define DLOG_SU 8 /* seguse accounting */
610 # define DLOG_CLEAN 9 /* cleaner routines */
611 # define DLOG_MOUNT 10 /* mount/unmount */
612 # define DLOG_PAGE 11 /* putpages/gop_write */
613 # define DLOG_DIROP 12 /* dirop accounting */
614 # define DLOG_MALLOC 13 /* lfs_malloc accounting */
615 # define DLOG_MAX 14 /* The terminator */
616 # define DLOG(a) lfs_debug_log a
617 # else /* ! DEBUG */
618 # define LFS_BCLEAN_LOG(fs, bp)
619 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp)->b_vp, (bp))
620 # define DLOG(a)
621 # endif /* ! DEBUG */
622 #else /* ! _KERNEL */
623 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp))
624 #endif /* _KERNEL */
625
626 #ifdef _KERNEL
627 /* This overlays the fid structure (see fstypes.h). */
628 struct ulfs_ufid {
629 u_int16_t ufid_len; /* Length of structure. */
630 u_int16_t ufid_pad; /* Force 32-bit alignment. */
631 u_int32_t ufid_ino; /* File number (ino). */
632 int32_t ufid_gen; /* Generation number. */
633 };
634 /* Filehandle structure for exported LFSes */
635 struct lfid {
636 struct ulfs_ufid lfid_ufid;
637 #define lfid_len lfid_ufid.ufid_len
638 #define lfid_ino lfid_ufid.ufid_ino
639 #define lfid_gen lfid_ufid.ufid_gen
640 uint32_t lfid_ident;
641 };
642 #endif /* _KERNEL */
643
644 /*
645 * "struct inode" associated definitions
646 */
647
648 /* Address calculations for metadata located in the inode */
649 #define S_INDIR(fs) -ULFS_NDADDR
650 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1)
651 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1)
652
653 /* For convenience */
654 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING)
655
656 #define LFS_SET_UINO(ip, flags) do { \
657 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \
658 ++(ip)->i_lfs->lfs_uinodes; \
659 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \
660 ++(ip)->i_lfs->lfs_uinodes; \
661 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \
662 ++(ip)->i_lfs->lfs_uinodes; \
663 (ip)->i_flag |= (flags); \
664 } while (0)
665
666 #define LFS_CLR_UINO(ip, flags) do { \
667 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \
668 --(ip)->i_lfs->lfs_uinodes; \
669 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \
670 --(ip)->i_lfs->lfs_uinodes; \
671 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \
672 --(ip)->i_lfs->lfs_uinodes; \
673 (ip)->i_flag &= ~(flags); \
674 if ((ip)->i_lfs->lfs_uinodes < 0) { \
675 panic("lfs_uinodes < 0"); \
676 } \
677 } while (0)
678
679 #define LFS_ITIMES(ip, acc, mod, cre) \
680 while ((ip)->i_flag & (IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY)) \
681 lfs_itimes(ip, acc, mod, cre)
682
683 /*
684 * "struct vnode" associated definitions
685 */
686
687 /* Heuristic emptiness measure */
688 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \
689 !(vp->v_type == VREG && (vp)->v_iflag & VI_ONWORKLST) &&\
690 VTOI(vp)->i_lfs_nbtree == 0)
691
692 #define WRITEINPROG(vp) ((vp)->v_numoutput > 0 || \
693 (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \
694 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING))))
695
696
697 /*
698 * On-disk and in-memory checkpoint segment usage structure.
699 */
700 typedef struct segusage SEGUSE;
701 struct segusage {
702 u_int32_t su_nbytes; /* 0: number of live bytes */
703 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */
704 u_int16_t su_nsums; /* 8: number of summaries in segment */
705 u_int16_t su_ninos; /* 10: number of inode blocks in seg */
706
707 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */
708 #define SEGUSE_DIRTY 0x02 /* segment has data in it */
709 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */
710 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */
711 #define SEGUSE_EMPTY 0x10 /* segment is empty */
712 #define SEGUSE_INVAL 0x20 /* segment is invalid */
713 u_int32_t su_flags; /* 12: segment flags */
714 u_int64_t su_lastmod; /* 16: last modified timestamp */
715 };
716
717 typedef struct segusage_v1 SEGUSE_V1;
718 struct segusage_v1 {
719 u_int32_t su_nbytes; /* 0: number of live bytes */
720 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */
721 u_int16_t su_nsums; /* 8: number of summaries in segment */
722 u_int16_t su_ninos; /* 10: number of inode blocks in seg */
723 u_int32_t su_flags; /* 12: segment flags */
724 };
725
726 #define SEGUPB(fs) (fs->lfs_sepb)
727 #define SEGTABSIZE_SU(fs) \
728 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb)
729
730 #ifdef _KERNEL
731 # define SHARE_IFLOCK(F) \
732 do { \
733 rw_enter(&(F)->lfs_iflock, RW_READER); \
734 } while(0)
735 # define UNSHARE_IFLOCK(F) \
736 do { \
737 rw_exit(&(F)->lfs_iflock); \
738 } while(0)
739 #else /* ! _KERNEL */
740 # define SHARE_IFLOCK(F)
741 # define UNSHARE_IFLOCK(F)
742 #endif /* ! _KERNEL */
743
744 /* Read in the block with a specific segment usage entry from the ifile. */
745 #define LFS_SEGENTRY(SP, F, IN, BP) do { \
746 int _e; \
747 SHARE_IFLOCK(F); \
748 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
749 if ((_e = bread((F)->lfs_ivnode, \
750 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \
751 (F)->lfs_bsize, NOCRED, 0, &(BP))) != 0) \
752 panic("lfs: ifile read: %d", _e); \
753 if ((F)->lfs_version == 1) \
754 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \
755 ((IN) & ((F)->lfs_sepb - 1))); \
756 else \
757 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \
758 UNSHARE_IFLOCK(F); \
759 } while (0)
760
761 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \
762 if ((SP)->su_nbytes == 0) \
763 (SP)->su_flags |= SEGUSE_EMPTY; \
764 else \
765 (SP)->su_flags &= ~SEGUSE_EMPTY; \
766 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \
767 LFS_BWRITE_LOG(BP); \
768 } while (0)
769
770 /*
771 * On-disk file information. One per file with data blocks in the segment.
772 */
773 typedef struct finfo FINFO;
774 struct finfo {
775 u_int32_t fi_nblocks; /* number of blocks */
776 u_int32_t fi_version; /* version number */
777 u_int32_t fi_ino; /* inode number */
778 u_int32_t fi_lastlength; /* length of last block in array */
779 int32_t fi_blocks[1]; /* array of logical block numbers */
780 };
781 /* sizeof FINFO except fi_blocks */
782 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t))
783
784 /*
785 * Index file inode entries.
786 */
787 typedef struct ifile IFILE;
788 struct ifile {
789 u_int32_t if_version; /* inode version number */
790 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */
791 int32_t if_daddr; /* inode disk address */
792 #define LFS_ORPHAN_NEXTFREE (~(u_int32_t)0) /* indicate orphaned file */
793 u_int32_t if_nextfree; /* next-unallocated inode */
794 u_int32_t if_atime_sec; /* Last access time, seconds */
795 u_int32_t if_atime_nsec; /* and nanoseconds */
796 };
797
798 typedef struct ifile_v1 IFILE_V1;
799 struct ifile_v1 {
800 u_int32_t if_version; /* inode version number */
801 int32_t if_daddr; /* inode disk address */
802 u_int32_t if_nextfree; /* next-unallocated inode */
803 #if LFS_ATIME_IFILE
804 struct timespec if_atime; /* Last access time */
805 #endif
806 };
807
808 /*
809 * LFSv1 compatibility code is not allowed to touch if_atime, since it
810 * may not be mapped!
811 */
812 /* Read in the block with a specific inode from the ifile. */
813 #define LFS_IENTRY(IP, F, IN, BP) do { \
814 int _e; \
815 SHARE_IFLOCK(F); \
816 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
817 if ((_e = bread((F)->lfs_ivnode, \
818 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \
819 (F)->lfs_bsize, NOCRED, 0, &(BP))) != 0) \
820 panic("lfs: ifile ino %d read %d", (int)(IN), _e); \
821 if ((F)->lfs_version == 1) \
822 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \
823 (IN) % (F)->lfs_ifpb); \
824 else \
825 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \
826 UNSHARE_IFLOCK(F); \
827 } while (0)
828
829 /*
830 * Cleaner information structure. This resides in the ifile and is used
831 * to pass information from the kernel to the cleaner.
832 */
833 typedef struct _cleanerinfo {
834 u_int32_t clean; /* number of clean segments */
835 u_int32_t dirty; /* number of dirty segments */
836 int32_t bfree; /* disk blocks free */
837 int32_t avail; /* disk blocks available */
838 u_int32_t free_head; /* head of the inode free list */
839 u_int32_t free_tail; /* tail of the inode free list */
840 #define LFS_CLEANER_MUST_CLEAN 0x01
841 u_int32_t flags; /* status word from the kernel */
842 } CLEANERINFO;
843
844 #define CLEANSIZE_SU(fs) \
845 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift)
846
847 /* Read in the block with the cleaner info from the ifile. */
848 #define LFS_CLEANERINFO(CP, F, BP) do { \
849 SHARE_IFLOCK(F); \
850 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
851 if (bread((F)->lfs_ivnode, \
852 (daddr_t)0, (F)->lfs_bsize, NOCRED, 0, &(BP))) \
853 panic("lfs: ifile read"); \
854 (CP) = (CLEANERINFO *)(BP)->b_data; \
855 UNSHARE_IFLOCK(F); \
856 } while (0)
857
858 /*
859 * Synchronize the Ifile cleaner info with current avail and bfree.
860 */
861 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \
862 mutex_enter(&lfs_lock); \
863 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \
864 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \
865 (fs)->lfs_favail) { \
866 (cip)->bfree = (fs)->lfs_bfree; \
867 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \
868 (fs)->lfs_favail; \
869 if (((bp)->b_flags & B_GATHERED) == 0) { \
870 (fs)->lfs_flags |= LFS_IFDIRTY; \
871 } \
872 mutex_exit(&lfs_lock); \
873 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \
874 } else { \
875 mutex_exit(&lfs_lock); \
876 brelse(bp, 0); \
877 } \
878 } while (0)
879
880 /*
881 * Get the head of the inode free list.
882 * Always called with the segment lock held.
883 */
884 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \
885 if ((FS)->lfs_version > 1) { \
886 LFS_CLEANERINFO((CIP), (FS), (BP)); \
887 (FS)->lfs_freehd = (CIP)->free_head; \
888 brelse(BP, 0); \
889 } \
890 *(FREEP) = (FS)->lfs_freehd; \
891 } while (0)
892
893 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \
894 (FS)->lfs_freehd = (VAL); \
895 if ((FS)->lfs_version > 1) { \
896 LFS_CLEANERINFO((CIP), (FS), (BP)); \
897 (CIP)->free_head = (VAL); \
898 LFS_BWRITE_LOG(BP); \
899 mutex_enter(&lfs_lock); \
900 (FS)->lfs_flags |= LFS_IFDIRTY; \
901 mutex_exit(&lfs_lock); \
902 } \
903 } while (0)
904
905 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \
906 LFS_CLEANERINFO((CIP), (FS), (BP)); \
907 *(FREEP) = (CIP)->free_tail; \
908 brelse(BP, 0); \
909 } while (0)
910
911 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \
912 LFS_CLEANERINFO((CIP), (FS), (BP)); \
913 (CIP)->free_tail = (VAL); \
914 LFS_BWRITE_LOG(BP); \
915 mutex_enter(&lfs_lock); \
916 (FS)->lfs_flags |= LFS_IFDIRTY; \
917 mutex_exit(&lfs_lock); \
918 } while (0)
919
920 /*
921 * On-disk segment summary information
922 */
923 typedef struct segsum_v1 SEGSUM_V1;
924 struct segsum_v1 {
925 u_int32_t ss_sumsum; /* 0: check sum of summary block */
926 u_int32_t ss_datasum; /* 4: check sum of data */
927 u_int32_t ss_magic; /* 8: segment summary magic number */
928 #define SS_MAGIC 0x061561
929 int32_t ss_next; /* 12: next segment */
930 u_int32_t ss_create; /* 16: creation time stamp */
931 u_int16_t ss_nfinfo; /* 20: number of file info structures */
932 u_int16_t ss_ninos; /* 22: number of inodes in summary */
933
934 #define SS_DIROP 0x01 /* segment begins a dirop */
935 #define SS_CONT 0x02 /* more partials to finish this write*/
936 #define SS_CLEAN 0x04 /* written by the cleaner */
937 #define SS_RFW 0x08 /* written by the roll-forward agent */
938 #define SS_RECLAIM 0x10 /* written by the roll-forward agent */
939 u_int16_t ss_flags; /* 24: used for directory operations */
940 u_int16_t ss_pad; /* 26: extra space */
941 /* FINFO's and inode daddr's... */
942 };
943
944 typedef struct segsum SEGSUM;
945 struct segsum {
946 u_int32_t ss_sumsum; /* 0: check sum of summary block */
947 u_int32_t ss_datasum; /* 4: check sum of data */
948 u_int32_t ss_magic; /* 8: segment summary magic number */
949 int32_t ss_next; /* 12: next segment */
950 u_int32_t ss_ident; /* 16: roll-forward fsid */
951 #define ss_ocreate ss_ident /* ident is where create was in v1 */
952 u_int16_t ss_nfinfo; /* 20: number of file info structures */
953 u_int16_t ss_ninos; /* 22: number of inodes in summary */
954 u_int16_t ss_flags; /* 24: used for directory operations */
955 u_int8_t ss_pad[2]; /* 26: extra space */
956 u_int32_t ss_reclino; /* 28: inode being reclaimed */
957 u_int64_t ss_serial; /* 32: serial number */
958 u_int64_t ss_create; /* 40: time stamp */
959 /* FINFO's and inode daddr's... */
960 };
961
962 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM))
963
964
965 /*
966 * On-disk super block.
967 */
968 struct dlfs {
969 #define LFS_MAGIC 0x070162
970 u_int32_t dlfs_magic; /* 0: magic number */
971 #define LFS_VERSION 2
972 u_int32_t dlfs_version; /* 4: version number */
973
974 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */
975 /* number of frags in fs (v2) */
976 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */
977 /* number of bytes per segment (v2) */
978 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */
979 u_int32_t dlfs_bsize; /* 20: file system block size */
980 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */
981 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */
982
983 /* Checkpoint region. */
984 u_int32_t dlfs_freehd; /* 32: start of the free list */
985 int32_t dlfs_bfree; /* 36: number of free disk blocks */
986 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */
987 int32_t dlfs_avail; /* 44: blocks available for writing */
988 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */
989 int32_t dlfs_idaddr; /* 52: inode file disk address */
990 u_int32_t dlfs_ifile; /* 56: inode file inode number */
991 int32_t dlfs_lastseg; /* 60: address of last segment written */
992 int32_t dlfs_nextseg; /* 64: address of next segment to write */
993 int32_t dlfs_curseg; /* 68: current segment being written */
994 int32_t dlfs_offset; /* 72: offset in curseg for next partial */
995 int32_t dlfs_lastpseg; /* 76: address of last partial written */
996 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */
997 #define dlfs_otstamp dlfs_inopf
998
999 /* These are configuration parameters. */
1000 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */
1001
1002 /* These fields can be computed from the others. */
1003 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */
1004 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */
1005 u_int32_t dlfs_inopb; /* 100: inodes per block */
1006 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */
1007 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */
1008 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */
1009 u_int32_t dlfs_nseg; /* 116: number of segments */
1010 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */
1011 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */
1012 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */
1013 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */
1014 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */
1015 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */
1016 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */
1017 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */
1018 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */
1019 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */
1020 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */
1021 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */
1022 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */
1023
1024 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */
1025 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */
1026 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */
1027 int32_t dlfs_sboffs[LFS_MAXNUMSB];
1028
1029 u_int32_t dlfs_nclean; /* 228: Number of clean segments */
1030 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */
1031 #define LFS_PF_CLEAN 0x1
1032 u_int16_t dlfs_pflags; /* 322: file system persistent flags */
1033 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */
1034 u_int32_t dlfs_minfreeseg; /* 328: segments not counted in bfree */
1035 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */
1036 u_int64_t dlfs_serial; /* 336: serial number */
1037 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */
1038 int32_t dlfs_start; /* 348: start of segment 0 */
1039 u_int64_t dlfs_tstamp; /* 352: time stamp */
1040 #define LFS_44INODEFMT 0
1041 #define LFS_MAXINODEFMT 0
1042 u_int32_t dlfs_inodefmt; /* 360: inode format version */
1043 u_int32_t dlfs_interleave; /* 364: segment interleave */
1044 u_int32_t dlfs_ident; /* 368: per-fs identifier */
1045 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */
1046 u_int32_t dlfs_resvseg; /* 376: segments reserved for the cleaner */
1047 int8_t dlfs_pad[128]; /* 380: round to 512 bytes */
1048 /* Checksum -- last valid disk field. */
1049 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */
1050 };
1051
1052 /* Type used for the inode bitmap */
1053 typedef u_int32_t lfs_bm_t;
1054
1055 /*
1056 * Linked list of segments whose byte count needs updating following a
1057 * file truncation.
1058 */
1059 struct segdelta {
1060 long segnum;
1061 size_t num;
1062 LIST_ENTRY(segdelta) list;
1063 };
1064
1065 /*
1066 * In-memory super block.
1067 */
1068 struct lfs {
1069 struct dlfs lfs_dlfs; /* on-disk parameters */
1070 #define lfs_magic lfs_dlfs.dlfs_magic
1071 #define lfs_version lfs_dlfs.dlfs_version
1072 #define lfs_size lfs_dlfs.dlfs_size
1073 #define lfs_ssize lfs_dlfs.dlfs_ssize
1074 #define lfs_dsize lfs_dlfs.dlfs_dsize
1075 #define lfs_bsize lfs_dlfs.dlfs_bsize
1076 #define lfs_fsize lfs_dlfs.dlfs_fsize
1077 #define lfs_frag lfs_dlfs.dlfs_frag
1078 #define lfs_freehd lfs_dlfs.dlfs_freehd
1079 #define lfs_bfree lfs_dlfs.dlfs_bfree
1080 #define lfs_nfiles lfs_dlfs.dlfs_nfiles
1081 #define lfs_avail lfs_dlfs.dlfs_avail
1082 #define lfs_uinodes lfs_dlfs.dlfs_uinodes
1083 #define lfs_idaddr lfs_dlfs.dlfs_idaddr
1084 #define lfs_ifile lfs_dlfs.dlfs_ifile
1085 #define lfs_lastseg lfs_dlfs.dlfs_lastseg
1086 #define lfs_nextseg lfs_dlfs.dlfs_nextseg
1087 #define lfs_curseg lfs_dlfs.dlfs_curseg
1088 #define lfs_offset lfs_dlfs.dlfs_offset
1089 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg
1090 #define lfs_otstamp lfs_dlfs.dlfs_inopf
1091 #define lfs_inopf lfs_dlfs.dlfs_inopf
1092 #define lfs_minfree lfs_dlfs.dlfs_minfree
1093 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize
1094 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg
1095 #define lfs_inopb lfs_dlfs.dlfs_inopb
1096 #define lfs_ifpb lfs_dlfs.dlfs_ifpb
1097 #define lfs_sepb lfs_dlfs.dlfs_sepb
1098 #define lfs_nindir lfs_dlfs.dlfs_nindir
1099 #define lfs_nseg lfs_dlfs.dlfs_nseg
1100 #define lfs_nspf lfs_dlfs.dlfs_nspf
1101 #define lfs_cleansz lfs_dlfs.dlfs_cleansz
1102 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz
1103 #define lfs_segmask lfs_dlfs.dlfs_segmask
1104 #define lfs_segshift lfs_dlfs.dlfs_segshift
1105 #define lfs_bmask lfs_dlfs.dlfs_bmask
1106 #define lfs_bshift lfs_dlfs.dlfs_bshift
1107 #define lfs_ffmask lfs_dlfs.dlfs_ffmask
1108 #define lfs_ffshift lfs_dlfs.dlfs_ffshift
1109 #define lfs_fbmask lfs_dlfs.dlfs_fbmask
1110 #define lfs_fbshift lfs_dlfs.dlfs_fbshift
1111 #define lfs_blktodb lfs_dlfs.dlfs_blktodb
1112 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb
1113 #define lfs_sushift lfs_dlfs.dlfs_sushift
1114 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen
1115 #define lfs_sboffs lfs_dlfs.dlfs_sboffs
1116 #define lfs_cksum lfs_dlfs.dlfs_cksum
1117 #define lfs_pflags lfs_dlfs.dlfs_pflags
1118 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt
1119 #define lfs_nclean lfs_dlfs.dlfs_nclean
1120 #define lfs_dmeta lfs_dlfs.dlfs_dmeta
1121 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg
1122 #define lfs_sumsize lfs_dlfs.dlfs_sumsize
1123 #define lfs_serial lfs_dlfs.dlfs_serial
1124 #define lfs_ibsize lfs_dlfs.dlfs_ibsize
1125 #define lfs_start lfs_dlfs.dlfs_start
1126 #define lfs_tstamp lfs_dlfs.dlfs_tstamp
1127 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt
1128 #define lfs_interleave lfs_dlfs.dlfs_interleave
1129 #define lfs_ident lfs_dlfs.dlfs_ident
1130 #define lfs_resvseg lfs_dlfs.dlfs_resvseg
1131
1132 /* These fields are set at mount time and are meaningless on disk. */
1133 struct segment *lfs_sp; /* current segment being written */
1134 struct vnode *lfs_ivnode; /* vnode for the ifile */
1135 u_int32_t lfs_seglock; /* single-thread the segment writer */
1136 pid_t lfs_lockpid; /* pid of lock holder */
1137 lwpid_t lfs_locklwp; /* lwp of lock holder */
1138 u_int32_t lfs_iocount; /* number of ios pending */
1139 u_int32_t lfs_writer; /* don't allow any dirops to start */
1140 u_int32_t lfs_dirops; /* count of active directory ops */
1141 u_int32_t lfs_dirvcount; /* count of VDIROP nodes in this fs */
1142 u_int32_t lfs_doifile; /* Write ifile blocks on next write */
1143 u_int32_t lfs_nactive; /* Number of segments since last ckp */
1144 int8_t lfs_fmod; /* super block modified flag */
1145 int8_t lfs_ronly; /* mounted read-only flag */
1146 #define LFS_NOTYET 0x01
1147 #define LFS_IFDIRTY 0x02
1148 #define LFS_WARNED 0x04
1149 #define LFS_UNDIROP 0x08
1150 int8_t lfs_flags; /* currently unused flag */
1151 u_int16_t lfs_activesb; /* toggle between superblocks */
1152 daddr_t lfs_sbactive; /* disk address of current sb write */
1153 struct vnode *lfs_flushvp; /* vnode being flushed */
1154 int lfs_flushvp_fakevref; /* fake vref count for flushvp */
1155 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */
1156 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */
1157 size_t lfs_devbsize; /* Device block size */
1158 size_t lfs_devbshift; /* Device block shift */
1159 krwlock_t lfs_fraglock;
1160 krwlock_t lfs_iflock; /* Ifile lock */
1161 kcondvar_t lfs_stopcv; /* Wrap lock */
1162 struct lwp *lfs_stoplwp;
1163 pid_t lfs_rfpid; /* Process ID of roll-forward agent */
1164 int lfs_nadirop; /* number of active dirop nodes */
1165 long lfs_ravail; /* blocks pre-reserved for writing */
1166 long lfs_favail; /* blocks pre-reserved for writing */
1167 res_t *lfs_resblk; /* Reserved memory for pageout */
1168 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */
1169 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */
1170 #define LFS_RESHASH_WIDTH 17
1171 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH];
1172 int lfs_pdflush; /* pagedaemon wants us to flush */
1173 u_int32_t **lfs_suflags; /* Segment use flags */
1174 #ifdef _KERNEL
1175 struct pool lfs_clpool; /* Pool for struct lfs_cluster */
1176 struct pool lfs_bpppool; /* Pool for bpp */
1177 struct pool lfs_segpool; /* Pool for struct segment */
1178 #endif /* _KERNEL */
1179 #define LFS_MAX_CLEANIND 64
1180 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */
1181 int lfs_cleanind; /* Index into intervals */
1182 int lfs_sleepers; /* # procs sleeping this fs */
1183 int lfs_pages; /* dirty pages blaming this fs */
1184 lfs_bm_t *lfs_ino_bitmap; /* Inuse inodes bitmap */
1185 int lfs_nowrap; /* Suspend log wrap */
1186 int lfs_wrappass; /* Allow first log wrap requester to pass */
1187 int lfs_wrapstatus; /* Wrap status */
1188 int lfs_reclino; /* Inode being reclaimed */
1189 int lfs_startseg; /* Segment we started writing at */
1190 LIST_HEAD(, segdelta) lfs_segdhd; /* List of pending trunc accounting events */
1191 };
1192
1193 /* NINDIR is the number of indirects in a file system block. */
1194 #define NINDIR(fs) ((fs)->lfs_nindir)
1195
1196 /* INOPB is the number of inodes in a secondary storage block. */
1197 #define INOPB(fs) ((fs)->lfs_inopb)
1198 /* INOPF is the number of inodes in a fragment. */
1199 #define INOPF(fs) ((fs)->lfs_inopf)
1200
1201 #define blksize(fs, ip, lbn) \
1202 (((lbn) >= ULFS_NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \
1203 ? (fs)->lfs_bsize \
1204 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size))))
1205 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask))
1206 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \
1207 ((int)((loc) & (fs)->lfs_ffmask))
1208
1209 #if defined (_KERNEL)
1210 #define fsbtodb(fs, b) ((b) << ((fs)->lfs_ffshift - DEV_BSHIFT))
1211 #define dbtofsb(fs, b) ((b) >> ((fs)->lfs_ffshift - DEV_BSHIFT))
1212 #else
1213 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb)
1214 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb)
1215 #endif
1216
1217 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift)
1218 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift)
1219
1220 #define fsbtob(fs, b) ((b) << (fs)->lfs_ffshift)
1221 #define btofsb(fs, b) ((b) >> (fs)->lfs_ffshift)
1222
1223 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \
1224 ((loc) >> (fs)->lfs_ffshift)
1225 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \
1226 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask)))
1227 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \
1228 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask)))
1229 #define fragstoblks(fs, frags)/* calculates (frags / fs->fs_frag) */ \
1230 ((frags) >> (fs)->lfs_fbshift)
1231 #define blkstofrags(fs, blks) /* calculates (blks * fs->fs_frag) */ \
1232 ((blks) << (fs)->lfs_fbshift)
1233 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \
1234 ((fsb) & ((fs)->lfs_frag - 1))
1235 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \
1236 ((fsb) &~ ((fs)->lfs_frag - 1))
1237 #define dblksize(fs, dp, lbn) \
1238 (((lbn) >= ULFS_NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\
1239 ? (fs)->lfs_bsize \
1240 : (fragroundup(fs, blkoff(fs, (dp)->di_size))))
1241
1242 #define segsize(fs) ((fs)->lfs_version == 1 ? \
1243 lblktosize((fs), (fs)->lfs_ssize) : \
1244 (fs)->lfs_ssize)
1245 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \
1246 (fs)->lfs_ssize << (fs)->lfs_blktodb : \
1247 btofsb((fs), (fs)->lfs_ssize)) * (seg))
1248 #define dtosn(fs, daddr) /* block address to segment number */ \
1249 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1)))
1250 #define sntod(fs, sn) /* segment number to disk address */ \
1251 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start))
1252
1253 /*
1254 * Structures used by lfs_bmapv and lfs_markv to communicate information
1255 * about inodes and data blocks.
1256 */
1257 typedef struct block_info {
1258 u_int32_t bi_inode; /* inode # */
1259 int32_t bi_lbn; /* logical block w/in file */
1260 int32_t bi_daddr; /* disk address of block */
1261 u_int64_t bi_segcreate; /* origin segment create time */
1262 int bi_version; /* file version number */
1263 void *bi_bp; /* data buffer */
1264 int bi_size; /* size of the block (if fragment) */
1265 } BLOCK_INFO;
1266
1267 /* Compatibility for 1.5 binaries */
1268 typedef struct block_info_15 {
1269 u_int32_t bi_inode; /* inode # */
1270 int32_t bi_lbn; /* logical block w/in file */
1271 int32_t bi_daddr; /* disk address of block */
1272 u_int32_t bi_segcreate; /* origin segment create time */
1273 int bi_version; /* file version number */
1274 void *bi_bp; /* data buffer */
1275 int bi_size; /* size of the block (if fragment) */
1276 } BLOCK_INFO_15;
1277
1278 /* In-memory description of a segment about to be written. */
1279 struct segment {
1280 struct lfs *fs; /* file system pointer */
1281 struct buf **bpp; /* pointer to buffer array */
1282 struct buf **cbpp; /* pointer to next available bp */
1283 struct buf **start_bpp; /* pointer to first bp in this set */
1284 struct buf *ibp; /* buffer pointer to inode page */
1285 struct ulfs1_dinode *idp; /* pointer to ifile dinode */
1286 struct finfo *fip; /* current fileinfo pointer */
1287 struct vnode *vp; /* vnode being gathered */
1288 void *segsum; /* segment summary info */
1289 u_int32_t ninodes; /* number of inodes in this segment */
1290 int32_t seg_bytes_left; /* bytes left in segment */
1291 int32_t sum_bytes_left; /* bytes left in summary block */
1292 u_int32_t seg_number; /* number of this segment */
1293 int32_t *start_lbp; /* beginning lbn for this set */
1294
1295 #define SEGM_CKP 0x0001 /* doing a checkpoint */
1296 #define SEGM_CLEAN 0x0002 /* cleaner call; don't sort */
1297 #define SEGM_SYNC 0x0004 /* wait for segment */
1298 #define SEGM_PROT 0x0008 /* don't inactivate at segunlock */
1299 #define SEGM_PAGEDAEMON 0x0010 /* pagedaemon called us */
1300 #define SEGM_WRITERD 0x0020 /* LFS writed called us */
1301 #define SEGM_FORCE_CKP 0x0040 /* Force checkpoint right away */
1302 #define SEGM_RECLAIM 0x0080 /* Writing to reclaim vnode */
1303 #define SEGM_SINGLE 0x0100 /* Opportunistic writevnodes */
1304 u_int16_t seg_flags; /* run-time flags for this segment */
1305 u_int32_t seg_iocount; /* number of ios pending */
1306 int ndupino; /* number of duplicate inodes */
1307 };
1308
1309 #ifdef _KERNEL
1310 struct lfs_cluster {
1311 size_t bufsize; /* Size of kept data */
1312 struct buf **bpp; /* Array of kept buffers */
1313 int bufcount; /* Number of kept buffers */
1314 #define LFS_CL_MALLOC 0x00000001
1315 #define LFS_CL_SHIFT 0x00000002
1316 #define LFS_CL_SYNC 0x00000004
1317 u_int32_t flags; /* Flags */
1318 struct lfs *fs; /* LFS that this belongs to */
1319 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */
1320 };
1321
1322 /*
1323 * Splay tree containing block numbers allocated through lfs_balloc.
1324 */
1325 struct lbnentry {
1326 SPLAY_ENTRY(lbnentry) entry;
1327 daddr_t lbn;
1328 };
1329 #endif /* _KERNEL */
1330
1331 /*
1332 * LFS inode extensions.
1333 */
1334 struct lfs_inode_ext {
1335 off_t lfs_osize; /* size of file on disk */
1336 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */
1337 size_t lfs_fragsize[ULFS_NDADDR]; /* size of on-disk direct blocks */
1338 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */
1339 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */
1340 #define LFSI_NO_GOP_WRITE 0x01
1341 #define LFSI_DELETED 0x02
1342 #define LFSI_WRAPBLOCK 0x04
1343 #define LFSI_WRAPWAIT 0x08
1344 #define LFSI_BMAP 0x10
1345 u_int32_t lfs_iflags; /* Inode flags */
1346 daddr_t lfs_hiblk; /* Highest lbn held by inode */
1347 #ifdef _KERNEL
1348 SPLAY_HEAD(lfs_splay, lbnentry) lfs_lbtree; /* Tree of balloc'd lbns */
1349 int lfs_nbtree; /* Size of tree */
1350 LIST_HEAD(, segdelta) lfs_segdhd;
1351 #endif
1352 int16_t lfs_odnlink; /* on-disk nlink count for cleaner */
1353 };
1354 #define i_lfs_osize inode_ext.lfs->lfs_osize
1355 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks
1356 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize
1357 #define i_lfs_dchain inode_ext.lfs->lfs_dchain
1358 #define i_lfs_pchain inode_ext.lfs->lfs_pchain
1359 #define i_lfs_iflags inode_ext.lfs->lfs_iflags
1360 #define i_lfs_hiblk inode_ext.lfs->lfs_hiblk
1361 #define i_lfs_lbtree inode_ext.lfs->lfs_lbtree
1362 #define i_lfs_nbtree inode_ext.lfs->lfs_nbtree
1363 #define i_lfs_segdhd inode_ext.lfs->lfs_segdhd
1364 #define i_lfs_odnlink inode_ext.lfs->lfs_odnlink
1365
1366 /*
1367 * Macros for determining free space on the disk, with the variable metadata
1368 * of segment summaries and inode blocks taken into account.
1369 */
1370 /*
1371 * Estimate number of clean blocks not available for writing because
1372 * they will contain metadata or overhead. This is calculated as
1373 *
1374 * E = ((C * M / D) * D + (0) * (T - D)) / T
1375 * or more simply
1376 * E = (C * M) / T
1377 *
1378 * where
1379 * C is the clean space,
1380 * D is the dirty space,
1381 * M is the dirty metadata, and
1382 * T = C + D is the total space on disk.
1383 *
1384 * This approximates the old formula of E = C * M / D when D is close to T,
1385 * but avoids falsely reporting "disk full" when the sample size (D) is small.
1386 */
1387 #define LFS_EST_CMETA(F) (int32_t)(( \
1388 ((F)->lfs_dmeta * (int64_t)(F)->lfs_nclean) / \
1389 ((F)->lfs_nseg)))
1390
1391 /* Estimate total size of the disk not including metadata */
1392 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F))
1393
1394 /* Estimate number of blocks actually available for writing */
1395 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) ? \
1396 (F)->lfs_bfree - LFS_EST_CMETA(F) : 0)
1397
1398 /* Amount of non-meta space not available to mortal man */
1399 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \
1400 (u_int64_t)(F)->lfs_minfree) / \
1401 100)
1402
1403 /* Can credential C write BB blocks */
1404 #define ISSPACE(F, BB, C) \
1405 ((((C) == NOCRED || kauth_cred_geteuid(C) == 0) && \
1406 LFS_EST_BFREE(F) >= (BB)) || \
1407 (kauth_cred_geteuid(C) != 0 && IS_FREESPACE(F, BB)))
1408
1409 /* Can an ordinary user write BB blocks */
1410 #define IS_FREESPACE(F, BB) \
1411 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F))
1412
1413 /*
1414 * The minimum number of blocks to create a new inode. This is:
1415 * directory direct block (1) + ULFS_NIADDR indirect blocks + inode block (1) +
1416 * ifile direct block (1) + ULFS_NIADDR indirect blocks = 3 + 2 * ULFS_NIADDR blocks.
1417 */
1418 #define LFS_NRESERVE(F) (btofsb((F), (2 * ULFS_NIADDR + 3) << (F)->lfs_bshift))
1419
1420 /* Statistics Counters */
1421 struct lfs_stats { /* Must match sysctl list in lfs_vfsops.h ! */
1422 u_int segsused;
1423 u_int psegwrites;
1424 u_int psyncwrites;
1425 u_int pcleanwrites;
1426 u_int blocktot;
1427 u_int cleanblocks;
1428 u_int ncheckpoints;
1429 u_int nwrites;
1430 u_int nsync_writes;
1431 u_int wait_exceeded;
1432 u_int write_exceeded;
1433 u_int flush_invoked;
1434 u_int vflush_invoked;
1435 u_int clean_inlocked;
1436 u_int clean_vnlocked;
1437 u_int segs_reclaimed;
1438 };
1439 #ifdef _KERNEL
1440 extern struct lfs_stats lfs_stats;
1441 #endif
1442
1443 /* Fcntls to take the place of the lfs syscalls */
1444 struct lfs_fcntl_markv {
1445 BLOCK_INFO *blkiov; /* blocks to relocate */
1446 int blkcnt; /* number of blocks */
1447 };
1448
1449 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 14, struct timeval)
1450 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 15, struct timeval)
1451 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv)
1452 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv)
1453 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4)
1454
1455 struct lfs_fhandle {
1456 char space[28]; /* FHANDLE_SIZE_COMPAT (but used from userland too) */
1457 };
1458 #define LFCNREWIND _FCNR_FSPRIV('L', 6, int)
1459 #define LFCNINVAL _FCNR_FSPRIV('L', 7, int)
1460 #define LFCNRESIZE _FCNR_FSPRIV('L', 8, int)
1461 #define LFCNWRAPSTOP _FCNR_FSPRIV('L', 9, int)
1462 #define LFCNWRAPGO _FCNR_FSPRIV('L', 10, int)
1463 #define LFCNIFILEFH _FCNW_FSPRIV('L', 11, struct lfs_fhandle)
1464 #define LFCNWRAPPASS _FCNR_FSPRIV('L', 12, int)
1465 # define LFS_WRAP_GOING 0x0
1466 # define LFS_WRAP_WAITING 0x1
1467 #define LFCNWRAPSTATUS _FCNW_FSPRIV('L', 13, int)
1468
1469 /*
1470 * Compat. Defined for kernel only. Userland always uses
1471 * "the one true version".
1472 */
1473 #ifdef _KERNEL
1474 #include <compat/sys/time_types.h>
1475
1476 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval50)
1477 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval50)
1478 #define LFCNIFILEFH_COMPAT _FCNW_FSPRIV('L', 5, struct lfs_fhandle)
1479 #define LFCNIFILEFH_COMPAT2 _FCN_FSPRIV(F_FSOUT, 'L', 11, 32)
1480 #define LFCNWRAPSTOP_COMPAT _FCNO_FSPRIV('L', 9)
1481 #define LFCNWRAPGO_COMPAT _FCNO_FSPRIV('L', 10)
1482 #define LFCNSEGWAITALL_COMPAT_50 _FCNR_FSPRIV('L', 0, struct timeval50)
1483 #define LFCNSEGWAIT_COMPAT_50 _FCNR_FSPRIV('L', 1, struct timeval50)
1484 #endif
1485
1486 #ifdef _KERNEL
1487 /* XXX MP */
1488 #define LFS_SEGLOCK_HELD(fs) \
1489 ((fs)->lfs_seglock != 0 && \
1490 (fs)->lfs_lockpid == curproc->p_pid && \
1491 (fs)->lfs_locklwp == curlwp->l_lid)
1492 #endif /* _KERNEL */
1493
1494 /* Debug segment lock */
1495 #ifdef notyet
1496 # define ASSERT_SEGLOCK(fs) KASSERT(LFS_SEGLOCK_HELD(fs))
1497 # define ASSERT_NO_SEGLOCK(fs) KASSERT(!LFS_SEGLOCK_HELD(fs))
1498 # define ASSERT_DUNNO_SEGLOCK(fs)
1499 # define ASSERT_MAYBE_SEGLOCK(fs)
1500 #else /* !notyet */
1501 # define ASSERT_DUNNO_SEGLOCK(fs) \
1502 DLOG((DLOG_SEG, "lfs func %s seglock wrong (%d)\n", __func__, \
1503 LFS_SEGLOCK_HELD(fs)))
1504 # define ASSERT_SEGLOCK(fs) do { \
1505 if (!LFS_SEGLOCK_HELD(fs)) { \
1506 DLOG((DLOG_SEG, "lfs func %s seglock wrong (0)\n", __func__)); \
1507 } \
1508 } while(0)
1509 # define ASSERT_NO_SEGLOCK(fs) do { \
1510 if (LFS_SEGLOCK_HELD(fs)) { \
1511 DLOG((DLOG_SEG, "lfs func %s seglock wrong (1)\n", __func__)); \
1512 } \
1513 } while(0)
1514 # define ASSERT_MAYBE_SEGLOCK(x)
1515 #endif /* !notyet */
1516
1517 /*
1518 * Arguments to mount LFS filesystems
1519 */
1520 struct ulfs_args {
1521 char *fspec; /* block special device to mount */
1522 };
1523
1524 __BEGIN_DECLS
1525 void lfs_itimes(struct inode *, const struct timespec *,
1526 const struct timespec *, const struct timespec *);
1527 __END_DECLS
1528
1529 #endif /* !_UFS_LFS_LFS_H_ */
1530