lfs.h revision 1.41 1 /* $NetBSD: lfs.h,v 1.41 2002/07/06 01:30:11 perseant Exp $ */
2
3 /*-
4 * Copyright (c) 1999, 2000 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. All advertising materials mentioning features or use of this software
51 * must display the following acknowledgement:
52 * This product includes software developed by the University of
53 * California, Berkeley and its contributors.
54 * 4. Neither the name of the University nor the names of its contributors
55 * may be used to endorse or promote products derived from this software
56 * without specific prior written permission.
57 *
58 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
59 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
60 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
61 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
62 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
63 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
64 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
65 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
66 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
67 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
68 * SUCH DAMAGE.
69 *
70 * @(#)lfs.h 8.9 (Berkeley) 5/8/95
71 */
72
73 /*
74 * Compile-time options for LFS.
75 */
76 #define LFS_IFIND_RETRIES 16
77 #define LFS_EAGAIN_FAIL /* markv fail with EAGAIN if ino is locked */
78 #define LFS_DEBUG_RFW /* print roll-forward debugging info */
79 #define LFS_NO_PAGEMOVE /* Use malloc/copy to write clusters */
80 #define LFS_AGGRESSIVE_SEGLOCK
81 #define LFS_LOGLENGTH 1024
82
83 /* #define DEBUG_LFS */ /* Intensive debugging of LFS subsystem */
84
85 #ifdef LFS_NO_PAGEMOVE
86 # define LFS_MALLOC_SUMMARY
87 #endif
88
89 /*
90 * Parameters and generic definitions
91 */
92 #define BW_CLEAN 1
93 #define MIN_FREE_SEGS 2
94 #define LFS_MAX_ACTIVE 10
95 #define LFS_MAXDIROP (desiredvnodes >> 2)
96 #ifndef LFS_ATIME_IFILE
97 # define LFS_ATIME_IFILE 0
98 #endif
99
100 /*
101 * #define WRITE_THRESHHOLD ((nbuf >> 1) - 10)
102 * #define WAIT_THRESHHOLD (nbuf - (nbuf >> 2) - 10)
103 */
104 #define LFS_MAX_BUFS ((nbuf >> 2) - 10)
105 #define LFS_WAIT_BUFS ((nbuf >> 1) - (nbuf >> 3) - 10)
106 /* These are new ... is LFS taking up too much memory in its buffers? */
107 #define LFS_MAX_BYTES (((bufpages >> 2) - 10) * NBPG)
108 #define LFS_WAIT_BYTES (((bufpages >> 1) - (bufpages >> 3) - 10) * NBPG)
109 #define LFS_BUFWAIT 2
110
111 #define LFS_LOCK_BUF(bp) do { \
112 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \
113 ++locked_queue_count; \
114 locked_queue_bytes += bp->b_bufsize; \
115 } \
116 (bp)->b_flags |= B_LOCKED; \
117 } while (0)
118
119 #define LFS_UNLOCK_BUF(bp) do { \
120 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \
121 --locked_queue_count; \
122 locked_queue_bytes -= bp->b_bufsize; \
123 if (locked_queue_count < LFS_WAIT_BUFS && \
124 locked_queue_bytes < LFS_WAIT_BYTES) \
125 wakeup(&locked_queue_count); \
126 } \
127 (bp)->b_flags &= ~B_LOCKED; \
128 } while (0)
129
130 #ifdef DEBUG_LOCKED_LIST
131 # define LFS_DEBUG_COUNTLOCKED(m) do { \
132 int _s; \
133 extern int locked_queue_count; \
134 extern long locked_queue_bytes; \
135 _s = splbio(); \
136 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \
137 splx(_s); \
138 wakeup(&locked_queue_count); \
139 } while (0)
140 #else
141 # define LFS_DEBUG_COUNTLOCKED(m)
142 #endif
143
144 /* For convenience */
145 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_ACCESSED|IN_CLEANING)
146
147 #define LFS_SET_UINO(ip, flags) do { \
148 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \
149 ++(ip)->i_lfs->lfs_uinodes; \
150 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \
151 ++(ip)->i_lfs->lfs_uinodes; \
152 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \
153 ++(ip)->i_lfs->lfs_uinodes; \
154 (ip)->i_flag |= (flags); \
155 } while (0)
156
157 #define LFS_CLR_UINO(ip, flags) do { \
158 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \
159 --(ip)->i_lfs->lfs_uinodes; \
160 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \
161 --(ip)->i_lfs->lfs_uinodes; \
162 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \
163 --(ip)->i_lfs->lfs_uinodes; \
164 (ip)->i_flag &= ~(flags); \
165 if ((ip)->i_lfs->lfs_uinodes < 0) { \
166 panic("lfs_uinodes < 0"); \
167 } \
168 } while (0)
169
170 #ifdef DEBUG
171 struct lfs_log_entry {
172 char *op;
173 char *file;
174 int line;
175 ufs_daddr_t block;
176 unsigned long flags;
177 };
178 extern int lfs_lognum;
179 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH];
180 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__)
181 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags) do { \
182 int _s; \
183 \
184 _s = splbio(); \
185 lfs_log[lfs_lognum].op = theop; \
186 lfs_log[lfs_lognum].file = thefile; \
187 lfs_log[lfs_lognum].line = (theline); \
188 lfs_log[lfs_lognum].block = (lbn); \
189 lfs_log[lfs_lognum].flags = (theflags); \
190 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \
191 splx(_s); \
192 } while (0)
193
194 # define LFS_BCLEAN_LOG(fs, bp) do { \
195 if ((bp)->b_vp == (fs)->lfs_ivnode) \
196 LFS_ENTER_LOG("clear", __FILE__, __LINE__, bp->b_lblkno, bp->b_flags); \
197 } while (0)
198 #else
199 # define LFS_BCLEAN_LOG(fs, bp)
200 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp))
201 #endif
202
203 #define LFS_ITIMES(ip, acc, mod, cre) do { \
204 struct lfs *_fs = (ip)->i_lfs; \
205 \
206 if ((ip)->i_flag & IN_ACCESS) { \
207 (ip)->i_ffs_atime = (acc)->tv_sec; \
208 (ip)->i_ffs_atimensec = (acc)->tv_nsec; \
209 if ((ip)->i_lfs->lfs_version > 1) { \
210 struct buf *ibp; \
211 IFILE *ifp; \
212 \
213 LFS_IENTRY(ifp, ip->i_lfs, ip->i_number, ibp); \
214 ifp->if_atime_sec = (acc)->tv_sec; \
215 ifp->if_atime_nsec = (acc)->tv_nsec; \
216 LFS_BWRITE_LOG(ibp); \
217 _fs->lfs_flags |= LFS_IFDIRTY; \
218 } else { \
219 LFS_SET_UINO(ip, IN_ACCESSED); \
220 } \
221 } \
222 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) { \
223 if ((ip)->i_flag & IN_UPDATE) { \
224 (ip)->i_ffs_mtime = (mod)->tv_sec; \
225 (ip)->i_ffs_mtimensec = (mod)->tv_nsec; \
226 (ip)->i_modrev++; \
227 } \
228 if ((ip)->i_flag & IN_CHANGE) { \
229 (ip)->i_ffs_ctime = (cre)->tv_sec; \
230 (ip)->i_ffs_ctimensec = (cre)->tv_nsec; \
231 } \
232 LFS_SET_UINO(ip, IN_MODIFIED); \
233 } \
234 (ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE); \
235 } while (0)
236
237 #define WRITEINPROG(vp) (vp->v_dirtyblkhd.lh_first && !(VTOI(vp)->i_flag & \
238 (IN_MODIFIED | IN_ACCESSED | IN_CLEANING)))
239
240 /* Here begins the berkeley code */
241
242 #define LFS_LABELPAD 8192 /* LFS label size */
243 #define LFS_SBPAD 8192 /* LFS superblock size */
244
245 /* On-disk and in-memory checkpoint segment usage structure. */
246 typedef struct segusage SEGUSE;
247 struct segusage {
248 u_int32_t su_nbytes; /* 0: number of live bytes */
249 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */
250 u_int16_t su_nsums; /* 8: number of summaries in segment */
251 u_int16_t su_ninos; /* 10: number of inode blocks in seg */
252
253 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */
254 #define SEGUSE_DIRTY 0x02 /* segment has data in it */
255 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */
256 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */
257 u_int32_t su_flags; /* 12: segment flags */
258 u_int64_t su_lastmod; /* 16: last modified timestamp */
259 };
260
261 typedef struct segusage_v1 SEGUSE_V1;
262 struct segusage_v1 {
263 u_int32_t su_nbytes; /* 0: number of live bytes */
264 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */
265 u_int16_t su_nsums; /* 8: number of summaries in segment */
266 u_int16_t su_ninos; /* 10: number of inode blocks in seg */
267 u_int32_t su_flags; /* 12: segment flags */
268 };
269
270 #define SEGUPB(fs) (fs->lfs_sepb)
271 #define SEGTABSIZE_SU(fs) \
272 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb)
273
274 /* On-disk file information. One per file with data blocks in the segment. */
275 typedef struct finfo FINFO;
276 struct finfo {
277 u_int32_t fi_nblocks; /* number of blocks */
278 u_int32_t fi_version; /* version number */
279 u_int32_t fi_ino; /* inode number */
280 u_int32_t fi_lastlength; /* length of last block in array */
281 ufs_daddr_t fi_blocks[1]; /* array of logical block numbers */
282 };
283
284
285 /* On-disk super block. */
286 struct dlfs {
287 #define LFS_MAGIC 0x070162
288 u_int32_t dlfs_magic; /* 0: magic number */
289 #define LFS_VERSION 2
290 u_int32_t dlfs_version; /* 4: version number */
291
292 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */
293 /* number of frags in fs (v2) */
294 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */
295 /* number of bytes per segment (v2) */
296 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */
297 u_int32_t dlfs_bsize; /* 20: file system block size */
298 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */
299 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */
300
301 /* Checkpoint region. */
302 u_int32_t dlfs_free; /* 32: start of the free list */
303 u_int32_t dlfs_bfree; /* 36: number of free disk blocks */
304 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */
305 int32_t dlfs_avail; /* 44: blocks available for writing */
306 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */
307 ufs_daddr_t dlfs_idaddr; /* 52: inode file disk address */
308 u_int32_t dlfs_ifile; /* 56: inode file inode number */
309 ufs_daddr_t dlfs_lastseg; /* 60: address of last segment written */
310 ufs_daddr_t dlfs_nextseg; /* 64: address of next segment to write */
311 ufs_daddr_t dlfs_curseg; /* 68: current segment being written */
312 ufs_daddr_t dlfs_offset; /* 72: offset in curseg for next partial */
313 ufs_daddr_t dlfs_lastpseg; /* 76: address of last partial written */
314 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */
315 #define dlfs_otstamp dlfs_inopf
316
317 /* These are configuration parameters. */
318 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */
319
320 /* These fields can be computed from the others. */
321 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */
322 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */
323 u_int32_t dlfs_inopb; /* 100: inodes per block */
324 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */
325 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */
326 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */
327 u_int32_t dlfs_nseg; /* 116: number of segments */
328 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */
329 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */
330 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */
331 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */
332 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */
333 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */
334 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */
335 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */
336 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */
337 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */
338 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */
339 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */
340 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */
341
342 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */
343 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */
344 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */
345 ufs_daddr_t dlfs_sboffs[LFS_MAXNUMSB];
346
347 u_int32_t dlfs_nclean; /* 228: Number of clean segments */
348 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */
349 #define LFS_PF_CLEAN 0x1
350 u_int16_t dlfs_pflags; /* 322: file system persistent flags */
351 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */
352 u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */
353 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */
354 u_int64_t dlfs_serial; /* 336: serial number */
355 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */
356 ufs_daddr_t dlfs_start; /* 348: start of segment 0 */
357 u_int64_t dlfs_tstamp; /* 352: time stamp */
358 #define LFS_44INODEFMT 0
359 #define LFS_MAXINODEFMT 0
360 u_int32_t dlfs_inodefmt; /* 360: inode format version */
361 u_int32_t dlfs_interleave; /* 364: segment interleave */
362 u_int32_t dlfs_ident; /* 368: per-fs identifier */
363 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */
364 int8_t dlfs_pad[132]; /* 376: round to 512 bytes */
365 /* Checksum -- last valid disk field. */
366 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */
367 };
368
369 /* Maximum number of io's we can have pending at once */
370 #define LFS_THROTTLE 32 /* XXX should be better paramtrized - ? */
371
372 /* In-memory super block. */
373 struct lfs {
374 struct dlfs lfs_dlfs; /* on-disk parameters */
375 #define lfs_magic lfs_dlfs.dlfs_magic
376 #define lfs_version lfs_dlfs.dlfs_version
377 #define lfs_size lfs_dlfs.dlfs_size
378 #define lfs_ssize lfs_dlfs.dlfs_ssize
379 #define lfs_dsize lfs_dlfs.dlfs_dsize
380 #define lfs_bsize lfs_dlfs.dlfs_bsize
381 #define lfs_fsize lfs_dlfs.dlfs_fsize
382 #define lfs_frag lfs_dlfs.dlfs_frag
383 #define lfs_free lfs_dlfs.dlfs_free
384 #define lfs_bfree lfs_dlfs.dlfs_bfree
385 #define lfs_nfiles lfs_dlfs.dlfs_nfiles
386 #define lfs_avail lfs_dlfs.dlfs_avail
387 #define lfs_uinodes lfs_dlfs.dlfs_uinodes
388 #define lfs_idaddr lfs_dlfs.dlfs_idaddr
389 #define lfs_ifile lfs_dlfs.dlfs_ifile
390 #define lfs_lastseg lfs_dlfs.dlfs_lastseg
391 #define lfs_nextseg lfs_dlfs.dlfs_nextseg
392 #define lfs_curseg lfs_dlfs.dlfs_curseg
393 #define lfs_offset lfs_dlfs.dlfs_offset
394 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg
395 #define lfs_otstamp lfs_dlfs.dlfs_inopf
396 #define lfs_inopf lfs_dlfs.dlfs_inopf
397 #define lfs_minfree lfs_dlfs.dlfs_minfree
398 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize
399 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg
400 #define lfs_inopb lfs_dlfs.dlfs_inopb
401 #define lfs_ifpb lfs_dlfs.dlfs_ifpb
402 #define lfs_sepb lfs_dlfs.dlfs_sepb
403 #define lfs_nindir lfs_dlfs.dlfs_nindir
404 #define lfs_nseg lfs_dlfs.dlfs_nseg
405 #define lfs_nspf lfs_dlfs.dlfs_nspf
406 #define lfs_cleansz lfs_dlfs.dlfs_cleansz
407 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz
408 #define lfs_segmask lfs_dlfs.dlfs_segmask
409 #define lfs_segshift lfs_dlfs.dlfs_segshift
410 #define lfs_bmask lfs_dlfs.dlfs_bmask
411 #define lfs_bshift lfs_dlfs.dlfs_bshift
412 #define lfs_ffmask lfs_dlfs.dlfs_ffmask
413 #define lfs_ffshift lfs_dlfs.dlfs_ffshift
414 #define lfs_fbmask lfs_dlfs.dlfs_fbmask
415 #define lfs_fbshift lfs_dlfs.dlfs_fbshift
416 #define lfs_blktodb lfs_dlfs.dlfs_blktodb
417 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb
418 #define lfs_sushift lfs_dlfs.dlfs_sushift
419 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen
420 #define lfs_sboffs lfs_dlfs.dlfs_sboffs
421 #define lfs_cksum lfs_dlfs.dlfs_cksum
422 #define lfs_pflags lfs_dlfs.dlfs_pflags
423 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt
424 #define lfs_nclean lfs_dlfs.dlfs_nclean
425 #define lfs_dmeta lfs_dlfs.dlfs_dmeta
426 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg
427 #define lfs_sumsize lfs_dlfs.dlfs_sumsize
428 #define lfs_serial lfs_dlfs.dlfs_serial
429 #define lfs_ibsize lfs_dlfs.dlfs_ibsize
430 #define lfs_start lfs_dlfs.dlfs_start
431 #define lfs_tstamp lfs_dlfs.dlfs_tstamp
432 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt
433 #define lfs_interleave lfs_dlfs.dlfs_interleave
434 #define lfs_ident lfs_dlfs.dlfs_ident
435
436 /* These fields are set at mount time and are meaningless on disk. */
437 struct segment *lfs_sp; /* current segment being written */
438 struct vnode *lfs_ivnode; /* vnode for the ifile */
439 u_int32_t lfs_seglock; /* single-thread the segment writer */
440 pid_t lfs_lockpid; /* pid of lock holder */
441 u_int32_t lfs_iocount; /* number of ios pending */
442 u_int32_t lfs_writer; /* don't allow any dirops to start */
443 u_int32_t lfs_dirops; /* count of active directory ops */
444 u_int32_t lfs_doifile; /* Write ifile blocks on next write */
445 u_int32_t lfs_nactive; /* Number of segments since last ckp */
446 int8_t lfs_fmod; /* super block modified flag */
447 int8_t lfs_ronly; /* mounted read-only flag */
448 #define LFS_NOTYET 0x01
449 #define LFS_IFDIRTY 0x02
450 #define LFS_WARNED 0x04
451 int8_t lfs_flags; /* currently unused flag */
452 u_int16_t lfs_activesb; /* toggle between superblocks */
453 #ifdef LFS_TRACK_IOS
454 daddr_t lfs_pending[LFS_THROTTLE]; /* daddrs of pending writes */
455 #endif /* LFS_TRACK_IOS */
456 daddr_t lfs_sbactive; /* disk address of in-progress sb write */
457 struct vnode *lfs_flushvp; /* vnode being flushed */
458 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */
459 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */
460 size_t lfs_devbsize; /* Device block size */
461 size_t lfs_devbshift; /* Device block shift */
462 struct lock lfs_freelock;
463 struct lock lfs_fraglock;
464 pid_t lfs_rfpid; /* Process ID of roll-forward agent */
465 int lfs_nadirop; /* number of active dirop nodes */
466 long lfs_ravail; /* blocks pre-reserved for writing */
467 };
468
469 /*
470 * Inode 0: out-of-band inode number
471 * Inode 1: IFILE inode number
472 * Inode 2: root inode
473 * Inode 3: lost+found inode number
474 */
475 #define LFS_UNUSED_INUM 0 /* out of band inode number */
476 #define LFS_IFILE_INUM 1 /* IFILE inode number */
477 #define LOSTFOUNDINO 3 /* lost+found inode number */
478 #define LFS_FIRST_INUM 4 /* first free inode number */
479
480 /* Address calculations for metadata located in the inode */
481 #define S_INDIR(fs) -NDADDR
482 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1)
483 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1)
484
485 /* Unassigned disk addresses. */
486 #define UNASSIGNED -1
487 #define UNWRITTEN -2
488
489 /* Unused logical block number */
490 #define LFS_UNUSED_LBN -1
491
492 typedef struct ifile IFILE;
493 struct ifile {
494 u_int32_t if_version; /* inode version number */
495 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */
496 ufs_daddr_t if_daddr; /* inode disk address */
497 ino_t if_nextfree; /* next-unallocated inode */
498 /* XXX - when inode format changes, this changes too */
499 u_int32_t if_atime_sec; /* Last access time, seconds */
500 u_int32_t if_atime_nsec; /* and nanoseconds */
501 };
502
503 typedef struct ifile_v1 IFILE_V1;
504 struct ifile_v1 {
505 u_int32_t if_version; /* inode version number */
506 ufs_daddr_t if_daddr; /* inode disk address */
507 ino_t if_nextfree; /* next-unallocated inode */
508 #if LFS_ATIME_IFILE
509 struct timespec if_atime; /* Last access time */
510 #endif
511 };
512
513 /*
514 * Cleaner information structure. This resides in the ifile and is used
515 * to pass information between the cleaner and the kernel.
516 */
517 typedef struct _cleanerinfo {
518 u_int32_t clean; /* number of clean segments */
519 u_int32_t dirty; /* number of dirty segments */
520 u_int32_t bfree; /* disk blocks free */
521 int32_t avail; /* disk blocks available */
522 u_int32_t free_head; /* head of the inode free list */
523 u_int32_t free_tail; /* tail of the inode free list */
524 } CLEANERINFO;
525
526 #define CLEANSIZE_SU(fs) \
527 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift)
528
529 /*
530 * All summary blocks are the same size, so we can always read a summary
531 * block easily from a segment.
532 */
533 #define LFS_V1_SUMMARY_SIZE 512
534 #define LFS_DFL_SUMMARY_SIZE 512
535
536 /* On-disk segment summary information */
537 typedef struct segsum_v1 SEGSUM_V1;
538 struct segsum_v1 {
539 u_int32_t ss_sumsum; /* 0: check sum of summary block */
540 u_int32_t ss_datasum; /* 4: check sum of data */
541 u_int32_t ss_magic; /* 8: segment summary magic number */
542 #define SS_MAGIC 0x061561
543 ufs_daddr_t ss_next; /* 12: next segment */
544 u_int32_t ss_create; /* 16: creation time stamp */
545 u_int16_t ss_nfinfo; /* 20: number of file info structures */
546 u_int16_t ss_ninos; /* 22: number of inodes in summary */
547
548 #define SS_DIROP 0x01 /* segment begins a dirop */
549 #define SS_CONT 0x02 /* more partials to finish this write*/
550 u_int16_t ss_flags; /* 24: used for directory operations */
551 u_int16_t ss_pad; /* 26: extra space */
552 /* FINFO's and inode daddr's... */
553 };
554
555 typedef struct segsum SEGSUM;
556 struct segsum {
557 u_int32_t ss_sumsum; /* 0: check sum of summary block */
558 u_int32_t ss_datasum; /* 4: check sum of data */
559 u_int32_t ss_magic; /* 8: segment summary magic number */
560 ufs_daddr_t ss_next; /* 12: next segment */
561 u_int32_t ss_ident; /* 16: roll-forward fsid */
562 #define ss_ocreate ss_ident /* ident is where create was in v1 */
563 u_int16_t ss_nfinfo; /* 20: number of file info structures */
564 u_int16_t ss_ninos; /* 22: number of inodes in summary */
565 u_int16_t ss_flags; /* 24: used for directory operations */
566 u_int8_t ss_pad[6]; /* 26: extra space */
567 u_int64_t ss_serial; /* 32: serial number */
568 u_int64_t ss_create; /* 40: time stamp */
569 /* FINFO's and inode daddr's... */
570 };
571
572 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM))
573
574 /* NINDIR is the number of indirects in a file system block. */
575 #define NINDIR(fs) ((fs)->lfs_nindir)
576
577 /* INOPB is the number of inodes in a secondary storage block. */
578 #define INOPB(fs) ((fs)->lfs_inopb)
579 /* INOPF is the number of inodes in a fragment. */
580 #define INOPF(fs) ((fs)->lfs_inopf)
581
582 #define blksize(fs, ip, lbn) \
583 (((lbn) >= NDADDR || (ip)->i_ffs_size >= ((lbn) + 1) << (fs)->lfs_bshift) \
584 ? (fs)->lfs_bsize \
585 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs_size))))
586 #define blkoff(fs, loc) ((int)(loc) & (fs)->lfs_bmask)
587 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \
588 ((int)((loc) & (fs)->lfs_ffmask))
589 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb)
590 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb)
591 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift))
592 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift))
593 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift)
594 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift)
595 /* Same as above, but named like dbtob(), btodb() */
596 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \
597 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb))
598 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \
599 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb))
600 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \
601 (fs)->lfs_fsbtodb))
602 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \
603 (fs)->lfs_fsbtodb))
604 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift)
605 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \
606 ((loc) >> (fs)->lfs_ffshift)
607 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \
608 ((int)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask)))
609 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \
610 ((int)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask)))
611 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \
612 ((frags) >> (fs)->lfs_fbshift)
613 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \
614 ((blks) << (fs)->lfs_fbshift)
615 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \
616 ((fsb) & ((fs)->lfs_frag - 1))
617 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \
618 ((fsb) &~ ((fs)->lfs_frag - 1))
619 #define dblksize(fs, dip, lbn) \
620 (((lbn) >= NDADDR || (dip)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\
621 ? (fs)->lfs_bsize \
622 : (fragroundup(fs, blkoff(fs, (dip)->di_size))))
623
624 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \
625 (fs)->lfs_ssize << (fs)->lfs_blktodb : \
626 btofsb((fs), (fs)->lfs_ssize)) * (seg))
627 #define dtosn(fs, daddr) /* block address to segment number */ \
628 (((daddr) - (fs)->lfs_start) / segtod((fs), 1))
629 #define sntod(fs, sn) /* segment number to disk address */ \
630 ((ufs_daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start))
631
632 /* Read in the block with the cleaner info from the ifile. */
633 #define LFS_CLEANERINFO(CP, F, BP) do { \
634 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
635 if (bread((F)->lfs_ivnode, \
636 (ufs_daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \
637 panic("lfs: ifile read"); \
638 (CP) = (CLEANERINFO *)(BP)->b_data; \
639 } while(0)
640
641 /* Synchronize the Ifile cleaner info with current avail and bfree */
642 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \
643 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \
644 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail) { \
645 (cip)->bfree = (fs)->lfs_bfree; \
646 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail; \
647 if (((bp)->b_flags & B_GATHERED) == 0) \
648 (fs)->lfs_flags |= LFS_IFDIRTY; \
649 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \
650 } else \
651 brelse(bp); \
652 } while (0)
653
654 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \
655 if ((FS)->lfs_version > 1) { \
656 LFS_CLEANERINFO((CIP), (FS), (BP)); \
657 (FS)->lfs_free = (CIP)->free_head; \
658 brelse(BP); \
659 } \
660 *(FREEP) = (FS)->lfs_free; \
661 } while (0)
662
663 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \
664 (FS)->lfs_free = (VAL); \
665 if ((FS)->lfs_version > 1) { \
666 LFS_CLEANERINFO((CIP), (FS), (BP)); \
667 (CIP)->free_head = (VAL); \
668 LFS_BWRITE_LOG(BP); \
669 (FS)->lfs_flags |= LFS_IFDIRTY; \
670 } \
671 } while (0)
672
673 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \
674 LFS_CLEANERINFO((CIP), (FS), (BP)); \
675 *(FREEP) = (CIP)->free_tail; \
676 brelse(BP); \
677 } while (0)
678
679 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \
680 LFS_CLEANERINFO((CIP), (FS), (BP)); \
681 (CIP)->free_tail = (VAL); \
682 LFS_BWRITE_LOG(BP); \
683 (FS)->lfs_flags |= LFS_IFDIRTY; \
684 } while (0)
685
686 /*
687 * XXX - v1 compatibility code is not allowed to touch if_atime, since it
688 * may not be mapped!
689 */
690 /* Read in the block with a specific inode from the ifile. */
691 #define LFS_IENTRY(IP, F, IN, BP) do { \
692 int _e; \
693 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
694 if ((_e = bread((F)->lfs_ivnode, \
695 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \
696 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \
697 panic("lfs: ifile read %d", _e); \
698 if ((F)->lfs_version == 1) \
699 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + (IN) % (F)->lfs_ifpb); \
700 else \
701 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \
702 } while(0)
703
704 /* Read in the block with a specific segment usage entry from the ifile. */
705 #define LFS_SEGENTRY(SP, F, IN, BP) do { \
706 int _e; \
707 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
708 if ((_e = bread((F)->lfs_ivnode, \
709 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \
710 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \
711 panic("lfs: ifile read: %d", _e); \
712 if ((F)->lfs_version == 1) \
713 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \
714 ((IN) & ((F)->lfs_sepb - 1))); \
715 else \
716 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \
717 } while(0)
718
719 /* Determine if a buffer belongs to the ifile */
720 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM)
721
722 /*
723 * Structures used by lfs_bmapv and lfs_markv to communicate information
724 * about inodes and data blocks.
725 */
726 typedef struct block_info {
727 ino_t bi_inode; /* inode # */
728 ufs_daddr_t bi_lbn; /* logical block w/in file */
729 ufs_daddr_t bi_daddr; /* disk address of block */
730 u_int64_t bi_segcreate; /* origin segment create time */
731 int bi_version; /* file version number */
732 void *bi_bp; /* data buffer */
733 int bi_size; /* size of the block (if fragment) */
734 } BLOCK_INFO;
735
736 /* Compatibility for 1.5 binaries */
737 typedef struct block_info_15 {
738 ino_t bi_inode; /* inode # */
739 ufs_daddr_t bi_lbn; /* logical block w/in file */
740 ufs_daddr_t bi_daddr; /* disk address of block */
741 u_int32_t bi_segcreate; /* origin segment create time */
742 int bi_version; /* file version number */
743 void *bi_bp; /* data buffer */
744 int bi_size; /* size of the block (if fragment) */
745 } BLOCK_INFO_15;
746
747 /* In-memory description of a segment about to be written. */
748 struct segment {
749 struct lfs *fs; /* file system pointer */
750 struct buf **bpp; /* pointer to buffer array */
751 struct buf **cbpp; /* pointer to next available bp */
752 struct buf **start_bpp; /* pointer to first bp in this set */
753 struct buf *ibp; /* buffer pointer to inode page */
754 struct dinode *idp; /* pointer to ifile dinode */
755 struct finfo *fip; /* current fileinfo pointer */
756 struct vnode *vp; /* vnode being gathered */
757 void *segsum; /* segment summary info */
758 u_int32_t ninodes; /* number of inodes in this segment */
759 u_int32_t seg_bytes_left; /* bytes left in segment */
760 u_int32_t sum_bytes_left; /* bytes left in summary block */
761 u_int32_t seg_number; /* number of this segment */
762 ufs_daddr_t *start_lbp; /* beginning lbn for this set */
763
764 #define SEGM_CKP 0x01 /* doing a checkpoint */
765 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */
766 #define SEGM_SYNC 0x04 /* wait for segment */
767 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */
768 u_int16_t seg_flags; /* run-time flags for this segment */
769 u_int32_t seg_iocount; /* number of ios pending */
770 int ndupino; /* number of duplicate inodes */
771 };
772
773 struct lfs_cluster {
774 struct buf **bpp; /* Array of kept buffers */
775 int bufcount; /* Number of kept buffers */
776 size_t bufsize; /* Size of kept data */
777 #define LFS_CL_MALLOC 0x00000001
778 #define LFS_CL_SHIFT 0x00000002
779 #define LFS_CL_SYNC 0x00000004
780 u_int32_t flags; /* Flags */
781 struct lfs *fs; /* LFS that this belongs to */
782 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */
783 void *saveaddr; /* Original contents of saveaddr */
784 char *olddata; /* Original b_data, if LFS_CL_MALLOC */
785 };
786
787 /*
788 * Macros for determining free space on the disk, with the variable metadata
789 * of segment summaries and inode blocks taken into account.
790 */
791 /* Estimate number of clean blocks not available for writing */
792 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \
793 (int64_t)(F)->lfs_nclean) / \
794 ((F)->lfs_nseg - (F)->lfs_nclean)))
795
796 /* Estimate total size of the disk not including metadata */
797 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F))
798
799 /* Estimate number of blocks actually available for writing */
800 #define LFS_EST_BFREE(F) ((F)->lfs_bfree - LFS_EST_CMETA(F) - (F)->lfs_dmeta)
801
802 /* Amount of non-meta space not available to mortal man */
803 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \
804 (u_int64_t)(F)->lfs_minfree) / \
805 100)
806
807 /* Can credential C write BB blocks */
808 #define ISSPACE(F, BB, C) \
809 ((((C) == NOCRED || (C)->cr_uid == 0) && \
810 LFS_EST_BFREE(F) >= (BB)) || \
811 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB)))
812
813 /* Can an ordinary user write BB blocks */
814 #define IS_FREESPACE(F, BB) \
815 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F))
816
817 /* Statistics Counters */
818 struct lfs_stats {
819 u_int segsused;
820 u_int psegwrites;
821 u_int psyncwrites;
822 u_int pcleanwrites;
823 u_int blocktot;
824 u_int cleanblocks;
825 u_int ncheckpoints;
826 u_int nwrites;
827 u_int nsync_writes;
828 u_int wait_exceeded;
829 u_int write_exceeded;
830 u_int flush_invoked;
831 u_int vflush_invoked;
832 };
833 extern struct lfs_stats lfs_stats;
834