lfs.h revision 1.45 1 /* $NetBSD: lfs.h,v 1.45 2003/01/29 13:14:33 yamt 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 #ifndef _UFS_LFS_LFS_H_
74 #define _UFS_LFS_LFS_H_
75
76 /*
77 * Compile-time options for LFS.
78 */
79 #define LFS_IFIND_RETRIES 16
80 #define LFS_EAGAIN_FAIL /* markv fail with EAGAIN if ino is locked */
81 #define LFS_DEBUG_RFW /* print roll-forward debugging info */
82 #define LFS_NO_PAGEMOVE /* Use malloc/copy to write clusters */
83 #define LFS_AGGRESSIVE_SEGLOCK
84 #define LFS_LOGLENGTH 1024
85
86 /* #define DEBUG_LFS */ /* Intensive debugging of LFS subsystem */
87
88 #ifdef LFS_NO_PAGEMOVE
89 # define LFS_MALLOC_SUMMARY
90 #endif
91
92 /*
93 * Parameters and generic definitions
94 */
95 #define BW_CLEAN 1
96 #define MIN_FREE_SEGS 2
97 #define LFS_MAX_ACTIVE 10
98 #define LFS_MAXDIROP (desiredvnodes >> 2)
99 #ifndef LFS_ATIME_IFILE
100 # define LFS_ATIME_IFILE 0
101 #endif
102
103 /*
104 * #define WRITE_THRESHHOLD ((nbuf >> 1) - 10)
105 * #define WAIT_THRESHHOLD (nbuf - (nbuf >> 2) - 10)
106 */
107 #define LFS_MAX_BUFS ((nbuf >> 2) - 10)
108 #define LFS_WAIT_BUFS ((nbuf >> 1) - (nbuf >> 3) - 10)
109 /* These are new ... is LFS taking up too much memory in its buffers? */
110 #define LFS_MAX_BYTES (((bufpages >> 2) - 10) * NBPG)
111 #define LFS_WAIT_BYTES (((bufpages >> 1) - (bufpages >> 3) - 10) * NBPG)
112 #define LFS_BUFWAIT 2
113
114 #define LFS_LOCK_BUF(bp) do { \
115 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \
116 ++locked_queue_count; \
117 locked_queue_bytes += bp->b_bufsize; \
118 } \
119 (bp)->b_flags |= B_LOCKED; \
120 } while (0)
121
122 #define LFS_UNLOCK_BUF(bp) do { \
123 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \
124 --locked_queue_count; \
125 locked_queue_bytes -= bp->b_bufsize; \
126 if (locked_queue_count < LFS_WAIT_BUFS && \
127 locked_queue_bytes < LFS_WAIT_BYTES) \
128 wakeup(&locked_queue_count); \
129 } \
130 (bp)->b_flags &= ~B_LOCKED; \
131 } while (0)
132
133 #ifdef DEBUG_LOCKED_LIST
134 # define LFS_DEBUG_COUNTLOCKED(m) do { \
135 int _s; \
136 extern int locked_queue_count; \
137 extern long locked_queue_bytes; \
138 _s = splbio(); \
139 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \
140 splx(_s); \
141 wakeup(&locked_queue_count); \
142 } while (0)
143 #else
144 # define LFS_DEBUG_COUNTLOCKED(m)
145 #endif
146
147 /* For convenience */
148 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_ACCESSED|IN_CLEANING)
149
150 #define LFS_SET_UINO(ip, flags) do { \
151 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \
152 ++(ip)->i_lfs->lfs_uinodes; \
153 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \
154 ++(ip)->i_lfs->lfs_uinodes; \
155 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \
156 ++(ip)->i_lfs->lfs_uinodes; \
157 (ip)->i_flag |= (flags); \
158 } while (0)
159
160 #define LFS_CLR_UINO(ip, flags) do { \
161 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \
162 --(ip)->i_lfs->lfs_uinodes; \
163 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \
164 --(ip)->i_lfs->lfs_uinodes; \
165 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \
166 --(ip)->i_lfs->lfs_uinodes; \
167 (ip)->i_flag &= ~(flags); \
168 if ((ip)->i_lfs->lfs_uinodes < 0) { \
169 panic("lfs_uinodes < 0"); \
170 } \
171 } while (0)
172
173 #ifdef DEBUG
174 struct lfs_log_entry {
175 char *op;
176 char *file;
177 int line;
178 daddr_t block;
179 unsigned long flags;
180 };
181 extern int lfs_lognum;
182 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH];
183 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__)
184 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags) do { \
185 int _s; \
186 \
187 _s = splbio(); \
188 lfs_log[lfs_lognum].op = theop; \
189 lfs_log[lfs_lognum].file = thefile; \
190 lfs_log[lfs_lognum].line = (theline); \
191 lfs_log[lfs_lognum].block = (lbn); \
192 lfs_log[lfs_lognum].flags = (theflags); \
193 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \
194 splx(_s); \
195 } while (0)
196
197 # define LFS_BCLEAN_LOG(fs, bp) do { \
198 if ((bp)->b_vp == (fs)->lfs_ivnode) \
199 LFS_ENTER_LOG("clear", __FILE__, __LINE__, bp->b_lblkno, bp->b_flags); \
200 } while (0)
201 #else
202 # define LFS_BCLEAN_LOG(fs, bp)
203 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp))
204 #endif
205
206 #define LFS_ITIMES(ip, acc, mod, cre) do { \
207 struct lfs *_fs = (ip)->i_lfs; \
208 \
209 if ((ip)->i_flag & IN_ACCESS) { \
210 (ip)->i_ffs_atime = (acc)->tv_sec; \
211 (ip)->i_ffs_atimensec = (acc)->tv_nsec; \
212 if ((ip)->i_lfs->lfs_version > 1) { \
213 struct buf *ibp; \
214 IFILE *ifp; \
215 \
216 LFS_IENTRY(ifp, ip->i_lfs, ip->i_number, ibp); \
217 ifp->if_atime_sec = (acc)->tv_sec; \
218 ifp->if_atime_nsec = (acc)->tv_nsec; \
219 LFS_BWRITE_LOG(ibp); \
220 _fs->lfs_flags |= LFS_IFDIRTY; \
221 } else { \
222 LFS_SET_UINO(ip, IN_ACCESSED); \
223 } \
224 } \
225 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) { \
226 if ((ip)->i_flag & IN_UPDATE) { \
227 (ip)->i_ffs_mtime = (mod)->tv_sec; \
228 (ip)->i_ffs_mtimensec = (mod)->tv_nsec; \
229 (ip)->i_modrev++; \
230 } \
231 if ((ip)->i_flag & IN_CHANGE) { \
232 (ip)->i_ffs_ctime = (cre)->tv_sec; \
233 (ip)->i_ffs_ctimensec = (cre)->tv_nsec; \
234 } \
235 LFS_SET_UINO(ip, IN_MODIFIED); \
236 } \
237 (ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE); \
238 } while (0)
239
240 #define WRITEINPROG(vp) (vp->v_dirtyblkhd.lh_first && !(VTOI(vp)->i_flag & \
241 (IN_MODIFIED | IN_ACCESSED | IN_CLEANING)))
242
243 /* Here begins the berkeley code */
244
245 #define LFS_LABELPAD 8192 /* LFS label size */
246 #define LFS_SBPAD 8192 /* LFS superblock size */
247
248 /* On-disk and in-memory checkpoint segment usage structure. */
249 typedef struct segusage SEGUSE;
250 struct segusage {
251 u_int32_t su_nbytes; /* 0: number of live bytes */
252 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */
253 u_int16_t su_nsums; /* 8: number of summaries in segment */
254 u_int16_t su_ninos; /* 10: number of inode blocks in seg */
255
256 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */
257 #define SEGUSE_DIRTY 0x02 /* segment has data in it */
258 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */
259 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */
260 u_int32_t su_flags; /* 12: segment flags */
261 u_int64_t su_lastmod; /* 16: last modified timestamp */
262 };
263
264 typedef struct segusage_v1 SEGUSE_V1;
265 struct segusage_v1 {
266 u_int32_t su_nbytes; /* 0: number of live bytes */
267 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */
268 u_int16_t su_nsums; /* 8: number of summaries in segment */
269 u_int16_t su_ninos; /* 10: number of inode blocks in seg */
270 u_int32_t su_flags; /* 12: segment flags */
271 };
272
273 #define SEGUPB(fs) (fs->lfs_sepb)
274 #define SEGTABSIZE_SU(fs) \
275 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb)
276
277 /* On-disk file information. One per file with data blocks in the segment. */
278 typedef struct finfo FINFO;
279 struct finfo {
280 u_int32_t fi_nblocks; /* number of blocks */
281 u_int32_t fi_version; /* version number */
282 u_int32_t fi_ino; /* inode number */
283 u_int32_t fi_lastlength; /* length of last block in array */
284 int32_t fi_blocks[1]; /* array of logical block numbers */
285 };
286
287 /* sizeof FINFO except fi_blocks */
288 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t))
289
290 /* On-disk super block. */
291 struct dlfs {
292 #define LFS_MAGIC 0x070162
293 u_int32_t dlfs_magic; /* 0: magic number */
294 #define LFS_VERSION 2
295 u_int32_t dlfs_version; /* 4: version number */
296
297 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */
298 /* number of frags in fs (v2) */
299 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */
300 /* number of bytes per segment (v2) */
301 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */
302 u_int32_t dlfs_bsize; /* 20: file system block size */
303 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */
304 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */
305
306 /* Checkpoint region. */
307 u_int32_t dlfs_free; /* 32: start of the free list */
308 u_int32_t dlfs_bfree; /* 36: number of free disk blocks */
309 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */
310 int32_t dlfs_avail; /* 44: blocks available for writing */
311 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */
312 int32_t dlfs_idaddr; /* 52: inode file disk address */
313 u_int32_t dlfs_ifile; /* 56: inode file inode number */
314 int32_t dlfs_lastseg; /* 60: address of last segment written */
315 int32_t dlfs_nextseg; /* 64: address of next segment to write */
316 int32_t dlfs_curseg; /* 68: current segment being written */
317 int32_t dlfs_offset; /* 72: offset in curseg for next partial */
318 int32_t dlfs_lastpseg; /* 76: address of last partial written */
319 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */
320 #define dlfs_otstamp dlfs_inopf
321
322 /* These are configuration parameters. */
323 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */
324
325 /* These fields can be computed from the others. */
326 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */
327 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */
328 u_int32_t dlfs_inopb; /* 100: inodes per block */
329 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */
330 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */
331 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */
332 u_int32_t dlfs_nseg; /* 116: number of segments */
333 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */
334 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */
335 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */
336 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */
337 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */
338 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */
339 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */
340 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */
341 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */
342 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */
343 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */
344 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */
345 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */
346
347 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */
348 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */
349 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */
350 int32_t dlfs_sboffs[LFS_MAXNUMSB];
351
352 u_int32_t dlfs_nclean; /* 228: Number of clean segments */
353 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */
354 #define LFS_PF_CLEAN 0x1
355 u_int16_t dlfs_pflags; /* 322: file system persistent flags */
356 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */
357 u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */
358 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */
359 u_int64_t dlfs_serial; /* 336: serial number */
360 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */
361 int32_t dlfs_start; /* 348: start of segment 0 */
362 u_int64_t dlfs_tstamp; /* 352: time stamp */
363 #define LFS_44INODEFMT 0
364 #define LFS_MAXINODEFMT 0
365 u_int32_t dlfs_inodefmt; /* 360: inode format version */
366 u_int32_t dlfs_interleave; /* 364: segment interleave */
367 u_int32_t dlfs_ident; /* 368: per-fs identifier */
368 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */
369 int8_t dlfs_pad[132]; /* 376: round to 512 bytes */
370 /* Checksum -- last valid disk field. */
371 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */
372 };
373
374 /* Maximum number of io's we can have pending at once */
375 #define LFS_THROTTLE 32 /* XXX should be better paramtrized - ? */
376
377 /* In-memory super block. */
378 struct lfs {
379 struct dlfs lfs_dlfs; /* on-disk parameters */
380 #define lfs_magic lfs_dlfs.dlfs_magic
381 #define lfs_version lfs_dlfs.dlfs_version
382 #define lfs_size lfs_dlfs.dlfs_size
383 #define lfs_ssize lfs_dlfs.dlfs_ssize
384 #define lfs_dsize lfs_dlfs.dlfs_dsize
385 #define lfs_bsize lfs_dlfs.dlfs_bsize
386 #define lfs_fsize lfs_dlfs.dlfs_fsize
387 #define lfs_frag lfs_dlfs.dlfs_frag
388 #define lfs_free lfs_dlfs.dlfs_free
389 #define lfs_bfree lfs_dlfs.dlfs_bfree
390 #define lfs_nfiles lfs_dlfs.dlfs_nfiles
391 #define lfs_avail lfs_dlfs.dlfs_avail
392 #define lfs_uinodes lfs_dlfs.dlfs_uinodes
393 #define lfs_idaddr lfs_dlfs.dlfs_idaddr
394 #define lfs_ifile lfs_dlfs.dlfs_ifile
395 #define lfs_lastseg lfs_dlfs.dlfs_lastseg
396 #define lfs_nextseg lfs_dlfs.dlfs_nextseg
397 #define lfs_curseg lfs_dlfs.dlfs_curseg
398 #define lfs_offset lfs_dlfs.dlfs_offset
399 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg
400 #define lfs_otstamp lfs_dlfs.dlfs_inopf
401 #define lfs_inopf lfs_dlfs.dlfs_inopf
402 #define lfs_minfree lfs_dlfs.dlfs_minfree
403 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize
404 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg
405 #define lfs_inopb lfs_dlfs.dlfs_inopb
406 #define lfs_ifpb lfs_dlfs.dlfs_ifpb
407 #define lfs_sepb lfs_dlfs.dlfs_sepb
408 #define lfs_nindir lfs_dlfs.dlfs_nindir
409 #define lfs_nseg lfs_dlfs.dlfs_nseg
410 #define lfs_nspf lfs_dlfs.dlfs_nspf
411 #define lfs_cleansz lfs_dlfs.dlfs_cleansz
412 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz
413 #define lfs_segmask lfs_dlfs.dlfs_segmask
414 #define lfs_segshift lfs_dlfs.dlfs_segshift
415 #define lfs_bmask lfs_dlfs.dlfs_bmask
416 #define lfs_bshift lfs_dlfs.dlfs_bshift
417 #define lfs_ffmask lfs_dlfs.dlfs_ffmask
418 #define lfs_ffshift lfs_dlfs.dlfs_ffshift
419 #define lfs_fbmask lfs_dlfs.dlfs_fbmask
420 #define lfs_fbshift lfs_dlfs.dlfs_fbshift
421 #define lfs_blktodb lfs_dlfs.dlfs_blktodb
422 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb
423 #define lfs_sushift lfs_dlfs.dlfs_sushift
424 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen
425 #define lfs_sboffs lfs_dlfs.dlfs_sboffs
426 #define lfs_cksum lfs_dlfs.dlfs_cksum
427 #define lfs_pflags lfs_dlfs.dlfs_pflags
428 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt
429 #define lfs_nclean lfs_dlfs.dlfs_nclean
430 #define lfs_dmeta lfs_dlfs.dlfs_dmeta
431 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg
432 #define lfs_sumsize lfs_dlfs.dlfs_sumsize
433 #define lfs_serial lfs_dlfs.dlfs_serial
434 #define lfs_ibsize lfs_dlfs.dlfs_ibsize
435 #define lfs_start lfs_dlfs.dlfs_start
436 #define lfs_tstamp lfs_dlfs.dlfs_tstamp
437 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt
438 #define lfs_interleave lfs_dlfs.dlfs_interleave
439 #define lfs_ident lfs_dlfs.dlfs_ident
440
441 /* These fields are set at mount time and are meaningless on disk. */
442 struct segment *lfs_sp; /* current segment being written */
443 struct vnode *lfs_ivnode; /* vnode for the ifile */
444 u_int32_t lfs_seglock; /* single-thread the segment writer */
445 pid_t lfs_lockpid; /* pid of lock holder */
446 u_int32_t lfs_iocount; /* number of ios pending */
447 u_int32_t lfs_writer; /* don't allow any dirops to start */
448 u_int32_t lfs_dirops; /* count of active directory ops */
449 u_int32_t lfs_doifile; /* Write ifile blocks on next write */
450 u_int32_t lfs_nactive; /* Number of segments since last ckp */
451 int8_t lfs_fmod; /* super block modified flag */
452 int8_t lfs_ronly; /* mounted read-only flag */
453 #define LFS_NOTYET 0x01
454 #define LFS_IFDIRTY 0x02
455 #define LFS_WARNED 0x04
456 int8_t lfs_flags; /* currently unused flag */
457 u_int16_t lfs_activesb; /* toggle between superblocks */
458 #ifdef LFS_TRACK_IOS
459 daddr_t lfs_pending[LFS_THROTTLE]; /* daddrs of pending writes */
460 #endif /* LFS_TRACK_IOS */
461 daddr_t lfs_sbactive; /* disk address of in-progress sb write */
462 struct vnode *lfs_flushvp; /* vnode being flushed */
463 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */
464 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */
465 size_t lfs_devbsize; /* Device block size */
466 size_t lfs_devbshift; /* Device block shift */
467 struct lock lfs_freelock;
468 struct lock lfs_fraglock;
469 pid_t lfs_rfpid; /* Process ID of roll-forward agent */
470 int lfs_nadirop; /* number of active dirop nodes */
471 long lfs_ravail; /* blocks pre-reserved for writing */
472 };
473
474 /*
475 * Inode 0: out-of-band inode number
476 * Inode 1: IFILE inode number
477 * Inode 2: root inode
478 * Inode 3: lost+found inode number
479 */
480 #define LFS_UNUSED_INUM 0 /* out of band inode number */
481 #define LFS_IFILE_INUM 1 /* IFILE inode number */
482 #define LOSTFOUNDINO 3 /* lost+found inode number */
483 #define LFS_FIRST_INUM 4 /* first free inode number */
484
485 /* Address calculations for metadata located in the inode */
486 #define S_INDIR(fs) -NDADDR
487 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1)
488 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1)
489
490 /* Unassigned disk addresses. */
491 #define UNASSIGNED -1
492 #define UNWRITTEN -2
493
494 /* Unused logical block number */
495 #define LFS_UNUSED_LBN -1
496
497 typedef struct ifile IFILE;
498 struct ifile {
499 u_int32_t if_version; /* inode version number */
500 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */
501 int32_t if_daddr; /* inode disk address */
502 ino_t if_nextfree; /* next-unallocated inode */
503 /* XXX - when inode format changes, this changes too */
504 u_int32_t if_atime_sec; /* Last access time, seconds */
505 u_int32_t if_atime_nsec; /* and nanoseconds */
506 };
507
508 typedef struct ifile_v1 IFILE_V1;
509 struct ifile_v1 {
510 u_int32_t if_version; /* inode version number */
511 int32_t if_daddr; /* inode disk address */
512 ino_t if_nextfree; /* next-unallocated inode */
513 #if LFS_ATIME_IFILE
514 struct timespec if_atime; /* Last access time */
515 #endif
516 };
517
518 /*
519 * Cleaner information structure. This resides in the ifile and is used
520 * to pass information between the cleaner and the kernel.
521 */
522 typedef struct _cleanerinfo {
523 u_int32_t clean; /* number of clean segments */
524 u_int32_t dirty; /* number of dirty segments */
525 u_int32_t bfree; /* disk blocks free */
526 int32_t avail; /* disk blocks available */
527 u_int32_t free_head; /* head of the inode free list */
528 u_int32_t free_tail; /* tail of the inode free list */
529 } CLEANERINFO;
530
531 #define CLEANSIZE_SU(fs) \
532 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift)
533
534 /*
535 * All summary blocks are the same size, so we can always read a summary
536 * block easily from a segment.
537 */
538 #define LFS_V1_SUMMARY_SIZE 512
539 #define LFS_DFL_SUMMARY_SIZE 512
540
541 /* On-disk segment summary information */
542 typedef struct segsum_v1 SEGSUM_V1;
543 struct segsum_v1 {
544 u_int32_t ss_sumsum; /* 0: check sum of summary block */
545 u_int32_t ss_datasum; /* 4: check sum of data */
546 u_int32_t ss_magic; /* 8: segment summary magic number */
547 #define SS_MAGIC 0x061561
548 int32_t ss_next; /* 12: next segment */
549 u_int32_t ss_create; /* 16: creation time stamp */
550 u_int16_t ss_nfinfo; /* 20: number of file info structures */
551 u_int16_t ss_ninos; /* 22: number of inodes in summary */
552
553 #define SS_DIROP 0x01 /* segment begins a dirop */
554 #define SS_CONT 0x02 /* more partials to finish this write*/
555 u_int16_t ss_flags; /* 24: used for directory operations */
556 u_int16_t ss_pad; /* 26: extra space */
557 /* FINFO's and inode daddr's... */
558 };
559
560 typedef struct segsum SEGSUM;
561 struct segsum {
562 u_int32_t ss_sumsum; /* 0: check sum of summary block */
563 u_int32_t ss_datasum; /* 4: check sum of data */
564 u_int32_t ss_magic; /* 8: segment summary magic number */
565 int32_t ss_next; /* 12: next segment */
566 u_int32_t ss_ident; /* 16: roll-forward fsid */
567 #define ss_ocreate ss_ident /* ident is where create was in v1 */
568 u_int16_t ss_nfinfo; /* 20: number of file info structures */
569 u_int16_t ss_ninos; /* 22: number of inodes in summary */
570 u_int16_t ss_flags; /* 24: used for directory operations */
571 u_int8_t ss_pad[6]; /* 26: extra space */
572 u_int64_t ss_serial; /* 32: serial number */
573 u_int64_t ss_create; /* 40: time stamp */
574 /* FINFO's and inode daddr's... */
575 };
576
577 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM))
578
579 /* NINDIR is the number of indirects in a file system block. */
580 #define NINDIR(fs) ((fs)->lfs_nindir)
581
582 /* INOPB is the number of inodes in a secondary storage block. */
583 #define INOPB(fs) ((fs)->lfs_inopb)
584 /* INOPF is the number of inodes in a fragment. */
585 #define INOPF(fs) ((fs)->lfs_inopf)
586
587 #define blksize(fs, ip, lbn) \
588 (((lbn) >= NDADDR || (ip)->i_ffs_size >= ((lbn) + 1) << (fs)->lfs_bshift) \
589 ? (fs)->lfs_bsize \
590 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs_size))))
591 #define blkoff(fs, loc) ((int)(loc) & (fs)->lfs_bmask)
592 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \
593 ((int)((loc) & (fs)->lfs_ffmask))
594 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb)
595 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb)
596 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift))
597 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift))
598 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift)
599 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift)
600 /* Same as above, but named like dbtob(), btodb() */
601 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \
602 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb))
603 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \
604 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb))
605 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \
606 (fs)->lfs_fsbtodb))
607 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \
608 (fs)->lfs_fsbtodb))
609 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift)
610 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \
611 ((loc) >> (fs)->lfs_ffshift)
612 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \
613 ((int)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask)))
614 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \
615 ((int)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask)))
616 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \
617 ((frags) >> (fs)->lfs_fbshift)
618 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \
619 ((blks) << (fs)->lfs_fbshift)
620 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \
621 ((fsb) & ((fs)->lfs_frag - 1))
622 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \
623 ((fsb) &~ ((fs)->lfs_frag - 1))
624 #define dblksize(fs, dip, lbn) \
625 (((lbn) >= NDADDR || (dip)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\
626 ? (fs)->lfs_bsize \
627 : (fragroundup(fs, blkoff(fs, (dip)->di_size))))
628
629 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \
630 (fs)->lfs_ssize << (fs)->lfs_blktodb : \
631 btofsb((fs), (fs)->lfs_ssize)) * (seg))
632 #define dtosn(fs, daddr) /* block address to segment number */ \
633 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1)))
634 #define sntod(fs, sn) /* segment number to disk address */ \
635 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start))
636
637 /* Read in the block with the cleaner info from the ifile. */
638 #define LFS_CLEANERINFO(CP, F, BP) do { \
639 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
640 if (bread((F)->lfs_ivnode, \
641 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \
642 panic("lfs: ifile read"); \
643 (CP) = (CLEANERINFO *)(BP)->b_data; \
644 } while(0)
645
646 /* Synchronize the Ifile cleaner info with current avail and bfree */
647 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \
648 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \
649 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail) { \
650 (cip)->bfree = (fs)->lfs_bfree; \
651 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail; \
652 if (((bp)->b_flags & B_GATHERED) == 0) \
653 (fs)->lfs_flags |= LFS_IFDIRTY; \
654 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \
655 } else \
656 brelse(bp); \
657 } while (0)
658
659 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \
660 if ((FS)->lfs_version > 1) { \
661 LFS_CLEANERINFO((CIP), (FS), (BP)); \
662 (FS)->lfs_free = (CIP)->free_head; \
663 brelse(BP); \
664 } \
665 *(FREEP) = (FS)->lfs_free; \
666 } while (0)
667
668 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \
669 (FS)->lfs_free = (VAL); \
670 if ((FS)->lfs_version > 1) { \
671 LFS_CLEANERINFO((CIP), (FS), (BP)); \
672 (CIP)->free_head = (VAL); \
673 LFS_BWRITE_LOG(BP); \
674 (FS)->lfs_flags |= LFS_IFDIRTY; \
675 } \
676 } while (0)
677
678 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \
679 LFS_CLEANERINFO((CIP), (FS), (BP)); \
680 *(FREEP) = (CIP)->free_tail; \
681 brelse(BP); \
682 } while (0)
683
684 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \
685 LFS_CLEANERINFO((CIP), (FS), (BP)); \
686 (CIP)->free_tail = (VAL); \
687 LFS_BWRITE_LOG(BP); \
688 (FS)->lfs_flags |= LFS_IFDIRTY; \
689 } while (0)
690
691 /*
692 * XXX - v1 compatibility code is not allowed to touch if_atime, since it
693 * may not be mapped!
694 */
695 /* Read in the block with a specific inode from the ifile. */
696 #define LFS_IENTRY(IP, F, IN, BP) do { \
697 int _e; \
698 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
699 if ((_e = bread((F)->lfs_ivnode, \
700 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \
701 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \
702 panic("lfs: ifile read %d", _e); \
703 if ((F)->lfs_version == 1) \
704 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + (IN) % (F)->lfs_ifpb); \
705 else \
706 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \
707 } while(0)
708
709 /* Read in the block with a specific segment usage entry from the ifile. */
710 #define LFS_SEGENTRY(SP, F, IN, BP) do { \
711 int _e; \
712 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
713 if ((_e = bread((F)->lfs_ivnode, \
714 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \
715 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \
716 panic("lfs: ifile read: %d", _e); \
717 if ((F)->lfs_version == 1) \
718 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \
719 ((IN) & ((F)->lfs_sepb - 1))); \
720 else \
721 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \
722 } while(0)
723
724 /* Determine if a buffer belongs to the ifile */
725 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM)
726
727 /*
728 * Structures used by lfs_bmapv and lfs_markv to communicate information
729 * about inodes and data blocks.
730 */
731 typedef struct block_info {
732 ino_t bi_inode; /* inode # */
733 /* XXX ondisk32 */
734 int32_t bi_lbn; /* logical block w/in file */
735 int32_t bi_daddr; /* disk address of block */
736 u_int64_t bi_segcreate; /* origin segment create time */
737 int bi_version; /* file version number */
738 void *bi_bp; /* data buffer */
739 int bi_size; /* size of the block (if fragment) */
740 } BLOCK_INFO;
741
742 /* Compatibility for 1.5 binaries */
743 typedef struct block_info_15 {
744 ino_t bi_inode; /* inode # */
745 /* XXX ondisk32 */
746 int32_t bi_lbn; /* logical block w/in file */
747 int32_t bi_daddr; /* disk address of block */
748 u_int32_t bi_segcreate; /* origin segment create time */
749 int bi_version; /* file version number */
750 void *bi_bp; /* data buffer */
751 int bi_size; /* size of the block (if fragment) */
752 } BLOCK_INFO_15;
753
754 /* In-memory description of a segment about to be written. */
755 struct segment {
756 struct lfs *fs; /* file system pointer */
757 struct buf **bpp; /* pointer to buffer array */
758 struct buf **cbpp; /* pointer to next available bp */
759 struct buf **start_bpp; /* pointer to first bp in this set */
760 struct buf *ibp; /* buffer pointer to inode page */
761 struct dinode *idp; /* pointer to ifile dinode */
762 struct finfo *fip; /* current fileinfo pointer */
763 struct vnode *vp; /* vnode being gathered */
764 void *segsum; /* segment summary info */
765 u_int32_t ninodes; /* number of inodes in this segment */
766 u_int32_t seg_bytes_left; /* bytes left in segment */
767 u_int32_t sum_bytes_left; /* bytes left in summary block */
768 u_int32_t seg_number; /* number of this segment */
769 /* XXX ondisk32 */
770 int32_t *start_lbp; /* beginning lbn for this set */
771
772 #define SEGM_CKP 0x01 /* doing a checkpoint */
773 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */
774 #define SEGM_SYNC 0x04 /* wait for segment */
775 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */
776 u_int16_t seg_flags; /* run-time flags for this segment */
777 u_int32_t seg_iocount; /* number of ios pending */
778 int ndupino; /* number of duplicate inodes */
779 };
780
781 struct lfs_cluster {
782 struct buf **bpp; /* Array of kept buffers */
783 int bufcount; /* Number of kept buffers */
784 size_t bufsize; /* Size of kept data */
785 #define LFS_CL_MALLOC 0x00000001
786 #define LFS_CL_SHIFT 0x00000002
787 #define LFS_CL_SYNC 0x00000004
788 u_int32_t flags; /* Flags */
789 struct lfs *fs; /* LFS that this belongs to */
790 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */
791 void *saveaddr; /* Original contents of saveaddr */
792 char *olddata; /* Original b_data, if LFS_CL_MALLOC */
793 };
794
795 /*
796 * Macros for determining free space on the disk, with the variable metadata
797 * of segment summaries and inode blocks taken into account.
798 */
799 /* Estimate number of clean blocks not available for writing */
800 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \
801 (int64_t)(F)->lfs_nclean) / \
802 ((F)->lfs_nseg - (F)->lfs_nclean)))
803
804 /* Estimate total size of the disk not including metadata */
805 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F))
806
807 /* Estimate number of blocks actually available for writing */
808 #define LFS_EST_BFREE(F) ((F)->lfs_bfree - LFS_EST_CMETA(F) - (F)->lfs_dmeta)
809
810 /* Amount of non-meta space not available to mortal man */
811 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \
812 (u_int64_t)(F)->lfs_minfree) / \
813 100)
814
815 /* Can credential C write BB blocks */
816 #define ISSPACE(F, BB, C) \
817 ((((C) == NOCRED || (C)->cr_uid == 0) && \
818 LFS_EST_BFREE(F) >= (BB)) || \
819 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB)))
820
821 /* Can an ordinary user write BB blocks */
822 #define IS_FREESPACE(F, BB) \
823 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F))
824
825 /* Statistics Counters */
826 struct lfs_stats {
827 u_int segsused;
828 u_int psegwrites;
829 u_int psyncwrites;
830 u_int pcleanwrites;
831 u_int blocktot;
832 u_int cleanblocks;
833 u_int ncheckpoints;
834 u_int nwrites;
835 u_int nsync_writes;
836 u_int wait_exceeded;
837 u_int write_exceeded;
838 u_int flush_invoked;
839 u_int vflush_invoked;
840 };
841 #ifdef _KERNEL
842 extern struct lfs_stats lfs_stats;
843 #endif
844
845 #endif /* !_UFS_LFS_LFS_H_ */
846