lfs.h revision 1.9 1 /* $NetBSD: lfs.h,v 1.9 1998/03/01 02:23:23 fvdl Exp $ */
2
3 /*-
4 * Copyright (c) 1991, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. All advertising materials mentioning features or use of this software
16 * must display the following acknowledgement:
17 * This product includes software developed by the University of
18 * California, Berkeley and its contributors.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 * @(#)lfs.h 8.9 (Berkeley) 5/8/95
36 */
37
38 #define LFS_LABELPAD 8192 /* LFS label size */
39 #define LFS_SBPAD 8192 /* LFS superblock size */
40
41 /*
42 * XXX
43 * This is a kluge and NEEDS to go away.
44 *
45 * Right now, ufs code handles most of the calls for directory operations
46 * such as create, mkdir, link, etc. As a result VOP_UPDATE is being
47 * called with waitfor set (since ffs does these things synchronously).
48 * Since LFS does not want to do these synchronously, we treat the last
49 * argument to lfs_update as a set of flags. If LFS_SYNC is set, then
50 * the update should be synchronous, if not, do it asynchronously.
51 * Unfortunately, this means that LFS won't work with NFS yet because
52 * NFS goes through paths that will make normal calls to ufs which will
53 * call lfs with a last argument of 1.
54 */
55 #define LFS_SYNC 0x02
56
57 /* On-disk and in-memory checkpoint segment usage structure. */
58 typedef struct segusage SEGUSE;
59 struct segusage {
60 u_int32_t su_nbytes; /* number of live bytes */
61 u_int32_t su_lastmod; /* SEGUSE last modified timestamp */
62 u_int16_t su_nsums; /* number of summaries in segment */
63 u_int16_t su_ninos; /* number of inode blocks in seg */
64
65 #define SEGUSE_ACTIVE 0x01 /* segment is currently being written */
66 #define SEGUSE_DIRTY 0x02 /* segment has data in it */
67 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */
68 u_int32_t su_flags;
69 };
70
71 #define SEGUPB(fs) (1 << (fs)->lfs_sushift)
72 #define SEGTABSIZE_SU(fs) \
73 (((fs)->lfs_nseg + SEGUPB(fs) - 1) >> (fs)->lfs_sushift)
74
75 /* On-disk file information. One per file with data blocks in the segment. */
76 typedef struct finfo FINFO;
77 struct finfo {
78 u_int32_t fi_nblocks; /* number of blocks */
79 u_int32_t fi_version; /* version number */
80 u_int32_t fi_ino; /* inode number */
81 u_int32_t fi_lastlength; /* length of last block in array */
82 ufs_daddr_t fi_blocks[1]; /* array of logical block numbers */
83 };
84
85 /* On-disk and in-memory super block. */
86 struct lfs {
87 #define LFS_MAGIC 0x070162
88 u_int32_t lfs_magic; /* magic number */
89 #define LFS_VERSION 1
90 u_int32_t lfs_version; /* version number */
91
92 u_int32_t lfs_size; /* number of blocks in fs */
93 u_int32_t lfs_ssize; /* number of blocks per segment */
94 u_int32_t lfs_dsize; /* number of disk blocks in fs */
95 u_int32_t lfs_bsize; /* file system block size */
96 u_int32_t lfs_fsize; /* size of frag blocks in fs */
97 u_int32_t lfs_frag; /* number of frags in a block in fs */
98
99 /* Checkpoint region. */
100 ino_t lfs_free; /* start of the free list */
101 u_int32_t lfs_bfree; /* number of free disk blocks */
102 u_int32_t lfs_nfiles; /* number of allocated inodes */
103 int32_t lfs_avail; /* blocks available for writing */
104 u_int32_t lfs_uinodes; /* inodes in cache not yet on disk */
105 ufs_daddr_t lfs_idaddr; /* inode file disk address */
106 ino_t lfs_ifile; /* inode file inode number */
107 ufs_daddr_t lfs_lastseg; /* address of last segment written */
108 ufs_daddr_t lfs_nextseg; /* address of next segment to write */
109 ufs_daddr_t lfs_curseg; /* current segment being written */
110 ufs_daddr_t lfs_offset; /* offset in curseg for next partial */
111 ufs_daddr_t lfs_lastpseg; /* address of last partial written */
112 u_int32_t lfs_tstamp; /* time stamp */
113
114 /* These are configuration parameters. */
115 u_int32_t lfs_minfree; /* minimum percentage of free blocks */
116
117 /* These fields can be computed from the others. */
118 u_int64_t lfs_maxfilesize; /* maximum representable file size */
119 u_int32_t lfs_dbpseg; /* disk blocks per segment */
120 u_int32_t lfs_inopb; /* inodes per block */
121 u_int32_t lfs_ifpb; /* IFILE entries per block */
122 u_int32_t lfs_sepb; /* SEGUSE entries per block */
123 u_int32_t lfs_nindir; /* indirect pointers per block */
124 u_int32_t lfs_nseg; /* number of segments */
125 u_int32_t lfs_nspf; /* number of sectors per fragment */
126 u_int32_t lfs_cleansz; /* cleaner info size in blocks */
127 u_int32_t lfs_segtabsz; /* segment table size in blocks */
128
129 u_int32_t lfs_segmask; /* calculate offset within a segment */
130 u_int32_t lfs_segshift; /* fast mult/div for segments */
131 u_int64_t lfs_bmask; /* calc block offset from file offset */
132 u_int32_t lfs_bshift; /* calc block number from file offset */
133 u_int64_t lfs_ffmask; /* calc frag offset from file offset */
134 u_int32_t lfs_ffshift; /* fast mult/div for frag from file */
135 u_int64_t lfs_fbmask; /* calc frag offset from block offset */
136 u_int32_t lfs_fbshift; /* fast mult/div for frag from block */
137 u_int32_t lfs_fsbtodb; /* fsbtodb and dbtofsb shift constant */
138 u_int32_t lfs_sushift; /* fast mult/div for segusage table */
139
140 int32_t lfs_maxsymlinklen; /* max length of an internal symlink */
141
142 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */
143 #define LFS_MAXNUMSB 10 /* superblock disk offsets */
144 ufs_daddr_t lfs_sboffs[LFS_MAXNUMSB];
145
146 /* Checksum -- last valid disk field. */
147 u_int32_t lfs_cksum; /* checksum for superblock checking */
148
149 /* These fields are set at mount time and are meaningless on disk. */
150 struct segment *lfs_sp; /* current segment being written */
151 struct vnode *lfs_ivnode; /* vnode for the ifile */
152 u_int32_t lfs_seglock; /* single-thread the segment writer */
153 pid_t lfs_lockpid; /* pid of lock holder */
154 u_int32_t lfs_iocount; /* number of ios pending */
155 u_int32_t lfs_writer; /* don't allow any dirops to start */
156 u_int32_t lfs_dirops; /* count of active directory ops */
157 u_int32_t lfs_doifile; /* Write ifile blocks on next write */
158 u_int32_t lfs_nactive; /* Number of segments since last ckp */
159 int8_t lfs_fmod; /* super block modified flag */
160 int8_t lfs_clean; /* file system is clean flag */
161 int8_t lfs_ronly; /* mounted read-only flag */
162 int8_t lfs_flags; /* currently unused flag */
163 u_char lfs_fsmnt[MNAMELEN]; /* name mounted on */
164
165 int32_t lfs_pad[40]; /* round to 512 bytes */
166 };
167
168 /*
169 * Inode 0: out-of-band inode number
170 * Inode 1: IFILE inode number
171 * Inode 2: root inode
172 * Inode 3: lost+found inode number
173 */
174 #define LFS_UNUSED_INUM 0 /* out of band inode number */
175 #define LFS_IFILE_INUM 1 /* IFILE inode number */
176 #define LOSTFOUNDINO 3 /* lost+found inode number */
177 #define LFS_FIRST_INUM 4 /* first free inode number */
178
179 /* Address calculations for metadata located in the inode */
180 #define S_INDIR(fs) -NDADDR
181 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1)
182 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1)
183
184 /* Unassigned disk address. */
185 #define UNASSIGNED -1
186
187 /* Unused logical block number */
188 #define LFS_UNUSED_LBN -1
189
190 typedef struct ifile IFILE;
191 struct ifile {
192 u_int32_t if_version; /* inode version number */
193 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */
194 ufs_daddr_t if_daddr; /* inode disk address */
195 ino_t if_nextfree; /* next-unallocated inode */
196 };
197
198 /*
199 * Cleaner information structure. This resides in the ifile and is used
200 * to pass information between the cleaner and the kernel.
201 */
202 typedef struct _cleanerinfo {
203 u_int32_t clean; /* K: number of clean segments */
204 u_int32_t dirty; /* K: number of dirty segments */
205 } CLEANERINFO;
206
207 #define CLEANSIZE_SU(fs) \
208 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift)
209
210 /*
211 * All summary blocks are the same size, so we can always read a summary
212 * block easily from a segment.
213 */
214 #define LFS_SUMMARY_SIZE 512
215
216 /* On-disk segment summary information */
217 typedef struct segsum SEGSUM;
218 struct segsum {
219 u_int32_t ss_sumsum; /* check sum of summary block */
220 u_int32_t ss_datasum; /* check sum of data */
221 u_int32_t ss_magic; /* segment summary magic number */
222 #define SS_MAGIC 0x061561
223 ufs_daddr_t ss_next; /* next segment */
224 u_int32_t ss_create; /* creation time stamp */
225 u_int16_t ss_nfinfo; /* number of file info structures */
226 u_int16_t ss_ninos; /* number of inodes in summary */
227
228 #define SS_DIROP 0x01 /* segment begins a dirop */
229 #define SS_CONT 0x02 /* more partials to finish this write*/
230 u_int16_t ss_flags; /* used for directory operations */
231 u_int16_t ss_pad; /* extra space */
232 /* FINFO's and inode daddr's... */
233 };
234
235 /* NINDIR is the number of indirects in a file system block. */
236 #define NINDIR(fs) ((fs)->lfs_nindir)
237
238 /* INOPB is the number of inodes in a secondary storage block. */
239 #define INOPB(fs) ((fs)->lfs_inopb)
240
241 #define blksize(fs, ip, lbn) \
242 (((lbn) >= NDADDR || (ip)->i_ffs_size >= ((lbn) + 1) << (fs)->lfs_bshift) \
243 ? (fs)->lfs_bsize \
244 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs_size))))
245 #define blkoff(fs, loc) ((int)(loc) & (fs)->lfs_bmask)
246 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \
247 ((int)((loc) & (fs)->lfs_ffmask))
248 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb)
249 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb)
250 #define fragstodb(fs, b) ((b) << ((fs)->lfs_fsbtodb - (fs)->lfs_fbshift))
251 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_fsbtodb - (fs)->lfs_fbshift))
252 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift)
253 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift)
254 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \
255 ((loc) >> (fs)->lfs_ffshift)
256 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \
257 ((int)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask)))
258 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \
259 ((int)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask)))
260 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \
261 ((frags) >> (fs)->lfs_fbshift)
262 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \
263 ((blks) << (fs)->lfs_fbshift)
264 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \
265 ((fsb) & ((fs)->lfs_frag - 1))
266 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \
267 ((fsb) &~ ((fs)->lfs_frag - 1))
268 #define dblksize(fs, dip, lbn) \
269 (((lbn) >= NDADDR || (dip)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\
270 ? (fs)->lfs_bsize \
271 : (fragroundup(fs, blkoff(fs, (dip)->di_size))))
272 #define datosn(fs, daddr) /* disk address to segment number */ \
273 (((daddr) - (fs)->lfs_sboffs[0]) / fsbtodb((fs), (fs)->lfs_ssize))
274 #define sntoda(fs, sn) /* segment number to disk address */ \
275 ((ufs_daddr_t)((sn) * ((fs)->lfs_ssize << (fs)->lfs_fsbtodb) + \
276 (fs)->lfs_sboffs[0]))
277
278 /* Read in the block with the cleaner info from the ifile. */
279 #define LFS_CLEANERINFO(CP, F, BP) { \
280 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
281 if (bread((F)->lfs_ivnode, \
282 (ufs_daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \
283 panic("lfs: ifile read"); \
284 (CP) = (CLEANERINFO *)(BP)->b_data; \
285 }
286
287 /* Read in the block with a specific inode from the ifile. */
288 #define LFS_IENTRY(IP, F, IN, BP) { \
289 int _e; \
290 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
291 if ((_e = bread((F)->lfs_ivnode, \
292 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz,\
293 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \
294 panic("lfs: ifile read %d", _e); \
295 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \
296 }
297
298 /* Read in the block with a specific segment usage entry from the ifile. */
299 #define LFS_SEGENTRY(SP, F, IN, BP) { \
300 int _e; \
301 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
302 if ((_e = bread((F)->lfs_ivnode, \
303 ((IN) >> (F)->lfs_sushift) + (F)->lfs_cleansz, \
304 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \
305 panic("lfs: ifile read: %d", _e); \
306 (SP) = (SEGUSE *)(BP)->b_data + ((IN) & ((F)->lfs_sepb - 1)); \
307 }
308
309 /*
310 * Determine if there is enough room currently available to write db
311 * disk blocks. We need enough blocks for the new blocks, the current,
312 * inode blocks, a summary block, plus potentially the ifile inode and
313 * the segment usage table, plus an ifile page.
314 */
315 #define LFS_FITS(fs, db) \
316 ((int32_t)((db + ((fs)->lfs_uinodes + INOPB((fs))) / \
317 INOPB((fs)) + fsbtodb(fs, 1) + LFS_SUMMARY_SIZE / DEV_BSIZE + \
318 (fs)->lfs_segtabsz)) < (fs)->lfs_avail)
319
320 /* Determine if a buffer belongs to the ifile */
321 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM)
322
323 /*
324 * Structures used by lfs_bmapv and lfs_markv to communicate information
325 * about inodes and data blocks.
326 */
327 typedef struct block_info {
328 ino_t bi_inode; /* inode # */
329 ufs_daddr_t bi_lbn; /* logical block w/in file */
330 ufs_daddr_t bi_daddr; /* disk address of block */
331 time_t bi_segcreate; /* origin segment create time */
332 int bi_version; /* file version number */
333 void *bi_bp; /* data buffer */
334 int bi_size; /* size of the block (if fragment) */
335 } BLOCK_INFO;
336
337 /* In-memory description of a segment about to be written. */
338 struct segment {
339 struct lfs *fs; /* file system pointer */
340 struct buf **bpp; /* pointer to buffer array */
341 struct buf **cbpp; /* pointer to next available bp */
342 struct buf **start_bpp; /* pointer to first bp in this set */
343 struct buf *ibp; /* buffer pointer to inode page */
344 struct finfo *fip; /* current fileinfo pointer */
345 struct vnode *vp; /* vnode being gathered */
346 void *segsum; /* segment summary info */
347 u_int32_t ninodes; /* number of inodes in this segment */
348 u_int32_t seg_bytes_left; /* bytes left in segment */
349 u_int32_t sum_bytes_left; /* bytes left in summary block */
350 u_int32_t seg_number; /* number of this segment */
351 ufs_daddr_t *start_lbp; /* beginning lbn for this set */
352
353 #define SEGM_CKP 0x01 /* doing a checkpoint */
354 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */
355 #define SEGM_SYNC 0x04 /* wait for segment */
356 u_int16_t seg_flags; /* run-time flags for this segment */
357 };
358
359 #define ISSPACE(F, BB, C) \
360 (((C)->cr_uid == 0 && (F)->lfs_bfree >= (BB)) || \
361 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB)))
362
363 #define IS_FREESPACE(F, BB) \
364 ((F)->lfs_bfree > ((F)->lfs_dsize * (F)->lfs_minfree / 100 + (BB)))
365
366 #define ISSPACE_XXX(F, BB) \
367 ((F)->lfs_bfree >= (BB))
368
369 #define DOSTATS
370 #ifdef DOSTATS
371 /* Statistics Counters */
372 struct lfs_stats {
373 u_int segsused;
374 u_int psegwrites;
375 u_int psyncwrites;
376 u_int pcleanwrites;
377 u_int blocktot;
378 u_int cleanblocks;
379 u_int ncheckpoints;
380 u_int nwrites;
381 u_int nsync_writes;
382 u_int wait_exceeded;
383 u_int write_exceeded;
384 u_int flush_invoked;
385 };
386 extern struct lfs_stats lfs_stats;
387 #endif
388