lfs.h revision 1.75 1 /* $NetBSD: lfs.h,v 1.75 2005/02/26 05:40:42 perseant Exp $ */
2
3 /*-
4 * Copyright (c) 1999, 2000, 2001, 2002, 2003 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This code is derived from software contributed to The NetBSD Foundation
8 * by Konrad E. Schroder <perseant (at) hhhh.org>.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the NetBSD
21 * Foundation, Inc. and its contributors.
22 * 4. Neither the name of The NetBSD Foundation nor the names of its
23 * contributors may be used to endorse or promote products derived
24 * from this software without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
27 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
28 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
29 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
30 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36 * POSSIBILITY OF SUCH DAMAGE.
37 */
38 /*-
39 * Copyright (c) 1991, 1993
40 * The Regents of the University of California. All rights reserved.
41 *
42 * Redistribution and use in source and binary forms, with or without
43 * modification, are permitted provided that the following conditions
44 * are met:
45 * 1. Redistributions of source code must retain the above copyright
46 * notice, this list of conditions and the following disclaimer.
47 * 2. Redistributions in binary form must reproduce the above copyright
48 * notice, this list of conditions and the following disclaimer in the
49 * documentation and/or other materials provided with the distribution.
50 * 3. Neither the name of the University nor the names of its contributors
51 * may be used to endorse or promote products derived from this software
52 * without specific prior written permission.
53 *
54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
64 * SUCH DAMAGE.
65 *
66 * @(#)lfs.h 8.9 (Berkeley) 5/8/95
67 */
68
69 #ifndef _UFS_LFS_LFS_H_
70 #define _UFS_LFS_LFS_H_
71
72 /*
73 * Compile-time options for LFS.
74 */
75 #define LFS_IFIND_RETRIES 16
76 #define LFS_EAGAIN_FAIL /* markv fail with EAGAIN if ino is locked */
77 #define LFS_DEBUG_RFW /* print roll-forward debugging info */
78 #define LFS_LOGLENGTH 1024 /* size of debugging log */
79 #define LFS_MAX_ACTIVE 10 /* Dirty segments before ckp forced */
80 #define LFS_PD /* pagedaemon codaemon */
81
82 /* #define DEBUG_LFS */ /* Intensive debugging of LFS subsystem */
83
84 /*
85 * Fixed filesystem layout parameters
86 */
87 #define LFS_LABELPAD 8192 /* LFS label size */
88 #define LFS_SBPAD 8192 /* LFS superblock size */
89
90 #define LFS_UNUSED_INUM 0 /* 0: out of band inode number */
91 #define LFS_IFILE_INUM 1 /* 1: IFILE inode number */
92 /* 2: Root inode number */
93 #define LOSTFOUNDINO 3 /* 3: lost+found inode number */
94 #define LFS_FIRST_INUM 4 /* 4: first free inode number */
95
96 #define LFS_V1_SUMMARY_SIZE 512 /* V1 fixed summary size */
97 #define LFS_DFL_SUMMARY_SIZE 512 /* Default summary size */
98
99 #define LFS_MAX_DADDR 0x7fffffff /* Highest addressable fsb */
100
101 /* Adjustable filesystem parameters */
102 #define MIN_FREE_SEGS 2
103 #ifndef LFS_ATIME_IFILE
104 # define LFS_ATIME_IFILE 0 /* Store atime info in ifile (optional in LFSv1) */
105 #endif
106 #define LFS_MARKV_MAXBLKCNT 65536 /* Max block count for lfs_markv() */
107
108 /* Misc. definitions */
109 #define BW_CLEAN 1 /* Flag for lfs_bwrite_ext() */
110 #define PG_DELWRI PG_PAGER1 /* Local def for delayed pageout */
111
112 /* Resource limits */
113 #define LFS_MAX_BUFS ((nbuf >> 2) - 10)
114 #define LFS_WAIT_BUFS ((nbuf >> 1) - (nbuf >> 3) - 10)
115 #define LFS_INVERSE_MAX_BUFS(n) (((n) + 10) << 2)
116 extern u_long bufmem_lowater, bufmem_hiwater; /* XXX */
117 #define LFS_MAX_BYTES ((bufmem_lowater >> 2) - 10 * PAGE_SIZE)
118 #define LFS_INVERSE_MAX_BYTES(n) (((n) + 10 * PAGE_SIZE) << 2)
119 #define LFS_WAIT_BYTES ((bufmem_lowater >> 1) - (bufmem_lowater >> 3) - 10 * PAGE_SIZE)
120 #define LFS_MAX_DIROP ((desiredvnodes >> 2) + (desiredvnodes >> 3))
121 #define LFS_MAX_PAGES \
122 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemin) >> 8)
123 #define LFS_WAIT_PAGES \
124 (((uvmexp.active + uvmexp.inactive + uvmexp.free) * uvmexp.filemax) >> 8)
125 #define LFS_BUFWAIT 2 /* How long to wait if over *_WAIT_* */
126
127 /*
128 * Reserved blocks for lfs_malloc
129 */
130
131 /* Structure to keep reserved blocks */
132 typedef struct lfs_res_blk {
133 void *p;
134 LIST_ENTRY(lfs_res_blk) res;
135 int size;
136 char inuse;
137 } res_t;
138
139 /* Types for lfs_newbuf and lfs_malloc */
140 #define LFS_NB_UNKNOWN -1
141 #define LFS_NB_SUMMARY 0
142 #define LFS_NB_SBLOCK 1
143 #define LFS_NB_IBLOCK 2
144 #define LFS_NB_CLUSTER 3
145 #define LFS_NB_CLEAN 4
146 #define LFS_NB_COUNT 5 /* always last */
147
148 /* Number of reserved memory blocks of each type */
149 #define LFS_N_SUMMARIES 2
150 #define LFS_N_SBLOCKS 1 /* Always 1, to throttle superblock writes */
151 #define LFS_N_IBLOCKS 16 /* In theory ssize/bsize; in practice around 2 */
152 #define LFS_N_CLUSTERS 16 /* In theory ssize/MAXPHYS */
153 #define LFS_N_CLEAN 0
154
155 /* Total count of "large" (non-pool) types */
156 #define LFS_N_TOTAL (LFS_N_SUMMARIES + LFS_N_SBLOCKS + LFS_N_IBLOCKS + \
157 LFS_N_CLUSTERS + LFS_N_CLEAN)
158
159 /* Counts for pool types */
160 #define LFS_N_CL LFS_N_CLUSTERS
161 #define LFS_N_BPP 2
162 #define LFS_N_SEG 2
163
164 /*
165 * "struct buf" associated definitions
166 */
167
168 /* Unassigned disk addresses. */
169 #define UNASSIGNED -1
170 #define UNWRITTEN -2
171
172 /* Unused logical block number */
173 #define LFS_UNUSED_LBN -1
174
175 /* Determine if a buffer belongs to the ifile */
176 #define IS_IFILE(bp) (VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM)
177
178 # define LFS_LOCK_BUF(bp) do { \
179 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == 0) { \
180 simple_lock(&lfs_subsys_lock); \
181 ++locked_queue_count; \
182 locked_queue_bytes += bp->b_bufsize; \
183 simple_unlock(&lfs_subsys_lock); \
184 } \
185 (bp)->b_flags |= B_LOCKED; \
186 } while (0)
187
188 # define LFS_UNLOCK_BUF(bp) do { \
189 if (((bp)->b_flags & (B_LOCKED | B_CALL)) == B_LOCKED) { \
190 simple_lock(&lfs_subsys_lock); \
191 --locked_queue_count; \
192 locked_queue_bytes -= bp->b_bufsize; \
193 if (locked_queue_count < LFS_WAIT_BUFS && \
194 locked_queue_bytes < LFS_WAIT_BYTES) \
195 wakeup(&locked_queue_count); \
196 simple_unlock(&lfs_subsys_lock); \
197 } \
198 (bp)->b_flags &= ~B_LOCKED; \
199 } while (0)
200
201 #ifdef _KERNEL
202 # define LFS_IS_MALLOC_BUF(bp) (((bp)->b_flags & B_CALL) && \
203 (bp)->b_iodone == lfs_callback)
204
205 # ifdef DEBUG_LOCKED_LIST
206 # define LFS_DEBUG_COUNTLOCKED(m) do { \
207 lfs_countlocked(&locked_queue_count, &locked_queue_bytes, (m)); \
208 wakeup(&locked_queue_count); \
209 } while (0)
210 # else
211 # define LFS_DEBUG_COUNTLOCKED(m)
212 # endif
213
214 /* log for debugging writes to the Ifile */
215 # ifdef DEBUG
216 struct lfs_log_entry {
217 char *op;
218 char *file;
219 int line;
220 daddr_t block;
221 unsigned long flags;
222 };
223 extern int lfs_lognum;
224 extern struct lfs_log_entry lfs_log[LFS_LOGLENGTH];
225 # define LFS_BWRITE_LOG(bp) lfs_bwrite_log((bp), __FILE__, __LINE__)
226 # define LFS_ENTER_LOG(theop, thefile, theline, lbn, theflags) do { \
227 int _s; \
228 \
229 _s = splbio(); \
230 lfs_log[lfs_lognum].op = theop; \
231 lfs_log[lfs_lognum].file = thefile; \
232 lfs_log[lfs_lognum].line = (theline); \
233 lfs_log[lfs_lognum].block = (lbn); \
234 lfs_log[lfs_lognum].flags = (theflags); \
235 lfs_lognum = (lfs_lognum + 1) % LFS_LOGLENGTH; \
236 splx(_s); \
237 } while (0)
238
239 # define LFS_BCLEAN_LOG(fs, bp) do { \
240 if ((bp)->b_vp == (fs)->lfs_ivnode) \
241 LFS_ENTER_LOG("clear", __FILE__, __LINE__, \
242 bp->b_lblkno, bp->b_flags); \
243 } while (0)
244 # else /* ! DEBUG */
245 # define LFS_BCLEAN_LOG(fs, bp)
246 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp))
247 # endif /* ! DEBUG */
248 #else /* ! _KERNEL */
249 # define LFS_BWRITE_LOG(bp) VOP_BWRITE((bp))
250 #endif /* _KERNEL */
251
252 #ifdef _KERNEL
253 /* Filehandle structure for exported LFSes */
254 struct lfid {
255 struct ufid lfid_ufid;
256 #define lfid_len lfid_ufid.ufid_len
257 #define lfid_ino lfid_ufid.ufid_ino
258 #define lfid_gen lfid_ufid.ufid_gen
259 uint32_t lfid_ident;
260 };
261 #endif /* _KERNEL */
262
263 /*
264 * "struct inode" associated definitions
265 */
266
267 /* Address calculations for metadata located in the inode */
268 #define S_INDIR(fs) -NDADDR
269 #define D_INDIR(fs) (S_INDIR(fs) - NINDIR(fs) - 1)
270 #define T_INDIR(fs) (D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1)
271
272 /* For convenience */
273 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_MODIFY|IN_ACCESSED|IN_CLEANING)
274
275 #define LFS_SET_UINO(ip, flags) do { \
276 simple_lock(&(ip)->i_lfs->lfs_interlock); \
277 if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED)) \
278 ++(ip)->i_lfs->lfs_uinodes; \
279 if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING)) \
280 ++(ip)->i_lfs->lfs_uinodes; \
281 if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED)) \
282 ++(ip)->i_lfs->lfs_uinodes; \
283 (ip)->i_flag |= (flags); \
284 simple_unlock(&(ip)->i_lfs->lfs_interlock); \
285 } while (0)
286
287 #define LFS_CLR_UINO(ip, flags) do { \
288 simple_lock(&(ip)->i_lfs->lfs_interlock); \
289 if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED)) \
290 --(ip)->i_lfs->lfs_uinodes; \
291 if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING)) \
292 --(ip)->i_lfs->lfs_uinodes; \
293 if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED)) \
294 --(ip)->i_lfs->lfs_uinodes; \
295 (ip)->i_flag &= ~(flags); \
296 if ((ip)->i_lfs->lfs_uinodes < 0) { \
297 panic("lfs_uinodes < 0"); \
298 } \
299 simple_unlock(&(ip)->i_lfs->lfs_interlock); \
300 } while (0)
301
302 #define LFS_ITIMES(ip, acc, mod, cre) do { \
303 struct lfs *_fs = (ip)->i_lfs; \
304 \
305 if ((ip)->i_flag & IN_ACCESS) { \
306 (ip)->i_ffs1_atime = (acc)->tv_sec; \
307 (ip)->i_ffs1_atimensec = (acc)->tv_nsec; \
308 if ((ip)->i_lfs->lfs_version > 1) { \
309 struct buf *_ibp; \
310 IFILE *_ifp; \
311 \
312 LFS_IENTRY(_ifp, ip->i_lfs, ip->i_number, _ibp); \
313 _ifp->if_atime_sec = (acc)->tv_sec; \
314 _ifp->if_atime_nsec = (acc)->tv_nsec; \
315 LFS_BWRITE_LOG(_ibp); \
316 _fs->lfs_flags |= LFS_IFDIRTY; \
317 } else { \
318 LFS_SET_UINO(ip, IN_ACCESSED); \
319 } \
320 } \
321 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE | IN_MODIFY)) { \
322 if ((ip)->i_flag & (IN_UPDATE | IN_MODIFY)) { \
323 (ip)->i_ffs1_mtime = (mod)->tv_sec; \
324 (ip)->i_ffs1_mtimensec = (mod)->tv_nsec; \
325 (ip)->i_modrev++; \
326 } \
327 if ((ip)->i_flag & (IN_CHANGE | IN_MODIFY)) { \
328 (ip)->i_ffs1_ctime = (cre)->tv_sec; \
329 (ip)->i_ffs1_ctimensec = (cre)->tv_nsec; \
330 } \
331 if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) \
332 LFS_SET_UINO(ip, IN_MODIFIED); \
333 if ((ip)->i_flag & IN_MODIFY) \
334 LFS_SET_UINO(ip, IN_ACCESSED); \
335 } \
336 (ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFY);\
337 } while (0)
338
339 /*
340 * "struct vnode" associated definitions
341 */
342
343 /* Heuristic emptiness measure */
344 #define VPISEMPTY(vp) (LIST_EMPTY(&(vp)->v_dirtyblkhd) && \
345 !((vp)->v_flag & VONWORKLST))
346
347 /* XXX Shouldn't we use v_numoutput instead? */
348 #define WRITEINPROG(vp) (!LIST_EMPTY(&(vp)->v_dirtyblkhd) && \
349 !(VTOI(vp)->i_flag & (IN_MODIFIED | IN_ACCESSED | IN_CLEANING)))
350
351
352 /*
353 * On-disk and in-memory checkpoint segment usage structure.
354 */
355 typedef struct segusage SEGUSE;
356 struct segusage {
357 u_int32_t su_nbytes; /* 0: number of live bytes */
358 u_int32_t su_olastmod; /* 4: SEGUSE last modified timestamp */
359 u_int16_t su_nsums; /* 8: number of summaries in segment */
360 u_int16_t su_ninos; /* 10: number of inode blocks in seg */
361
362 #define SEGUSE_ACTIVE 0x01 /* segment currently being written */
363 #define SEGUSE_DIRTY 0x02 /* segment has data in it */
364 #define SEGUSE_SUPERBLOCK 0x04 /* segment contains a superblock */
365 #define SEGUSE_ERROR 0x08 /* cleaner: do not clean segment */
366 #define SEGUSE_EMPTY 0x10 /* segment is empty */
367 u_int32_t su_flags; /* 12: segment flags */
368 u_int64_t su_lastmod; /* 16: last modified timestamp */
369 };
370
371 typedef struct segusage_v1 SEGUSE_V1;
372 struct segusage_v1 {
373 u_int32_t su_nbytes; /* 0: number of live bytes */
374 u_int32_t su_lastmod; /* 4: SEGUSE last modified timestamp */
375 u_int16_t su_nsums; /* 8: number of summaries in segment */
376 u_int16_t su_ninos; /* 10: number of inode blocks in seg */
377 u_int32_t su_flags; /* 12: segment flags */
378 };
379
380 #define SEGUPB(fs) (fs->lfs_sepb)
381 #define SEGTABSIZE_SU(fs) \
382 (((fs)->lfs_nseg + SEGUPB(fs) - 1) / (fs)->lfs_sepb)
383
384 /* Read in the block with a specific segment usage entry from the ifile. */
385 #define LFS_SEGENTRY(SP, F, IN, BP) do { \
386 int _e; \
387 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
388 if ((_e = bread((F)->lfs_ivnode, \
389 ((IN) / (F)->lfs_sepb) + (F)->lfs_cleansz, \
390 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \
391 panic("lfs: ifile read: %d", _e); \
392 if ((F)->lfs_version == 1) \
393 (SP) = (SEGUSE *)((SEGUSE_V1 *)(BP)->b_data + \
394 ((IN) & ((F)->lfs_sepb - 1))); \
395 else \
396 (SP) = (SEGUSE *)(BP)->b_data + ((IN) % (F)->lfs_sepb); \
397 } while (0)
398
399 #define LFS_WRITESEGENTRY(SP, F, IN, BP) do { \
400 if ((SP)->su_nbytes == 0) \
401 (SP)->su_flags |= SEGUSE_EMPTY; \
402 else \
403 (SP)->su_flags &= ~SEGUSE_EMPTY; \
404 (F)->lfs_suflags[(F)->lfs_activesb][(IN)] = (SP)->su_flags; \
405 LFS_BWRITE_LOG(BP); \
406 } while (0)
407
408 /*
409 * On-disk file information. One per file with data blocks in the segment.
410 */
411 typedef struct finfo FINFO;
412 struct finfo {
413 u_int32_t fi_nblocks; /* number of blocks */
414 u_int32_t fi_version; /* version number */
415 u_int32_t fi_ino; /* inode number */
416 u_int32_t fi_lastlength; /* length of last block in array */
417 int32_t fi_blocks[1]; /* array of logical block numbers */
418 };
419 /* sizeof FINFO except fi_blocks */
420 #define FINFOSIZE (sizeof(FINFO) - sizeof(int32_t))
421
422 /*
423 * Index file inode entries.
424 */
425 typedef struct ifile IFILE;
426 struct ifile {
427 u_int32_t if_version; /* inode version number */
428 #define LFS_UNUSED_DADDR 0 /* out-of-band daddr */
429 int32_t if_daddr; /* inode disk address */
430 ino_t if_nextfree; /* next-unallocated inode */
431 u_int32_t if_atime_sec; /* Last access time, seconds */
432 u_int32_t if_atime_nsec; /* and nanoseconds */
433 };
434
435 typedef struct ifile_v1 IFILE_V1;
436 struct ifile_v1 {
437 u_int32_t if_version; /* inode version number */
438 int32_t if_daddr; /* inode disk address */
439 ino_t if_nextfree; /* next-unallocated inode */
440 #if LFS_ATIME_IFILE
441 struct timespec if_atime; /* Last access time */
442 #endif
443 };
444
445 /*
446 * LFSv1 compatibility code is not allowed to touch if_atime, since it
447 * may not be mapped!
448 */
449 /* Read in the block with a specific inode from the ifile. */
450 #define LFS_IENTRY(IP, F, IN, BP) do { \
451 int _e; \
452 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
453 if ((_e = bread((F)->lfs_ivnode, \
454 (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz, \
455 (F)->lfs_bsize, NOCRED, &(BP))) != 0) \
456 panic("lfs: ifile read %d", _e); \
457 if ((F)->lfs_version == 1) \
458 (IP) = (IFILE *)((IFILE_V1 *)(BP)->b_data + \
459 (IN) % (F)->lfs_ifpb); \
460 else \
461 (IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb; \
462 } while (0)
463
464 /*
465 * Cleaner information structure. This resides in the ifile and is used
466 * to pass information from the kernel to the cleaner.
467 */
468 typedef struct _cleanerinfo {
469 u_int32_t clean; /* number of clean segments */
470 u_int32_t dirty; /* number of dirty segments */
471 int32_t bfree; /* disk blocks free */
472 int32_t avail; /* disk blocks available */
473 u_int32_t free_head; /* head of the inode free list */
474 u_int32_t free_tail; /* tail of the inode free list */
475 } CLEANERINFO;
476
477 #define CLEANSIZE_SU(fs) \
478 ((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift)
479
480 /* Read in the block with the cleaner info from the ifile. */
481 #define LFS_CLEANERINFO(CP, F, BP) do { \
482 VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS; \
483 if (bread((F)->lfs_ivnode, \
484 (daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP))) \
485 panic("lfs: ifile read"); \
486 (CP) = (CLEANERINFO *)(BP)->b_data; \
487 } while (0)
488
489 /* Synchronize the Ifile cleaner info with current avail and bfree */
490 #define LFS_SYNC_CLEANERINFO(cip, fs, bp, w) do { \
491 if ((w) || (cip)->bfree != (fs)->lfs_bfree || \
492 (cip)->avail != (fs)->lfs_avail - (fs)->lfs_ravail - \
493 (fs)->lfs_favail) { \
494 (cip)->bfree = (fs)->lfs_bfree; \
495 (cip)->avail = (fs)->lfs_avail - (fs)->lfs_ravail - \
496 (fs)->lfs_favail; \
497 if (((bp)->b_flags & B_GATHERED) == 0) \
498 (fs)->lfs_flags |= LFS_IFDIRTY; \
499 (void) LFS_BWRITE_LOG(bp); /* Ifile */ \
500 } else \
501 brelse(bp); \
502 } while (0)
503
504 #define LFS_GET_HEADFREE(FS, CIP, BP, FREEP) do { \
505 if ((FS)->lfs_version > 1) { \
506 LFS_CLEANERINFO((CIP), (FS), (BP)); \
507 (FS)->lfs_freehd = (CIP)->free_head; \
508 brelse(BP); \
509 } \
510 *(FREEP) = (FS)->lfs_freehd; \
511 } while (0)
512
513 #define LFS_PUT_HEADFREE(FS, CIP, BP, VAL) do { \
514 (FS)->lfs_freehd = (VAL); \
515 if ((FS)->lfs_version > 1) { \
516 LFS_CLEANERINFO((CIP), (FS), (BP)); \
517 (CIP)->free_head = (VAL); \
518 LFS_BWRITE_LOG(BP); \
519 (FS)->lfs_flags |= LFS_IFDIRTY; \
520 } \
521 } while (0)
522
523 #define LFS_GET_TAILFREE(FS, CIP, BP, FREEP) do { \
524 LFS_CLEANERINFO((CIP), (FS), (BP)); \
525 *(FREEP) = (CIP)->free_tail; \
526 brelse(BP); \
527 } while (0)
528
529 #define LFS_PUT_TAILFREE(FS, CIP, BP, VAL) do { \
530 LFS_CLEANERINFO((CIP), (FS), (BP)); \
531 (CIP)->free_tail = (VAL); \
532 LFS_BWRITE_LOG(BP); \
533 (FS)->lfs_flags |= LFS_IFDIRTY; \
534 } while (0)
535
536 /*
537 * On-disk segment summary information
538 */
539 typedef struct segsum_v1 SEGSUM_V1;
540 struct segsum_v1 {
541 u_int32_t ss_sumsum; /* 0: check sum of summary block */
542 u_int32_t ss_datasum; /* 4: check sum of data */
543 u_int32_t ss_magic; /* 8: segment summary magic number */
544 #define SS_MAGIC 0x061561
545 int32_t ss_next; /* 12: next segment */
546 u_int32_t ss_create; /* 16: creation time stamp */
547 u_int16_t ss_nfinfo; /* 20: number of file info structures */
548 u_int16_t ss_ninos; /* 22: number of inodes in summary */
549
550 #define SS_DIROP 0x01 /* segment begins a dirop */
551 #define SS_CONT 0x02 /* more partials to finish this write*/
552 u_int16_t ss_flags; /* 24: used for directory operations */
553 u_int16_t ss_pad; /* 26: extra space */
554 /* FINFO's and inode daddr's... */
555 };
556
557 typedef struct segsum SEGSUM;
558 struct segsum {
559 u_int32_t ss_sumsum; /* 0: check sum of summary block */
560 u_int32_t ss_datasum; /* 4: check sum of data */
561 u_int32_t ss_magic; /* 8: segment summary magic number */
562 int32_t ss_next; /* 12: next segment */
563 u_int32_t ss_ident; /* 16: roll-forward fsid */
564 #define ss_ocreate ss_ident /* ident is where create was in v1 */
565 u_int16_t ss_nfinfo; /* 20: number of file info structures */
566 u_int16_t ss_ninos; /* 22: number of inodes in summary */
567 u_int16_t ss_flags; /* 24: used for directory operations */
568 u_int8_t ss_pad[6]; /* 26: extra space */
569 u_int64_t ss_serial; /* 32: serial number */
570 u_int64_t ss_create; /* 40: time stamp */
571 /* FINFO's and inode daddr's... */
572 };
573
574 #define SEGSUM_SIZE(fs) ((fs)->lfs_version == 1 ? sizeof(SEGSUM_V1) : sizeof(SEGSUM))
575
576
577 /*
578 * On-disk super block.
579 */
580 struct dlfs {
581 #define LFS_MAGIC 0x070162
582 u_int32_t dlfs_magic; /* 0: magic number */
583 #define LFS_VERSION 2
584 u_int32_t dlfs_version; /* 4: version number */
585
586 u_int32_t dlfs_size; /* 8: number of blocks in fs (v1) */
587 /* number of frags in fs (v2) */
588 u_int32_t dlfs_ssize; /* 12: number of blocks per segment (v1) */
589 /* number of bytes per segment (v2) */
590 u_int32_t dlfs_dsize; /* 16: number of disk blocks in fs */
591 u_int32_t dlfs_bsize; /* 20: file system block size */
592 u_int32_t dlfs_fsize; /* 24: size of frag blocks in fs */
593 u_int32_t dlfs_frag; /* 28: number of frags in a block in fs */
594
595 /* Checkpoint region. */
596 u_int32_t dlfs_freehd; /* 32: start of the free list */
597 int32_t dlfs_bfree; /* 36: number of free disk blocks */
598 u_int32_t dlfs_nfiles; /* 40: number of allocated inodes */
599 int32_t dlfs_avail; /* 44: blocks available for writing */
600 int32_t dlfs_uinodes; /* 48: inodes in cache not yet on disk */
601 int32_t dlfs_idaddr; /* 52: inode file disk address */
602 u_int32_t dlfs_ifile; /* 56: inode file inode number */
603 int32_t dlfs_lastseg; /* 60: address of last segment written */
604 int32_t dlfs_nextseg; /* 64: address of next segment to write */
605 int32_t dlfs_curseg; /* 68: current segment being written */
606 int32_t dlfs_offset; /* 72: offset in curseg for next partial */
607 int32_t dlfs_lastpseg; /* 76: address of last partial written */
608 u_int32_t dlfs_inopf; /* 80: v1: time stamp; v2: inodes per frag */
609 #define dlfs_otstamp dlfs_inopf
610
611 /* These are configuration parameters. */
612 u_int32_t dlfs_minfree; /* 84: minimum percentage of free blocks */
613
614 /* These fields can be computed from the others. */
615 u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */
616 u_int32_t dlfs_fsbpseg; /* 96: fsb per segment */
617 u_int32_t dlfs_inopb; /* 100: inodes per block */
618 u_int32_t dlfs_ifpb; /* 104: IFILE entries per block */
619 u_int32_t dlfs_sepb; /* 108: SEGUSE entries per block */
620 u_int32_t dlfs_nindir; /* 112: indirect pointers per block */
621 u_int32_t dlfs_nseg; /* 116: number of segments */
622 u_int32_t dlfs_nspf; /* 120: number of sectors per fragment */
623 u_int32_t dlfs_cleansz; /* 124: cleaner info size in blocks */
624 u_int32_t dlfs_segtabsz; /* 128: segment table size in blocks */
625 u_int32_t dlfs_segmask; /* 132: calculate offset within a segment */
626 u_int32_t dlfs_segshift; /* 136: fast mult/div for segments */
627 u_int32_t dlfs_bshift; /* 140: calc block number from file offset */
628 u_int32_t dlfs_ffshift; /* 144: fast mult/div for frag from file */
629 u_int32_t dlfs_fbshift; /* 148: fast mult/div for frag from block */
630 u_int64_t dlfs_bmask; /* 152: calc block offset from file offset */
631 u_int64_t dlfs_ffmask; /* 160: calc frag offset from file offset */
632 u_int64_t dlfs_fbmask; /* 168: calc frag offset from block offset */
633 u_int32_t dlfs_blktodb; /* 176: blktodb and dbtoblk shift constant */
634 u_int32_t dlfs_sushift; /* 180: fast mult/div for segusage table */
635
636 int32_t dlfs_maxsymlinklen; /* 184: max length of an internal symlink */
637 #define LFS_MIN_SBINTERVAL 5 /* minimum superblock segment spacing */
638 #define LFS_MAXNUMSB 10 /* 188: superblock disk offsets */
639 int32_t dlfs_sboffs[LFS_MAXNUMSB];
640
641 u_int32_t dlfs_nclean; /* 228: Number of clean segments */
642 u_char dlfs_fsmnt[MNAMELEN]; /* 232: name mounted on */
643 #define LFS_PF_CLEAN 0x1
644 u_int16_t dlfs_pflags; /* 322: file system persistent flags */
645 int32_t dlfs_dmeta; /* 324: total number of dirty summaries */
646 u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */
647 u_int32_t dlfs_sumsize; /* 332: size of summary blocks */
648 u_int64_t dlfs_serial; /* 336: serial number */
649 u_int32_t dlfs_ibsize; /* 344: size of inode blocks */
650 int32_t dlfs_start; /* 348: start of segment 0 */
651 u_int64_t dlfs_tstamp; /* 352: time stamp */
652 #define LFS_44INODEFMT 0
653 #define LFS_MAXINODEFMT 0
654 u_int32_t dlfs_inodefmt; /* 360: inode format version */
655 u_int32_t dlfs_interleave; /* 364: segment interleave */
656 u_int32_t dlfs_ident; /* 368: per-fs identifier */
657 u_int32_t dlfs_fsbtodb; /* 372: fsbtodb abd dbtodsb shift constant */
658 int8_t dlfs_pad[132]; /* 376: round to 512 bytes */
659 /* Checksum -- last valid disk field. */
660 u_int32_t dlfs_cksum; /* 508: checksum for superblock checking */
661 };
662
663 /*
664 * In-memory super block.
665 */
666 struct lfs {
667 struct dlfs lfs_dlfs; /* on-disk parameters */
668 #define lfs_magic lfs_dlfs.dlfs_magic
669 #define lfs_version lfs_dlfs.dlfs_version
670 #define lfs_size lfs_dlfs.dlfs_size
671 #define lfs_ssize lfs_dlfs.dlfs_ssize
672 #define lfs_dsize lfs_dlfs.dlfs_dsize
673 #define lfs_bsize lfs_dlfs.dlfs_bsize
674 #define lfs_fsize lfs_dlfs.dlfs_fsize
675 #define lfs_frag lfs_dlfs.dlfs_frag
676 #define lfs_freehd lfs_dlfs.dlfs_freehd
677 #define lfs_bfree lfs_dlfs.dlfs_bfree
678 #define lfs_nfiles lfs_dlfs.dlfs_nfiles
679 #define lfs_avail lfs_dlfs.dlfs_avail
680 #define lfs_uinodes lfs_dlfs.dlfs_uinodes
681 #define lfs_idaddr lfs_dlfs.dlfs_idaddr
682 #define lfs_ifile lfs_dlfs.dlfs_ifile
683 #define lfs_lastseg lfs_dlfs.dlfs_lastseg
684 #define lfs_nextseg lfs_dlfs.dlfs_nextseg
685 #define lfs_curseg lfs_dlfs.dlfs_curseg
686 #define lfs_offset lfs_dlfs.dlfs_offset
687 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg
688 #define lfs_otstamp lfs_dlfs.dlfs_inopf
689 #define lfs_inopf lfs_dlfs.dlfs_inopf
690 #define lfs_minfree lfs_dlfs.dlfs_minfree
691 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize
692 #define lfs_fsbpseg lfs_dlfs.dlfs_fsbpseg
693 #define lfs_inopb lfs_dlfs.dlfs_inopb
694 #define lfs_ifpb lfs_dlfs.dlfs_ifpb
695 #define lfs_sepb lfs_dlfs.dlfs_sepb
696 #define lfs_nindir lfs_dlfs.dlfs_nindir
697 #define lfs_nseg lfs_dlfs.dlfs_nseg
698 #define lfs_nspf lfs_dlfs.dlfs_nspf
699 #define lfs_cleansz lfs_dlfs.dlfs_cleansz
700 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz
701 #define lfs_segmask lfs_dlfs.dlfs_segmask
702 #define lfs_segshift lfs_dlfs.dlfs_segshift
703 #define lfs_bmask lfs_dlfs.dlfs_bmask
704 #define lfs_bshift lfs_dlfs.dlfs_bshift
705 #define lfs_ffmask lfs_dlfs.dlfs_ffmask
706 #define lfs_ffshift lfs_dlfs.dlfs_ffshift
707 #define lfs_fbmask lfs_dlfs.dlfs_fbmask
708 #define lfs_fbshift lfs_dlfs.dlfs_fbshift
709 #define lfs_blktodb lfs_dlfs.dlfs_blktodb
710 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb
711 #define lfs_sushift lfs_dlfs.dlfs_sushift
712 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen
713 #define lfs_sboffs lfs_dlfs.dlfs_sboffs
714 #define lfs_cksum lfs_dlfs.dlfs_cksum
715 #define lfs_pflags lfs_dlfs.dlfs_pflags
716 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt
717 #define lfs_nclean lfs_dlfs.dlfs_nclean
718 #define lfs_dmeta lfs_dlfs.dlfs_dmeta
719 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg
720 #define lfs_sumsize lfs_dlfs.dlfs_sumsize
721 #define lfs_serial lfs_dlfs.dlfs_serial
722 #define lfs_ibsize lfs_dlfs.dlfs_ibsize
723 #define lfs_start lfs_dlfs.dlfs_start
724 #define lfs_tstamp lfs_dlfs.dlfs_tstamp
725 #define lfs_inodefmt lfs_dlfs.dlfs_inodefmt
726 #define lfs_interleave lfs_dlfs.dlfs_interleave
727 #define lfs_ident lfs_dlfs.dlfs_ident
728
729 /* These fields are set at mount time and are meaningless on disk. */
730 struct segment *lfs_sp; /* current segment being written */
731 struct vnode *lfs_ivnode; /* vnode for the ifile */
732 u_int32_t lfs_seglock; /* single-thread the segment writer */
733 pid_t lfs_lockpid; /* pid of lock holder */
734 u_int32_t lfs_iocount; /* number of ios pending */
735 u_int32_t lfs_writer; /* don't allow any dirops to start */
736 u_int32_t lfs_dirops; /* count of active directory ops */
737 u_int32_t lfs_doifile; /* Write ifile blocks on next write */
738 u_int32_t lfs_nactive; /* Number of segments since last ckp */
739 int8_t lfs_fmod; /* super block modified flag */
740 int8_t lfs_ronly; /* mounted read-only flag */
741 #define LFS_NOTYET 0x01
742 #define LFS_IFDIRTY 0x02
743 #define LFS_WARNED 0x04
744 #define LFS_UNDIROP 0x08
745 int8_t lfs_flags; /* currently unused flag */
746 u_int16_t lfs_activesb; /* toggle between superblocks */
747 daddr_t lfs_sbactive; /* disk address of current sb write */
748 struct vnode *lfs_flushvp; /* vnode being flushed */
749 struct vnode *lfs_unlockvp; /* being inactivated in lfs_segunlock */
750 u_int32_t lfs_diropwait; /* # procs waiting on dirop flush */
751 size_t lfs_devbsize; /* Device block size */
752 size_t lfs_devbshift; /* Device block shift */
753 struct lock lfs_fraglock;
754 pid_t lfs_rfpid; /* Process ID of roll-forward agent */
755 int lfs_nadirop; /* number of active dirop nodes */
756 long lfs_ravail; /* blocks pre-reserved for writing */
757 long lfs_favail; /* blocks pre-reserved for writing */
758 res_t *lfs_resblk; /* Reserved memory for pageout */
759 TAILQ_HEAD(, inode) lfs_dchainhd; /* dirop vnodes */
760 TAILQ_HEAD(, inode) lfs_pchainhd; /* paging vnodes */
761 #define LFS_RESHASH_WIDTH 17
762 LIST_HEAD(, lfs_res_blk) lfs_reshash[LFS_RESHASH_WIDTH];
763 int lfs_pdflush; /* pagedaemon wants us to flush */
764 u_int32_t **lfs_suflags; /* Segment use flags */
765 #ifdef _KERNEL
766 struct pool lfs_clpool; /* Pool for struct lfs_cluster */
767 struct pool lfs_bpppool; /* Pool for bpp */
768 struct pool lfs_segpool; /* Pool for struct segment */
769 #endif /* _KERNEL */
770 #define LFS_MAX_CLEANIND 64
771 int32_t lfs_cleanint[LFS_MAX_CLEANIND]; /* Active cleaning intervals */
772 int lfs_cleanind; /* Index into intervals */
773 struct simplelock lfs_interlock; /* lock for lfs_seglock */
774 int lfs_sleepers; /* # procs sleeping this fs */
775 int lfs_pages; /* dirty pages blaming this fs */
776 };
777
778 /* NINDIR is the number of indirects in a file system block. */
779 #define NINDIR(fs) ((fs)->lfs_nindir)
780
781 /* INOPB is the number of inodes in a secondary storage block. */
782 #define INOPB(fs) ((fs)->lfs_inopb)
783 /* INOPF is the number of inodes in a fragment. */
784 #define INOPF(fs) ((fs)->lfs_inopf)
785
786 #define blksize(fs, ip, lbn) \
787 (((lbn) >= NDADDR || (ip)->i_ffs1_size >= ((lbn) + 1) << (fs)->lfs_bshift) \
788 ? (fs)->lfs_bsize \
789 : (fragroundup(fs, blkoff(fs, (ip)->i_ffs1_size))))
790 #define blkoff(fs, loc) ((int)((loc) & (fs)->lfs_bmask))
791 #define fragoff(fs, loc) /* calculates (loc % fs->lfs_fsize) */ \
792 ((int)((loc) & (fs)->lfs_ffmask))
793 #define fsbtodb(fs, b) ((b) << (fs)->lfs_fsbtodb)
794 #define dbtofsb(fs, b) ((b) >> (fs)->lfs_fsbtodb)
795 #define fragstodb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift))
796 #define dbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift))
797 #define lblkno(fs, loc) ((loc) >> (fs)->lfs_bshift)
798 #define lblktosize(fs, blk) ((blk) << (fs)->lfs_bshift)
799 /* Same as above, but named like dbtob(), btodb() */
800 #define fsbtob(fs, b) ((b) << ((fs)->lfs_bshift - \
801 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb))
802 #define btofsb(fs, b) ((b) >> ((fs)->lfs_bshift - \
803 (fs)->lfs_blktodb + (fs)->lfs_fsbtodb))
804 #define fsbtofrags(fs, b) ((b) >> ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \
805 (fs)->lfs_fsbtodb))
806 #define fragstofsb(fs, b) ((b) << ((fs)->lfs_blktodb - (fs)->lfs_fbshift - \
807 (fs)->lfs_fsbtodb))
808 #define btofrags(fs, b) ((b) >> (fs)->lfs_ffshift)
809 #define numfrags(fs, loc) /* calculates (loc / fs->lfs_fsize) */ \
810 ((loc) >> (fs)->lfs_ffshift)
811 #define blkroundup(fs, size) /* calculates roundup(size, fs->lfs_bsize) */ \
812 ((off_t)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask)))
813 #define fragroundup(fs, size) /* calculates roundup(size, fs->lfs_fsize) */ \
814 ((off_t)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask)))
815 #define fragstoblks(fs, frags) /* calculates (frags / fs->lfs_frag) */ \
816 ((frags) >> (fs)->lfs_fbshift)
817 #define blkstofrags(fs, blks) /* calculates (blks * fs->lfs_frag) */ \
818 ((blks) << (fs)->lfs_fbshift)
819 #define fragnum(fs, fsb) /* calculates (fsb % fs->lfs_frag) */ \
820 ((fsb) & ((fs)->lfs_frag - 1))
821 #define blknum(fs, fsb) /* calculates rounddown(fsb, fs->lfs_frag) */ \
822 ((fsb) &~ ((fs)->lfs_frag - 1))
823 #define dblksize(fs, dp, lbn) \
824 (((lbn) >= NDADDR || (dp)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\
825 ? (fs)->lfs_bsize \
826 : (fragroundup(fs, blkoff(fs, (dp)->di_size))))
827
828 #define segsize(fs) ((fs)->lfs_version == 1 ? \
829 lblktosize((fs), (fs)->lfs_ssize) : \
830 (fs)->lfs_ssize)
831 #define segtod(fs, seg) (((fs)->lfs_version == 1 ? \
832 (fs)->lfs_ssize << (fs)->lfs_blktodb : \
833 btofsb((fs), (fs)->lfs_ssize)) * (seg))
834 #define dtosn(fs, daddr) /* block address to segment number */ \
835 ((uint32_t)(((daddr) - (fs)->lfs_start) / segtod((fs), 1)))
836 #define sntod(fs, sn) /* segment number to disk address */ \
837 ((daddr_t)(segtod((fs), (sn)) + (fs)->lfs_start))
838
839 /*
840 * Structures used by lfs_bmapv and lfs_markv to communicate information
841 * about inodes and data blocks.
842 */
843 typedef struct block_info {
844 ino_t bi_inode; /* inode # */
845 int32_t bi_lbn; /* logical block w/in file */
846 int32_t bi_daddr; /* disk address of block */
847 u_int64_t bi_segcreate; /* origin segment create time */
848 int bi_version; /* file version number */
849 void *bi_bp; /* data buffer */
850 int bi_size; /* size of the block (if fragment) */
851 } BLOCK_INFO;
852
853 /* Compatibility for 1.5 binaries */
854 typedef struct block_info_15 {
855 ino_t bi_inode; /* inode # */
856 int32_t bi_lbn; /* logical block w/in file */
857 int32_t bi_daddr; /* disk address of block */
858 u_int32_t bi_segcreate; /* origin segment create time */
859 int bi_version; /* file version number */
860 void *bi_bp; /* data buffer */
861 int bi_size; /* size of the block (if fragment) */
862 } BLOCK_INFO_15;
863
864 /* In-memory description of a segment about to be written. */
865 struct segment {
866 struct lfs *fs; /* file system pointer */
867 struct buf **bpp; /* pointer to buffer array */
868 struct buf **cbpp; /* pointer to next available bp */
869 struct buf **start_bpp; /* pointer to first bp in this set */
870 struct buf *ibp; /* buffer pointer to inode page */
871 struct ufs1_dinode *idp; /* pointer to ifile dinode */
872 struct finfo *fip; /* current fileinfo pointer */
873 struct vnode *vp; /* vnode being gathered */
874 void *segsum; /* segment summary info */
875 u_int32_t ninodes; /* number of inodes in this segment */
876 int32_t seg_bytes_left; /* bytes left in segment */
877 int32_t sum_bytes_left; /* bytes left in summary block */
878 u_int32_t seg_number; /* number of this segment */
879 int32_t *start_lbp; /* beginning lbn for this set */
880
881 #define SEGM_CKP 0x01 /* doing a checkpoint */
882 #define SEGM_CLEAN 0x02 /* cleaner call; don't sort */
883 #define SEGM_SYNC 0x04 /* wait for segment */
884 #define SEGM_PROT 0x08 /* don't inactivate at segunlock */
885 #define SEGM_PAGEDAEMON 0x10 /* pagedaemon called us */
886 #define SEGM_WRITERD 0x20 /* LFS writed called us */
887 #define SEGM_FORCE_CKP 0x40 /* Force checkpoint right away */
888 u_int16_t seg_flags; /* run-time flags for this segment */
889 u_int32_t seg_iocount; /* number of ios pending */
890 int ndupino; /* number of duplicate inodes */
891 };
892
893 #ifdef _KERNEL
894 struct lfs_cluster {
895 size_t bufsize; /* Size of kept data */
896 struct buf **bpp; /* Array of kept buffers */
897 int bufcount; /* Number of kept buffers */
898 #define LFS_CL_MALLOC 0x00000001
899 #define LFS_CL_SHIFT 0x00000002
900 #define LFS_CL_SYNC 0x00000004
901 u_int32_t flags; /* Flags */
902 struct lfs *fs; /* LFS that this belongs to */
903 struct segment *seg; /* Segment structure, for LFS_CL_SYNC */
904 };
905 #endif /* _KERNEL */
906
907 /*
908 * List containing block numbers allocated through lfs_balloc.
909 */
910 struct lbnentry {
911 LIST_ENTRY(lbnentry) entry;
912 daddr_t lbn;
913 };
914
915 /*
916 * LFS inode extensions.
917 */
918 struct lfs_inode_ext {
919 off_t lfs_osize; /* size of file on disk */
920 u_int32_t lfs_effnblocks; /* number of blocks when i/o completes */
921 size_t lfs_fragsize[NDADDR]; /* size of on-disk direct blocks */
922 TAILQ_ENTRY(inode) lfs_dchain; /* Dirop chain. */
923 TAILQ_ENTRY(inode) lfs_pchain; /* Paging chain. */
924 /* Blocks allocated for write */
925 #define LFS_BLIST_HASH_WIDTH 17
926 LIST_HEAD(, lbnentry) lfs_blist[LFS_BLIST_HASH_WIDTH];
927 #define LFSI_NO_GOP_WRITE 0x01
928 u_int32_t lfs_iflags; /* Inode flags */
929 };
930 #define i_lfs_osize inode_ext.lfs->lfs_osize
931 #define i_lfs_effnblks inode_ext.lfs->lfs_effnblocks
932 #define i_lfs_fragsize inode_ext.lfs->lfs_fragsize
933 #define i_lfs_dchain inode_ext.lfs->lfs_dchain
934 #define i_lfs_blist inode_ext.lfs->lfs_blist
935 #define i_lfs_iflags inode_ext.lfs->lfs_iflags
936
937 /*
938 * Macros for determining free space on the disk, with the variable metadata
939 * of segment summaries and inode blocks taken into account.
940 */
941 /* Estimate number of clean blocks not available for writing */
942 #define LFS_EST_CMETA(F) (int32_t)((((F)->lfs_dmeta * \
943 (int64_t)(F)->lfs_nclean) / \
944 ((F)->lfs_nseg - (F)->lfs_nclean)))
945
946 /* Estimate total size of the disk not including metadata */
947 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F))
948
949 /* Estimate number of blocks actually available for writing */
950 #define LFS_EST_BFREE(F) ((F)->lfs_bfree > LFS_EST_CMETA(F) + (F)->lfs_dmeta ? (F)->lfs_bfree - LFS_EST_CMETA(F) - (F)->lfs_dmeta : 0)
951
952 /* Amount of non-meta space not available to mortal man */
953 #define LFS_EST_RSVD(F) (int32_t)((LFS_EST_NONMETA(F) * \
954 (u_int64_t)(F)->lfs_minfree) / \
955 100)
956
957 /* Can credential C write BB blocks */
958 #define ISSPACE(F, BB, C) \
959 ((((C) == NOCRED || (C)->cr_uid == 0) && \
960 LFS_EST_BFREE(F) >= (BB)) || \
961 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB)))
962
963 /* Can an ordinary user write BB blocks */
964 #define IS_FREESPACE(F, BB) \
965 (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F))
966
967 /*
968 * The minimum number of blocks to create a new inode. This is:
969 * directory direct block (1) + NIADDR indirect blocks + inode block (1) +
970 * ifile direct block (1) + NIADDR indirect blocks = 3 + 2 * NIADDR blocks.
971 */
972 #define LFS_NRESERVE(F) (btofsb((F), (2 * NIADDR + 3) << (F)->lfs_bshift))
973
974 /* Statistics Counters */
975 struct lfs_stats {
976 u_int segsused;
977 u_int psegwrites;
978 u_int psyncwrites;
979 u_int pcleanwrites;
980 u_int blocktot;
981 u_int cleanblocks;
982 u_int ncheckpoints;
983 u_int nwrites;
984 u_int nsync_writes;
985 u_int wait_exceeded;
986 u_int write_exceeded;
987 u_int flush_invoked;
988 u_int vflush_invoked;
989 };
990 #ifdef _KERNEL
991 extern struct lfs_stats lfs_stats;
992 #endif
993
994 /* Fcntls to take the place of the lfs syscalls */
995 struct lfs_fcntl_markv {
996 BLOCK_INFO *blkiov; /* blocks to relocate */
997 int blkcnt; /* number of blocks */
998 };
999
1000 #define LFCNSEGWAITALL _FCNR_FSPRIV('L', 0, struct timeval)
1001 #define LFCNSEGWAIT _FCNR_FSPRIV('L', 1, struct timeval)
1002 #define LFCNBMAPV _FCNRW_FSPRIV('L', 2, struct lfs_fcntl_markv)
1003 #define LFCNMARKV _FCNRW_FSPRIV('L', 3, struct lfs_fcntl_markv)
1004 #define LFCNRECLAIM _FCNO_FSPRIV('L', 4)
1005 #define LFCNIFILEFH _FCNW_FSPRIV('L', 5, struct fhandle)
1006 /* Compat for NetBSD 2.x error */
1007 #define LFCNSEGWAITALL_COMPAT _FCNW_FSPRIV('L', 0, struct timeval)
1008 #define LFCNSEGWAIT_COMPAT _FCNW_FSPRIV('L', 1, struct timeval)
1009
1010 #ifdef _KERNEL
1011 /* XXX MP */
1012 #define LFS_SEGLOCK_HELD(fs) \
1013 ((fs)->lfs_seglock != 0 && (fs)->lfs_lockpid == curproc->p_pid)
1014 #endif /* _KERNEL */
1015
1016 #endif /* !_UFS_LFS_LFS_H_ */
1017