xref: /illumos-gate/usr/src/boot/sys/ufs/ffs/fs.h (revision 55fea89dcaa64928bed4327112404dcb3e07b79f)
1 /*-
2  * Copyright (c) 1982, 1986, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)fs.h	8.13 (Berkeley) 3/21/95
30  * $FreeBSD$
31  */
32 
33 #ifndef _UFS_FFS_FS_H_
34 #define	_UFS_FFS_FS_H_
35 
36 #include <sys/mount.h>
37 #include <ufs/ufs/dinode.h>
38 
39 /*
40  * Each disk drive contains some number of filesystems.
41  * A filesystem consists of a number of cylinder groups.
42  * Each cylinder group has inodes and data.
43  *
44  * A filesystem is described by its super-block, which in turn
45  * describes the cylinder groups.  The super-block is critical
46  * data and is replicated in each cylinder group to protect against
47  * catastrophic loss.  This is done at `newfs' time and the critical
48  * super-block data does not change, so the copies need not be
49  * referenced further unless disaster strikes.
50  *
51  * For filesystem fs, the offsets of the various blocks of interest
52  * are given in the super block as:
53  *	[fs->fs_sblkno]		Super-block
54  *	[fs->fs_cblkno]		Cylinder group block
55  *	[fs->fs_iblkno]		Inode blocks
56  *	[fs->fs_dblkno]		Data blocks
57  * The beginning of cylinder group cg in fs, is given by
58  * the ``cgbase(fs, cg)'' macro.
59  *
60  * Depending on the architecture and the media, the superblock may
61  * reside in any one of four places. For tiny media where every block
62  * counts, it is placed at the very front of the partition. Historically,
63  * UFS1 placed it 8K from the front to leave room for the disk label and
64  * a small bootstrap. For UFS2 it got moved to 64K from the front to leave
65  * room for the disk label and a bigger bootstrap, and for really piggy
66  * systems we check at 256K from the front if the first three fail. In
67  * all cases the size of the superblock will be SBLOCKSIZE. All values are
68  * given in byte-offset form, so they do not imply a sector size. The
69  * SBLOCKSEARCH specifies the order in which the locations should be searched.
70  */
71 #define	SBLOCK_FLOPPY	     0
72 #define	SBLOCK_UFS1	  8192
73 #define	SBLOCK_UFS2	 65536
74 #define	SBLOCK_PIGGY	262144
75 #define	SBLOCKSIZE	  8192
76 #define	SBLOCKSEARCH \
77 	{ SBLOCK_UFS2, SBLOCK_UFS1, SBLOCK_FLOPPY, SBLOCK_PIGGY, -1 }
78 
79 /*
80  * Max number of fragments per block. This value is NOT tweakable.
81  */
82 #define	MAXFRAG 	8
83 
84 /*
85  * Addresses stored in inodes are capable of addressing fragments
86  * of `blocks'. File system blocks of at most size MAXBSIZE can
87  * be optionally broken into 2, 4, or 8 pieces, each of which is
88  * addressable; these pieces may be DEV_BSIZE, or some multiple of
89  * a DEV_BSIZE unit.
90  *
91  * Large files consist of exclusively large data blocks.  To avoid
92  * undue wasted disk space, the last data block of a small file may be
93  * allocated as only as many fragments of a large block as are
94  * necessary.  The filesystem format retains only a single pointer
95  * to such a fragment, which is a piece of a single large block that
96  * has been divided.  The size of such a fragment is determinable from
97  * information in the inode, using the ``blksize(fs, ip, lbn)'' macro.
98  *
99  * The filesystem records space availability at the fragment level;
100  * to determine block availability, aligned fragments are examined.
101  */
102 
103 /*
104  * MINBSIZE is the smallest allowable block size.
105  * In order to insure that it is possible to create files of size
106  * 2^32 with only two levels of indirection, MINBSIZE is set to 4096.
107  * MINBSIZE must be big enough to hold a cylinder group block,
108  * thus changes to (struct cg) must keep its size within MINBSIZE.
109  * Note that super blocks are always of size SBLOCKSIZE,
110  * and that both SBLOCKSIZE and MAXBSIZE must be >= MINBSIZE.
111  */
112 #define	MINBSIZE	4096
113 
114 /*
115  * The path name on which the filesystem is mounted is maintained
116  * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in
117  * the super block for this name.
118  */
119 #define	MAXMNTLEN	468
120 
121 /*
122  * The volume name for this filesystem is maintained in fs_volname.
123  * MAXVOLLEN defines the length of the buffer allocated.
124  */
125 #define	MAXVOLLEN	32
126 
127 /*
128  * There is a 128-byte region in the superblock reserved for in-core
129  * pointers to summary information. Originally this included an array
130  * of pointers to blocks of struct csum; now there are just a few
131  * pointers and the remaining space is padded with fs_ocsp[].
132  *
133  * NOCSPTRS determines the size of this padding. One pointer (fs_csp)
134  * is taken away to point to a contiguous array of struct csum for
135  * all cylinder groups; a second (fs_maxcluster) points to an array
136  * of cluster sizes that is computed as cylinder groups are inspected,
137  * and the third points to an array that tracks the creation of new
138  * directories. A fourth pointer, fs_active, is used when creating
139  * snapshots; it points to a bitmap of cylinder groups for which the
140  * free-block bitmap has changed since the snapshot operation began.
141  */
142 #define	NOCSPTRS	((128 / sizeof(void *)) - 4)
143 
144 /*
145  * A summary of contiguous blocks of various sizes is maintained
146  * in each cylinder group. Normally this is set by the initial
147  * value of fs_maxcontig. To conserve space, a maximum summary size
148  * is set by FS_MAXCONTIG.
149  */
150 #define	FS_MAXCONTIG	16
151 
152 /*
153  * MINFREE gives the minimum acceptable percentage of filesystem
154  * blocks which may be free. If the freelist drops below this level
155  * only the superuser may continue to allocate blocks. This may
156  * be set to 0 if no reserve of free blocks is deemed necessary,
157  * however throughput drops by fifty percent if the filesystem
158  * is run at between 95% and 100% full; thus the minimum default
159  * value of fs_minfree is 5%. However, to get good clustering
160  * performance, 10% is a better choice. hence we use 10% as our
161  * default value. With 10% free space, fragmentation is not a
162  * problem, so we choose to optimize for time.
163  */
164 #define	MINFREE		8
165 #define	DEFAULTOPT	FS_OPTTIME
166 
167 /*
168  * Grigoriy Orlov <gluk@ptci.ru> has done some extensive work to fine
169  * tune the layout preferences for directories within a filesystem.
170  * His algorithm can be tuned by adjusting the following parameters
171  * which tell the system the average file size and the average number
172  * of files per directory. These defaults are well selected for typical
173  * filesystems, but may need to be tuned for odd cases like filesystems
174  * being used for squid caches or news spools.
175  */
176 #define	AVFILESIZ	16384	/* expected average file size */
177 #define	AFPDIR		64	/* expected number of files per directory */
178 
179 /*
180  * The maximum number of snapshot nodes that can be associated
181  * with each filesystem. This limit affects only the number of
182  * snapshot files that can be recorded within the superblock so
183  * that they can be found when the filesystem is mounted. However,
184  * maintaining too many will slow the filesystem performance, so
185  * having this limit is a good idea.
186  */
187 #define	FSMAXSNAP 20
188 
189 /*
190  * Used to identify special blocks in snapshots:
191  *
192  * BLK_NOCOPY - A block that was unallocated at the time the snapshot
193  *	was taken, hence does not need to be copied when written.
194  * BLK_SNAP - A block held by another snapshot that is not needed by this
195  *	snapshot. When the other snapshot is freed, the BLK_SNAP entries
196  *	are converted to BLK_NOCOPY. These are needed to allow fsck to
197  *	identify blocks that are in use by other snapshots (which are
198  *	expunged from this snapshot).
199  */
200 #define	BLK_NOCOPY ((ufs2_daddr_t)(1))
201 #define	BLK_SNAP ((ufs2_daddr_t)(2))
202 
203 /*
204  * Sysctl values for the fast filesystem.
205  */
206 #define	FFS_ADJ_REFCNT		 1	/* adjust inode reference count */
207 #define	FFS_ADJ_BLKCNT		 2	/* adjust inode used block count */
208 #define	FFS_BLK_FREE		 3	/* free range of blocks in map */
209 #define	FFS_DIR_FREE		 4	/* free specified dir inodes in map */
210 #define	FFS_FILE_FREE		 5	/* free specified file inodes in map */
211 #define	FFS_SET_FLAGS		 6	/* set filesystem flags */
212 #define	FFS_ADJ_NDIR		 7	/* adjust number of directories */
213 #define	FFS_ADJ_NBFREE		 8	/* adjust number of free blocks */
214 #define	FFS_ADJ_NIFREE		 9	/* adjust number of free inodes */
215 #define	FFS_ADJ_NFFREE		10 	/* adjust number of free frags */
216 #define	FFS_ADJ_NUMCLUSTERS	11	/* adjust number of free clusters */
217 #define	FFS_SET_CWD		12	/* set current directory */
218 #define	FFS_SET_DOTDOT		13	/* set inode number for ".." */
219 #define	FFS_UNLINK		14	/* remove a name in the filesystem */
220 #define	FFS_SET_INODE		15	/* update an on-disk inode */
221 #define	FFS_SET_BUFOUTPUT	16	/* set buffered writing on descriptor */
222 #define	FFS_MAXID		16	/* number of valid ffs ids */
223 
224 /*
225  * Command structure passed in to the filesystem to adjust filesystem values.
226  */
227 #define	FFS_CMD_VERSION		0x19790518	/* version ID */
228 struct fsck_cmd {
229 	int32_t	version;	/* version of command structure */
230 	int32_t	handle;		/* reference to filesystem to be changed */
231 	int64_t	value;		/* inode or block number to be affected */
232 	int64_t	size;		/* amount or range to be adjusted */
233 	int64_t	spare;		/* reserved for future use */
234 };
235 
236 /*
237  * Per cylinder group information; summarized in blocks allocated
238  * from first cylinder group data blocks.  These blocks have to be
239  * read in from fs_csaddr (size fs_cssize) in addition to the
240  * super block.
241  */
242 struct csum {
243 	int32_t	cs_ndir;		/* number of directories */
244 	int32_t	cs_nbfree;		/* number of free blocks */
245 	int32_t	cs_nifree;		/* number of free inodes */
246 	int32_t	cs_nffree;		/* number of free frags */
247 };
248 struct csum_total {
249 	int64_t	cs_ndir;		/* number of directories */
250 	int64_t	cs_nbfree;		/* number of free blocks */
251 	int64_t	cs_nifree;		/* number of free inodes */
252 	int64_t	cs_nffree;		/* number of free frags */
253 	int64_t	cs_numclusters;		/* number of free clusters */
254 	int64_t	cs_spare[3];		/* future expansion */
255 };
256 
257 /*
258  * Super block for an FFS filesystem.
259  */
260 struct fs {
261 	int32_t	 fs_firstfield;		/* historic filesystem linked list, */
262 	int32_t	 fs_unused_1;		/*     used for incore super blocks */
263 	int32_t	 fs_sblkno;		/* offset of super-block in filesys */
264 	int32_t	 fs_cblkno;		/* offset of cyl-block in filesys */
265 	int32_t	 fs_iblkno;		/* offset of inode-blocks in filesys */
266 	int32_t	 fs_dblkno;		/* offset of first data after cg */
267 	int32_t	 fs_old_cgoffset;	/* cylinder group offset in cylinder */
268 	int32_t	 fs_old_cgmask;		/* used to calc mod fs_ntrak */
269 	int32_t  fs_old_time;		/* last time written */
270 	int32_t	 fs_old_size;		/* number of blocks in fs */
271 	int32_t	 fs_old_dsize;		/* number of data blocks in fs */
272 	u_int32_t fs_ncg;		/* number of cylinder groups */
273 	int32_t	 fs_bsize;		/* size of basic blocks in fs */
274 	int32_t	 fs_fsize;		/* size of frag blocks in fs */
275 	int32_t	 fs_frag;		/* number of frags in a block in fs */
276 /* these are configuration parameters */
277 	int32_t	 fs_minfree;		/* minimum percentage of free blocks */
278 	int32_t	 fs_old_rotdelay;	/* num of ms for optimal next block */
279 	int32_t	 fs_old_rps;		/* disk revolutions per second */
280 /* these fields can be computed from the others */
281 	int32_t	 fs_bmask;		/* ``blkoff'' calc of blk offsets */
282 	int32_t	 fs_fmask;		/* ``fragoff'' calc of frag offsets */
283 	int32_t	 fs_bshift;		/* ``lblkno'' calc of logical blkno */
284 	int32_t	 fs_fshift;		/* ``numfrags'' calc number of frags */
285 /* these are configuration parameters */
286 	int32_t	 fs_maxcontig;		/* max number of contiguous blks */
287 	int32_t	 fs_maxbpg;		/* max number of blks per cyl group */
288 /* these fields can be computed from the others */
289 	int32_t	 fs_fragshift;		/* block to frag shift */
290 	int32_t	 fs_fsbtodb;		/* fsbtodb and dbtofsb shift constant */
291 	int32_t	 fs_sbsize;		/* actual size of super block */
292 	int32_t	 fs_spare1[2];		/* old fs_csmask */
293 					/* old fs_csshift */
294 	int32_t	 fs_nindir;		/* value of NINDIR */
295 	u_int32_t fs_inopb;		/* value of INOPB */
296 	int32_t	 fs_old_nspf;		/* value of NSPF */
297 /* yet another configuration parameter */
298 	int32_t	 fs_optim;		/* optimization preference, see below */
299 	int32_t	 fs_old_npsect;		/* # sectors/track including spares */
300 	int32_t	 fs_old_interleave;	/* hardware sector interleave */
301 	int32_t	 fs_old_trackskew;	/* sector 0 skew, per track */
302 	int32_t	 fs_id[2];		/* unique filesystem id */
303 /* sizes determined by number of cylinder groups and their sizes */
304 	int32_t	 fs_old_csaddr;		/* blk addr of cyl grp summary area */
305 	int32_t	 fs_cssize;		/* size of cyl grp summary area */
306 	int32_t	 fs_cgsize;		/* cylinder group size */
307 	int32_t	 fs_spare2;		/* old fs_ntrak */
308 	int32_t	 fs_old_nsect;		/* sectors per track */
309 	int32_t  fs_old_spc;		/* sectors per cylinder */
310 	int32_t	 fs_old_ncyl;		/* cylinders in filesystem */
311 	int32_t	 fs_old_cpg;		/* cylinders per group */
312 	u_int32_t fs_ipg;		/* inodes per group */
313 	int32_t	 fs_fpg;		/* blocks per group * fs_frag */
314 /* this data must be re-computed after crashes */
315 	struct	csum fs_old_cstotal;	/* cylinder summary information */
316 /* these fields are cleared at mount time */
317 	int8_t   fs_fmod;		/* super block modified flag */
318 	int8_t   fs_clean;		/* filesystem is clean flag */
319 	int8_t 	 fs_ronly;		/* mounted read-only flag */
320 	int8_t   fs_old_flags;		/* old FS_ flags */
321 	u_char	 fs_fsmnt[MAXMNTLEN];	/* name mounted on */
322 	u_char	 fs_volname[MAXVOLLEN];	/* volume name */
323 	u_int64_t fs_swuid;		/* system-wide uid */
324 	int32_t  fs_pad;		/* due to alignment of fs_swuid */
325 /* these fields retain the current block allocation info */
326 	int32_t	 fs_cgrotor;		/* last cg searched */
327 	void 	*fs_ocsp[NOCSPTRS];	/* padding; was list of fs_cs buffers */
328 	u_int8_t *fs_contigdirs;	/* (u) # of contig. allocated dirs */
329 	struct	csum *fs_csp;		/* (u) cg summary info buffer */
330 	int32_t	*fs_maxcluster;		/* (u) max cluster in each cyl group */
331 	u_int	*fs_active;		/* (u) used by snapshots to track fs */
332 	int32_t	 fs_old_cpc;		/* cyl per cycle in postbl */
333 	int32_t	 fs_maxbsize;		/* maximum blocking factor permitted */
334 	int64_t	 fs_unrefs;		/* number of unreferenced inodes */
335 	int64_t  fs_providersize;	/* size of underlying GEOM provider */
336 	int64_t	 fs_metaspace;		/* size of area reserved for metadata */
337 	int64_t	 fs_sparecon64[14];	/* old rotation block list head */
338 	int64_t	 fs_sblockloc;		/* byte offset of standard superblock */
339 	struct	csum_total fs_cstotal;	/* (u) cylinder summary information */
340 	ufs_time_t fs_time;		/* last time written */
341 	int64_t	 fs_size;		/* number of blocks in fs */
342 	int64_t	 fs_dsize;		/* number of data blocks in fs */
343 	ufs2_daddr_t fs_csaddr;		/* blk addr of cyl grp summary area */
344 	int64_t	 fs_pendingblocks;	/* (u) blocks being freed */
345 	u_int32_t fs_pendinginodes;	/* (u) inodes being freed */
346 	uint32_t fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */
347 	u_int32_t fs_avgfilesize;	/* expected average file size */
348 	u_int32_t fs_avgfpdir;		/* expected # of files per directory */
349 	int32_t	 fs_save_cgsize;	/* save real cg size to use fs_bsize */
350 	ufs_time_t fs_mtime;		/* Last mount or fsck time. */
351 	int32_t  fs_sujfree;		/* SUJ free list */
352 	int32_t	 fs_sparecon32[23];	/* reserved for future constants */
353 	int32_t  fs_flags;		/* see FS_ flags below */
354 	int32_t	 fs_contigsumsize;	/* size of cluster summary array */
355 	int32_t	 fs_maxsymlinklen;	/* max length of an internal symlink */
356 	int32_t	 fs_old_inodefmt;	/* format of on-disk inodes */
357 	u_int64_t fs_maxfilesize;	/* maximum representable file size */
358 	int64_t	 fs_qbmask;		/* ~fs_bmask for use with 64-bit size */
359 	int64_t	 fs_qfmask;		/* ~fs_fmask for use with 64-bit size */
360 	int32_t	 fs_state;		/* validate fs_clean field */
361 	int32_t	 fs_old_postblformat;	/* format of positional layout tables */
362 	int32_t	 fs_old_nrpos;		/* number of rotational positions */
363 	int32_t	 fs_spare5[2];		/* old fs_postbloff */
364 					/* old fs_rotbloff */
365 	int32_t	 fs_magic;		/* magic number */
366 };
367 
368 /* Sanity checking. */
369 #ifdef CTASSERT
370 CTASSERT(sizeof(struct fs) == 1376);
371 #endif
372 
373 /*
374  * Filesystem identification
375  */
376 #define	FS_UFS1_MAGIC	0x011954	/* UFS1 fast filesystem magic number */
377 #define	FS_UFS2_MAGIC	0x19540119	/* UFS2 fast filesystem magic number */
378 #define	FS_BAD_MAGIC	0x19960408	/* UFS incomplete newfs magic number */
379 #define	FS_OKAY		0x7c269d38	/* superblock checksum */
380 #define	FS_42INODEFMT	-1		/* 4.2BSD inode format */
381 #define	FS_44INODEFMT	2		/* 4.4BSD inode format */
382 
383 /*
384  * Preference for optimization.
385  */
386 #define	FS_OPTTIME	0	/* minimize allocation time */
387 #define	FS_OPTSPACE	1	/* minimize disk fragmentation */
388 
389 /*
390  * Filesystem flags.
391  *
392  * The FS_UNCLEAN flag is set by the kernel when the filesystem was
393  * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates
394  * that the filesystem should be managed by the soft updates code.
395  * Note that the FS_NEEDSFSCK flag is set and cleared only by the
396  * fsck utility. It is set when background fsck finds an unexpected
397  * inconsistency which requires a traditional foreground fsck to be
398  * run. Such inconsistencies should only be found after an uncorrectable
399  * disk error. A foreground fsck will clear the FS_NEEDSFSCK flag when
400  * it has successfully cleaned up the filesystem. The kernel uses this
401  * flag to enforce that inconsistent filesystems be mounted read-only.
402  * The FS_INDEXDIRS flag when set indicates that the kernel maintains
403  * on-disk auxiliary indexes (such as B-trees) for speeding directory
404  * accesses. Kernels that do not support auxiliary indicies clear the
405  * flag to indicate that the indicies need to be rebuilt (by fsck) before
406  * they can be used.
407  *
408  * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled
409  * for the file system, so they should be loaded from extended attributes,
410  * observed for access control purposes, and be administered by object
411  * owners.  FS_NFS4ACLS indicates that NFSv4 ACLs are administratively
412  * enabled.  This flag is mutually exclusive with FS_ACLS.  FS_MULTILABEL
413  * indicates that the TrustedBSD MAC Framework should attempt to back MAC
414  * labels into extended attributes on the file system rather than maintain
415  * a single mount label for all objects.
416  */
417 #define	FS_UNCLEAN	0x0001	/* filesystem not clean at mount */
418 #define	FS_DOSOFTDEP	0x0002	/* filesystem using soft dependencies */
419 #define	FS_NEEDSFSCK	0x0004	/* filesystem needs sync fsck before mount */
420 #define	FS_SUJ       	0x0008	/* Filesystem using softupdate journal */
421 #define	FS_ACLS		0x0010	/* file system has POSIX.1e ACLs enabled */
422 #define	FS_MULTILABEL	0x0020	/* file system is MAC multi-label */
423 #define	FS_GJOURNAL	0x0040	/* gjournaled file system */
424 #define	FS_FLAGS_UPDATED 0x0080	/* flags have been moved to new location */
425 #define	FS_NFS4ACLS	0x0100	/* file system has NFSv4 ACLs enabled */
426 #define	FS_INDEXDIRS	0x0200	/* kernel supports indexed directories */
427 #define	FS_TRIM		0x0400	/* issue BIO_DELETE for deleted blocks */
428 
429 /*
430  * Macros to access bits in the fs_active array.
431  */
432 #define	ACTIVECGNUM(fs, cg)	((fs)->fs_active[(cg) / (NBBY * sizeof(int))])
433 #define	ACTIVECGOFF(cg)		(1 << ((cg) % (NBBY * sizeof(int))))
434 #define	ACTIVESET(fs, cg)	do {					\
435 	if ((fs)->fs_active)						\
436 		ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg));		\
437 } while (0)
438 #define	ACTIVECLEAR(fs, cg)	do {					\
439 	if ((fs)->fs_active)						\
440 		ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg));		\
441 } while (0)
442 
443 /*
444  * The size of a cylinder group is calculated by CGSIZE. The maximum size
445  * is limited by the fact that cylinder groups are at most one block.
446  * Its size is derived from the size of the maps maintained in the
447  * cylinder group and the (struct cg) size.
448  */
449 #define	CGSIZE(fs) \
450     /* base cg */	(sizeof(struct cg) + sizeof(int32_t) + \
451     /* old btotoff */	(fs)->fs_old_cpg * sizeof(int32_t) + \
452     /* old boff */	(fs)->fs_old_cpg * sizeof(u_int16_t) + \
453     /* inode map */	howmany((fs)->fs_ipg, NBBY) + \
454     /* block map */	howmany((fs)->fs_fpg, NBBY) +\
455     /* if present */	((fs)->fs_contigsumsize <= 0 ? 0 : \
456     /* cluster sum */	(fs)->fs_contigsumsize * sizeof(int32_t) + \
457     /* cluster map */	howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY)))
458 
459 /*
460  * The minimal number of cylinder groups that should be created.
461  */
462 #define	MINCYLGRPS	4
463 
464 /*
465  * Convert cylinder group to base address of its global summary info.
466  */
467 #define	fs_cs(fs, indx) fs_csp[indx]
468 
469 /*
470  * Cylinder group block for a filesystem.
471  */
472 #define	CG_MAGIC	0x090255
473 struct cg {
474 	int32_t	 cg_firstfield;		/* historic cyl groups linked list */
475 	int32_t	 cg_magic;		/* magic number */
476 	int32_t  cg_old_time;		/* time last written */
477 	u_int32_t cg_cgx;		/* we are the cgx'th cylinder group */
478 	int16_t	 cg_old_ncyl;		/* number of cyl's this cg */
479 	int16_t  cg_old_niblk;		/* number of inode blocks this cg */
480 	u_int32_t cg_ndblk;		/* number of data blocks this cg */
481 	struct	 csum cg_cs;		/* cylinder summary information */
482 	u_int32_t cg_rotor;		/* position of last used block */
483 	u_int32_t cg_frotor;		/* position of last used frag */
484 	u_int32_t cg_irotor;		/* position of last used inode */
485 	u_int32_t cg_frsum[MAXFRAG];	/* counts of available frags */
486 	int32_t	 cg_old_btotoff;	/* (int32) block totals per cylinder */
487 	int32_t	 cg_old_boff;		/* (u_int16) free block positions */
488 	u_int32_t cg_iusedoff;		/* (u_int8) used inode map */
489 	u_int32_t cg_freeoff;		/* (u_int8) free block map */
490 	u_int32_t cg_nextfreeoff;	/* (u_int8) next available space */
491 	u_int32_t cg_clustersumoff;	/* (u_int32) counts of avail clusters */
492 	u_int32_t cg_clusteroff;		/* (u_int8) free cluster map */
493 	u_int32_t cg_nclusterblks;	/* number of clusters this cg */
494 	u_int32_t cg_niblk;		/* number of inode blocks this cg */
495 	u_int32_t cg_initediblk;		/* last initialized inode */
496 	u_int32_t cg_unrefs;		/* number of unreferenced inodes */
497 	int32_t	 cg_sparecon32[2];	/* reserved for future use */
498 	ufs_time_t cg_time;		/* time last written */
499 	int64_t	 cg_sparecon64[3];	/* reserved for future use */
500 	u_int8_t cg_space[1];		/* space for cylinder group maps */
501 /* actually longer */
502 };
503 
504 /*
505  * Macros for access to cylinder group array structures
506  */
507 #define	cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC)
508 #define	cg_inosused(cgp) \
509     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_iusedoff))
510 #define	cg_blksfree(cgp) \
511     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_freeoff))
512 #define	cg_clustersfree(cgp) \
513     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_clusteroff))
514 #define	cg_clustersum(cgp) \
515     ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff))
516 
517 /*
518  * Turn filesystem block numbers into disk block addresses.
519  * This maps filesystem blocks to device size blocks.
520  */
521 #define	fsbtodb(fs, b)	((daddr_t)(b) << (fs)->fs_fsbtodb)
522 #define	dbtofsb(fs, b)	((b) >> (fs)->fs_fsbtodb)
523 
524 /*
525  * Cylinder group macros to locate things in cylinder groups.
526  * They calc filesystem addresses of cylinder group data structures.
527  */
528 #define	cgbase(fs, c)	(((ufs2_daddr_t)(fs)->fs_fpg) * (c))
529 #define	cgdata(fs, c)	(cgdmin(fs, c) + (fs)->fs_metaspace)	/* data zone */
530 #define	cgmeta(fs, c)	(cgdmin(fs, c))				/* meta data */
531 #define	cgdmin(fs, c)	(cgstart(fs, c) + (fs)->fs_dblkno)	/* 1st data */
532 #define	cgimin(fs, c)	(cgstart(fs, c) + (fs)->fs_iblkno)	/* inode blk */
533 #define	cgsblock(fs, c)	(cgstart(fs, c) + (fs)->fs_sblkno)	/* super blk */
534 #define	cgtod(fs, c)	(cgstart(fs, c) + (fs)->fs_cblkno)	/* cg block */
535 #define	cgstart(fs, c)							\
536        ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) :		\
537        (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask))))
538 
539 /*
540  * Macros for handling inode numbers:
541  *     inode number to filesystem block offset.
542  *     inode number to cylinder group number.
543  *     inode number to filesystem block address.
544  */
545 #define	ino_to_cg(fs, x)	(((ino_t)(x)) / (fs)->fs_ipg)
546 #define	ino_to_fsba(fs, x)						\
547 	((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) +		\
548 	    (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs))))))
549 #define	ino_to_fsbo(fs, x)	(((ino_t)(x)) % INOPB(fs))
550 
551 /*
552  * Give cylinder group number for a filesystem block.
553  * Give cylinder group block number for a filesystem block.
554  */
555 #define	dtog(fs, d)	((d) / (fs)->fs_fpg)
556 #define	dtogd(fs, d)	((d) % (fs)->fs_fpg)
557 
558 /*
559  * Extract the bits for a block from a map.
560  * Compute the cylinder and rotational position of a cyl block addr.
561  */
562 #define	blkmap(fs, map, loc) \
563     (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag)))
564 
565 /*
566  * The following macros optimize certain frequently calculated
567  * quantities by using shifts and masks in place of divisions
568  * modulos and multiplications.
569  */
570 #define	blkoff(fs, loc)		/* calculates (loc % fs->fs_bsize) */ \
571 	((loc) & ~((fs)->fs_bmask))
572 #define	fragoff(fs, loc)	/* calculates (loc % fs->fs_fsize) */ \
573 	((loc) & ~((fs)->fs_fmask))
574 #define	lfragtosize(fs, frag)	/* calculates ((off_t)frag * fs->fs_fsize) */ \
575 	(((off_t)(frag)) << (fs)->fs_fshift)
576 #define	lblktosize(fs, blk)	/* calculates ((off_t)blk * fs->fs_bsize) */ \
577 	(((off_t)(blk)) << (fs)->fs_bshift)
578 /* Use this only when `blk' is known to be small, e.g., < NDADDR. */
579 #define	smalllblktosize(fs, blk)    /* calculates (blk * fs->fs_bsize) */ \
580 	((blk) << (fs)->fs_bshift)
581 #define	lblkno(fs, loc)		/* calculates (loc / fs->fs_bsize) */ \
582 	((loc) >> (fs)->fs_bshift)
583 #define	numfrags(fs, loc)	/* calculates (loc / fs->fs_fsize) */ \
584 	((loc) >> (fs)->fs_fshift)
585 #define	blkroundup(fs, size)	/* calculates roundup(size, fs->fs_bsize) */ \
586 	(((size) + (fs)->fs_bsize - 1) & (fs)->fs_bmask)
587 #define	fragroundup(fs, size)	/* calculates roundup(size, fs->fs_fsize) */ \
588 	(((size) + (fs)->fs_fsize - 1) & (fs)->fs_fmask)
589 #define	fragstoblks(fs, frags)	/* calculates (frags / fs->fs_frag) */ \
590 	((frags) >> (fs)->fs_fragshift)
591 #define	blkstofrags(fs, blks)	/* calculates (blks * fs->fs_frag) */ \
592 	((blks) << (fs)->fs_fragshift)
593 #define	fragnum(fs, fsb)	/* calculates (fsb % fs->fs_frag) */ \
594 	((fsb) & ((fs)->fs_frag - 1))
595 #define	blknum(fs, fsb)		/* calculates rounddown(fsb, fs->fs_frag) */ \
596 	((fsb) &~ ((fs)->fs_frag - 1))
597 
598 /*
599  * Determine the number of available frags given a
600  * percentage to hold in reserve.
601  */
602 #define	freespace(fs, percentreserved) \
603 	(blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \
604 	(fs)->fs_cstotal.cs_nffree - \
605 	(((off_t)((fs)->fs_dsize)) * (percentreserved) / 100))
606 
607 /*
608  * Determining the size of a file block in the filesystem.
609  */
610 #define	blksize(fs, ip, lbn) \
611 	(((lbn) >= NDADDR || (ip)->i_size >= smalllblktosize(fs, (lbn) + 1)) \
612 	    ? (fs)->fs_bsize \
613 	    : (fragroundup(fs, blkoff(fs, (ip)->i_size))))
614 #define	sblksize(fs, size, lbn) \
615 	(((lbn) >= NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \
616 	  ? (fs)->fs_bsize \
617 	  : (fragroundup(fs, blkoff(fs, (size)))))
618 
619 /*
620  * Number of indirects in a filesystem block.
621  */
622 #define	NINDIR(fs)	((fs)->fs_nindir)
623 
624 /*
625  * Indirect lbns are aligned on NDADDR addresses where single indirects
626  * are the negated address of the lowest lbn reachable, double indirects
627  * are this lbn - 1 and triple indirects are this lbn - 2.  This yields
628  * an unusual bit order to determine level.
629  */
630 static inline int
lbn_level(ufs_lbn_t lbn)631 lbn_level(ufs_lbn_t lbn)
632 {
633 	if (lbn >= 0)
634 		return 0;
635 	switch (lbn & 0x3) {
636 	case 0:
637 		return (0);
638 	case 1:
639 		break;
640 	case 2:
641 		return (2);
642 	case 3:
643 		return (1);
644 	default:
645 		break;
646 	}
647 	return (-1);
648 }
649 
650 static inline ufs_lbn_t
lbn_offset(struct fs * fs,int level)651 lbn_offset(struct fs *fs, int level)
652 {
653 	ufs_lbn_t res;
654 
655 	for (res = 1; level > 0; level--)
656 		res *= NINDIR(fs);
657 	return (res);
658 }
659 
660 /*
661  * Number of inodes in a secondary storage block/fragment.
662  */
663 #define	INOPB(fs)	((fs)->fs_inopb)
664 #define	INOPF(fs)	((fs)->fs_inopb >> (fs)->fs_fragshift)
665 
666 /*
667  * Softdep journal record format.
668  */
669 
670 #define	JOP_ADDREF	1	/* Add a reference to an inode. */
671 #define	JOP_REMREF	2	/* Remove a reference from an inode. */
672 #define	JOP_NEWBLK	3	/* Allocate a block. */
673 #define	JOP_FREEBLK	4	/* Free a block or a tree of blocks. */
674 #define	JOP_MVREF	5	/* Move a reference from one off to another. */
675 #define	JOP_TRUNC	6	/* Partial truncation record. */
676 #define	JOP_SYNC	7	/* fsync() complete record. */
677 
678 #define	JREC_SIZE	32	/* Record and segment header size. */
679 
680 #define	SUJ_MIN		(4 * 1024 * 1024)	/* Minimum journal size */
681 #define	SUJ_MAX		(32 * 1024 * 1024)	/* Maximum journal size */
682 #define	SUJ_FILE	".sujournal"		/* Journal file name */
683 
684 /*
685  * Size of the segment record header.  There is at most one for each disk
686  * block in the journal.  The segment header is followed by an array of
687  * records.  fsck depends on the first element in each record being 'op'
688  * and the second being 'ino'.  Segments may span multiple disk blocks but
689  * the header is present on each.
690  */
691 struct jsegrec {
692 	uint64_t	jsr_seq;	/* Our sequence number */
693 	uint64_t	jsr_oldest;	/* Oldest valid sequence number */
694 	uint16_t	jsr_cnt;	/* Count of valid records */
695 	uint16_t	jsr_blocks;	/* Count of device bsize blocks. */
696 	uint32_t	jsr_crc;	/* 32bit crc of the valid space */
697 	ufs_time_t	jsr_time;	/* timestamp for mount instance */
698 };
699 
700 /*
701  * Reference record.  Records a single link count modification.
702  */
703 struct jrefrec {
704 	uint32_t	jr_op;
705 	uint32_t	jr_ino;
706 	uint32_t	jr_parent;
707 	uint16_t	jr_nlink;
708 	uint16_t	jr_mode;
709 	int64_t		jr_diroff;
710 	uint64_t	jr_unused;
711 };
712 
713 /*
714  * Move record.  Records a reference moving within a directory block.  The
715  * nlink is unchanged but we must search both locations.
716  */
717 struct jmvrec {
718 	uint32_t	jm_op;
719 	uint32_t	jm_ino;
720 	uint32_t	jm_parent;
721 	uint16_t	jm_unused;
722 	int64_t		jm_oldoff;
723 	int64_t		jm_newoff;
724 };
725 
726 /*
727  * Block record.  A set of frags or tree of blocks starting at an indirect are
728  * freed or a set of frags are allocated.
729  */
730 struct jblkrec {
731 	uint32_t	jb_op;
732 	uint32_t	jb_ino;
733 	ufs2_daddr_t	jb_blkno;
734 	ufs_lbn_t	jb_lbn;
735 	uint16_t	jb_frags;
736 	uint16_t	jb_oldfrags;
737 	uint32_t	jb_unused;
738 };
739 
740 /*
741  * Truncation record.  Records a partial truncation so that it may be
742  * completed at check time.  Also used for sync records.
743  */
744 struct jtrncrec {
745 	uint32_t	jt_op;
746 	uint32_t	jt_ino;
747 	int64_t		jt_size;
748 	uint32_t	jt_extsize;
749 	uint32_t	jt_pad[3];
750 };
751 
752 union jrec {
753 	struct jsegrec	rec_jsegrec;
754 	struct jrefrec	rec_jrefrec;
755 	struct jmvrec	rec_jmvrec;
756 	struct jblkrec	rec_jblkrec;
757 	struct jtrncrec	rec_jtrncrec;
758 };
759 
760 #ifdef CTASSERT
761 CTASSERT(sizeof(struct jsegrec) == JREC_SIZE);
762 CTASSERT(sizeof(struct jrefrec) == JREC_SIZE);
763 CTASSERT(sizeof(struct jmvrec) == JREC_SIZE);
764 CTASSERT(sizeof(struct jblkrec) == JREC_SIZE);
765 CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE);
766 CTASSERT(sizeof(union jrec) == JREC_SIZE);
767 #endif
768 
769 extern int inside[], around[];
770 extern u_char *fragtbl[];
771 
772 /*
773  * IOCTLs used for filesystem write suspension.
774  */
775 #define	UFSSUSPEND	_IOW('U', 1, fsid_t)
776 #define	UFSRESUME	_IO('U', 2)
777 
778 #endif
779