xref: /freebsd/sys/ufs/ffs/fs.h (revision 123af6ec70016f5556da5972d4d63c7d175c06d3)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1982, 1986, 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. Neither the name of the University nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  *
31  *	@(#)fs.h	8.13 (Berkeley) 3/21/95
32  * $FreeBSD$
33  */
34 
35 #ifndef _UFS_FFS_FS_H_
36 #define	_UFS_FFS_FS_H_
37 
38 #include <sys/mount.h>
39 #include <ufs/ufs/dinode.h>
40 
41 /*
42  * Each disk drive contains some number of filesystems.
43  * A filesystem consists of a number of cylinder groups.
44  * Each cylinder group has inodes and data.
45  *
46  * A filesystem is described by its super-block, which in turn
47  * describes the cylinder groups.  The super-block is critical
48  * data and is replicated in each cylinder group to protect against
49  * catastrophic loss.  This is done at `newfs' time and the critical
50  * super-block data does not change, so the copies need not be
51  * referenced further unless disaster strikes.
52  *
53  * For filesystem fs, the offsets of the various blocks of interest
54  * are given in the super block as:
55  *	[fs->fs_sblkno]		Super-block
56  *	[fs->fs_cblkno]		Cylinder group block
57  *	[fs->fs_iblkno]		Inode blocks
58  *	[fs->fs_dblkno]		Data blocks
59  * The beginning of cylinder group cg in fs, is given by
60  * the ``cgbase(fs, cg)'' macro.
61  *
62  * Depending on the architecture and the media, the superblock may
63  * reside in any one of four places. For tiny media where every block
64  * counts, it is placed at the very front of the partition. Historically,
65  * UFS1 placed it 8K from the front to leave room for the disk label and
66  * a small bootstrap. For UFS2 it got moved to 64K from the front to leave
67  * room for the disk label and a bigger bootstrap, and for really piggy
68  * systems we check at 256K from the front if the first three fail. In
69  * all cases the size of the superblock will be SBLOCKSIZE. All values are
70  * given in byte-offset form, so they do not imply a sector size. The
71  * SBLOCKSEARCH specifies the order in which the locations should be searched.
72  */
73 #define	SBLOCK_FLOPPY	     0
74 #define	SBLOCK_UFS1	  8192
75 #define	SBLOCK_UFS2	 65536
76 #define	SBLOCK_PIGGY	262144
77 #define	SBLOCKSIZE	  8192
78 #define	SBLOCKSEARCH \
79 	{ SBLOCK_UFS2, SBLOCK_UFS1, SBLOCK_FLOPPY, SBLOCK_PIGGY, -1 }
80 
81 /*
82  * Max number of fragments per block. This value is NOT tweakable.
83  */
84 #define	MAXFRAG 	8
85 
86 /*
87  * Addresses stored in inodes are capable of addressing fragments
88  * of `blocks'. File system blocks of at most size MAXBSIZE can
89  * be optionally broken into 2, 4, or 8 pieces, each of which is
90  * addressable; these pieces may be DEV_BSIZE, or some multiple of
91  * a DEV_BSIZE unit.
92  *
93  * Large files consist of exclusively large data blocks.  To avoid
94  * undue wasted disk space, the last data block of a small file may be
95  * allocated as only as many fragments of a large block as are
96  * necessary.  The filesystem format retains only a single pointer
97  * to such a fragment, which is a piece of a single large block that
98  * has been divided.  The size of such a fragment is determinable from
99  * information in the inode, using the ``blksize(fs, ip, lbn)'' macro.
100  *
101  * The filesystem records space availability at the fragment level;
102  * to determine block availability, aligned fragments are examined.
103  */
104 
105 /*
106  * MINBSIZE is the smallest allowable block size.
107  * In order to insure that it is possible to create files of size
108  * 2^32 with only two levels of indirection, MINBSIZE is set to 4096.
109  * MINBSIZE must be big enough to hold a cylinder group block,
110  * thus changes to (struct cg) must keep its size within MINBSIZE.
111  * Note that super blocks are always of size SBLOCKSIZE,
112  * and that both SBLOCKSIZE and MAXBSIZE must be >= MINBSIZE.
113  */
114 #define	MINBSIZE	4096
115 
116 /*
117  * The path name on which the filesystem is mounted is maintained
118  * in fs_fsmnt. MAXMNTLEN defines the amount of space allocated in
119  * the super block for this name.
120  */
121 #define	MAXMNTLEN	468
122 
123 /*
124  * The volume name for this filesystem is maintained in fs_volname.
125  * MAXVOLLEN defines the length of the buffer allocated.
126  */
127 #define	MAXVOLLEN	32
128 
129 /*
130  * There is a 128-byte region in the superblock reserved for in-core
131  * pointers to summary information. Originally this included an array
132  * of pointers to blocks of struct csum; now there are just a few
133  * pointers and the remaining space is padded with fs_ocsp[].
134  *
135  * NOCSPTRS determines the size of this padding. One pointer (fs_csp)
136  * is taken away to point to a contiguous array of struct csum for
137  * all cylinder groups; a second (fs_maxcluster) points to an array
138  * of cluster sizes that is computed as cylinder groups are inspected,
139  * and the third points to an array that tracks the creation of new
140  * directories. A fourth pointer, fs_active, is used when creating
141  * snapshots; it points to a bitmap of cylinder groups for which the
142  * free-block bitmap has changed since the snapshot operation began.
143  */
144 #define	NOCSPTRS	((128 / sizeof(void *)) - 4)
145 
146 /*
147  * A summary of contiguous blocks of various sizes is maintained
148  * in each cylinder group. Normally this is set by the initial
149  * value of fs_maxcontig. To conserve space, a maximum summary size
150  * is set by FS_MAXCONTIG.
151  */
152 #define	FS_MAXCONTIG	16
153 
154 /*
155  * MINFREE gives the minimum acceptable percentage of filesystem
156  * blocks which may be free. If the freelist drops below this level
157  * only the superuser may continue to allocate blocks. This may
158  * be set to 0 if no reserve of free blocks is deemed necessary,
159  * however throughput drops by fifty percent if the filesystem
160  * is run at between 95% and 100% full; thus the minimum default
161  * value of fs_minfree is 5%. However, to get good clustering
162  * performance, 10% is a better choice. hence we use 10% as our
163  * default value. With 10% free space, fragmentation is not a
164  * problem, so we choose to optimize for time.
165  */
166 #define	MINFREE		8
167 #define	DEFAULTOPT	FS_OPTTIME
168 
169 /*
170  * Grigoriy Orlov <gluk@ptci.ru> has done some extensive work to fine
171  * tune the layout preferences for directories within a filesystem.
172  * His algorithm can be tuned by adjusting the following parameters
173  * which tell the system the average file size and the average number
174  * of files per directory. These defaults are well selected for typical
175  * filesystems, but may need to be tuned for odd cases like filesystems
176  * being used for squid caches or news spools.
177  */
178 #define	AVFILESIZ	16384	/* expected average file size */
179 #define	AFPDIR		64	/* expected number of files per directory */
180 
181 /*
182  * The maximum number of snapshot nodes that can be associated
183  * with each filesystem. This limit affects only the number of
184  * snapshot files that can be recorded within the superblock so
185  * that they can be found when the filesystem is mounted. However,
186  * maintaining too many will slow the filesystem performance, so
187  * having this limit is a good idea.
188  */
189 #define	FSMAXSNAP 20
190 
191 /*
192  * Used to identify special blocks in snapshots:
193  *
194  * BLK_NOCOPY - A block that was unallocated at the time the snapshot
195  *	was taken, hence does not need to be copied when written.
196  * BLK_SNAP - A block held by another snapshot that is not needed by this
197  *	snapshot. When the other snapshot is freed, the BLK_SNAP entries
198  *	are converted to BLK_NOCOPY. These are needed to allow fsck to
199  *	identify blocks that are in use by other snapshots (which are
200  *	expunged from this snapshot).
201  */
202 #define	BLK_NOCOPY ((ufs2_daddr_t)(1))
203 #define	BLK_SNAP ((ufs2_daddr_t)(2))
204 
205 /*
206  * Sysctl values for the fast filesystem.
207  */
208 #define	FFS_ADJ_REFCNT		 1	/* adjust inode reference count */
209 #define	FFS_ADJ_BLKCNT		 2	/* adjust inode used block count */
210 #define	FFS_BLK_FREE		 3	/* free range of blocks in map */
211 #define	FFS_DIR_FREE		 4	/* free specified dir inodes in map */
212 #define	FFS_FILE_FREE		 5	/* free specified file inodes in map */
213 #define	FFS_SET_FLAGS		 6	/* set filesystem flags */
214 #define	FFS_ADJ_NDIR		 7	/* adjust number of directories */
215 #define	FFS_ADJ_NBFREE		 8	/* adjust number of free blocks */
216 #define	FFS_ADJ_NIFREE		 9	/* adjust number of free inodes */
217 #define	FFS_ADJ_NFFREE		10 	/* adjust number of free frags */
218 #define	FFS_ADJ_NUMCLUSTERS	11	/* adjust number of free clusters */
219 #define	FFS_SET_CWD		12	/* set current directory */
220 #define	FFS_SET_DOTDOT		13	/* set inode number for ".." */
221 #define	FFS_UNLINK		14	/* remove a name in the filesystem */
222 #define	FFS_SET_INODE		15	/* update an on-disk inode */
223 #define	FFS_SET_BUFOUTPUT	16	/* set buffered writing on descriptor */
224 #define	FFS_SET_SIZE		17	/* set inode size */
225 #define	FFS_MAXID		17	/* number of valid ffs ids */
226 
227 /*
228  * Command structure passed in to the filesystem to adjust filesystem values.
229  */
230 #define	FFS_CMD_VERSION		0x19790518	/* version ID */
231 struct fsck_cmd {
232 	int32_t	version;	/* version of command structure */
233 	int32_t	handle;		/* reference to filesystem to be changed */
234 	int64_t	value;		/* inode or block number to be affected */
235 	int64_t	size;		/* amount or range to be adjusted */
236 	int64_t	spare;		/* reserved for future use */
237 };
238 
239 /*
240  * A recovery structure placed at the end of the boot block area by newfs
241  * that can be used by fsck to search for alternate superblocks.
242  */
243 struct fsrecovery {
244 	int32_t	fsr_magic;	/* magic number */
245 	int32_t	fsr_fsbtodb;	/* fsbtodb and dbtofsb shift constant */
246 	int32_t	fsr_sblkno;	/* offset of super-block in filesys */
247 	int32_t	fsr_fpg;	/* blocks per group * fs_frag */
248 	u_int32_t fsr_ncg;	/* number of cylinder groups */
249 };
250 
251 /*
252  * Per cylinder group information; summarized in blocks allocated
253  * from first cylinder group data blocks.  These blocks have to be
254  * read in from fs_csaddr (size fs_cssize) in addition to the
255  * super block.
256  */
257 struct csum {
258 	int32_t	cs_ndir;		/* number of directories */
259 	int32_t	cs_nbfree;		/* number of free blocks */
260 	int32_t	cs_nifree;		/* number of free inodes */
261 	int32_t	cs_nffree;		/* number of free frags */
262 };
263 struct csum_total {
264 	int64_t	cs_ndir;		/* number of directories */
265 	int64_t	cs_nbfree;		/* number of free blocks */
266 	int64_t	cs_nifree;		/* number of free inodes */
267 	int64_t	cs_nffree;		/* number of free frags */
268 	int64_t	cs_numclusters;		/* number of free clusters */
269 	int64_t	cs_spare[3];		/* future expansion */
270 };
271 
272 /*
273  * Super block for an FFS filesystem.
274  */
275 struct fs {
276 	int32_t	 fs_firstfield;		/* historic filesystem linked list, */
277 	int32_t	 fs_unused_1;		/*     used for incore super blocks */
278 	int32_t	 fs_sblkno;		/* offset of super-block in filesys */
279 	int32_t	 fs_cblkno;		/* offset of cyl-block in filesys */
280 	int32_t	 fs_iblkno;		/* offset of inode-blocks in filesys */
281 	int32_t	 fs_dblkno;		/* offset of first data after cg */
282 	int32_t	 fs_old_cgoffset;	/* cylinder group offset in cylinder */
283 	int32_t	 fs_old_cgmask;		/* used to calc mod fs_ntrak */
284 	int32_t  fs_old_time;		/* last time written */
285 	int32_t	 fs_old_size;		/* number of blocks in fs */
286 	int32_t	 fs_old_dsize;		/* number of data blocks in fs */
287 	u_int32_t fs_ncg;		/* number of cylinder groups */
288 	int32_t	 fs_bsize;		/* size of basic blocks in fs */
289 	int32_t	 fs_fsize;		/* size of frag blocks in fs */
290 	int32_t	 fs_frag;		/* number of frags in a block in fs */
291 /* these are configuration parameters */
292 	int32_t	 fs_minfree;		/* minimum percentage of free blocks */
293 	int32_t	 fs_old_rotdelay;	/* num of ms for optimal next block */
294 	int32_t	 fs_old_rps;		/* disk revolutions per second */
295 /* these fields can be computed from the others */
296 	int32_t	 fs_bmask;		/* ``blkoff'' calc of blk offsets */
297 	int32_t	 fs_fmask;		/* ``fragoff'' calc of frag offsets */
298 	int32_t	 fs_bshift;		/* ``lblkno'' calc of logical blkno */
299 	int32_t	 fs_fshift;		/* ``numfrags'' calc number of frags */
300 /* these are configuration parameters */
301 	int32_t	 fs_maxcontig;		/* max number of contiguous blks */
302 	int32_t	 fs_maxbpg;		/* max number of blks per cyl group */
303 /* these fields can be computed from the others */
304 	int32_t	 fs_fragshift;		/* block to frag shift */
305 	int32_t	 fs_fsbtodb;		/* fsbtodb and dbtofsb shift constant */
306 	int32_t	 fs_sbsize;		/* actual size of super block */
307 	int32_t	 fs_spare1[2];		/* old fs_csmask */
308 					/* old fs_csshift */
309 	int32_t	 fs_nindir;		/* value of NINDIR */
310 	u_int32_t fs_inopb;		/* value of INOPB */
311 	int32_t	 fs_old_nspf;		/* value of NSPF */
312 /* yet another configuration parameter */
313 	int32_t	 fs_optim;		/* optimization preference, see below */
314 	int32_t	 fs_old_npsect;		/* # sectors/track including spares */
315 	int32_t	 fs_old_interleave;	/* hardware sector interleave */
316 	int32_t	 fs_old_trackskew;	/* sector 0 skew, per track */
317 	int32_t	 fs_id[2];		/* unique filesystem id */
318 /* sizes determined by number of cylinder groups and their sizes */
319 	int32_t	 fs_old_csaddr;		/* blk addr of cyl grp summary area */
320 	int32_t	 fs_cssize;		/* size of cyl grp summary area */
321 	int32_t	 fs_cgsize;		/* cylinder group size */
322 	int32_t	 fs_spare2;		/* old fs_ntrak */
323 	int32_t	 fs_old_nsect;		/* sectors per track */
324 	int32_t  fs_old_spc;		/* sectors per cylinder */
325 	int32_t	 fs_old_ncyl;		/* cylinders in filesystem */
326 	int32_t	 fs_old_cpg;		/* cylinders per group */
327 	u_int32_t fs_ipg;		/* inodes per group */
328 	int32_t	 fs_fpg;		/* blocks per group * fs_frag */
329 /* this data must be re-computed after crashes */
330 	struct	csum fs_old_cstotal;	/* cylinder summary information */
331 /* these fields are cleared at mount time */
332 	int8_t   fs_fmod;		/* super block modified flag */
333 	int8_t   fs_clean;		/* filesystem is clean flag */
334 	int8_t 	 fs_ronly;		/* mounted read-only flag */
335 	int8_t   fs_old_flags;		/* old FS_ flags */
336 	u_char	 fs_fsmnt[MAXMNTLEN];	/* name mounted on */
337 	u_char	 fs_volname[MAXVOLLEN];	/* volume name */
338 	u_int64_t fs_swuid;		/* system-wide uid */
339 	int32_t  fs_pad;		/* due to alignment of fs_swuid */
340 /* these fields retain the current block allocation info */
341 	int32_t	 fs_cgrotor;		/* last cg searched */
342 	void 	*fs_ocsp[NOCSPTRS];	/* padding; was list of fs_cs buffers */
343 	u_int8_t *fs_contigdirs;	/* (u) # of contig. allocated dirs */
344 	struct	csum *fs_csp;		/* (u) cg summary info buffer */
345 	int32_t	*fs_maxcluster;		/* (u) max cluster in each cyl group */
346 	u_int	*fs_active;		/* (u) used by snapshots to track fs */
347 	int32_t	 fs_old_cpc;		/* cyl per cycle in postbl */
348 	int32_t	 fs_maxbsize;		/* maximum blocking factor permitted */
349 	int64_t	 fs_unrefs;		/* number of unreferenced inodes */
350 	int64_t  fs_providersize;	/* size of underlying GEOM provider */
351 	int64_t	 fs_metaspace;		/* size of area reserved for metadata */
352 	int64_t	 fs_sparecon64[13];	/* old rotation block list head */
353 	int64_t	 fs_sblockactualloc;	/* byte offset of this superblock */
354 	int64_t	 fs_sblockloc;		/* byte offset of standard superblock */
355 	struct	csum_total fs_cstotal;	/* (u) cylinder summary information */
356 	ufs_time_t fs_time;		/* last time written */
357 	int64_t	 fs_size;		/* number of blocks in fs */
358 	int64_t	 fs_dsize;		/* number of data blocks in fs */
359 	ufs2_daddr_t fs_csaddr;		/* blk addr of cyl grp summary area */
360 	int64_t	 fs_pendingblocks;	/* (u) blocks being freed */
361 	u_int32_t fs_pendinginodes;	/* (u) inodes being freed */
362 	uint32_t fs_snapinum[FSMAXSNAP];/* list of snapshot inode numbers */
363 	u_int32_t fs_avgfilesize;	/* expected average file size */
364 	u_int32_t fs_avgfpdir;		/* expected # of files per directory */
365 	int32_t	 fs_save_cgsize;	/* save real cg size to use fs_bsize */
366 	ufs_time_t fs_mtime;		/* Last mount or fsck time. */
367 	int32_t  fs_sujfree;		/* SUJ free list */
368 	int32_t	 fs_sparecon32[21];	/* reserved for future constants */
369 	u_int32_t fs_ckhash;		/* if CK_SUPERBLOCK, its check-hash */
370 	u_int32_t fs_metackhash;	/* metadata check-hash, see CK_ below */
371 	int32_t  fs_flags;		/* see FS_ flags below */
372 	int32_t	 fs_contigsumsize;	/* size of cluster summary array */
373 	int32_t	 fs_maxsymlinklen;	/* max length of an internal symlink */
374 	int32_t	 fs_old_inodefmt;	/* format of on-disk inodes */
375 	u_int64_t fs_maxfilesize;	/* maximum representable file size */
376 	int64_t	 fs_qbmask;		/* ~fs_bmask for use with 64-bit size */
377 	int64_t	 fs_qfmask;		/* ~fs_fmask for use with 64-bit size */
378 	int32_t	 fs_state;		/* validate fs_clean field */
379 	int32_t	 fs_old_postblformat;	/* format of positional layout tables */
380 	int32_t	 fs_old_nrpos;		/* number of rotational positions */
381 	int32_t	 fs_spare5[2];		/* old fs_postbloff */
382 					/* old fs_rotbloff */
383 	int32_t	 fs_magic;		/* magic number */
384 };
385 
386 /* Sanity checking. */
387 #ifdef CTASSERT
388 CTASSERT(sizeof(struct fs) == 1376);
389 #endif
390 
391 /*
392  * Filesystem identification
393  */
394 #define	FS_UFS1_MAGIC	0x011954	/* UFS1 fast filesystem magic number */
395 #define	FS_UFS2_MAGIC	0x19540119	/* UFS2 fast filesystem magic number */
396 #define	FS_BAD_MAGIC	0x19960408	/* UFS incomplete newfs magic number */
397 #define	FS_42INODEFMT	-1		/* 4.2BSD inode format */
398 #define	FS_44INODEFMT	2		/* 4.4BSD inode format */
399 
400 /*
401  * Preference for optimization.
402  */
403 #define	FS_OPTTIME	0	/* minimize allocation time */
404 #define	FS_OPTSPACE	1	/* minimize disk fragmentation */
405 
406 /*
407  * Filesystem flags.
408  *
409  * The FS_UNCLEAN flag is set by the kernel when the filesystem was
410  * mounted with fs_clean set to zero. The FS_DOSOFTDEP flag indicates
411  * that the filesystem should be managed by the soft updates code.
412  * Note that the FS_NEEDSFSCK flag is set and cleared only by the
413  * fsck utility. It is set when background fsck finds an unexpected
414  * inconsistency which requires a traditional foreground fsck to be
415  * run. Such inconsistencies should only be found after an uncorrectable
416  * disk error. A foreground fsck will clear the FS_NEEDSFSCK flag when
417  * it has successfully cleaned up the filesystem. The kernel uses this
418  * flag to enforce that inconsistent filesystems be mounted read-only.
419  * The FS_INDEXDIRS flag when set indicates that the kernel maintains
420  * on-disk auxiliary indexes (such as B-trees) for speeding directory
421  * accesses. Kernels that do not support auxiliary indices clear the
422  * flag to indicate that the indices need to be rebuilt (by fsck) before
423  * they can be used. When a filesystem is mounted, any flags not
424  * included in FS_SUPPORTED are cleared. This lets newer features
425  * know that the filesystem has been run on an older version of the
426  * filesystem and thus that data structures associated with those
427  * features are out-of-date and need to be rebuilt.
428  *
429  * FS_ACLS indicates that POSIX.1e ACLs are administratively enabled
430  * for the file system, so they should be loaded from extended attributes,
431  * observed for access control purposes, and be administered by object
432  * owners.  FS_NFS4ACLS indicates that NFSv4 ACLs are administratively
433  * enabled.  This flag is mutually exclusive with FS_ACLS.  FS_MULTILABEL
434  * indicates that the TrustedBSD MAC Framework should attempt to back MAC
435  * labels into extended attributes on the file system rather than maintain
436  * a single mount label for all objects.
437  */
438 #define	FS_UNCLEAN	0x00000001 /* filesystem not clean at mount */
439 #define	FS_DOSOFTDEP	0x00000002 /* filesystem using soft dependencies */
440 #define	FS_NEEDSFSCK	0x00000004 /* filesystem needs sync fsck before mount */
441 #define	FS_SUJ       	0x00000008 /* Filesystem using softupdate journal */
442 #define	FS_ACLS		0x00000010 /* file system has POSIX.1e ACLs enabled */
443 #define	FS_MULTILABEL	0x00000020 /* file system is MAC multi-label */
444 #define	FS_GJOURNAL	0x00000040 /* gjournaled file system */
445 #define	FS_FLAGS_UPDATED 0x0000080 /* flags have been moved to new location */
446 #define	FS_NFS4ACLS	0x00000100 /* file system has NFSv4 ACLs enabled */
447 #define	FS_METACKHASH	0x00000200 /* kernel supports metadata check hashes */
448 #define	FS_TRIM		0x00000400 /* issue BIO_DELETE for deleted blocks */
449 #define	FS_SUPPORTED	0x00FFFFFF /* supported flags, others cleared at mount*/
450 /*
451  * Things that we may someday support, but currently do not.
452  * These flags are all cleared so we know if we ran on a kernel
453  * that does not support them.
454  */
455 #define	FS_INDEXDIRS	0x01000000 /* kernel supports indexed directories */
456 #define	FS_VARBLKSIZE	0x02000000 /* kernel supports variable block sizes */
457 #define	FS_COOLOPT1	0x04000000 /* kernel supports cool option 1 */
458 #define	FS_COOLOPT2	0x08000000 /* kernel supports cool option 2 */
459 #define	FS_COOLOPT3	0x10000000 /* kernel supports cool option 3 */
460 #define	FS_COOLOPT4	0x20000000 /* kernel supports cool option 4 */
461 #define	FS_COOLOPT5	0x40000000 /* kernel supports cool option 5 */
462 #define	FS_COOLOPT6	0x80000000 /* kernel supports cool option 6 */
463 
464 /*
465  * The fs_metackhash field indicates the types of metadata check-hash
466  * that are maintained for a filesystem. Not all filesystems check-hash
467  * all metadata.
468  */
469 #define	CK_SUPERBLOCK	0x0001	/* the superblock */
470 #define	CK_CYLGRP	0x0002	/* the cylinder groups */
471 #define	CK_INODE	0x0004	/* inodes */
472 #define	CK_INDIR	0x0008	/* indirect blocks */
473 #define	CK_DIR		0x0010	/* directory contents */
474 /*
475  * The BX_FSPRIV buffer b_xflags are used to track types of data in buffers.
476  */
477 #define	BX_SUPERBLOCK	0x00010000	/* superblock */
478 #define	BX_CYLGRP	0x00020000	/* cylinder groups */
479 #define	BX_INODE	0x00040000	/* inodes */
480 #define	BX_INDIR	0x00080000	/* indirect blocks */
481 #define	BX_DIR		0x00100000	/* directory contents */
482 
483 #define	PRINT_UFS_BUF_XFLAGS "\20\25dir\24indir\23inode\22cylgrp\21superblock"
484 
485 /*
486  * Macros to access bits in the fs_active array.
487  */
488 #define	ACTIVECGNUM(fs, cg)	((fs)->fs_active[(cg) / (NBBY * sizeof(int))])
489 #define	ACTIVECGOFF(cg)		(1 << ((cg) % (NBBY * sizeof(int))))
490 #define	ACTIVESET(fs, cg)	do {					\
491 	if ((fs)->fs_active)						\
492 		ACTIVECGNUM((fs), (cg)) |= ACTIVECGOFF((cg));		\
493 } while (0)
494 #define	ACTIVECLEAR(fs, cg)	do {					\
495 	if ((fs)->fs_active)						\
496 		ACTIVECGNUM((fs), (cg)) &= ~ACTIVECGOFF((cg));		\
497 } while (0)
498 
499 /*
500  * The size of a cylinder group is calculated by CGSIZE. The maximum size
501  * is limited by the fact that cylinder groups are at most one block.
502  * Its size is derived from the size of the maps maintained in the
503  * cylinder group and the (struct cg) size.
504  */
505 #define	CGSIZE(fs) \
506     /* base cg */	(sizeof(struct cg) + sizeof(int32_t) + \
507     /* old btotoff */	(fs)->fs_old_cpg * sizeof(int32_t) + \
508     /* old boff */	(fs)->fs_old_cpg * sizeof(u_int16_t) + \
509     /* inode map */	howmany((fs)->fs_ipg, NBBY) + \
510     /* block map */	howmany((fs)->fs_fpg, NBBY) +\
511     /* if present */	((fs)->fs_contigsumsize <= 0 ? 0 : \
512     /* cluster sum */	(fs)->fs_contigsumsize * sizeof(int32_t) + \
513     /* cluster map */	howmany(fragstoblks(fs, (fs)->fs_fpg), NBBY)))
514 
515 /*
516  * The minimal number of cylinder groups that should be created.
517  */
518 #define	MINCYLGRPS	4
519 
520 /*
521  * Convert cylinder group to base address of its global summary info.
522  */
523 #define	fs_cs(fs, indx) fs_csp[indx]
524 
525 /*
526  * Cylinder group block for a filesystem.
527  */
528 #define	CG_MAGIC	0x090255
529 struct cg {
530 	int32_t	 cg_firstfield;		/* historic cyl groups linked list */
531 	int32_t	 cg_magic;		/* magic number */
532 	int32_t  cg_old_time;		/* time last written */
533 	u_int32_t cg_cgx;		/* we are the cgx'th cylinder group */
534 	int16_t	 cg_old_ncyl;		/* number of cyl's this cg */
535 	int16_t  cg_old_niblk;		/* number of inode blocks this cg */
536 	u_int32_t cg_ndblk;		/* number of data blocks this cg */
537 	struct	 csum cg_cs;		/* cylinder summary information */
538 	u_int32_t cg_rotor;		/* position of last used block */
539 	u_int32_t cg_frotor;		/* position of last used frag */
540 	u_int32_t cg_irotor;		/* position of last used inode */
541 	u_int32_t cg_frsum[MAXFRAG];	/* counts of available frags */
542 	int32_t	 cg_old_btotoff;	/* (int32) block totals per cylinder */
543 	int32_t	 cg_old_boff;		/* (u_int16) free block positions */
544 	u_int32_t cg_iusedoff;		/* (u_int8) used inode map */
545 	u_int32_t cg_freeoff;		/* (u_int8) free block map */
546 	u_int32_t cg_nextfreeoff;	/* (u_int8) next available space */
547 	u_int32_t cg_clustersumoff;	/* (u_int32) counts of avail clusters */
548 	u_int32_t cg_clusteroff;		/* (u_int8) free cluster map */
549 	u_int32_t cg_nclusterblks;	/* number of clusters this cg */
550 	u_int32_t cg_niblk;		/* number of inode blocks this cg */
551 	u_int32_t cg_initediblk;		/* last initialized inode */
552 	u_int32_t cg_unrefs;		/* number of unreferenced inodes */
553 	int32_t	 cg_sparecon32[1];	/* reserved for future use */
554 	u_int32_t cg_ckhash;		/* check-hash of this cg */
555 	ufs_time_t cg_time;		/* time last written */
556 	int64_t	 cg_sparecon64[3];	/* reserved for future use */
557 	u_int8_t cg_space[1];		/* space for cylinder group maps */
558 /* actually longer */
559 };
560 
561 /*
562  * Macros for access to cylinder group array structures
563  */
564 #define	cg_chkmagic(cgp) ((cgp)->cg_magic == CG_MAGIC)
565 #define	cg_inosused(cgp) \
566     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_iusedoff))
567 #define	cg_blksfree(cgp) \
568     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_freeoff))
569 #define	cg_clustersfree(cgp) \
570     ((u_int8_t *)((u_int8_t *)(cgp) + (cgp)->cg_clusteroff))
571 #define	cg_clustersum(cgp) \
572     ((int32_t *)((uintptr_t)(cgp) + (cgp)->cg_clustersumoff))
573 
574 /*
575  * Turn filesystem block numbers into disk block addresses.
576  * This maps filesystem blocks to device size blocks.
577  */
578 #define	fsbtodb(fs, b)	((daddr_t)(b) << (fs)->fs_fsbtodb)
579 #define	dbtofsb(fs, b)	((b) >> (fs)->fs_fsbtodb)
580 
581 /*
582  * Cylinder group macros to locate things in cylinder groups.
583  * They calc filesystem addresses of cylinder group data structures.
584  */
585 #define	cgbase(fs, c)	(((ufs2_daddr_t)(fs)->fs_fpg) * (c))
586 #define	cgdata(fs, c)	(cgdmin(fs, c) + (fs)->fs_metaspace)	/* data zone */
587 #define	cgmeta(fs, c)	(cgdmin(fs, c))				/* meta data */
588 #define	cgdmin(fs, c)	(cgstart(fs, c) + (fs)->fs_dblkno)	/* 1st data */
589 #define	cgimin(fs, c)	(cgstart(fs, c) + (fs)->fs_iblkno)	/* inode blk */
590 #define	cgsblock(fs, c)	(cgstart(fs, c) + (fs)->fs_sblkno)	/* super blk */
591 #define	cgtod(fs, c)	(cgstart(fs, c) + (fs)->fs_cblkno)	/* cg block */
592 #define	cgstart(fs, c)							\
593        ((fs)->fs_magic == FS_UFS2_MAGIC ? cgbase(fs, c) :		\
594        (cgbase(fs, c) + (fs)->fs_old_cgoffset * ((c) & ~((fs)->fs_old_cgmask))))
595 
596 /*
597  * Macros for handling inode numbers:
598  *     inode number to filesystem block offset.
599  *     inode number to cylinder group number.
600  *     inode number to filesystem block address.
601  */
602 #define	ino_to_cg(fs, x)	(((ino_t)(x)) / (fs)->fs_ipg)
603 #define	ino_to_fsba(fs, x)						\
604 	((ufs2_daddr_t)(cgimin(fs, ino_to_cg(fs, (ino_t)(x))) +		\
605 	    (blkstofrags((fs), ((((ino_t)(x)) % (fs)->fs_ipg) / INOPB(fs))))))
606 #define	ino_to_fsbo(fs, x)	(((ino_t)(x)) % INOPB(fs))
607 
608 /*
609  * Give cylinder group number for a filesystem block.
610  * Give cylinder group block number for a filesystem block.
611  */
612 #define	dtog(fs, d)	((d) / (fs)->fs_fpg)
613 #define	dtogd(fs, d)	((d) % (fs)->fs_fpg)
614 
615 /*
616  * Extract the bits for a block from a map.
617  * Compute the cylinder and rotational position of a cyl block addr.
618  */
619 #define	blkmap(fs, map, loc) \
620     (((map)[(loc) / NBBY] >> ((loc) % NBBY)) & (0xff >> (NBBY - (fs)->fs_frag)))
621 
622 /*
623  * The following macros optimize certain frequently calculated
624  * quantities by using shifts and masks in place of divisions
625  * modulos and multiplications.
626  */
627 #define	blkoff(fs, loc)		/* calculates (loc % fs->fs_bsize) */ \
628 	((loc) & (fs)->fs_qbmask)
629 #define	fragoff(fs, loc)	/* calculates (loc % fs->fs_fsize) */ \
630 	((loc) & (fs)->fs_qfmask)
631 #define	lfragtosize(fs, frag)	/* calculates ((off_t)frag * fs->fs_fsize) */ \
632 	(((off_t)(frag)) << (fs)->fs_fshift)
633 #define	lblktosize(fs, blk)	/* calculates ((off_t)blk * fs->fs_bsize) */ \
634 	(((off_t)(blk)) << (fs)->fs_bshift)
635 /* Use this only when `blk' is known to be small, e.g., < UFS_NDADDR. */
636 #define	smalllblktosize(fs, blk)    /* calculates (blk * fs->fs_bsize) */ \
637 	((blk) << (fs)->fs_bshift)
638 #define	lblkno(fs, loc)		/* calculates (loc / fs->fs_bsize) */ \
639 	((loc) >> (fs)->fs_bshift)
640 #define	numfrags(fs, loc)	/* calculates (loc / fs->fs_fsize) */ \
641 	((loc) >> (fs)->fs_fshift)
642 #define	blkroundup(fs, size)	/* calculates roundup(size, fs->fs_bsize) */ \
643 	(((size) + (fs)->fs_qbmask) & (fs)->fs_bmask)
644 #define	fragroundup(fs, size)	/* calculates roundup(size, fs->fs_fsize) */ \
645 	(((size) + (fs)->fs_qfmask) & (fs)->fs_fmask)
646 #define	fragstoblks(fs, frags)	/* calculates (frags / fs->fs_frag) */ \
647 	((frags) >> (fs)->fs_fragshift)
648 #define	blkstofrags(fs, blks)	/* calculates (blks * fs->fs_frag) */ \
649 	((blks) << (fs)->fs_fragshift)
650 #define	fragnum(fs, fsb)	/* calculates (fsb % fs->fs_frag) */ \
651 	((fsb) & ((fs)->fs_frag - 1))
652 #define	blknum(fs, fsb)		/* calculates rounddown(fsb, fs->fs_frag) */ \
653 	((fsb) &~ ((fs)->fs_frag - 1))
654 
655 /*
656  * Determine the number of available frags given a
657  * percentage to hold in reserve.
658  */
659 #define	freespace(fs, percentreserved) \
660 	(blkstofrags((fs), (fs)->fs_cstotal.cs_nbfree) + \
661 	(fs)->fs_cstotal.cs_nffree - \
662 	(((off_t)((fs)->fs_dsize)) * (percentreserved) / 100))
663 
664 /*
665  * Determining the size of a file block in the filesystem.
666  */
667 #define	blksize(fs, ip, lbn) \
668 	(((lbn) >= UFS_NDADDR || (ip)->i_size >= \
669 	    (uint64_t)smalllblktosize(fs, (lbn) + 1)) \
670 	    ? (fs)->fs_bsize \
671 	    : (fragroundup(fs, blkoff(fs, (ip)->i_size))))
672 #define	sblksize(fs, size, lbn) \
673 	(((lbn) >= UFS_NDADDR || (size) >= ((lbn) + 1) << (fs)->fs_bshift) \
674 	  ? (fs)->fs_bsize \
675 	  : (fragroundup(fs, blkoff(fs, (size)))))
676 
677 /*
678  * Number of indirects in a filesystem block.
679  */
680 #define	NINDIR(fs)	((fs)->fs_nindir)
681 
682 /*
683  * Indirect lbns are aligned on UFS_NDADDR addresses where single indirects
684  * are the negated address of the lowest lbn reachable, double indirects
685  * are this lbn - 1 and triple indirects are this lbn - 2.  This yields
686  * an unusual bit order to determine level.
687  */
688 static inline int
689 lbn_level(ufs_lbn_t lbn)
690 {
691 	if (lbn >= 0)
692 		return 0;
693 	switch (lbn & 0x3) {
694 	case 0:
695 		return (0);
696 	case 1:
697 		break;
698 	case 2:
699 		return (2);
700 	case 3:
701 		return (1);
702 	default:
703 		break;
704 	}
705 	return (-1);
706 }
707 
708 static inline ufs_lbn_t
709 lbn_offset(struct fs *fs, int level)
710 {
711 	ufs_lbn_t res;
712 
713 	for (res = 1; level > 0; level--)
714 		res *= NINDIR(fs);
715 	return (res);
716 }
717 
718 /*
719  * Number of inodes in a secondary storage block/fragment.
720  */
721 #define	INOPB(fs)	((fs)->fs_inopb)
722 #define	INOPF(fs)	((fs)->fs_inopb >> (fs)->fs_fragshift)
723 
724 /*
725  * Softdep journal record format.
726  */
727 
728 #define	JOP_ADDREF	1	/* Add a reference to an inode. */
729 #define	JOP_REMREF	2	/* Remove a reference from an inode. */
730 #define	JOP_NEWBLK	3	/* Allocate a block. */
731 #define	JOP_FREEBLK	4	/* Free a block or a tree of blocks. */
732 #define	JOP_MVREF	5	/* Move a reference from one off to another. */
733 #define	JOP_TRUNC	6	/* Partial truncation record. */
734 #define	JOP_SYNC	7	/* fsync() complete record. */
735 
736 #define	JREC_SIZE	32	/* Record and segment header size. */
737 
738 #define	SUJ_MIN		(4 * 1024 * 1024)	/* Minimum journal size */
739 #define	SUJ_MAX		(32 * 1024 * 1024)	/* Maximum journal size */
740 #define	SUJ_FILE	".sujournal"		/* Journal file name */
741 
742 /*
743  * Size of the segment record header.  There is at most one for each disk
744  * block in the journal.  The segment header is followed by an array of
745  * records.  fsck depends on the first element in each record being 'op'
746  * and the second being 'ino'.  Segments may span multiple disk blocks but
747  * the header is present on each.
748  */
749 struct jsegrec {
750 	uint64_t	jsr_seq;	/* Our sequence number */
751 	uint64_t	jsr_oldest;	/* Oldest valid sequence number */
752 	uint16_t	jsr_cnt;	/* Count of valid records */
753 	uint16_t	jsr_blocks;	/* Count of device bsize blocks. */
754 	uint32_t	jsr_crc;	/* 32bit crc of the valid space */
755 	ufs_time_t	jsr_time;	/* timestamp for mount instance */
756 };
757 
758 /*
759  * Reference record.  Records a single link count modification.
760  */
761 struct jrefrec {
762 	uint32_t	jr_op;
763 	uint32_t	jr_ino;
764 	uint32_t	jr_parent;
765 	uint16_t	jr_nlink;
766 	uint16_t	jr_mode;
767 	int64_t		jr_diroff;
768 	uint64_t	jr_unused;
769 };
770 
771 /*
772  * Move record.  Records a reference moving within a directory block.  The
773  * nlink is unchanged but we must search both locations.
774  */
775 struct jmvrec {
776 	uint32_t	jm_op;
777 	uint32_t	jm_ino;
778 	uint32_t	jm_parent;
779 	uint16_t	jm_unused;
780 	int64_t		jm_oldoff;
781 	int64_t		jm_newoff;
782 };
783 
784 /*
785  * Block record.  A set of frags or tree of blocks starting at an indirect are
786  * freed or a set of frags are allocated.
787  */
788 struct jblkrec {
789 	uint32_t	jb_op;
790 	uint32_t	jb_ino;
791 	ufs2_daddr_t	jb_blkno;
792 	ufs_lbn_t	jb_lbn;
793 	uint16_t	jb_frags;
794 	uint16_t	jb_oldfrags;
795 	uint32_t	jb_unused;
796 };
797 
798 /*
799  * Truncation record.  Records a partial truncation so that it may be
800  * completed at check time.  Also used for sync records.
801  */
802 struct jtrncrec {
803 	uint32_t	jt_op;
804 	uint32_t	jt_ino;
805 	int64_t		jt_size;
806 	uint32_t	jt_extsize;
807 	uint32_t	jt_pad[3];
808 };
809 
810 union jrec {
811 	struct jsegrec	rec_jsegrec;
812 	struct jrefrec	rec_jrefrec;
813 	struct jmvrec	rec_jmvrec;
814 	struct jblkrec	rec_jblkrec;
815 	struct jtrncrec	rec_jtrncrec;
816 };
817 
818 #ifdef CTASSERT
819 CTASSERT(sizeof(struct jsegrec) == JREC_SIZE);
820 CTASSERT(sizeof(struct jrefrec) == JREC_SIZE);
821 CTASSERT(sizeof(struct jmvrec) == JREC_SIZE);
822 CTASSERT(sizeof(struct jblkrec) == JREC_SIZE);
823 CTASSERT(sizeof(struct jtrncrec) == JREC_SIZE);
824 CTASSERT(sizeof(union jrec) == JREC_SIZE);
825 #endif
826 
827 extern int inside[], around[];
828 extern u_char *fragtbl[];
829 
830 /*
831  * IOCTLs used for filesystem write suspension.
832  */
833 #define	UFSSUSPEND	_IOW('U', 1, fsid_t)
834 #define	UFSRESUME	_IO('U', 2)
835 
836 #endif
837