Home | History | Annotate | Line # | Download | only in lfs
lfs.h revision 1.29
      1 /*	$NetBSD: lfs.h,v 1.29 2000/07/05 22:25:43 perseant Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 1999 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by Konrad E. Schroder <perseant (at) hhhh.org>.
      9  *
     10  * Redistribution and use in source and binary forms, with or without
     11  * modification, are permitted provided that the following conditions
     12  * are met:
     13  * 1. Redistributions of source code must retain the above copyright
     14  *    notice, this list of conditions and the following disclaimer.
     15  * 2. Redistributions in binary form must reproduce the above copyright
     16  *    notice, this list of conditions and the following disclaimer in the
     17  *    documentation and/or other materials provided with the distribution.
     18  * 3. All advertising materials mentioning features or use of this software
     19  *    must display the following acknowledgement:
     20  *      This product includes software developed by the NetBSD
     21  *      Foundation, Inc. and its contributors.
     22  * 4. Neither the name of The NetBSD Foundation nor the names of its
     23  *    contributors may be used to endorse or promote products derived
     24  *    from this software without specific prior written permission.
     25  *
     26  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     27  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     28  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     29  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     30  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     31  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     32  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     33  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     34  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     35  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     36  * POSSIBILITY OF SUCH DAMAGE.
     37  */
     38 /*-
     39  * Copyright (c) 1991, 1993
     40  *	The Regents of the University of California.  All rights reserved.
     41  *
     42  * Redistribution and use in source and binary forms, with or without
     43  * modification, are permitted provided that the following conditions
     44  * are met:
     45  * 1. Redistributions of source code must retain the above copyright
     46  *    notice, this list of conditions and the following disclaimer.
     47  * 2. Redistributions in binary form must reproduce the above copyright
     48  *    notice, this list of conditions and the following disclaimer in the
     49  *    documentation and/or other materials provided with the distribution.
     50  * 3. All advertising materials mentioning features or use of this software
     51  *    must display the following acknowledgement:
     52  *	This product includes software developed by the University of
     53  *	California, Berkeley and its contributors.
     54  * 4. Neither the name of the University nor the names of its contributors
     55  *    may be used to endorse or promote products derived from this software
     56  *    without specific prior written permission.
     57  *
     58  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     59  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     60  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     61  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     62  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     63  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     64  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     65  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     66  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     67  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     68  * SUCH DAMAGE.
     69  *
     70  *	@(#)lfs.h	8.9 (Berkeley) 5/8/95
     71  */
     72 
     73 /*
     74  * Compile-time options for LFS.
     75  */
     76 #define LFS_EAGAIN_FAIL          /* markv fail with EAGAIN if ino is locked */
     77 #define LFS_TRACK_IOS            /* attempt to avoid cleaning segments not yet fully written to disk */
     78 #define LFS_CANNOT_ROLLFW       /* No roll-forward agent exists */
     79 
     80 /* #define DEBUG_LFS */              /* Intensive debugging of LFS subsystem */
     81 
     82 /* #define LFS_ATIME_IFILE */         /* Store atime in Ifile, don't push */
     83 
     84 /*
     85  * Parameters and generic definitions
     86  */
     87 #define BW_CLEAN	1
     88 #define MIN_FREE_SEGS	2
     89 #define LFS_MAX_ACTIVE	10
     90 #define LFS_MAXDIROP	(desiredvnodes>>2)
     91 
     92 /* For convenience */
     93 #define IN_ALLMOD (IN_MODIFIED|IN_ACCESS|IN_CHANGE|IN_UPDATE|IN_ACCESSED|IN_CLEANING)
     94 #define LFS_SET_UINO(ip, flags) do {                                    \
     95         if (((flags) & IN_ACCESSED) && !((ip)->i_flag & IN_ACCESSED))   \
     96                 ++(ip)->i_lfs->lfs_uinodes;                             \
     97         if (((flags) & IN_CLEANING) && !((ip)->i_flag & IN_CLEANING))   \
     98                 ++(ip)->i_lfs->lfs_uinodes;                             \
     99         if (((flags) & IN_MODIFIED) && !((ip)->i_flag & IN_MODIFIED))   \
    100                 ++(ip)->i_lfs->lfs_uinodes;                             \
    101         (ip)->i_flag |= (flags);                                        \
    102 } while(0)
    103 
    104 #define LFS_CLR_UINO(ip, flags) do {                                    \
    105         if (((flags) & IN_ACCESSED) && ((ip)->i_flag & IN_ACCESSED))    \
    106                 --(ip)->i_lfs->lfs_uinodes;                             \
    107         if (((flags) & IN_CLEANING) && ((ip)->i_flag & IN_CLEANING))    \
    108                 --(ip)->i_lfs->lfs_uinodes;                             \
    109         if (((flags) & IN_MODIFIED) && ((ip)->i_flag & IN_MODIFIED))    \
    110                 --(ip)->i_lfs->lfs_uinodes;                             \
    111         (ip)->i_flag &= ~(flags);                                       \
    112 	if ((ip)->i_lfs->lfs_uinodes < 0) {                             \
    113 		panic("lfs_uinodes < 0");                               \
    114 	}                                                               \
    115 } while(0)
    116 
    117 
    118 #ifndef LFS_ATIME_IFILE
    119 #define	LFS_ITIMES(ip, acc, mod, cre) {					\
    120 	if ((ip)->i_flag & IN_ACCESS) {					\
    121 		(ip)->i_ffs_atime = (acc)->tv_sec;			\
    122 		(ip)->i_ffs_atimensec = (acc)->tv_nsec;			\
    123 		LFS_SET_UINO(ip, IN_ACCESSED);				\
    124 	}								\
    125 	if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) {			\
    126 		if ((ip)->i_flag & IN_UPDATE) {				\
    127 			(ip)->i_ffs_mtime = (mod)->tv_sec;		\
    128 			(ip)->i_ffs_mtimensec = (mod)->tv_nsec;		\
    129 			(ip)->i_modrev++;				\
    130 		}							\
    131 		if ((ip)->i_flag & IN_CHANGE) {				\
    132 			(ip)->i_ffs_ctime = (cre)->tv_sec;		\
    133 			(ip)->i_ffs_ctimensec = (cre)->tv_nsec;		\
    134 		}							\
    135 		LFS_SET_UINO(ip, IN_MODIFIED);				\
    136 	}								\
    137 	(ip)->i_flag &= ~(IN_ACCESS | IN_CHANGE | IN_UPDATE);		\
    138 }
    139 #else
    140 # define LFS_ITIMES(ip, acc, mod, cre) {                                \
    141 	struct buf *ibp;						\
    142 	IFILE *ifp;							\
    143 									\
    144         if ((ip)->i_flag & IN_ACCESS) {                         	\
    145 		LFS_IENTRY(ifp, ip->i_lfs, ip->i_number, ibp);		\
    146        		ifp->if_atime = (mod);					\
    147        		VOP_BWRITE(bp);						\
    148 		(ip)->i_flag &= ~IN_ACCESS;				\
    149         }                                                       	\
    150         if ((ip)->i_flag & (IN_CHANGE | IN_UPDATE)) {       		\
    151 		LFS_SET_UINO(ip, IN_MODIFIED);				\
    152                 if ((ip)->i_flag & IN_UPDATE) {                         \
    153                         (ip)->i_ffs_mtime = (mod)->tv_sec;		\
    154                         (ip)->i_ffs_mtimensec = (mod)->tv_nsec;         \
    155                         (ip)->i_modrev++;                               \
    156                 }                                                       \
    157                 if ((ip)->i_flag & IN_CHANGE) {                         \
    158                         (ip)->i_ffs_ctime = (cre)->tv_sec;		\
    159                         (ip)->i_ffs_ctimensec = (cre)->tv_nsec;         \
    160                 }                                                       \
    161                 (ip)->i_flag &= ~(IN_CHANGE | IN_UPDATE);   		\
    162         }                                                               \
    163 }
    164 #endif
    165 
    166 #define WRITEINPROG(vp) (vp->v_dirtyblkhd.lh_first && !(VTOI(vp)->i_flag & (IN_MODIFIED|IN_ACCESSED|IN_CLEANING)))
    167 
    168 /* Here begins the berkeley code */
    169 
    170 #define	LFS_LABELPAD	8192		/* LFS label size */
    171 #define	LFS_SBPAD	8192		/* LFS superblock size */
    172 
    173 /* On-disk and in-memory checkpoint segment usage structure. */
    174 typedef struct segusage SEGUSE;
    175 struct segusage {
    176 	u_int32_t su_nbytes;		/* number of live bytes */
    177 	u_int32_t su_lastmod;		/* SEGUSE last modified timestamp */
    178 	u_int16_t su_nsums;		/* number of summaries in segment */
    179 	u_int16_t su_ninos;		/* number of inode blocks in seg */
    180 
    181 #define	SEGUSE_ACTIVE		0x01	/* segment is currently being written */
    182 #define	SEGUSE_DIRTY		0x02	/* segment has data in it */
    183 #define	SEGUSE_SUPERBLOCK	0x04	/* segment contains a superblock */
    184 	u_int32_t su_flags;
    185 };
    186 
    187 #define	SEGUPB(fs)	(1 << (fs)->lfs_sushift)
    188 #define	SEGTABSIZE_SU(fs)						\
    189 	(((fs)->lfs_nseg + SEGUPB(fs) - 1) >> (fs)->lfs_sushift)
    190 
    191 /* On-disk file information.  One per file with data blocks in the segment. */
    192 typedef struct finfo FINFO;
    193 struct finfo {
    194 	u_int32_t fi_nblocks;		/* number of blocks */
    195 	u_int32_t fi_version;		/* version number */
    196 	u_int32_t fi_ino;		/* inode number */
    197 	u_int32_t fi_lastlength;	/* length of last block in array */
    198 	ufs_daddr_t	  fi_blocks[1];		/* array of logical block numbers */
    199 };
    200 
    201 /* On-disk super block. */
    202 struct dlfs {
    203 #define	LFS_MAGIC	0x070162
    204         u_int32_t dlfs_magic;     /* 0: magic number */
    205 #define	LFS_VERSION	1
    206         u_int32_t dlfs_version;   /* 4: version number */
    207 
    208         u_int32_t dlfs_size;      /* 8: number of blocks in fs */
    209         u_int32_t dlfs_ssize;     /* 12: number of blocks per segment */
    210         u_int32_t dlfs_dsize;     /* 16: number of disk blocks in fs */
    211         u_int32_t dlfs_bsize;     /* 20: file system block size */
    212         u_int32_t dlfs_fsize;     /* 24: size of frag blocks in fs */
    213         u_int32_t dlfs_frag;      /* 28: number of frags in a block in fs */
    214 
    215 /* Checkpoint region. */
    216         u_int32_t dlfs_free;      /* 32: start of the free list */
    217         u_int32_t dlfs_bfree;     /* 36: number of free disk blocks */
    218         u_int32_t dlfs_nfiles;    /* 40: number of allocated inodes */
    219         int32_t   dlfs_avail;     /* 44: blocks available for writing */
    220         int32_t   dlfs_uinodes;   /* 48: inodes in cache not yet on disk */
    221         ufs_daddr_t  dlfs_idaddr; /* 52: inode file disk address */
    222         u_int32_t dlfs_ifile;     /* 56: inode file inode number */
    223         ufs_daddr_t  dlfs_lastseg; /* 60: address of last segment written */
    224         ufs_daddr_t  dlfs_nextseg; /* 64: address of next segment to write */
    225         ufs_daddr_t  dlfs_curseg; /* 68: current segment being written */
    226         ufs_daddr_t  dlfs_offset; /* 72: offset in curseg for next partial */
    227         ufs_daddr_t  dlfs_lastpseg; /* 76: address of last partial written */
    228         u_int32_t dlfs_tstamp;    /* 80: time stamp */
    229 
    230 /* These are configuration parameters. */
    231         u_int32_t dlfs_minfree;   /* 84: minimum percentage of free blocks */
    232 
    233 /* These fields can be computed from the others. */
    234         u_int64_t dlfs_maxfilesize; /* 88: maximum representable file size */
    235         u_int32_t dlfs_dbpseg;    /* 96: disk blocks per segment */
    236         u_int32_t dlfs_inopb;     /* 100: inodes per block */
    237         u_int32_t dlfs_ifpb;      /* 104: IFILE entries per block */
    238         u_int32_t dlfs_sepb;      /* 108: SEGUSE entries per block */
    239         u_int32_t dlfs_nindir;    /* 112: indirect pointers per block */
    240         u_int32_t dlfs_nseg;      /* 116: number of segments */
    241         u_int32_t dlfs_nspf;      /* 120: number of sectors per fragment */
    242         u_int32_t dlfs_cleansz;   /* 124: cleaner info size in blocks */
    243         u_int32_t dlfs_segtabsz;  /* 128: segment table size in blocks */
    244         u_int32_t dlfs_segmask;   /* 132: calculate offset within a segment */
    245         u_int32_t dlfs_segshift;  /* 136: fast mult/div for segments */
    246         u_int32_t dlfs_bshift;    /* 140: calc block number from file offset */
    247         u_int32_t dlfs_ffshift;   /* 144: fast mult/div for frag from file */
    248         u_int32_t dlfs_fbshift;   /* 148: fast mult/div for frag from block */
    249         u_int64_t dlfs_bmask;     /* 152: calc block offset from file offset */
    250         u_int64_t dlfs_ffmask;    /* 160: calc frag offset from file offset */
    251         u_int64_t dlfs_fbmask;    /* 168: calc frag offset from block offset */
    252         u_int32_t dlfs_fsbtodb;   /* 176: fsbtodb and dbtofsb shift constant */
    253         u_int32_t dlfs_sushift;   /* 180: fast mult/div for segusage table */
    254 
    255         int32_t   dlfs_maxsymlinklen; /* 184: max length of an internal symlink */
    256 
    257 #define	LFS_MIN_SBINTERVAL	5  /* minimum superblock segment spacing */
    258 #define LFS_MAXNUMSB            10 /* 188: superblock disk offsets */
    259         ufs_daddr_t       dlfs_sboffs[LFS_MAXNUMSB];
    260 
    261 	u_int32_t dlfs_nclean;    /* 228: Number of clean segments */
    262 	u_char	  dlfs_fsmnt[MNAMELEN];	 /* 232: name mounted on */
    263 	/* XXX this is 2 bytes only to pad to a quad boundary */
    264 	u_int16_t dlfs_clean;     /* 322: file system is clean flag */
    265 	u_int32_t dlfs_dmeta;     /* 324: total number of dirty summaries */
    266 	u_int32_t dlfs_minfreeseg; /* 328: segs reserved for cleaner */
    267         int8_t    dlfs_pad[176];  /* 332: round to 512 bytes */
    268 /* Checksum -- last valid disk field. */
    269         u_int32_t dlfs_cksum;     /* 508: checksum for superblock checking */
    270 };
    271 
    272 /* Maximum number of io's we can have pending at once */
    273 #define LFS_THROTTLE  16 /* XXX should be better paramtrized - ? */
    274 
    275 /* In-memory super block. */
    276 struct lfs {
    277         struct dlfs lfs_dlfs;           /* on-disk parameters */
    278 #define lfs_magic lfs_dlfs.dlfs_magic
    279 #define lfs_version lfs_dlfs.dlfs_version
    280 #define lfs_size lfs_dlfs.dlfs_size
    281 #define lfs_ssize lfs_dlfs.dlfs_ssize
    282 #define lfs_dsize lfs_dlfs.dlfs_dsize
    283 #define lfs_bsize lfs_dlfs.dlfs_bsize
    284 #define lfs_fsize lfs_dlfs.dlfs_fsize
    285 #define lfs_frag lfs_dlfs.dlfs_frag
    286 #define lfs_free lfs_dlfs.dlfs_free
    287 #define lfs_bfree lfs_dlfs.dlfs_bfree
    288 #define lfs_nfiles lfs_dlfs.dlfs_nfiles
    289 #define lfs_avail lfs_dlfs.dlfs_avail
    290 #define lfs_uinodes lfs_dlfs.dlfs_uinodes
    291 #define lfs_idaddr lfs_dlfs.dlfs_idaddr
    292 #define lfs_ifile lfs_dlfs.dlfs_ifile
    293 #define lfs_lastseg lfs_dlfs.dlfs_lastseg
    294 #define lfs_nextseg lfs_dlfs.dlfs_nextseg
    295 #define lfs_curseg lfs_dlfs.dlfs_curseg
    296 #define lfs_offset lfs_dlfs.dlfs_offset
    297 #define lfs_lastpseg lfs_dlfs.dlfs_lastpseg
    298 #define lfs_tstamp lfs_dlfs.dlfs_tstamp
    299 #define lfs_minfree lfs_dlfs.dlfs_minfree
    300 #define lfs_maxfilesize lfs_dlfs.dlfs_maxfilesize
    301 #define lfs_dbpseg lfs_dlfs.dlfs_dbpseg
    302 #define lfs_inopb lfs_dlfs.dlfs_inopb
    303 #define lfs_ifpb lfs_dlfs.dlfs_ifpb
    304 #define lfs_sepb lfs_dlfs.dlfs_sepb
    305 #define lfs_nindir lfs_dlfs.dlfs_nindir
    306 #define lfs_nseg lfs_dlfs.dlfs_nseg
    307 #define lfs_nspf lfs_dlfs.dlfs_nspf
    308 #define lfs_cleansz lfs_dlfs.dlfs_cleansz
    309 #define lfs_segtabsz lfs_dlfs.dlfs_segtabsz
    310 #define lfs_segmask lfs_dlfs.dlfs_segmask
    311 #define lfs_segshift lfs_dlfs.dlfs_segshift
    312 #define lfs_bmask lfs_dlfs.dlfs_bmask
    313 #define lfs_bshift lfs_dlfs.dlfs_bshift
    314 #define lfs_ffmask lfs_dlfs.dlfs_ffmask
    315 #define lfs_ffshift lfs_dlfs.dlfs_ffshift
    316 #define lfs_fbmask lfs_dlfs.dlfs_fbmask
    317 #define lfs_fbshift lfs_dlfs.dlfs_fbshift
    318 #define lfs_fsbtodb lfs_dlfs.dlfs_fsbtodb
    319 #define lfs_sushift lfs_dlfs.dlfs_sushift
    320 #define lfs_maxsymlinklen lfs_dlfs.dlfs_maxsymlinklen
    321 #define lfs_sboffs lfs_dlfs.dlfs_sboffs
    322 #define lfs_cksum lfs_dlfs.dlfs_cksum
    323 #define lfs_clean lfs_dlfs.dlfs_clean
    324 #define lfs_fsmnt lfs_dlfs.dlfs_fsmnt
    325 #define lfs_nclean lfs_dlfs.dlfs_nclean
    326 #define lfs_dmeta lfs_dlfs.dlfs_dmeta
    327 #define lfs_minfreeseg lfs_dlfs.dlfs_minfreeseg
    328 
    329 /* These fields are set at mount time and are meaningless on disk. */
    330 	struct segment *lfs_sp;		/* current segment being written */
    331 	struct vnode *lfs_ivnode;	/* vnode for the ifile */
    332 	u_int32_t  lfs_seglock;		/* single-thread the segment writer */
    333 	pid_t	  lfs_lockpid;		/* pid of lock holder */
    334 	u_int32_t lfs_iocount;		/* number of ios pending */
    335 	u_int32_t lfs_writer;		/* don't allow any dirops to start */
    336 	u_int32_t lfs_dirops;		/* count of active directory ops */
    337 	u_int32_t lfs_doifile;		/* Write ifile blocks on next write */
    338 	u_int32_t lfs_nactive;		/* Number of segments since last ckp */
    339 	int8_t	  lfs_fmod;		/* super block modified flag */
    340 	int8_t	  lfs_ronly;		/* mounted read-only flag */
    341 #define LFS_NOTYET 0x01
    342 	int8_t	  lfs_flags;		/* currently unused flag */
    343 	u_int16_t lfs_activesb;         /* toggle between superblocks */
    344 #ifdef LFS_TRACK_IOS
    345 	daddr_t   lfs_pending[LFS_THROTTLE]; /* daddrs of pending writes */
    346 #endif /* LFS_TRACK_IOS */
    347 #ifdef LFS_CANNOT_ROLLFW
    348 	daddr_t   lfs_sbactive;         /* disk address of in-progress sb write */
    349 #endif
    350 	struct vnode *lfs_flushvp;      /* vnode being flushed */
    351 	struct vnode *lfs_unlockvp;     /* being inactivated in lfs_segunlock */
    352 	u_int32_t lfs_diropwait;	/* # procs waiting on dirop flush */
    353 	struct lock lfs_freelock;
    354 	pid_t lfs_rfpid;		/* Process ID of roll-forward agent */
    355 	int       lfs_nadirop;		/* number of active dirop nodes */
    356 };
    357 
    358 /*
    359  * Inode 0:	out-of-band inode number
    360  * Inode 1:	IFILE inode number
    361  * Inode 2:	root inode
    362  * Inode 3:	lost+found inode number
    363  */
    364 #define	LFS_UNUSED_INUM	0		/* out of band inode number */
    365 #define	LFS_IFILE_INUM	1		/* IFILE inode number */
    366 #define	LOSTFOUNDINO	3		/* lost+found inode number */
    367 #define	LFS_FIRST_INUM	4		/* first free inode number */
    368 
    369 /* Address calculations for metadata located in the inode */
    370 #define	S_INDIR(fs)	-NDADDR
    371 #define	D_INDIR(fs)	(S_INDIR(fs) - NINDIR(fs) - 1)
    372 #define	T_INDIR(fs)	(D_INDIR(fs) - NINDIR(fs) * NINDIR(fs) - 1)
    373 
    374 /* Unassigned disk addresses. */
    375 #define	UNASSIGNED	-1
    376 #define UNWRITTEN       -2
    377 
    378 /* Unused logical block number */
    379 #define LFS_UNUSED_LBN	-1
    380 
    381 typedef struct ifile IFILE;
    382 struct ifile {
    383 	u_int32_t if_version;		/* inode version number */
    384 #define	LFS_UNUSED_DADDR	0	/* out-of-band daddr */
    385 	ufs_daddr_t if_daddr;		/* inode disk address */
    386 	ino_t	  if_nextfree;		/* next-unallocated inode */
    387 #ifdef LFS_ATIME_IFILE
    388 	struct timespec if_atime;	/* Last access time */
    389 #endif
    390 };
    391 
    392 /*
    393  * Cleaner information structure.  This resides in the ifile and is used
    394  * to pass information between the cleaner and the kernel.
    395  */
    396 typedef struct _cleanerinfo {
    397 	u_int32_t clean;		/* number of clean segments */
    398 	u_int32_t dirty;		/* number of dirty segments */
    399 	u_int32_t bfree;		/* disk blocks free */
    400 	u_int32_t avail;		/* disk blocks available */
    401 } CLEANERINFO;
    402 
    403 #define	CLEANSIZE_SU(fs)						\
    404 	((sizeof(CLEANERINFO) + (fs)->lfs_bsize - 1) >> (fs)->lfs_bshift)
    405 
    406 /*
    407  * All summary blocks are the same size, so we can always read a summary
    408  * block easily from a segment.
    409  */
    410 #define	LFS_SUMMARY_SIZE	512
    411 
    412 /* On-disk segment summary information */
    413 typedef struct segsum SEGSUM;
    414 struct segsum {
    415 	u_int32_t ss_sumsum;		/* check sum of summary block */
    416 	u_int32_t ss_datasum;		/* check sum of data */
    417 	u_int32_t ss_magic;		/* segment summary magic number */
    418 #define SS_MAGIC	0x061561
    419 	ufs_daddr_t ss_next;		/* next segment */
    420 	u_int32_t ss_create;		/* creation time stamp */
    421 	u_int16_t ss_nfinfo;		/* number of file info structures */
    422 	u_int16_t ss_ninos;		/* number of inodes in summary */
    423 
    424 #define	SS_DIROP	0x01		/* segment begins a dirop */
    425 #define	SS_CONT		0x02		/* more partials to finish this write*/
    426 	u_int16_t ss_flags;		/* used for directory operations */
    427 	u_int16_t ss_pad;		/* extra space */
    428 	/* FINFO's and inode daddr's... */
    429 };
    430 
    431 /* NINDIR is the number of indirects in a file system block. */
    432 #define	NINDIR(fs)	((fs)->lfs_nindir)
    433 
    434 /* INOPB is the number of inodes in a secondary storage block. */
    435 #define	INOPB(fs)	((fs)->lfs_inopb)
    436 
    437 #define	blksize(fs, ip, lbn) \
    438 	(((lbn) >= NDADDR || (ip)->i_ffs_size >= ((lbn) + 1) << (fs)->lfs_bshift) \
    439 	    ? (fs)->lfs_bsize \
    440 	    : (fragroundup(fs, blkoff(fs, (ip)->i_ffs_size))))
    441 #define	blkoff(fs, loc)		((int)(loc) & (fs)->lfs_bmask)
    442 #define fragoff(fs, loc)    /* calculates (loc % fs->lfs_fsize) */ \
    443     ((int)((loc) & (fs)->lfs_ffmask))
    444 #define	fsbtodb(fs, b)		((b) << (fs)->lfs_fsbtodb)
    445 #define	dbtofsb(fs, b)		((b) >> (fs)->lfs_fsbtodb)
    446 #define fragstodb(fs, b)	((b) << ((fs)->lfs_fsbtodb - (fs)->lfs_fbshift))
    447 #define dbtofrags(fs, b)	((b) >> ((fs)->lfs_fsbtodb - (fs)->lfs_fbshift))
    448 #define	lblkno(fs, loc)		((loc) >> (fs)->lfs_bshift)
    449 #define	lblktosize(fs, blk)	((blk) << (fs)->lfs_bshift)
    450 #define numfrags(fs, loc)	/* calculates (loc / fs->lfs_fsize) */	\
    451 	((loc) >> (fs)->lfs_ffshift)
    452 #define blkroundup(fs, size)	/* calculates roundup(size, fs->lfs_bsize) */ \
    453 	((int)(((size) + (fs)->lfs_bmask) & (~(fs)->lfs_bmask)))
    454 #define fragroundup(fs, size)	/* calculates roundup(size, fs->lfs_fsize) */ \
    455 	((int)(((size) + (fs)->lfs_ffmask) & (~(fs)->lfs_ffmask)))
    456 #define fragstoblks(fs, frags)	/* calculates (frags / fs->lfs_frag) */ \
    457 	((frags) >> (fs)->lfs_fbshift)
    458 #define blkstofrags(fs, blks)	/* calculates (blks * fs->lfs_frag) */ \
    459 	((blks) << (fs)->lfs_fbshift)
    460 #define fragnum(fs, fsb)	/* calculates (fsb % fs->lfs_frag) */ \
    461 	((fsb) & ((fs)->lfs_frag - 1))
    462 #define blknum(fs, fsb)		/* calculates rounddown(fsb, fs->lfs_frag) */ \
    463 	((fsb) &~ ((fs)->lfs_frag - 1))
    464 #define dblksize(fs, dip, lbn) \
    465 	(((lbn) >= NDADDR || (dip)->di_size >= ((lbn) + 1) << (fs)->lfs_bshift)\
    466 	    ? (fs)->lfs_bsize \
    467 	    : (fragroundup(fs, blkoff(fs, (dip)->di_size))))
    468 #define	datosn(fs, daddr)	/* disk address to segment number */	\
    469 	(((daddr) - (fs)->lfs_sboffs[0]) / fsbtodb((fs), (fs)->lfs_ssize))
    470 #define sntoda(fs, sn) 		/* segment number to disk address */	\
    471 	((ufs_daddr_t)((sn) * ((fs)->lfs_ssize << (fs)->lfs_fsbtodb) +	\
    472 	    (fs)->lfs_sboffs[0]))
    473 
    474 /* Read in the block with the cleaner info from the ifile. */
    475 #define LFS_CLEANERINFO(CP, F, BP) {					\
    476 	VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS;			\
    477 	if (bread((F)->lfs_ivnode,					\
    478 	    (ufs_daddr_t)0, (F)->lfs_bsize, NOCRED, &(BP)))		\
    479 		panic("lfs: ifile read");				\
    480 	(CP) = (CLEANERINFO *)(BP)->b_data;				\
    481 }
    482 
    483 /* Read in the block with a specific inode from the ifile. */
    484 #define	LFS_IENTRY(IP, F, IN, BP) {					\
    485 	int _e;								\
    486 	VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS;			\
    487 	if ((_e = bread((F)->lfs_ivnode,				\
    488 	    (IN) / (F)->lfs_ifpb + (F)->lfs_cleansz + (F)->lfs_segtabsz,\
    489 	    (F)->lfs_bsize, NOCRED, &(BP))) != 0)			\
    490 		panic("lfs: ifile read %d", _e);			\
    491 	(IP) = (IFILE *)(BP)->b_data + (IN) % (F)->lfs_ifpb;		\
    492 }
    493 
    494 /* Read in the block with a specific segment usage entry from the ifile. */
    495 #define	LFS_SEGENTRY(SP, F, IN, BP) {					\
    496 	int _e;								\
    497 	VTOI((F)->lfs_ivnode)->i_flag |= IN_ACCESS;			\
    498 	if ((_e = bread((F)->lfs_ivnode,				\
    499 	    ((IN) >> (F)->lfs_sushift) + (F)->lfs_cleansz,		\
    500 	    (F)->lfs_bsize, NOCRED, &(BP))) != 0)			\
    501 		panic("lfs: ifile read: %d", _e);			\
    502 	(SP) = (SEGUSE *)(BP)->b_data + ((IN) & ((F)->lfs_sepb - 1));	\
    503 }
    504 
    505 /* Determine if a buffer belongs to the ifile */
    506 #define IS_IFILE(bp)	(VTOI(bp->b_vp)->i_number == LFS_IFILE_INUM)
    507 
    508 /*
    509  * Structures used by lfs_bmapv and lfs_markv to communicate information
    510  * about inodes and data blocks.
    511  */
    512 typedef struct block_info {
    513 	ino_t	bi_inode;		/* inode # */
    514 	ufs_daddr_t bi_lbn;		/* logical block w/in file */
    515 	ufs_daddr_t bi_daddr;		/* disk address of block */
    516 	time_t	bi_segcreate;		/* origin segment create time */
    517 	int	bi_version;		/* file version number */
    518 	void	*bi_bp;			/* data buffer */
    519 	int     bi_size;		/* size of the block (if fragment) */
    520 } BLOCK_INFO;
    521 
    522 /* In-memory description of a segment about to be written. */
    523 struct segment {
    524 	struct lfs	 *fs;		/* file system pointer */
    525 	struct buf	**bpp;		/* pointer to buffer array */
    526 	struct buf	**cbpp;		/* pointer to next available bp */
    527 	struct buf	**start_bpp;	/* pointer to first bp in this set */
    528 	struct buf	 *ibp;		/* buffer pointer to inode page */
    529 	struct dinode    *idp;          /* pointer to ifile dinode */
    530 	struct finfo	 *fip;		/* current fileinfo pointer */
    531 	struct vnode	 *vp;		/* vnode being gathered */
    532 	void	 *segsum;		/* segment summary info */
    533 	u_int32_t ninodes;		/* number of inodes in this segment */
    534 	u_int32_t seg_bytes_left;	/* bytes left in segment */
    535 	u_int32_t sum_bytes_left;	/* bytes left in summary block */
    536 	u_int32_t seg_number;		/* number of this segment */
    537 	ufs_daddr_t *start_lbp;		/* beginning lbn for this set */
    538 
    539 #define	SEGM_CKP	0x01		/* doing a checkpoint */
    540 #define	SEGM_CLEAN	0x02		/* cleaner call; don't sort */
    541 #define	SEGM_SYNC	0x04		/* wait for segment */
    542 #define	SEGM_PROT	0x08		/* don't inactivate at segunlock */
    543 	u_int16_t seg_flags;		/* run-time flags for this segment */
    544 };
    545 
    546 /*
    547  * Macros for determining free space on the disk, with the variable metadata
    548  * of segment summaries and inode blocks taken into account.
    549  */
    550 /* Estimate number of clean blocks not available for writing */
    551 #define LFS_EST_CMETA(F) ((((F)->lfs_dmeta * (u_int64_t)(F)->lfs_nclean) /  \
    552 				      ((F)->lfs_nseg - (F)->lfs_nclean)))
    553 
    554 /* Estimate total size of the disk not including metadata */
    555 #define LFS_EST_NONMETA(F) ((F)->lfs_dsize - (F)->lfs_dmeta - LFS_EST_CMETA(F))
    556 
    557 /* Estimate number of blocks actually available for writing */
    558 #define LFS_EST_BFREE(F) ((F)->lfs_bfree - LFS_EST_CMETA(F) - (F)->lfs_dmeta)
    559 
    560 /* Amount of non-meta space not available to mortal man */
    561 #define LFS_EST_RSVD(F) ((LFS_EST_NONMETA(F) * (u_int64_t)(F)->lfs_minfree) / \
    562 			  100)
    563 
    564 /* Can credential C write BB blocks */
    565 #define ISSPACE(F, BB, C)						\
    566 	((((C) == NOCRED || (C)->cr_uid == 0) &&			\
    567           LFS_EST_BFREE(F) >= (BB)) ||					\
    568 	 ((C)->cr_uid != 0 && IS_FREESPACE(F, BB)))
    569 
    570 /* Can an ordinary user write BB blocks */
    571 #define IS_FREESPACE(F, BB)						\
    572           (LFS_EST_BFREE(F) >= (BB) + LFS_EST_RSVD(F))
    573 
    574 /* Statistics Counters */
    575 struct lfs_stats {
    576 	u_int	segsused;
    577 	u_int	psegwrites;
    578 	u_int	psyncwrites;
    579 	u_int	pcleanwrites;
    580 	u_int	blocktot;
    581 	u_int	cleanblocks;
    582 	u_int	ncheckpoints;
    583 	u_int	nwrites;
    584 	u_int	nsync_writes;
    585 	u_int	wait_exceeded;
    586 	u_int	write_exceeded;
    587 	u_int	flush_invoked;
    588 	u_int	vflush_invoked;
    589 };
    590 extern struct lfs_stats lfs_stats;
    591