bin/pax/tables.h

   1 /*-
   2  * Copyright (c) 1992 Keith Muller.
   3  * Copyright (c) 1992, 1993
   4  *      The Regents of the University of California.  All rights reserved.
   5  *
   6  * This code is derived from software contributed to Berkeley by
   7  * Keith Muller of the University of California, San Diego.
   8  *
   9  * Redistribution and use in source and binary forms, with or without
  10  * modification, are permitted provided that the following conditions
  11  * are met:
  12  * 1. Redistributions of source code must retain the above copyright
  13  *    notice, this list of conditions and the following disclaimer.
  14  * 2. Redistributions in binary form must reproduce the above copyright
  15  *    notice, this list of conditions and the following disclaimer in the
  16  *    documentation and/or other materials provided with the distribution.
  17  * 3. All advertising materials mentioning features or use of this software
  18  *    must display the following acknowledgement:
  19  *      This product includes software developed by the University of
  20  *      California, Berkeley and its contributors.
  21  * 4. Neither the name of the University nor the names of its contributors
  22  *    may be used to endorse or promote products derived from this software
  23  *    without specific prior written permission.
  24  *
  25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  35  * SUCH DAMAGE.
  36  *
  37  *      @(#)tables.h    8.1 (Berkeley) 5/31/93
  38  * $FreeBSD: src/bin/pax/tables.h,v 1.7 1999/08/27 23:14:47 peter Exp $
  39  * $DragonFly: src/bin/pax/tables.h,v 1.3 2006/09/27 21:58:08 pavalos Exp $
  40  */
  41
  42 /*
  43  * data structures and constants used by the different databases kept by pax
  44  */
  45
  46 /*
  47  * Hash Table Sizes MUST BE PRIME, if set too small performance suffers.
  48  * Probably safe to expect 500000 inodes per tape. Assuming good key
  49  * distribution (inodes) chains of under 50 long (worst case) is ok.
  50  */
  51 #define L_TAB_SZ        2503            /* hard link hash table size */
  52 #define F_TAB_SZ        50503           /* file time hash table size */
  53 #define N_TAB_SZ        541             /* interactive rename hash table */
  54 #define D_TAB_SZ        317             /* unique device mapping table */
  55 #define A_TAB_SZ        317             /* ftree dir access time reset table */
  56 #define MAXKEYLEN       64              /* max number of chars for hash */
  57
  58 /*
  59  * file hard link structure (hashed by dev/ino and chained) used to find the
  60  * hard links in a file system or with some archive formats (cpio)
  61  */
  62 typedef struct hrdlnk {
  63         char            *name;  /* name of first file seen with this ino/dev */
  64         dev_t           dev;    /* files device number */
  65         ino_t           ino;    /* files inode number */
  66         u_long          nlink;  /* expected link count */
  67         struct hrdlnk   *fow;
  68 } HRDLNK;
  69
  70 /*
  71  * Archive write update file time table (the -u, -C flag), hashed by filename.
  72  * Filenames are stored in a scratch file at seek offset into the file. The
  73  * file time (mod time) and the file name length (for a quick check) are
  74  * stored in a hash table node. We were forced to use a scratch file because
  75  * with -u, the mtime for every node in the archive must always be available
  76  * to compare against (and this data can get REALLY large with big archives).
  77  * By being careful to read only when we have a good chance of a match, the
  78  * performance loss is not measurable (and the size of the archive we can
  79  * handle is greatly increased).
  80  */
  81 typedef struct ftm {
  82         int             namelen;        /* file name length */
  83         time_t          mtime;          /* files last modification time */
  84         off_t           seek;           /* location in scratch file */
  85         struct ftm      *fow;
  86 } FTM;
  87
  88 /*
  89  * Interactive rename table (-i flag), hashed by orig filename.
  90  * We assume this will not be a large table as this mapping data can only be
  91  * obtained through interactive input by the user. Nobody is going to type in
  92  * changes for 500000 files? We use chaining to resolve collisions.
  93  */
  94
  95 typedef struct namt {
  96         char            *oname;         /* old name */
  97         char            *nname;         /* new name typed in by the user */
  98         struct namt     *fow;
  99 } NAMT;
 100
 101 /*
 102  * Unique device mapping tables. Some protocols (e.g. cpio) require that the
 103  * <c_dev,c_ino> pair will uniquely identify a file in an archive unless they
 104  * are links to the same file. Appending to archives can break this. For those
 105  * protocols that have this requirement we map c_dev to a unique value not seen
 106  * in the archive when we append. We also try to handle inode truncation with
 107  * this table. (When the inode field in the archive header are too small, we
 108  * remap the dev on writes to remove accidental collisions).
 109  *
 110  * The list is hashed by device number using chain collision resolution. Off of
 111  * each DEVT are linked the various remaps for this device based on those bits
 112  * in the inode which were truncated. For example if we are just remapping to
 113  * avoid a device number during an update append, off the DEVT we would have
 114  * only a single DLIST that has a truncation id of 0 (no inode bits were
 115  * stripped for this device so far). When we spot inode truncation we create
 116  * a new mapping based on the set of bits in the inode which were stripped off.
 117  * so if the top four bits of the inode are stripped and they have a pattern of
 118  * 0110...... (where . are those bits not truncated) we would have a mapping
 119  * assigned for all inodes that has the same 0110.... pattern (with this dev
 120  * number of course). This keeps the mapping sparse and should be able to store
 121  * close to the limit of files which can be represented by the optimal
 122  * combination of dev and inode bits, and without creating a fouled up archive.
 123  * Note we also remap truncated devs in the same way (an exercise for the
 124  * dedicated reader; always wanted to say that...:)
 125  */
 126
 127 typedef struct devt {
 128         dev_t           dev;    /* the orig device number we now have to map */
 129         struct devt     *fow;   /* new device map list */
 130         struct dlist    *list;  /* map list based on inode truncation bits */
 131 } DEVT;
 132
 133 typedef struct dlist {
 134         ino_t trunc_bits;       /* truncation pattern for a specific map */
 135         dev_t dev;              /* the new device id we use */
 136         struct dlist *fow;
 137 } DLIST;
 138
 139 /*
 140  * ftree directory access time reset table. When we are done with a
 141  * subtree we reset the access and mod time of the directory when the tflag is
 142  * set. Not really explicitly specified in the pax spec, but easy and fast to
 143  * do (and this may have even been intended in the spec, it is not clear).
 144  * table is hashed by inode with chaining.
 145  */
 146
 147 typedef struct atdir {
 148         char *name;     /* name of directory to reset */
 149         dev_t dev;      /* dev and inode for fast lookup */
 150         ino_t ino;
 151         time_t mtime;   /* access and mod time to reset to */
 152         time_t atime;
 153         struct atdir *fow;
 154 } ATDIR;
 155
 156 /*
 157  * created directory time and mode storage entry. After pax is finished during
 158  * extraction or copy, we must reset directory access modes and times that
 159  * may have been modified after creation (they no longer have the specified
 160  * times and/or modes). We must reset time in the reverse order of creation,
 161  * because entries are added  from the top of the file tree to the bottom.
 162  * We MUST reset times from leaf to root (it will not work the other
 163  * direction).  Entries are recorded into a spool file to make reverse
 164  * reading faster.
 165  */
 166
 167 typedef struct dirdata {
 168         int nlen;       /* length of the directory name (includes \0) */
 169         off_t npos;     /* position in file where this dir name starts */
 170         mode_t mode;    /* file mode to restore */
 171         time_t mtime;   /* mtime to set */
 172         time_t atime;   /* atime to set */
 173         int frc_mode;   /* do we force mode settings? */
 174 } DIRDATA;