paxmirabilis/src/tables.h

/**     $MirOS: src/bin/pax/tables.h,v 1.3 2007/10/23 20:07:42 tg Exp $ */
/*      $OpenBSD: tables.h,v 1.8 2006/08/05 23:05:13 ray Exp $  */
/*      $NetBSD: tables.h,v 1.3 1995/03/21 09:07:47 cgd Exp $   */

/*-
 * Copyright (c) 2005 Thorsten Glaser <tg@66h.42h.de>
 * Copyright (c) 1992 Keith Muller.
 * Copyright (c) 1992, 1993
 *      The Regents of the University of California.  All rights reserved.
 *
 * This code is derived from software contributed to Berkeley by
 * Keith Muller of the University of California, San Diego.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *      @(#)tables.h    8.1 (Berkeley) 5/31/93
 */

/*
 * data structures and constants used by the different databases kept by pax
 */

/*
 * Hash Table Sizes MUST BE PRIME, if set too small performance suffers.
 * Probably safe to expect 500000 inodes per tape. Assuming good key
 * distribution (inodes) chains of under 50 long (worst case) is ok.
 */
#define L_TAB_SZ        2503            /* hard link hash table size */
#define F_TAB_SZ        50503           /* file time hash table size */
#define N_TAB_SZ        541             /* interactive rename hash table */
#define D_TAB_SZ        317             /* unique device mapping table */
#define A_TAB_SZ        317             /* ftree dir access time reset table */
#define MAXKEYLEN       64              /* max number of chars for hash */
#define DIRP_SIZE       64              /* initial size of created dir table */

/*
 * file hard link structure (hashed by dev/ino and chained) used to find the
 * hard links in a file system or with some archive formats (cpio)
 */
typedef struct hrdlnk {
        char            *name;  /* name of first file seen with this ino/dev */
        dev_t           dev;    /* files device number */
        ino_t           ino;    /* files inode number */
        u_long          nlink;  /* expected link count */
        struct hrdlnk   *fow;
} HRDLNK;

/*
 * Archive write update file time table (the -u, -C flag), hashed by filename.
 * Filenames are stored in a scratch file at seek offset into the file. The
 * file time (mod time) and the file name length (for a quick check) are
 * stored in a hash table node. We were forced to use a scratch file because
 * with -u, the mtime for every node in the archive must always be available
 * to compare against (and this data can get REALLY large with big archives).
 * By being careful to read only when we have a good chance of a match, the
 * performance loss is not measurable (and the size of the archive we can
 * handle is greatly increased).
 */
typedef struct ftm {
        int             namelen;        /* file name length */
        time_t          mtime;          /* files last modification time */
        off_t           seek;           /* location in scratch file */
        struct ftm      *fow;
} FTM;

/*
 * Interactive rename table (-i flag), hashed by orig filename.
 * We assume this will not be a large table as this mapping data can only be
 * obtained through interactive input by the user. Nobody is going to type in
 * changes for 500000 files? We use chaining to resolve collisions.
 */

typedef struct namt {
        char            *oname;         /* old name */
        char            *nname;         /* new name typed in by the user */
        struct namt     *fow;
} NAMT;

/*
 * Unique device mapping tables. Some protocols (e.g. cpio) require that the
 * <c_dev,c_ino> pair will uniquely identify a file in an archive unless they
 * are links to the same file. Appending to archives can break this. For those
 * protocols that have this requirement we map c_dev to a unique value not seen
 * in the archive when we append. We also try to handle inode truncation with
 * this table. (When the inode field in the archive header are too small, we
 * remap the dev on writes to remove accidental collisions).
 *
 * The list is hashed by device number using chain collision resolution. Off of
 * each DEVT are linked the various remaps for this device based on those bits
 * in the inode which were truncated. For example if we are just remapping to
 * avoid a device number during an update append, off the DEVT we would have
 * only a single DLIST that has a truncation id of 0 (no inode bits were
 * stripped for this device so far). When we spot inode truncation we create
 * a new mapping based on the set of bits in the inode which were stripped off.
 * so if the top four bits of the inode are stripped and they have a pattern of
 * 0110...... (where . are those bits not truncated) we would have a mapping
 * assigned for all inodes that has the same 0110.... pattern (with this dev
 * number of course). This keeps the mapping sparse and should be able to store
 * close to the limit of files which can be represented by the optimal
 * combination of dev and inode bits, and without creating a fouled up archive.
 * Note we also remap truncated devs in the same way (an exercise for the
 * dedicated reader; always wanted to say that...:)
 */

typedef struct devt {
        dev_t           dev;    /* the orig device number we now have to map */
        struct devt     *fow;   /* new device map list */
        struct dlist    *list;  /* map list based on inode truncation bits */
} DEVT;

typedef struct dlist {
        ino_t trunc_bits;       /* truncation pattern for a specific map */
        dev_t dev;              /* the new device id we use */
        struct dlist *fow;
} DLIST;

/*
 * ftree directory access time reset table. When we are done with a
 * subtree we reset the access and mod time of the directory when the tflag is
 * set. Not really explicitly specified in the pax spec, but easy and fast to
 * do (and this may have even been intended in the spec, it is not clear).
 * table is hashed by inode with chaining.
 */

typedef struct atdir {
        char *name;     /* name of directory to reset */
        dev_t dev;      /* dev and inode for fast lookup */
        ino_t ino;
        time_t mtime;   /* access and mod time to reset to */
        time_t atime;
        struct atdir *fow;
} ATDIR;

/*
 * created directory time and mode storage entry. After pax is finished during
 * extraction or copy, we must reset directory access modes and times that
 * may have been modified after creation (they no longer have the specified
 * times and/or modes). We must reset time in the reverse order of creation,
 * because entries are added  from the top of the file tree to the bottom.
 * We MUST reset times from leaf to root (it will not work the other
 * direction).
 */

typedef struct dirdata {
        char *name;     /* file name */
        time_t mtime;   /* mtime to set */
        time_t atime;   /* atime to set */
        u_int16_t mode; /* file mode to restore */
        u_int16_t frc_mode;     /* do we force mode settings? */
} DIRDATA;

/*
 * file hard link structure (hashed by dev/ino and chained) for anonymisation
 */
typedef struct hrdflnk {
        dev_t           dev;    /* files device number */
        ino_t           ino;    /* files inode number */
        u_long          nlink;  /* expected link count */
        ino_t           newi;   /* new inode number */
        struct hrdflnk  *fow;
} HRDFLNK;
// Log In
// CVSweb
Project: FreeWRT
// Summary // Activity // Search // Tracker // Lists // News // SCM // Wiki
1	/** $MirOS: src/bin/pax/tables.h,v 1.3 2007/10/23 20:07:42 tg Exp $ */
2	/* $OpenBSD: tables.h,v 1.8 2006/08/05 23:05:13 ray Exp $ */
3	/* $NetBSD: tables.h,v 1.3 1995/03/21 09:07:47 cgd Exp $ */
4
5	/*-
6	* Copyright (c) 2005 Thorsten Glaser <tg@66h.42h.de>
7	* Copyright (c) 1992 Keith Muller.
8	* Copyright (c) 1992, 1993
9	* The Regents of the University of California. All rights reserved.
10	*
11	* This code is derived from software contributed to Berkeley by
12	* Keith Muller of the University of California, San Diego.
13	*
14	* Redistribution and use in source and binary forms, with or without
15	* modification, are permitted provided that the following conditions
16	* are met:
17	* 1. Redistributions of source code must retain the above copyright
18	* notice, this list of conditions and the following disclaimer.
19	* 2. Redistributions in binary form must reproduce the above copyright
20	* notice, this list of conditions and the following disclaimer in the
21	* documentation and/or other materials provided with the distribution.
22	* 3. Neither the name of the University nor the names of its contributors
23	* may be used to endorse or promote products derived from this software
24	* without specific prior written permission.
25	*
26	* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29	* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36	* SUCH DAMAGE.
37	*
38	* @(#)tables.h 8.1 (Berkeley) 5/31/93
39	*/
40
41	/*
42	* data structures and constants used by the different databases kept by pax
43	*/
44
45	/*
46	* Hash Table Sizes MUST BE PRIME, if set too small performance suffers.
47	* Probably safe to expect 500000 inodes per tape. Assuming good key
48	* distribution (inodes) chains of under 50 long (worst case) is ok.
49	*/
50	#define L_TAB_SZ 2503 /* hard link hash table size */
51	#define F_TAB_SZ 50503 /* file time hash table size */
52	#define N_TAB_SZ 541 /* interactive rename hash table */
53	#define D_TAB_SZ 317 /* unique device mapping table */
54	#define A_TAB_SZ 317 /* ftree dir access time reset table */
55	#define MAXKEYLEN 64 /* max number of chars for hash */
56	#define DIRP_SIZE 64 /* initial size of created dir table */
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).
164	*/
165
166	typedef struct dirdata {
167	char name; / file name */
168	time_t mtime; /* mtime to set */
169	time_t atime; /* atime to set */
170	u_int16_t mode; /* file mode to restore */
171	u_int16_t frc_mode; /* do we force mode settings? */
172	} DIRDATA;
173
174	/*
175	* file hard link structure (hashed by dev/ino and chained) for anonymisation
176	*/
177	typedef struct hrdflnk {
178	dev_t dev; /* files device number */
179	ino_t ino; /* files inode number */
180	u_long nlink; /* expected link count */
181	ino_t newi; /* new inode number */
182	struct hrdflnk *fow;
183	} HRDFLNK;