summaryrefslogtreecommitdiff
path: root/bin
diff options
context:
space:
mode:
authorPhilip Guenther <guenther@cvs.openbsd.org>2016-08-26 04:23:45 +0000
committerPhilip Guenther <guenther@cvs.openbsd.org>2016-08-26 04:23:45 +0000
commit7f92656de5a3fb4f68ee19df85bd8e477fd72fdb (patch)
treea8a0e4bdad67d267521d9fd41caae7a81b983777 /bin
parent515a30957abf36784c2dc22d44fb5d47c5c62c25 (diff)
tables.h is only used by tables.c; merge it into the .c file
<sys/time.h> is unnecessary; sort the #includes
Diffstat (limited to 'bin')
-rw-r--r--bin/pax/tables.c137
-rw-r--r--bin/pax/tables.h165
2 files changed, 132 insertions, 170 deletions
diff --git a/bin/pax/tables.c b/bin/pax/tables.c
index 74f01023903..3f29f6fba8d 100644
--- a/bin/pax/tables.c
+++ b/bin/pax/tables.c
@@ -1,4 +1,4 @@
-/* $OpenBSD: tables.c,v 1.48 2016/08/14 18:30:33 guenther Exp $ */
+/* $OpenBSD: tables.c,v 1.49 2016/08/26 04:23:44 guenther Exp $ */
/* $NetBSD: tables.c,v 1.4 1995/03/21 09:07:45 cgd Exp $ */
/*-
@@ -35,18 +35,17 @@
*/
#include <sys/types.h>
-#include <sys/time.h>
#include <sys/stat.h>
+#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <signal.h>
#include <stdio.h>
+#include <stdlib.h>
#include <string.h>
#include <unistd.h>
-#include <errno.h>
-#include <stdlib.h>
+
#include "pax.h"
-#include "tables.h"
#include "extern.h"
/*
@@ -64,6 +63,134 @@
* time well spent.
*/
+/*
+ * 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 SL_TAB_SZ 317 /* escape symlink tables */
+#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 {
+ ino_t ino; /* files inode number */
+ char *name; /* name of first file seen with this ino/dev */
+ dev_t dev; /* files device 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 {
+ off_t seek; /* location in scratch file */
+ struct timespec mtim; /* files last modification time */
+ struct ftm *fow;
+ int namelen; /* file name length */
+} 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 {
+ struct file_times ft;
+ 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 {
+ struct file_times ft;
+ u_int16_t mode; /* file mode to restore */
+ u_int16_t frc_mode; /* do we force mode settings? */
+} DIRDATA;
+
static HRDLNK **ltab = NULL; /* hard link table for detecting hard links */
static FTM **ftab = NULL; /* file time table for updating arch */
static NAMT **ntab = NULL; /* interactive rename storage table */
diff --git a/bin/pax/tables.h b/bin/pax/tables.h
deleted file mode 100644
index 55eeb959284..00000000000
--- a/bin/pax/tables.h
+++ /dev/null
@@ -1,165 +0,0 @@
-/* $OpenBSD: tables.h,v 1.16 2015/03/19 05:14:24 guenther Exp $ */
-/* $NetBSD: tables.h,v 1.3 1995/03/21 09:07:47 cgd Exp $ */
-
-/*-
- * 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 SL_TAB_SZ 317 /* escape symlink tables */
-#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 {
- ino_t ino; /* files inode number */
- char *name; /* name of first file seen with this ino/dev */
- dev_t dev; /* files device 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 {
- off_t seek; /* location in scratch file */
- struct timespec mtim; /* files last modification time */
- struct ftm *fow;
- int namelen; /* file name length */
-} 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 {
- struct file_times ft;
- 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 {
- struct file_times ft;
- u_int16_t mode; /* file mode to restore */
- u_int16_t frc_mode; /* do we force mode settings? */
-} DIRDATA;