Re: [PATCH] dump -U to specify dumpdates device

[Emmanuel Dreyfus <manu%netbsd.org@localhost>]

On Wed, Mar 20, 2019 at 04:20:09PM +0000, David Holland wrote:
> I have been meaning sometime to hack dump to allow using dumpdates
> with subtree or partial-volume dumps, which requires something like
> this as a prereq, so yes please.

Here is the -U flag patch augmented with a -H hash_pattern option,
which lets you partitionize given files-to-dump. Here is an example:

dump -H '[0123]*' -f dump0123 /data/*
dump -H '[4567]*' -f dump4567 /data/*
dump -H '[89ab]*' -f dump89ab /data/*
dump -H '[cdef]*' -f dumpcdef /data/*

It would make my backups scripts simplier, but it may be too 
much specialized. What do you think?

Index: dump.8
===================================================================
RCS file: /cvsroot/src/sbin/dump/dump.8,v
retrieving revision 1.69
diff -U4 -r1.69 dump.8
--- dump.8	15 Jul 2018 06:14:13 -0000	1.69
+++ dump.8	21 Mar 2019 13:03:34 -0000
@@ -43,15 +43,17 @@
 .Op Fl B Ar records
 .Op Fl b Ar blocksize
 .Op Fl d Ar density
 .Op Fl f Ar file
+.Op Fl H Ar hash_pattern
 .Op Fl h Ar level
 .Op Fl k Ar read-blocksize
 .Op Fl L Ar label
 .Op Fl l Ar timeout
 .Op Fl r Ar cachesize
 .Op Fl s Ar feet
 .Op Fl T Ar date
+.Op Fl U Ar dumpdev
 .Op Fl x Ar snap-backup
 .Ar files-to-dump
 .Nm
 .Op Fl W Li \&| Fl w
@@ -187,8 +189,19 @@
 .Ev RCMD_CMD .
 See
 .Xr rcmd 3
 for more details.
+.It Fl H Ar hash_pattern
+Within the list of given 
+.Ar files-to-dump
+only dump the one whose name's SHA1 match
+.Ar hash_pattern
+pattern. This is usefull to perform partial dumps. Here is an example
+to dump one quarter of paths matching
+.Pa /home/*
+.Bd -literal -offset indent
+/sbin/dump -0f /dev/nrst1 -H '[0123]*' /home/*
+.Ed
 .It Fl h Ar level
 Honor the user
 .Qq nodump
 flag
@@ -298,12 +311,24 @@
 The file
 .Pa /etc/dumpdates
 may be edited to change any of the fields,
 if necessary.
-If a list of files or subdirectories is being dumped
+If the
+.Fl T
+option is used or if a list of files or subdirectories is being dumped
 (as opposed to an entire file system), then
 .Fl u
 is ignored.
+.It Fl U Ar dumpdev
+Same as
+.Fl u
+but specifies the device in
+.Pa /etc/dumpdates
+as
+.Ar dumpdev .
+This option can be used with subdir dumps and with the
+.Fl T
+option.
 .It Fl W
 .Nm
 tells the operator what file systems need to be dumped.
 This information is gleaned from the files
Index: dump.h
===================================================================
RCS file: /cvsroot/src/sbin/dump/dump.h,v
retrieving revision 1.56
diff -U4 -r1.56 dump.h
--- dump.h	1 Mar 2019 16:42:11 -0000	1.56
+++ dump.h	21 Mar 2019 13:03:34 -0000
@@ -114,8 +114,9 @@
 const char *temp;	/* name of the file for doing rewrite of dumpdates */
 char	lastlevel;	/* dump level of previous dump */
 char	level;		/* dump level of this dump */
 int	uflag;		/* update flag */
+const char *dumpdev;	/* device name in dumpdates */
 int	eflag;		/* eject flag */
 int	lflag;		/* autoload flag */
 int	diskfd;		/* disk file descriptor */
 int	tapefd;		/* tape file descriptor */
Index: itime.c
===================================================================
RCS file: /cvsroot/src/sbin/dump/itime.c,v
retrieving revision 1.21
diff -U4 -r1.21 itime.c
--- itime.c	1 Mar 2019 16:42:11 -0000	1.21
+++ itime.c	21 Mar 2019 13:03:34 -0000
@@ -128,11 +128,11 @@
 getdumptime(void)
 {
 	struct dumpdates *ddp;
 	int i;
-	char *fname;
+	const char *fname;
 
-	fname = disk;
+	fname = dumpdev ? dumpdev : disk;
 #ifdef FDEBUG
 	msg("Looking for name %s in dumpdates = %s for level = %c\n",
 		fname, dumpdates, level);
 #endif
@@ -169,17 +169,17 @@
 	FILE *df;
 	struct dumpdates *dtwalk, *dtfound;
 	int i;
 	int fd;
-	char *fname;
+	const char *fname;
 
-	if(uflag == 0)
+	if(uflag == 0 && dumpdev == NULL)
 		return;
 	if ((df = fopen(dumpdates, "r+")) == NULL)
 		quite(errno, "cannot rewrite %s", dumpdates);
 	fd = fileno(df);
 	(void) flock(fd, LOCK_EX);
-	fname = disk;
+	fname = dumpdev ? dumpdev : disk;
 	free((char *)ddatev);
 	ddatev = 0;
 	nddates = 0;
 	readdumptimes(df);
Index: main.c
===================================================================
RCS file: /cvsroot/src/sbin/dump/main.c,v
retrieving revision 1.74
diff -U4 -r1.74 main.c
--- main.c	1 Mar 2019 16:42:11 -0000	1.74
+++ main.c	21 Mar 2019 13:03:34 -0000
@@ -55,9 +55,11 @@
 #include <ctype.h>
 #include <err.h>
 #include <errno.h>
 #include <fcntl.h>
+#include <fnmatch.h>
 #include <fstab.h>
+#include <sha1.h>
 #include <signal.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -102,8 +104,9 @@
 	int snap_internal = 0;
 	ino_t maxino;
 	time_t tnow, date;
 	int dirc;
+	char *hash_pattern = NULL;
 	char *mountpoint;
 	int just_estimate = 0;
 	char labelstr[LBLSIZE];
 	char buf[MAXPATHLEN], rbuf[MAXPATHLEN];
@@ -132,9 +135,9 @@
 		usage();
 
 	obsolete(&argc, &argv);
 	while ((ch = getopt(argc, argv,
-	    "0123456789aB:b:cd:eFf:h:ik:l:L:nr:s:StT:uWwx:X")) != -1)
+	    "0123456789aB:b:cd:eFf:H:h:ik:l:L:nr:s:StT:uU:Wwx:X")) != -1)
 		switch (ch) {
 		/* dump level */
 		case '0': case '1': case '2': case '3': case '4':
 		case '5': case '6': case '7': case '8': case '9':
@@ -175,8 +178,11 @@
 		case 'f':		/* output file */
 			tape = optarg;
 			break;
 
+		case 'H':		/* file to dump hash pattern */
+			hash_pattern = optarg;
+			break;
 		case 'h':
 			honorlevel = numarg("honor level", 0L, 10L);
 			break;
 
@@ -244,8 +250,12 @@
 		case 'u':		/* update /etc/dumpdates */
 			uflag = 1;
 			break;
 
+		case 'U':		/* dump device in /etc/dumpdates */
+			dumpdev = optarg;
+			break;
+
 		case 'W':		/* what to do */
 		case 'w':
 			lastdump(ch);
 			exit(X_FINOK);	/* do nothing else */
@@ -319,8 +329,24 @@
 		} else {
 			if (strcmp(mountpoint, fsbuf.f_mntonname) != 0)
 				quit("%s is not on %s", argv[i], mountpoint);
 		}
+
+		if (hash_pattern) {
+			char hash[SHA1_DIGEST_STRING_LENGTH];
+			uint8_t *arg = (uint8_t *)argv[i];
+			int j;
+
+			(void)SHA1Data(arg, strlen(argv[i]), hash);
+
+			if (fnmatch(hash_pattern, hash, 0) != 0) {
+				for (j = i; j < argc - 1; j++)
+					argv[j] = argv[j + 1];
+				argc--;
+				continue;
+			}
+		}
+
 		msg("Dumping file/directory %s\n", argv[i]);
 		dirc++;
 	}
 	if (mountpoint)
@@ -671,11 +697,11 @@
 	const char *prog = getprogname();
 
 	(void)fprintf(stderr,
 "usage: %s [-0123456789aceFinStuX] [-B records] [-b blocksize]\n"
-"            [-d density] [-f file] [-h level] [-k read-blocksize]\n"
-"            [-L label] [-l timeout] [-r cachesize] [-s feet]\n"
-"            [-T date] [-x snap-backup] files-to-dump\n"
+"            [-d density] [-f file] [-h level] [-H hash_pattern]\n"
+"            [-k read-blocksize] [-L label] [-l timeout] [-r cachesize]\n"
+"            [-s feet] [-T date] [-U dumpdev] [-x snap-backup] files-to-dump\n"
 "       %s [-W | -w]\n", prog, prog);
 	exit(X_STARTUP);
 }
 

-- 
Emmanuel Dreyfus
manu%netbsd.org@localhost