A little more cleanup.

[rsync/rsync-patches.git] / checksum4mirrors.diff

Optimize the ability of a mirror to send checksums.

This adds a sender optimization feature that allows a cache of checksums
to be used when the client specifies the --checksum option.  The checksum
files (.rsyncsums) must be created by some other process (see the perl
script in the support dir for one way).

This option should be used by mirrors that contain files that get created and
not changed.  There is a minimal amount of sanity-check information in the
.rsyncsums file (size and mtime) so that the sum files can be shared with your
mirror network.

To use this patch, run these commands for a successful build:

    patch -p1 <patches/checksum4mirrors.diff
    ./configure                               (optional if already run)
    make

--- old/flist.c
+++ new/flist.c
@@ -117,6 +117,7 @@ static char empty_sum[MAX_DIGEST_LEN];
 static int flist_count_offset; /* for --delete --progress */
 static int dir_count = 0;
 static int high_hlink_ndx;
+static struct file_list *checksum_flist = NULL;
 
 static void clean_flist(struct file_list *flist, int strip_root);
 static void output_flist(struct file_list *flist);
@@ -304,6 +305,186 @@ static void flist_done_allocating(struct
 		flist->pool_boundary = ptr;
 }
 
+/* The len count is the length of the basename + 1 for the null. */
+static int add_checksum(const char *dirname, const char *basename, int len,
+			OFF_T file_length, time_t mtime, const char *sum)
+{
+	struct file_struct *file;
+	int alloc_len, extra_len;
+	char *bp;
+
+	if (len == 10+1 && *basename == '.' && strcmp(basename, ".rsyncsums") == 0)
+		return 0;
+	if (file_length == 0)
+		return 0;
+
+	extra_len = (file_extra_cnt + (file_length > 0xFFFFFFFFu) + SUM_EXTRA_CNT)
+		  * EXTRA_LEN;
+#if EXTRA_ROUNDING > 0
+	if (extra_len & (EXTRA_ROUNDING * EXTRA_LEN))
+		extra_len = (extra_len | (EXTRA_ROUNDING * EXTRA_LEN)) + EXTRA_LEN;
+#endif
+	alloc_len = FILE_STRUCT_LEN + extra_len + len;
+	bp = pool_alloc(checksum_flist->file_pool, alloc_len, "add_checksum");
+
+	memset(bp, 0, extra_len + FILE_STRUCT_LEN);
+	bp += extra_len;
+	file = (struct file_struct *)bp;
+	bp += FILE_STRUCT_LEN;
+
+	memcpy(bp, basename, len);
+
+	file->mode = S_IFREG;
+	file->modtime = mtime;
+	file->len32 = (uint32)file_length;
+	if (file_length > 0xFFFFFFFFu) {
+		file->flags |= FLAG_LENGTH64;
+		OPT_EXTRA(file, 0)->unum = (uint32)(file_length >> 32);
+	}
+	file->dirname = dirname;
+	bp = F_SUM(file);
+	memcpy(bp, sum, checksum_len);
+
+	flist_expand(checksum_flist, 1);
+	checksum_flist->files[checksum_flist->used++] = file;
+
+	checksum_flist->sorted = checksum_flist->files;
+
+	return 1;
+}
+
+/* The direname value must remain unchanged during the lifespan of the
+ * created checksum_flist object because we use it directly. */
+static void read_checksums(const char *dirname)
+{
+	char line[MAXPATHLEN+1024], fbuf[MAXPATHLEN], sum[MAX_DIGEST_LEN];
+	OFF_T file_length;
+	time_t mtime;
+	int len, dlen, i;
+	char *cp;
+	FILE *fp;
+
+	if (checksum_flist) {
+		/* Reset the pool memory and empty the file-list array. */
+		pool_free_old(checksum_flist->file_pool,
+			      pool_boundary(checksum_flist->file_pool, 0));
+		checksum_flist->used = 0;
+	} else
+		checksum_flist = flist_new(FLIST_TEMP, "read_checksums");
+
+	checksum_flist->low = 0;
+	checksum_flist->high = -1;
+
+	if (!dirname)
+		return;
+
+	dlen = strlcpy(fbuf, dirname, sizeof fbuf);
+	if (dlen >= (int)sizeof fbuf)
+		return;
+	if (dlen)
+		fbuf[dlen++] = '/';
+	else
+		dirname = NULL;
+	strlcpy(fbuf+dlen, ".rsyncsums", sizeof fbuf - dlen);
+	if (!(fp = fopen(fbuf, "r")))
+		return;
+
+	while (fgets(line, sizeof line, fp)) {
+		cp = line;
+		if (protocol_version >= 30) {
+			char *alt_sum = cp;
+			if (*cp == '=')
+				while (*++cp == '=') {}
+			else
+				while (isXDigit(cp)) cp++;
+			if (cp - alt_sum != MD4_DIGEST_LEN*2 || *cp != ' ')
+				break;
+			while (*++cp == ' ') {}
+		}
+
+		if (*cp == '=') {
+			continue;
+		} else {
+			for (i = 0; i < checksum_len*2; i++, cp++) {
+				int x;
+				if (isXDigit(cp)) {
+					if (isDigit(cp))
+						x = *cp - '0';
+					else
+						x = (*cp & 0xF) + 9;
+				} else {
+					cp = "";
+					break;
+				}
+				if (i & 1)
+					sum[i/2] |= x;
+				else
+					sum[i/2] = x << 4;
+			}
+		}
+		if (*cp != ' ')
+			break;
+		while (*++cp == ' ') {}
+
+		if (protocol_version < 30) {
+			char *alt_sum = cp;
+			if (*cp == '=')
+				while (*++cp == '=') {}
+			else
+				while (isXDigit(cp)) cp++;
+			if (cp - alt_sum != MD5_DIGEST_LEN*2 || *cp != ' ')
+				break;
+			while (*++cp == ' ') {}
+		}
+
+		file_length = 0;
+		while (isDigit(cp))
+			file_length = file_length * 10 + *cp++ - '0';
+		if (*cp != ' ')
+			break;
+		while (*++cp == ' ') {}
+
+		mtime = 0;
+		while (isDigit(cp))
+			mtime = mtime * 10 + *cp++ - '0';
+		if (*cp != ' ')
+			break;
+		while (*++cp == ' ') {}
+
+		/* Ignore ctime. */
+		while (isDigit(cp))
+			cp++;
+		if (*cp != ' ')
+			break;
+		while (*++cp == ' ') {}
+
+		/* Ignore inode. */
+		while (isDigit(cp))
+			cp++;
+		if (*cp != ' ')
+			break;
+		while (*++cp == ' ') {}
+
+		len = strlen(cp);
+		while (len && (cp[len-1] == '\n' || cp[len-1] == '\r'))
+			len--;
+		if (!len)
+			break;
+		cp[len++] = '\0'; /* len now counts the null */
+		if (strchr(cp, '/'))
+			break;
+		if (len > MAXPATHLEN)
+			continue;
+
+		strlcpy(fbuf+dlen, cp, sizeof fbuf - dlen);
+
+		add_checksum(dirname, cp, len, file_length, mtime, sum);
+	}
+	fclose(fp);
+
+	clean_flist(checksum_flist, 0);
+}
+
 int push_pathname(const char *dir, int len)
 {
 	if (dir == pathname)
@@ -989,7 +1170,7 @@ struct file_struct *make_file(const char
 			      STRUCT_STAT *stp, int flags, int filter_level)
 {
 	static char *lastdir;
-	static int lastdir_len = -1;
+	static int lastdir_len = -2;
 	struct file_struct *file;
 	char thisname[MAXPATHLEN];
 	char linkname[MAXPATHLEN];
@@ -1119,9 +1300,16 @@ struct file_struct *make_file(const char
 			memcpy(lastdir, thisname, len);
 			lastdir[len] = '\0';
 			lastdir_len = len;
+			if (always_checksum && am_sender && flist)
+				read_checksums(lastdir);
 		}
-	} else
+	} else {
 		basename = thisname;
+		if (always_checksum && am_sender && flist && lastdir_len == -2) {
+			lastdir_len = -1;
+			read_checksums("");
+		}
+	}
 	basename_len = strlen(basename) + 1; /* count the '\0' */
 
 #ifdef SUPPORT_LINKS
@@ -1197,11 +1385,21 @@ struct file_struct *make_file(const char
 	}
 #endif
 
-	if (always_checksum && am_sender && S_ISREG(st.st_mode))
-		file_checksum(thisname, tmp_sum, st.st_size);
-
 	F_PATHNAME(file) = pathname;
 
+	if (always_checksum && am_sender && S_ISREG(st.st_mode)) {
+		int j;
+		if (flist && (j = flist_find(checksum_flist, file)) >= 0) {
+			struct file_struct *fp = checksum_flist->sorted[j];
+			if (F_LENGTH(fp) == st.st_size
+			 && fp->modtime == st.st_mtime)
+				memcpy(tmp_sum, F_SUM(fp), MAX_DIGEST_LEN);
+			else
+				file_checksum(thisname, tmp_sum, st.st_size);
+		} else
+			file_checksum(thisname, tmp_sum, st.st_size);
+	}
+
 	/* This code is only used by the receiver when it is building
 	 * a list of files for a delete pass. */
 	if (keep_dirlinks && linkname_len && flist) {
@@ -2051,7 +2249,11 @@ struct file_list *send_file_list(int f, 
 			 * file-list to check if this is a 1-file xfer. */
 			send_extra_file_list(f, 1);
 		}
-	}
+	} else
+		flist_eof = 1;
+	
+	if (checksum_updating && always_checksum && flist_eof)
+		read_checksums(NULL);
 
 	return flist;
 }
--- old/ifuncs.h
+++ new/ifuncs.h
@@ -64,6 +64,12 @@ isDigit(const char *ptr)
 }
 
 static inline int
+isXDigit(const char *ptr)
+{
+	return isxdigit(*(unsigned char *)ptr);
+}
+
+static inline int
 isPrint(const char *ptr)
 {
 	return isprint(*(unsigned char *)ptr);
--- old/support/rsyncsums
+++ new/support/rsyncsums
@@ -0,0 +1,183 @@
+#!/usr/bin/perl -w
+use strict;
+
+use Getopt::Long;
+use Cwd qw(abs_path cwd);
+use Digest::MD4;
+use Digest::MD5;
+
+our $SUMS_FILE = '.rsyncsums';
+
+&Getopt::Long::Configure('bundling');
+&usage if !&GetOptions(
+    'simple-cmp|s' => \( my $ignore_ctime_and_inode ),
+    'recurse|r' => \( my $recurse_opt ),
+    'verbose|v+' => \( my $verbosity = 0 ),
+    'help|h' => \( my $help_opt ),
+);
+&usage if $help_opt;
+
+my $start_dir = cwd();
+
+my @dirs = @ARGV;
+@dirs = '.' unless @dirs;
+foreach (@dirs) {
+    $_ = abs_path($_);
+}
+
+$| = 1;
+
+my $md4 = Digest::MD4->new;
+my $md5 = Digest::MD5->new;
+
+while (@dirs) {
+    my $dir = shift @dirs;
+
+    if (!chdir($dir)) {
+	warn "Unable to chdir to $dir: $!\n";
+	next;
+    }
+    if (!opendir(DP, '.')) {
+	warn "Unable to opendir $dir: $!\n";
+	next;
+    }
+
+    if ($verbosity) {
+	my $reldir = $dir;
+	$reldir =~ s#^$start_dir(/|$)# $1 ? '' : '.' #eo;
+	print "$reldir ... ";
+    }
+
+    my $sums_file_exists = -e $SUMS_FILE;
+    my %cache;
+    my @subdirs;
+    my $cnt = 0;
+    while (defined(my $fn = readdir(DP))) {
+	next if $fn =~ /^\.\.?$/ || $fn =~ /^\Q$SUMS_FILE\E$/o || -l $fn;
+	if (-d _) {
+	    push(@subdirs, "$dir/$fn") unless $fn =~ /^(CVS|\.svn|\.git|\.bzr)$/;
+	    next;
+	}
+	next unless -f _;
+
+	my($size,$mtime,$ctime,$inode) = (stat(_))[7,9,10,1];
+	next if $size == 0;
+
+	$cache{$fn} = [ $size, $mtime, $ctime & 0xFFFFFFFF, $inode & 0xFFFFFFFF ];
+	$cnt++;
+    }
+
+    closedir DP;
+
+    unshift(@dirs, sort @subdirs) if $recurse_opt;
+
+    if (!$cnt) {
+	if ($sums_file_exists) {
+	    print "(removed $SUMS_FILE) " if $verbosity;
+	    unlink($SUMS_FILE);
+	}
+	print "empty\n" if $verbosity;
+	next;
+    }
+
+    if (open(FP, '+<', $SUMS_FILE)) {
+	while (<FP>) {
+	    chomp;
+	    my($sum4, $sum5, $size, $mtime, $ctime, $inode, $fn) = split(' ', $_, 7);
+	    my $ref = $cache{$fn};
+	    if (defined $ref) {
+		if ($ignore_ctime_and_inode) {
+		    $ctime = $$ref[2];
+		    $inode = $$ref[3];
+		}
+		if ($$ref[0] == $size
+		 && $$ref[1] == $mtime
+		 && $$ref[2] == $ctime
+		 && $$ref[3] == $inode
+		 && $sum4 !~ /=/ && $sum5 !~ /=/) {
+		    $$ref[4] = $sum4;
+		    $$ref[5] = $sum5;
+		    $cnt--;
+		} else {
+		    $$ref[4] = $$ref[5] = undef;
+		}
+	    } else {
+		$cnt = -1; # Force rewrite due to removed line.
+	    }
+	}
+    } else {
+	open(FP, '>', $SUMS_FILE) or die "Unable to write $dir/$SUMS_FILE: $!\n";
+	$cnt = -1;
+    }
+
+    if ($cnt) {
+	print "UPDATING\n" if $verbosity;
+	while (my($fn, $ref) = each %cache) {
+	    next if defined $$ref[3] && defined $$ref[4];
+	    if (!open(IN, $fn)) {
+		print STDERR "Unable to read $fn: $!\n";
+		delete $cache{$fn};
+		next;
+	    }
+
+	    my($size,$mtime,$ctime,$inode) = (stat(IN))[7,9,10,1];
+	    if ($size == 0) {
+		close IN;
+		next;
+	    }
+
+	    my($sum4, $sum5);
+	    while (1) {
+		while (sysread(IN, $_, 64*1024)) {
+		    $md4->add($_);
+		    $md5->add($_);
+		}
+		$sum4 = $md4->hexdigest;
+		$sum5 = $md5->hexdigest;
+		print " $sum4 $sum5" if $verbosity > 2;
+		print " $fn\n" if $verbosity > 1;
+		my($size2,$mtime2,$ctime2,$inode2) = (stat(IN))[7,9,10,1];
+		if ($ignore_ctime_and_inode) {
+		    $ctime = $ctime2;
+		    $inode = $inode2;
+		}
+		last if $size == $size2 && $mtime == $mtime2
+		     && $ctime == $ctime2 && $inode == $inode2;
+		$size = $size2;
+		$mtime = $mtime2;
+		$ctime = $ctime2;
+		$inode = $inode2;
+		sysseek(IN, 0, 0);
+	    }
+	    
+	    close IN;
+
+	    $cache{$fn} = [ $size, $mtime, $ctime, $inode, $sum4, $sum5 ];
+	}
+
+	seek(FP, 0, 0);
+	foreach my $fn (sort keys %cache) {
+	    my $ref = $cache{$fn};
+	    my($size, $mtime, $ctime, $inode, $sum4, $sum5) = @$ref;
+	    printf FP '%s %s %10d %10d %10d %10d %s' . "\n", $sum4, $sum5, $size, $mtime, $ctime, $inode, $fn;
+	}
+	truncate(FP, tell(FP));
+    } else {
+	print "ok\n" if $verbosity;
+    }
+
+    close FP;
+}
+
+sub usage
+{
+    die <<EOT;
+Usage: rsyncsums [OPTIONS] [DIRS]
+
+Options:
+ -r, --recurse     Update $SUMS_FILE files in subdirectories too.
+ -s, --simple-cmp  Ignore ctime and inode values when comparing identicality.
+ -v, --verbose     Mention what we're doing.  Repeat for more info.
+ -h, --help        Display this help message.
+EOT
+}