Doc.

do_hard_links() actually only looks at the global hardlink table, so
it can be a (void) fn.  (Another gcc warning...)

.cvsignore

@@ -1,3 +1,4 @@
+testtmp.*
 ID
 Makefile
 config.cache

Makefile.in

@@ -62,8 +62,11 @@ install-strip:
 	$(MAKE) INSTALLCMD='$(INSTALLCMD) -s' install
 
 rsync: $(OBJS)
+	@echo "Please ignore warnings below about mktemp -- it is used in a safe way"
 	$(CC) $(CFLAGS) $(LDFLAGS) -o rsync $(OBJS) $(LIBS)
 
+$(OBJS): config.h
+
 tls: $(tls_OBJ)
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(tls_OBJ) $(LIBS)
 

NEWS

@@ -1,114 +1,29 @@
-rsync 2.5.0 (2001-11-30)		-*- indented-text -*- 
+rsync 2.5.2 (???)
 
-  ANNOUNCEMENTS
+  SECURITY FIXES:
 
-    * Martin Pool <mbp@samba.org> is now a co-maintainer.
-
-  NEW FEATURES
-
-    * Support for LSB-compliant packaging <http://www.linuxbase.org/>
-
-    * Shell wildcards are allowed in "auth users" lines.
-
-    * Merged UNC rsync+ patch to support creation of standalone patch
-      sets.  By Bert J. Dempsey and Debra Weiss, updated by Jos
-      Backus.  <http://www.ils.unc.edu/i2dsi/unc_rsync+.html>
-
-    * IPv6 support based on a patch from KAME.net, on systems
-      including modern versions of Linux, Solaris, and HP-UX.  Also
-      includes IPv6 compatibility functions for old OSs by the
-      Internet Software Consortium, Paul Vixie, the OpenSSH
-      portability project, and OpenBSD.
-
-  ENHANCEMENTS
-
-    * Include/exclude cluestick: with -vv, print out whether files are
-      included or excluded and why.
-
-    * Many error messages have more friendly explanations and more
-      details.
-
-    * Manual page improvements plus scanty protocol documentation.
-
-    * When running as --daemon in the background and using a "log
-      file" rsyncd.conf directive, close the log file every time it is
-      open when going to sleep on the socket.  This allows the log
-      file to get cleaned out by another process.
-
-    * Change to using libpopt rather than getopt for processing
-      options.	This makes the code cleaner and the behaviour more
-      consistent across platforms.  popt is included and built if not
-      installed on the platform.
-
-    * More details in --version, including note about whether 64-bit
-      files, symlinks and hardlinks are supported.
-
-    * MD4 code may use less CPU cycles.
-
-    * Use mkstemp on systems where it is secure.  If we use mktemp,
-      explain that we do it in a secure way.
-
-    * --whole-file is the default when source and target are on the
-	local machine.
+    * Signedness security patch from Sebastian Krahmer
+      <krahmer@suse.de> -- in some cases we were not sufficiently
+      careful about reading integers from the network.
 
   BUG FIXES:
 
-    * Fix for various bugs causing rsync to hang.
+    * Fix possible string mangling in log files.
 
-    * Attempt to fix Large File Summit support on AIX.
+    * Fix for setting local address of outgoing sockets.
 
-    * Attempt to fix error handling lockup bug.
+    * Better handling of hardlinks and devices on platforms with
+      64-bit dev_t or ino_t.
 
-    * Give a non-0 exit code if *any* of the files we have been asked
-      to transfer fail to transfer 
+  ENHANCEMENTS:
 
-    * For log messages containing ridiculously long strings that might
-      overflow a buffer rsync no longer aborts, but rather prints an
-      ellipsis at the end of the string.  (Patch from Ed Santiago.)
+    * --statistics now shows memory heap usage on platforms that
+        support mallinfo().
 
-  PLATFORMS:
+    * "The Ted T'so school of program optimization": make progress
+      visible and people will think it's faster.  (With --progress,
+      rsync will show you how many files it has seen as it builds the
+      file_list, giving some indication that it has not hung.)
 
-    * Improved support for UNICOS (tested on Cray T3E and Cray SV1)
-
-    * autoconf2.52 (or later) is now required to rebuild the autoconf
-      scripts.	It is not required to simply build rsync.
-
-    * Platforms thought to work in this release:
-
-		Cray SV1 UNICOS 10.0.0.8 cc
-		Debian Linux 2.2 UltraSparc gcc
-		Debian Linux testing/unstable ARM gcc
-		FreeBSD 3.3-RELEASE i386 cc
-		FreeBSD 4.1.1-RELEASE i386 cc
-		FreeBSD 4.3-STABLE i386 cc
-		HP PA-RISC HP-UX 10.20 gcc
-		HP PA-RISC HP-UX 11.11 cc
-		IRIX 6.5 MIPS cc
-		IRIX 6.5 MIPS gcc
-		Mac OS X PPC (--disable-ipv6) cc
-		NetBSD 1.5 i386 gcc
-		NetBSD Current i386 cc
-		OpenBSD 2.5 Sparc gcc
-		OpenBSD 2.9 i386 cc
-		OpenBSD Current i386 cc
-		RedHat 6.2 i386 gcc
-		RedHat 6.2 i386 insure++
-		RedHat 7.0 i386 gcc
-		RedHat 7.1 i386 (Kernel 2.4.10) gcc
-		Slackware 8.0 i686 (Kernel 2.4.10)
-		Solaris 8 UltraSparc cc
-		Solaris 8 UltraSparc gcc
-		Solaris 8 i386 gcc
-		SuSE 7.1 i386 gcc2.95.2
-		SuSE 7.1 ppc gcc2.95.2
-		i386-pc-sco3.2v5.0.5 cc
-		i386-pc-sco3.2v5.0.5 gcc
-		powerpc-ibm-aix4.3.3.0 cc
-		i686-unknown-sysv5UnixWare7.1.0 gcc
-		i686-unknown-sysv5UnixWare7.1.0 cc
-
-  TESTING:
-
-    * The existing test.sh script by Phil Hands has been merged into a
-      test framework that works from both "make check" and the Samba
-      build farm.
+    * Improvements to batch mode support.  This is still experimental
+      but testing would be welcome.   (Jos Backus)

OLDNEWS

@@ -0,0 +1,148 @@
+rsync 2.5.1 (2002-01-03)
+
+  BUG FIXES: 
+   
+    * Fix for segfault in --daemon mode configuration parser.  (Paul
+      Mackerras)
+
+    * Correct string<->address parsing for both IPv4 and 6.
+      (YOSHIFUJI Hideaki, SUMIKAWA Munechika and Jun-ichiro "itojun"
+      Hagino)
+
+    * Various fixes for IPv6 support.  (Dave Dykstra)
+
+    * rsync.1 typo fix.  (Matt Kraai)
+
+    * Test suite typo fixes.  (Tom Schmidt)
+
+    * rsync.1 grammar and clarity improvements.  (Edward
+      Welbourne)
+
+    * Correction to ./configure tests for inet_ntop.  (Jeff Garzik)
+
+  ENHANCEMENTS:
+
+    * --progress and -P now show estimated data transfer rate (in a
+        multiple of bytes/s) and estimated time to completion.  (Rik
+        Faith)
+   
+    * --no-detach option, required to run as a W32 service and also
+        useful when running on Unix under daemontools, AIX's SRC, or a
+        debugger.  (Max Bowsher, Jos Backus)
+
+    * Clearer error messages for some conditions.
+
+rsync 2.5.0 (2001-11-30)
+
+  ANNOUNCEMENTS
+
+    * Martin Pool <mbp@samba.org> is now a co-maintainer.
+
+  NEW FEATURES
+
+    * Support for LSB-compliant packaging <http://www.linuxbase.org/>
+
+    * Shell wildcards are allowed in "auth users" lines.
+
+    * Merged UNC rsync+ patch to support creation of standalone patch
+      sets.  By Bert J. Dempsey and Debra Weiss, updated by Jos
+      Backus.  <http://www.ils.unc.edu/i2dsi/unc_rsync+.html>
+
+    * IPv6 support based on a patch from KAME.net, on systems
+      including modern versions of Linux, Solaris, and HP-UX.  Also
+      includes IPv6 compatibility functions for old OSs by the
+      Internet Software Consortium, Paul Vixie, the OpenSSH
+      portability project, and OpenBSD.
+
+  ENHANCEMENTS
+
+    * Include/exclude cluestick: with -vv, print out whether files are
+      included or excluded and why.
+
+    * Many error messages have more friendly explanations and more
+      details.
+
+    * Manual page improvements plus scanty protocol documentation.
+
+    * When running as --daemon in the background and using a "log
+      file" rsyncd.conf directive, close the log file every time it is
+      open when going to sleep on the socket.  This allows the log
+      file to get cleaned out by another process.
+
+    * Change to using libpopt rather than getopt for processing
+      options.	This makes the code cleaner and the behaviour more
+      consistent across platforms.  popt is included and built if not
+      installed on the platform.
+
+    * More details in --version, including note about whether 64-bit
+      files, symlinks and hardlinks are supported.
+
+    * MD4 code may use less CPU cycles.
+
+    * Use mkstemp on systems where it is secure.  If we use mktemp,
+      explain that we do it in a secure way.
+
+    * --whole-file is the default when source and target are on the
+	local machine.
+
+  BUG FIXES:
+
+    * Fix for various bugs causing rsync to hang.
+
+    * Attempt to fix Large File Summit support on AIX.
+
+    * Attempt to fix error handling lockup bug.
+
+    * Give a non-0 exit code if *any* of the files we have been asked
+      to transfer fail to transfer 
+
+    * For log messages containing ridiculously long strings that might
+      overflow a buffer rsync no longer aborts, but rather prints an
+      ellipsis at the end of the string.  (Patch from Ed Santiago.)
+
+  PLATFORMS:
+
+    * Improved support for UNICOS (tested on Cray T3E and Cray SV1)
+
+    * autoconf2.52 (or later) is now required to rebuild the autoconf
+      scripts.	It is not required to simply build rsync.
+
+    * Platforms thought to work in this release:
+
+		Cray SV1 UNICOS 10.0.0.8 cc
+		Debian Linux 2.2 UltraSparc gcc
+		Debian Linux testing/unstable ARM gcc
+		FreeBSD 3.3-RELEASE i386 cc
+		FreeBSD 4.1.1-RELEASE i386 cc
+		FreeBSD 4.3-STABLE i386 cc
+		HP PA-RISC HP-UX 10.20 gcc
+		HP PA-RISC HP-UX 11.11 cc
+		IRIX 6.5 MIPS cc
+		IRIX 6.5 MIPS gcc
+		Mac OS X PPC (--disable-ipv6) cc
+		NetBSD 1.5 i386 gcc
+		NetBSD Current i386 cc
+		OpenBSD 2.5 Sparc gcc
+		OpenBSD 2.9 i386 cc
+		OpenBSD Current i386 cc
+		RedHat 6.2 i386 gcc
+		RedHat 6.2 i386 insure++
+		RedHat 7.0 i386 gcc
+		RedHat 7.1 i386 (Kernel 2.4.10) gcc
+		Slackware 8.0 i686 (Kernel 2.4.10)
+		Solaris 8 UltraSparc cc
+		Solaris 8 UltraSparc gcc
+		Solaris 8 i386 gcc
+		SuSE 7.1 i386 gcc2.95.2
+		SuSE 7.1 ppc gcc2.95.2
+		i386-pc-sco3.2v5.0.5 cc
+		i386-pc-sco3.2v5.0.5 gcc
+		powerpc-ibm-aix4.3.3.0 cc
+		i686-unknown-sysv5UnixWare7.1.0 gcc
+		i686-unknown-sysv5UnixWare7.1.0 cc
+
+  TESTING:
+
+    * The existing test.sh script by Phil Hands has been merged into a
+      test framework that works from both "make check" and the Samba
+      build farm.

README

@@ -143,9 +143,7 @@ This will give you access to the bug tracking system used by the
 developers of rsync and will allow you to look at other bug reports or
 submit a new bug report.
 
-If you don't have web access then mail bug reports to
-rsync-bugs@samba.org or (if you think it will be of interest to lots
-of people) send it to rsync@samba.org
+If you don't have web access then mail bug reports to rsync@samba.org.
 
 
 CVS TREE
@@ -167,11 +165,12 @@ details.
 COPYRIGHT
 ---------
 
-Rsync was written by Andrew Tridgell and Paul Mackerras, and is
-available under the GNU General Public License.
+rsync was originally written by Andrew Tridgell and has been improved
+by many developers around the world.  rsync may be used, modified and
+redistributed only under the terms of the GNU General Public License,
+found in the file COPYING in this distribution, or at 
 
-tridge@samba.org
-paulus@cs.anu.edu.au
+  http://www.fsf.org/licenses/gpl.html
 
 
 AVAILABILITY

TODO

@@ -32,15 +32,152 @@ use chroot
   for people who want to generate the file list using a find(1)
   command or a script.
 
+
 Performance
 
   Traverse just one directory at a time.  Tridge says it's possible.
- 
-  Can possibly also be smarter about memory use while looking for hard
-  links by reducing the refcount as we find alternative names.
+
+  At the moment rsync reads the whole file list into memory at the
+  start, which makes us use a lot of memory and also not pipeline
+  network access as much as we could.
+
+
+Handling duplicate names
+
+  We need to be careful of duplicate names getting into the file list.
+  See clean_flist().  This could happen if multiple arguments include
+  the same file.  Bad.
+
+  I think duplicates are only a problem if they're both flowing
+  through the pipeline at the same time.  For example we might have
+  updated the first occurrence after reading the checksums for the
+  second.  So possibly we just need to make sure that we don't have
+  both in the pipeline at the same time.  
+
+  Possibly if we did one directory at a time that would be sufficient.
+
+  Alternatively we could pre-process the arguments to make sure no
+  duplicates will ever be inserted.  There could be some bad cases
+  when we're collapsing symlinks.
+
+  We could have a hash table.
+
+  The root of the problem is that we do not want more than one file
+  list entry referring to the same file.  At first glance there are
+  several ways this could happen: symlinks, hardlinks, and repeated
+  names on the command line.
+
+  If names are repeated on the command line, they may be present in
+  different forms, perhaps by traversing directory paths in different
+  ways, traversing paths including symlinks.  Also we need to allow
+  for expansion of globs by rsync.
+
+  At the moment, clean_flist() requires having the entire file list in
+  memory.  Duplicate names are detected just by a string comparison.
+
+  We don't need to worry about hard links causing duplicates because
+  files are never updated in place.  Similarly for symlinks.
+
+  I think even if we're using a different symlink mode we don't need
+  to worry.
+
+  Unless we're really clever this will introduce a protocol
+  incompatibility, so we need to be able to accept the old format as
+  well.
+
+
+Memory accounting
+
+  At exit, show how much memory was used for the file list, etc.
+
+  Also we do a wierd exponential-growth allocation in flist.c.  I'm
+  not sure this makes sense with modern mallocs.  At any rate it will
+  make us allocate a huge amount of memory for large file lists.
+
+  We can try using the GNU/SVID/XPG mallinfo() function to get some
+  heap statistics.
+
+
+Hard-link handling
+
+  At the moment hardlink handling is very expensive, so it's off by
+  default.  It does not need to be so.  
+
+  Since most of the solutions are rather intertwined with the file
+  list it is probably better to fix that first, although fixing
+  hardlinks is possibly simpler.
+
+  We can rule out hardlinked directories since they will probably
+  screw us up in all kinds of ways.  They simply should not be used.
+
+  At the moment rsync only cares about hardlinks to regular files.  I
+  guess you could also use them for sockets, devices and other beasts,
+  but I have not seen them.
+
+  When trying to reproduce hard links, we only need to worry about
+  files that have more than one name (nlinks>1 && !S_ISDIR). 
+
+  The basic point of this is to discover alternate names that refer to
+  the same file.  All operations, including creating the file and
+  writing modifications to it need only to be done for the first name.
+  For all later names, we just create the link and then leave it
+  alone.
+
+  If hard links are to be preserved:
+
+    Before the generator/receiver fork, the list of files is received
+    from the sender (recv_file_list), and a table for detecting hard
+    links is built.
+
+    The generator looks for hard links within the file list and does
+    not send checksums for them, though it does send other metadata.
+
+    The sender sends the device number and inode with file entries, so
+    that files are uniquely identified.
+
+    The receiver goes through and creates hard links (do_hard_links)
+    after all data has been written, but before directory permissions
+    are set.
+
+  At the moment device and inum are sent as 4-byte integers, which
+  will probably cause problems on large filesystems.  On Linux the
+  kernel uses 64-bit ino_t's internally, and people will soon have
+  filesystems big enough to use them.  We ought to follow NFS4 in
+  using 64-bit device and inode identification, perhaps with a
+  protocol version bump.
+
+  Once we've seen all the names for a particular file, we no longer
+  need to think about it and we can deallocate the memory.
+
+  We can also have the case where there are links to a file that are
+  not in the tree being transferred.  There's nothing we can do about
+  that.  Because we rename the destination into place after writing,
+  any hardlinks to the old file are always going to be orphaned.  In
+  fact that is almost necessary because otherwise we'd get really
+  confused if we were generating checksums for one name of a file and
+  modifying another.
+
+  At the moment the code seems to make a whole second copy of the file
+  list, which seems unnecessary.
+
+  We should have a test case that exercises hard links.  Since it
+  might be hard to compare ./tls output where the inodes change we
+  might need a little program to check whether several names refer to
+  the same file.
 
 IPv6
 
+  Implement suggestions from http://www.kame.net/newsletter/19980604/
+  and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt
+
+  If a host has multiple addresses, then listen try to connect to all
+  in order until we get through.  (getaddrinfo may return multiple
+  addresses.)  This is kind of implemented already.
+
+  Possibly also when starting as a server we may need to listen on
+  multiple passive addresses.  This might be a bit harder, because we
+  may need to select on all of them.  Hm.
+
   Define a syntax for IPv6 literal addresses.  Since they include
   colons, they tend to break most naming systems, including ours.
   Based on the HTTP IPv6 syntax, I think we should use
@@ -61,6 +198,48 @@ Errors
   eof" give a message that is more detailed if possible and also more
   helpful.  
 
+File attributes
+
+  Device major/minor numbers should be at least 32 bits each.  See
+  http://lists.samba.org/pipermail/rsync/2001-November/005357.html
+
+  Transfer ACLs.  Need to think of a standard representation.
+  Probably better not to even try to convert between NT and POSIX.
+  Possibly can share some code with Samba.
+
+Empty directories
+
+  With the current common --include '*/' --exclude '*' pattern, people
+  can end up with many empty directories.  We might avoid this by
+  lazily creating such directories.
+
+zlib
+
+  Perhaps don't use our own zlib.  Will we actually be incompatible,
+  or just be slightly less efficient?
+
+logging
+
+  Perhaps flush stdout after each filename, so that people trying to
+  monitor progress in a log file can do so more easily.  See
+  http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108
+
+rsyncd over ssh
+
+  There are already some patches to do this.
+
+proxy authentication
+
+  Allow RSYNC_PROXY to be http://user:pass@proxy.foo:3128/, and do
+  HTTP Basic Proxy-Authentication.  
+
+  Multiple schemes are possible, up to and including the insanity that
+  is NTLM, but Basic probably covers most cases.
+
+SOCKS
+
+  Add --with-socks, and then perhaps a command-line option to put them
+  on or off.  This might be more reliable than LD_PRELOAD hacks.
 
 PLATFORMS ------------------------------------------------------------
 
@@ -77,6 +256,10 @@ Win32
   we are correct to call close(), because shutdown() discards
   untransmitted data.
 
+DOCUMENTATION --------------------------------------------------------
+
+Update README
+
 BUILD FARM -----------------------------------------------------------
 
 Add machines
@@ -99,7 +282,9 @@ NICE -----------------------------------------------------------------
 
 hang/timeout friendliness
 
-  On 
+verbose output
+  
+  Indicate whether files are new, updated, or deleted
 
 internationalization
 
@@ -118,3 +303,4 @@ rsyncsh
    current host, directory and so on.  We can probably even do
    completion of remote filenames.
 
+%K%

clientserver.c

@@ -45,6 +45,7 @@ int start_socket_client(char *host, char *path, int argc, char *argv[])
 	extern char *shell_cmd;
 	extern int kludge_around_eof;
 	extern char *bind_address;
+	extern int default_af_hint;
        
 	if (argc == 0 && !am_sender) {
 		extern int list_only;
@@ -79,7 +80,7 @@ int start_socket_client(char *host, char *path, int argc, char *argv[])
 	if (!user) user = getenv("LOGNAME");
 
 	fd = open_socket_out_wrapped (host, rsync_port, bind_address,
-				      global_opts.af_hint);
+				      default_af_hint);
 	if (fd == -1) {
 		exit_cleanup(RERR_SOCKETIO);
 	}
@@ -486,6 +487,7 @@ int daemon_main(void)
 	extern char *config_file;
 	extern int orig_umask;
 	char *pid_file;
+	extern int no_detach;
 
 	if (is_a_socket(STDIN_FILENO)) {
 		int i;
@@ -501,7 +503,8 @@ int daemon_main(void)
 		return start_daemon(STDIN_FILENO);
 	}
 
-	become_daemon();
+	if (!no_detach)
+	    become_daemon();
 
 	if (!lp_load(config_file, 1)) {
 		exit_cleanup(RERR_SYNTAX);
@@ -513,7 +516,8 @@ int daemon_main(void)
 		RSYNC_VERSION,
                 rsync_port);
         /* TODO: If listening on a particular address, then show that
-         * address too. */
+         * address too.  In fact, why not just do inet_ntop on the
+         * local address??? */
 
 	if (((pid_file = lp_pid_file()) != NULL) && (*pid_file != '\0')) {
 		char pidbuf[16];

configure.in

@@ -5,7 +5,7 @@ AC_CONFIG_SRCDIR([byteorder.h])
 AC_CONFIG_HEADER(config.h)
 AC_PREREQ(2.52)
 
-RSYNC_VERSION=2.5.0
+RSYNC_VERSION=2.5.2pre1
 AC_SUBST(RSYNC_VERSION)
 AC_MSG_NOTICE([Configuring rsync $RSYNC_VERSION])
 
@@ -22,6 +22,9 @@ AC_PROG_INSTALL
 AC_PROG_CC_STDC
 AC_SUBST(SHELL)
 
+AC_DEFINE([_GNU_SOURCE], 1,
+          [Define _GNU_SOURCE so that we get all necessary prototypes])
+
 if test "$xac_cv_prog_cc_stdc" = xno
 then
 	AC_MSG_WARN([rsync requires an ANSI C compiler and you don't seem to have one])
@@ -30,6 +33,9 @@ fi
 # compile with optimisation and without debugging by default, unless
 # --debug is given.  We must decide this before testing the compiler.
 
+# Please allow this to default to yes, so that your users have more
+# chance of getting a useful stack trace if problems occur.
+
 AC_MSG_CHECKING([whether to include debugging symbols])
 AC_ARG_ENABLE(debug,
 	AC_HELP_STRING([--enable-debug],
@@ -43,12 +49,33 @@ then
 else
     AC_MSG_RESULT([yes])
     # leave CFLAGS alone; AC_PROG_CC will try to include -g if it can
-    AC_DEFINE(DEBUG, 1, [Define to turn on debugging code that may slow normal operation])
+    dnl AC_DEFINE(DEBUG, 1, [Define to turn on debugging code that may slow normal operation])
     dnl CFLAGS=${CFLAGS-"-g"}
 fi
 
+
+
+
+AC_ARG_ENABLE(profile,
+	AC_HELP_STRING([--enable-profile],	
+		[turn on CPU profiling (default no)],
+		[], []))
+if test x"$enable_profile" = xyes
+then
+	CFLAGS="$CFLAGS -pg"
+fi
+
+
+# This is needed for our included version of popt.  Kind of silly, but
+# I don't want our version too far out of sync.
 CFLAGS="$CFLAGS -DHAVE_CONFIG_H"
 
+# If GCC, turn on warnings.
+if test "x$GCC" = "xyes"
+then
+	CFLAGS="$CFLAGS -Wall -W"
+fi
+
 AC_ARG_WITH(included-popt,
         [  --with-included-popt    use bundled popt library, not from system])
 
@@ -204,6 +231,7 @@ AC_CHECK_HEADERS(compat.h sys/param.h ctype.h sys/wait.h sys/ioctl.h)
 AC_CHECK_HEADERS(sys/filio.h string.h stdlib.h sys/socket.h sys/mode.h)
 AC_CHECK_HEADERS(glob.h alloca.h mcheck.h sys/sysctl.h arpa/inet.h arpa/nameser.h)
 AC_CHECK_HEADERS(netdb.h)
+AC_CHECK_HEADERS(malloc.h)
 
 AC_CHECK_SIZEOF(int)
 AC_CHECK_SIZEOF(long)
@@ -268,13 +296,13 @@ fi
 
 AC_CHECK_LIB(resolv, inet_ntop)
 
-AC_MSG_NOTICE([Looking in libraries: $LIBS])
+dnl AC_MSG_NOTICE([Looking in libraries: $LIBS])
 
-AC_CHECK_FUNC(inet_ntop, , AC_LIBOBJ(lib/inet_ntop)) 
-AC_CHECK_FUNC(inet_pton, , AC_LIBOBJ(lib/inet_pton))
+AC_CHECK_FUNCS(inet_ntop, , AC_LIBOBJ(lib/inet_ntop)) 
+AC_CHECK_FUNCS(inet_pton, , AC_LIBOBJ(lib/inet_pton))
 
-AC_CHECK_FUNC(getaddrinfo, , AC_LIBOBJ(lib/getaddrinfo)) 
-AC_CHECK_FUNC(getnameinfo, , AC_LIBOBJ(lib/getnameinfo))
+AC_CHECK_FUNCS(getaddrinfo, , AC_LIBOBJ(lib/getaddrinfo)) 
+AC_CHECK_FUNCS(getnameinfo, , AC_LIBOBJ(lib/getnameinfo))
 
 AC_CHECK_MEMBER([struct sockaddr.sa_len], 
 		[ AC_DEFINE(HAVE_SOCKADDR_LEN) ],
@@ -284,6 +312,15 @@ AC_CHECK_MEMBER([struct sockaddr.sa_len],
 #include <sys/socket.h>
 ])
 
+AC_MSG_CHECKING(struct sockaddr_storage)
+AC_TRY_COMPILE([#include <sys/types.h>
+#include <sys/socket.h>],
+[struct sockaddr_storage x;],
+	AC_MSG_RESULT(yes)
+	AC_DEFINE(HAVE_SOCKADDR_STORAGE, 1, 
+		[Define if you have strct sockaddr_storage.] ),
+	AC_MSG_RESULT(no))
+
 # if we can't find strcasecmp, look in -lresolv (for Unixware at least)
 #
 AC_CHECK_FUNCS(strcasecmp)
@@ -300,7 +337,7 @@ AC_FUNC_UTIME_NULL
 AC_CHECK_FUNCS(waitpid wait4 getcwd strdup strerror chown chmod mknod)
 AC_CHECK_FUNCS(fchmod fstat strchr readlink link utime utimes strftime)
 AC_CHECK_FUNCS(memmove lchown vsnprintf snprintf asprintf setsid glob strpbrk)
-AC_CHECK_FUNCS(strlcat strlcpy mtrace)
+AC_CHECK_FUNCS(strlcat strlcpy mtrace mallinfo)
 
 AC_CACHE_CHECK([for working socketpair],rsync_cv_HAVE_SOCKETPAIR,[
 AC_TRY_RUN([

exclude.c

@@ -299,7 +299,8 @@ void send_exclude_list(int f)
 void recv_exclude_list(int f)
 {
 	char line[MAXPATHLEN];
-	int l;
+	unsigned int l;
+
 	while ((l=read_int(f))) {
 		if (l >= MAXPATHLEN) overflow("recv_exclude_list");
 		read_sbuf(f,line,l);

fileio.c

@@ -36,7 +36,7 @@ int sparse_end(int f)
 }
 
 
-static int write_sparse(int f,char *buf,int len)
+static int write_sparse(int f,char *buf,size_t len)
 {
 	int l1=0,l2=0;
 	int ret;
@@ -69,7 +69,7 @@ static int write_sparse(int f,char *buf,int len)
 
 
 
-int write_file(int f,char *buf,int len)
+int write_file(int f,char *buf,size_t len)
 {
 	int ret = 0;
 

flist.c

@@ -1,7 +1,7 @@
 /* 
    Copyright (C) Andrew Tridgell 1996
    Copyright (C) Paul Mackerras 1996
-   Copyright (C) 2001 by Martin Pool <mbp@samba.org>
+   Copyright (C) 2001, 2002 by Martin Pool <mbp@samba.org>
    
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -18,7 +18,14 @@
    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */
 
-/* generate and receive file lists */
+/** @file flist.c
+ * Generate and receive file lists
+ *
+ * @todo Get rid of the string_area optimization.  Efficiently
+ * allocating blocks is the responsibility of the system's malloc
+ * library, not of rsync.
+ *
+ **/
 
 #include "rsync.h"
 
@@ -59,6 +66,35 @@ static struct file_struct null_file;
 
 static void clean_flist(struct file_list *flist, int strip_root);
 
+
+static int show_build_progress_p(void)
+{
+	extern int do_progress;
+	
+	return do_progress && verbose && recurse && !am_server;
+}
+
+/**
+ * True if we're local, etc, and should emit progress emssages.
+ **/
+static void emit_build_progress(const struct file_list *flist)
+{
+	rprintf(FINFO,
+		" %d files...\r",
+		flist->count);
+}
+
+
+static void finish_build_progress(const struct file_list *flist)
+{
+	if (verbose && recurse && !am_server) {
+		/* This overwrites the progress line, if any. */
+		rprintf(FINFO, RSYNC_NAME ": %d files to consider.\n",
+			flist->count);
+	}
+}
+
+
 static struct string_area *string_area_new(int size)
 {
 	struct string_area *a;
@@ -297,8 +333,15 @@ static void send_file_entry(struct file_struct *file,int f,unsigned base_flags)
 
 #if SUPPORT_HARD_LINKS
 	if (preserve_hard_links && S_ISREG(file->mode)) {
-		write_int(f,(int)file->dev);
-		write_int(f,(int)file->inode);
+		if (remote_version < 26) {
+			/* 32-bit dev_t and ino_t */
+			write_int(f,(int)file->dev);
+			write_int(f,(int)file->inode);
+		} else {
+			/* 64-bit dev_t and ino_t */
+			write_longint(f, file->dev);
+			write_longint(f, file->inode);
+		}
 	}
 #endif
 
@@ -332,7 +375,7 @@ static void receive_file_entry(struct file_struct **fptr,
 	static gid_t last_gid;
 	static char lastname[MAXPATHLEN];
 	char thisname[MAXPATHLEN];
-	int l1=0,l2=0;
+	unsigned int l1=0,l2=0;
 	char *p;
 	struct file_struct *file;
 
@@ -399,6 +442,10 @@ static void receive_file_entry(struct file_struct **fptr,
 
 	if (preserve_links && S_ISLNK(file->mode)) {
 		int l = read_int(f);
+		if (l < 0) {
+			rprintf(FERROR,"overflow: l=%d\n", l);
+			overflow("receive_file_entry");
+		}
 		file->link = (char *)malloc(l+1);
 		if (!file->link) out_of_memory("receive_file_entry 2");
 		read_sbuf(f,file->link,l);
@@ -409,8 +456,13 @@ static void receive_file_entry(struct file_struct **fptr,
 
 #if SUPPORT_HARD_LINKS
 	if (preserve_hard_links && S_ISREG(file->mode)) {
-		file->dev = read_int(f);
-		file->inode = read_int(f);
+		if (remote_version < 26) {
+			file->dev = read_int(f);
+			file->inode = read_int(f);
+		} else {
+			file->dev = read_longint(f);
+			file->inode = read_longint(f);
+		}
 	}
 #endif
   
@@ -464,7 +516,8 @@ static int skip_filesystem(char *fname, STRUCT_STAT *st)
 }
 
 #define STRDUP(ap, p)	(ap ? string_area_strdup(ap, p) : strdup(p))
-#define MALLOC(ap, i)	(ap ? string_area_malloc(ap, i) : malloc(i))
+/* IRIX cc cares that the operands to the ternary have the same type. */
+#define MALLOC(ap, i)	(ap ? (void*) string_area_malloc(ap, i) : malloc(i))
 
 /* create a file_struct for a named file */
 struct file_struct *make_file(int f, char *fname, struct string_area **ap,
@@ -606,7 +659,10 @@ void send_file_name(int f,struct file_list *flist,char *fname,
 
   file = make_file(f,fname, &flist->string_area, 0);
 
-  if (!file) return;  
+  if (!file) return;
+
+  if (show_build_progress_p() & !(flist->count % 100))
+	  emit_build_progress(flist);
   
   if (flist->count >= flist->malloced) {
 	  if (flist->malloced < 1000)
@@ -697,6 +753,11 @@ static void send_directory(int f,struct file_list *flist,char *dir)
 }
 
 
+/*
+ *
+ * I *think* f==-1 means that the list should just be built in memory
+ * and not transmitted.  But who can tell? -- mbp
+ */
 struct file_list *send_file_list(int f,int argc,char *argv[])
 {
 	int i,l;
@@ -707,7 +768,7 @@ struct file_list *send_file_list(int f,int argc,char *argv[])
 	int64 start_write;
 
 	if (verbose && recurse && !am_server && f != -1) {
-		rprintf(FINFO,"building file list ... ");
+		rprintf(FINFO, RSYNC_NAME ": building file list...\n");
                 if (verbose > 1)
                         rprintf(FINFO, "\n");
 		rflush(FINFO);
@@ -830,8 +891,7 @@ struct file_list *send_file_list(int f,int argc,char *argv[])
 		send_file_entry(NULL,f,0);
 	}
 
-	if (verbose && recurse && !am_server && f != -1)
-		rprintf(FINFO,"done\n");
+	finish_build_progress(flist);
 	
 	clean_flist(flist, 0);
 	
@@ -962,6 +1022,10 @@ oom:
 }
 
 
+/*
+ * XXX: This is currently the hottest function while building the file
+ * list, because building f_name()s every time is expensive.
+ **/
 int file_compare(struct file_struct **f1,struct file_struct **f2)
 {
 	if (!(*f1)->basename && !(*f2)->basename) return 0;
@@ -1084,6 +1148,10 @@ static void clean_flist(struct file_list *flist, int strip_root)
 		} 
 	}
 
+	/* FIXME: There is a bug here when filenames are repeated more
+	 * than once, because we don't handle freed files when doing
+	 * the comparison. */
+
 	if (strip_root) {
 		/* we need to strip off the root directory in the case
 		   of relative paths, but this must be done _after_
@@ -1119,6 +1187,10 @@ static void clean_flist(struct file_list *flist, int strip_root)
 
 /*
  * return the full filename of a flist entry
+ *
+ * This function is too expensive at the moment, because it copies
+ * strings when often we only want to compare them.  In any case,
+ * using strlcat is silly because it will walk the string repeatedly.
  */
 char *f_name(struct file_struct *f)
 {
@@ -1131,9 +1203,11 @@ char *f_name(struct file_struct *f)
 	n = (n+1)%10;
 
 	if (f->dirname) {
-		strlcpy(p, f->dirname, MAXPATHLEN);
-		strlcat(p, "/", MAXPATHLEN);
-		strlcat(p, f->basename, MAXPATHLEN);
+		int off;
+
+		off = strlcpy(p, f->dirname, MAXPATHLEN);
+		off += strlcpy(p+off, "/", MAXPATHLEN-off);
+		off += strlcpy(p+off, f->basename, MAXPATHLEN-off);
 	} else {
 		strlcpy(p, f->basename, MAXPATHLEN);
 	}

generator.c

@@ -1,7 +1,10 @@
 /* -*- c-file-style: "linux" -*-
+
+   rsync -- fast file replication program
    
    Copyright (C) 1996-2000 by Andrew Tridgell 
    Copyright (C) Paul Mackerras 1996
+   Copyright (C) 2002 by Martin Pool <mbp@samba.org>
    
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -99,25 +102,30 @@ static int adapt_block_size(struct file_struct *file, int bsize)
 /*
   send a sums struct down a fd
   */
-static void send_sums(struct sum_struct *s,int f_out)
+static void send_sums(struct sum_struct *s, int f_out)
 {
-	int i;
-	
-	/* tell the other guy how many we are going to be doing and how many
-	   bytes there are in the last chunk */
-	write_int(f_out,s?s->count:0);
-	write_int(f_out,s?s->n:block_size);
-	write_int(f_out,s?s->remainder:0);
+	if (s) {
+		size_t i;
 
-	if (!s) return;
+		/* tell the other guy how many we are going to be
+		   doing and how many bytes there are in the last
+		   chunk */
+		write_int(f_out, s->count);
+		write_int(f_out, s->n);
+		write_int(f_out, s->remainder);
 
-	for (i=0;i<s->count;i++) {
-		write_int(f_out,s->sums[i].sum1);
-		write_buf(f_out,s->sums[i].sum2,csum_length);
+		for (i = 0; i < s->count; i++) {
+			write_int(f_out, s->sums[i].sum1);
+			write_buf(f_out, s->sums[i].sum2, csum_length);
+		}
+	} else {
+		/* we don't have checksums */
+		write_int(f_out, 0);
+		write_int(f_out, block_size);
+		write_int(f_out, 0);
 	}
 }
 
-
 /*
   generate a stream of signatures/checksums that describe a buffer
 
@@ -210,7 +218,8 @@ void recv_generator(char *fname,struct file_list *flist,int i,int f_out)
 
 	if (only_existing && statret == -1 && errno == ENOENT) {
 		/* we only want to update existing files */
-		if (verbose > 1) rprintf(FINFO,"not creating %s\n",fname);
+		if (verbose > 1) rprintf(FINFO, RSYNC_NAME
+					 ": not creating new file \"%s\"\n",fname);
 		return;
 	}
 
@@ -230,10 +239,11 @@ void recv_generator(char *fname,struct file_list *flist,int i,int f_out)
                  * we need to delete it.  If it doesn't exist, then
                  * recursively create it. */
           
-		if (dry_run) return;
+		if (dry_run) return; /* XXXX -- might cause inaccuracies?? -- mbp */
 		if (statret == 0 && !S_ISDIR(st.st_mode)) {
 			if (robust_unlink(fname) != 0) {
-				rprintf(FERROR,"recv_generator: unlink %s: %s\n",
+				rprintf(FERROR, RSYNC_NAME
+					": recv_generator: unlink \"%s\" to make room for directory: %s\n",
                                         fname,strerror(errno));
 				return;
 			}
@@ -243,11 +253,13 @@ void recv_generator(char *fname,struct file_list *flist,int i,int f_out)
 			if (!(relative_paths && errno==ENOENT && 
 			      create_directory_path(fname)==0 && 
 			      do_mkdir(fname,file->mode)==0)) {
-				rprintf(FERROR,"recv_generator: mkdir %s: %s (2)\n",
+				rprintf(FERROR, RSYNC_NAME ": recv_generator: mkdir \"%s\": %s (2)\n",
 					fname,strerror(errno));
 			}
 		}
-		if (set_perms(fname,file,NULL,0) && verbose) 
+		/* f_out is set to -1 when doing final directory 
+		   permission and modification time repair */
+		if (set_perms(fname,file,NULL,0) && verbose && (f_out != -1)) 
 			rprintf(FINFO,"%s/\n",fname);
 		return;
 	}
@@ -260,7 +272,7 @@ void recv_generator(char *fname,struct file_list *flist,int i,int f_out)
 
 		if (safe_symlinks && unsafe_symlink(file->link, fname)) {
 			if (verbose) {
-				rprintf(FINFO,"ignoring unsafe symlink %s -> %s\n",
+				rprintf(FINFO,RSYNC_NAME ": ignoring unsafe symlink \"%s\" -> \"%s\"\n",
 					fname,file->link);
 			}
 			return;
@@ -269,20 +281,26 @@ void recv_generator(char *fname,struct file_list *flist,int i,int f_out)
 			l = readlink(fname,lnk,MAXPATHLEN-1);
 			if (l > 0) {
 				lnk[l] = 0;
+				/* A link already pointing to the
+				 * right place -- no further action
+				 * required. */
 				if (strcmp(lnk,file->link) == 0) {
 					set_perms(fname,file,&st,1);
 					return;
 				}
-			}
+			}  
+			/* Not a symlink, so delete whatever's
+			 * already there and put a new symlink
+			 * in place. */			   
 			delete_file(fname);
 		}
 		if (do_symlink(file->link,fname) != 0) {
-			rprintf(FERROR,"symlink %s -> %s : %s\n",
+			rprintf(FERROR,RSYNC_NAME": symlink \"%s\" -> \"%s\": %s\n",
 				fname,file->link,strerror(errno));
 		} else {
 			set_perms(fname,file,NULL,0);
 			if (verbose) {
-				rprintf(FINFO,"%s -> %s\n",
+				rprintf(FINFO,RSYNC_NAME": %s -> %s\n",
 					fname,file->link);
 			}
 		}
@@ -315,12 +333,14 @@ void recv_generator(char *fname,struct file_list *flist,int i,int f_out)
 
 	if (preserve_hard_links && check_hard_link(file)) {
 		if (verbose > 1)
-			rprintf(FINFO,"%s is a hard link\n",f_name(file));
+			rprintf(FINFO, RSYNC_NAME
+				": \"%s\" is a hard link\n",f_name(file));
 		return;
 	}
 
 	if (!S_ISREG(file->mode)) {
-		rprintf(FINFO,"skipping non-regular file %s\n",fname);
+		rprintf(FINFO, RSYNC_NAME
+			": skipping non-regular file \"%s\"\n",fname);
 		return;
 	}
 
@@ -345,7 +365,9 @@ void recv_generator(char *fname,struct file_list *flist,int i,int f_out)
 			if (!dry_run) send_sums(NULL,f_out);
 		} else {
 			if (verbose > 1)
-				rprintf(FERROR,"recv_generator failed to open %s\n",fname);
+				rprintf(FERROR, RSYNC_NAME
+					": recv_generator failed to open \"%s\": %s\n",
+					fname, strerror(errno));
 		}
 		return;
 	}
@@ -388,7 +410,7 @@ void recv_generator(char *fname,struct file_list *flist,int i,int f_out)
 	fd = do_open(fnamecmp, O_RDONLY, 0);
 
 	if (fd == -1) {
-		rprintf(FERROR,"failed to open %s, continuing : %s\n",fnamecmp,strerror(errno));
+		rprintf(FERROR,RSYNC_NAME": failed to open \"%s\", continuing : %s\n",fnamecmp,strerror(errno));
 		/* pretend the file didn't exist */
 		write_int(f_out,i);
 		send_sums(NULL,f_out);

hlink.c

@@ -1,6 +1,7 @@
 /* 
    Copyright (C) Andrew Tridgell 1996
    Copyright (C) Paul Mackerras 1996
+   Copyright (C) 2002 by Martin Pool <mbp@samba.org>
    
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -23,19 +24,22 @@ extern int dry_run;
 extern int verbose;
 
 #if SUPPORT_HARD_LINKS
-static int hlink_compare(struct file_struct *f1,struct file_struct *f2)
+static int hlink_compare(struct file_struct *f1, struct file_struct *f2)
 {
-  if (!S_ISREG(f1->mode) && !S_ISREG(f2->mode)) return 0;
-  if (!S_ISREG(f1->mode)) return -1;
-  if (!S_ISREG(f2->mode)) return 1;
+	if (!S_ISREG(f1->mode) && !S_ISREG(f2->mode))
+		return 0;
+	if (!S_ISREG(f1->mode))
+		return -1;
+	if (!S_ISREG(f2->mode))
+		return 1;
 
-  if (f1->dev != f2->dev) 
-    return (int)(f1->dev>f2->dev?1:-1);
+	if (f1->dev != f2->dev)
+		return (int) (f1->dev > f2->dev ? 1 : -1);
 
-  if (f1->inode != f2->inode) 
-    return (int)(f1->inode>f2->inode?1:-1);
+	if (f1->inode != f2->inode)
+		return (int) (f1->inode > f2->inode ? 1 : -1);
 
-  return file_compare(&f1,&f2);
+	return file_compare(&f1, &f2);
 }
 
 
@@ -47,22 +51,25 @@ void init_hard_links(struct file_list *flist)
 {
 #if SUPPORT_HARD_LINKS
 	int i;
-	if (flist->count < 2) return;
+	if (flist->count < 2)
+		return;
 
-	if (hlink_list) free(hlink_list);
-    
-	if (!(hlink_list = 
-	      (struct file_struct *)malloc(sizeof(hlink_list[0])*flist->count)))
+	if (hlink_list)
+		free(hlink_list);
+
+	if (!(hlink_list =
+	      (struct file_struct *) malloc(sizeof(hlink_list[0]) *
+					    flist->count)))
 		out_of_memory("init_hard_links");
 
 	for (i = 0; i < flist->count; i++)
-		memcpy(&hlink_list[i], flist->files[i], sizeof(hlink_list[0]));
+		memcpy(&hlink_list[i], flist->files[i],
+		       sizeof(hlink_list[0]));
 
-	qsort(hlink_list,flist->count,
-	      sizeof(hlink_list[0]),
-	      (int (*)())hlink_compare);
+	qsort(hlink_list, flist->count,
+	      sizeof(hlink_list[0]), (int (*)()) hlink_compare);
 
-	hlink_count=flist->count;
+	hlink_count = flist->count;
 #endif
 }
 
@@ -71,86 +78,102 @@ void init_hard_links(struct file_list *flist)
 int check_hard_link(struct file_struct *file)
 {
 #if SUPPORT_HARD_LINKS
-  int low=0,high=hlink_count-1;
-  int ret=0;
+	int low = 0, high = hlink_count - 1;
+	int ret = 0;
 
-  if (!hlink_list || !S_ISREG(file->mode)) return 0;
+	if (!hlink_list || !S_ISREG(file->mode))
+		return 0;
 
-  while (low != high) {
-    int mid = (low+high)/2;
-    ret = hlink_compare(&hlink_list[mid],file);
-    if (ret == 0) {
-	    low = mid;
-	    break;
-    }
-    if (ret > 0) 
-      high=mid;
-    else
-      low=mid+1;
-  }
+	while (low != high) {
+		int mid = (low + high) / 2;
+		ret = hlink_compare(&hlink_list[mid], file);
+		if (ret == 0) {
+			low = mid;
+			break;
+		}
+		if (ret > 0)
+			high = mid;
+		else
+			low = mid + 1;
+	}
 
-  if (hlink_compare(&hlink_list[low],file) != 0) return 0;
+	if (hlink_compare(&hlink_list[low], file) != 0)
+		return 0;
 
-  if (low > 0 &&
-      S_ISREG(hlink_list[low-1].mode) &&
-      file->dev == hlink_list[low-1].dev &&
-      file->inode == hlink_list[low-1].inode)
-    return 1;
+	if (low > 0 &&
+	    S_ISREG(hlink_list[low - 1].mode) &&
+	    file->dev == hlink_list[low - 1].dev &&
+	    file->inode == hlink_list[low - 1].inode)
+		return 1;
 #endif
 
-  return 0;
+	return 0;
 }
 
 
 #if SUPPORT_HARD_LINKS
 static void hard_link_one(int i)
 {
-	STRUCT_STAT st1,st2;
+	STRUCT_STAT st1, st2;
 
-	if (link_stat(f_name(&hlink_list[i-1]),&st1) != 0) return;
+	if (link_stat(f_name(&hlink_list[i - 1]), &st1) != 0)
+		return;
 
-	if (link_stat(f_name(&hlink_list[i]),&st2) != 0) {
-		if (do_link(f_name(&hlink_list[i-1]),f_name(&hlink_list[i])) != 0) {
+	if (link_stat(f_name(&hlink_list[i]), &st2) != 0) {
+		if (do_link
+		    (f_name(&hlink_list[i - 1]),
+		     f_name(&hlink_list[i])) != 0) {
 			if (verbose > 0)
-				rprintf(FINFO,"link %s => %s : %s\n",
+				rprintf(FINFO, "link %s => %s : %s\n",
 					f_name(&hlink_list[i]),
-					f_name(&hlink_list[i-1]),strerror(errno));
+					f_name(&hlink_list[i - 1]),
+					strerror(errno));
 			return;
 		}
 	} else {
-		if (st2.st_dev == st1.st_dev && st2.st_ino == st1.st_ino) return;
-		
+		if (st2.st_dev == st1.st_dev && st2.st_ino == st1.st_ino)
+			return;
+
 		if (robust_unlink(f_name(&hlink_list[i])) != 0 ||
-		    do_link(f_name(&hlink_list[i-1]),f_name(&hlink_list[i])) != 0) {
+		    do_link(f_name(&hlink_list[i - 1]),
+			    f_name(&hlink_list[i])) != 0) {
 			if (verbose > 0)
-				rprintf(FINFO,"link %s => %s : %s\n",
+				rprintf(FINFO, "link %s => %s : %s\n",
 					f_name(&hlink_list[i]),
-					f_name(&hlink_list[i-1]),strerror(errno));
+					f_name(&hlink_list[i - 1]),
+					strerror(errno));
 			return;
 		}
 	}
 	if (verbose > 0)
-		rprintf(FINFO,"%s => %s\n",
-			f_name(&hlink_list[i]),f_name(&hlink_list[i-1]));
+		rprintf(FINFO, "%s => %s\n",
+			f_name(&hlink_list[i]),
+			f_name(&hlink_list[i - 1]));
 }
 #endif
 
-/* create any hard links in the flist */
-void do_hard_links(struct file_list *flist)
+
+
+/**
+ * Create any hard links in the global hlink_list.  They were put
+ * there by running init_hard_links on the filelist.
+ **/
+void do_hard_links(void)
 {
 #if SUPPORT_HARD_LINKS
 	int i;
-  
-	if (!hlink_list) return;
 
-	for (i=1;i<hlink_count;i++) {
+	if (!hlink_list)
+		return;
+
+	for (i = 1; i < hlink_count; i++) {
 		if (S_ISREG(hlink_list[i].mode) &&
-		    S_ISREG(hlink_list[i-1].mode) &&
-		    hlink_list[i].basename && hlink_list[i-1].basename &&
-		    hlink_list[i].dev == hlink_list[i-1].dev &&
-		    hlink_list[i].inode == hlink_list[i-1].inode) {
+		    S_ISREG(hlink_list[i - 1].mode) &&
+		    hlink_list[i].basename && hlink_list[i - 1].basename &&
+		    hlink_list[i].dev == hlink_list[i - 1].dev &&
+		    hlink_list[i].inode == hlink_list[i - 1].inode) {
 			hard_link_one(i);
-		}	
+		}
 	}
 #endif
 }

io.c

@@ -49,7 +49,7 @@ int kludge_around_eof = False;
 
 static int io_error_fd = -1;
 
-static void read_loop(int fd, char *buf, int len);
+static void read_loop(int fd, char *buf, size_t len);
 
 static void check_timeout(void)
 {
@@ -86,7 +86,7 @@ void io_set_error_fd(int fd)
 static void read_error_fd(void)
 {
 	char buf[200];
-	int n;
+	size_t n;
 	int fd = io_error_fd;
 	int tag, len;
 
@@ -103,7 +103,8 @@ static void read_error_fd(void)
 
 	while (len) {
 		n = len;
-		if (n > (sizeof(buf)-1)) n = sizeof(buf)-1;
+		if (n > (sizeof(buf)-1))
+			n = sizeof(buf)-1;
 		read_loop(fd, buf, n);
 		rwrite((enum logcode)tag, buf, n);
 		len -= n;
@@ -163,7 +164,7 @@ static void die_from_readerr (int err)
  * give a better explanation.  We can tell whether the connection has
  * started by looking e.g. at whether the remote version is known yet.
  */
-static int read_timeout (int fd, char *buf, int len)
+static int read_timeout (int fd, char *buf, size_t len)
 {
 	int n, ret=0;
 
@@ -236,7 +237,7 @@ static int read_timeout (int fd, char *buf, int len)
 
 /*! Continue trying to read len bytes - don't return until len has
   been read.   */
-static void read_loop (int fd, char *buf, int len)
+static void read_loop (int fd, char *buf, size_t len)
 {
 	while (len) {
 		int n = read_timeout(fd, buf, len);
@@ -253,10 +254,10 @@ static void read_loop (int fd, char *buf, int len)
  * 
  * Never returns <= 0. 
  */
-static int read_unbuffered(int fd, char *buf, int len)
+static int read_unbuffered(int fd, char *buf, size_t len)
 {
-	static int remaining;
-	int tag, ret=0;
+	static size_t remaining;
+	int tag, ret = 0;
 	char line[1024];
 
 	if (!io_multiplexing_in || fd != multiplex_in_fd)
@@ -271,23 +272,24 @@ static int read_unbuffered(int fd, char *buf, int len)
 			continue;
 		}
 
-		read_loop (fd, line, 4);
+		read_loop(fd, line, 4);
 		tag = IVAL(line, 0);
 
 		remaining = tag & 0xFFFFFF;
 		tag = tag >> 24;
 
-		if (tag == MPLEX_BASE) continue;
+		if (tag == MPLEX_BASE)
+			continue;
 
 		tag -= MPLEX_BASE;
 
 		if (tag != FERROR && tag != FINFO) {
-			rprintf(FERROR,"unexpected tag %d\n", tag);
+			rprintf(FERROR, "unexpected tag %d\n", tag);
 			exit_cleanup(RERR_STREAMIO);
 		}
 
-		if (remaining > sizeof(line)-1) {
-			rprintf(FERROR,"multiplexing overflow %d\n\n", 
+		if (remaining > sizeof(line) - 1) {
+			rprintf(FERROR, "multiplexing overflow %d\n\n",
 				remaining);
 			exit_cleanup(RERR_STREAMIO);
 		}
@@ -295,7 +297,7 @@ static int read_unbuffered(int fd, char *buf, int len)
 		read_loop(fd, line, remaining);
 		line[remaining] = 0;
 
-		rprintf((enum logcode)tag,"%s", line);
+		rprintf((enum logcode) tag, "%s", line);
 		remaining = 0;
 	}
 
@@ -303,12 +305,13 @@ static int read_unbuffered(int fd, char *buf, int len)
 }
 
 
+
 /* do a buffered read from fd. don't return until all N bytes
    have been read. If all N can't be read then exit with an error */
-static void readfd (int fd, char *buffer, int N)
+static void readfd (int fd, char *buffer, size_t N)
 {
 	int  ret;
-	int total=0;  
+	size_t total=0;  
 	
 	while (total < N) {
 		io_flush();
@@ -356,12 +359,12 @@ int64 read_longint(int f)
 	return ret;
 }
 
-void read_buf(int f,char *buf,int len)
+void read_buf(int f,char *buf,size_t len)
 {
 	readfd(f,buf,len);
 }
 
-void read_sbuf(int f,char *buf,int len)
+void read_sbuf(int f,char *buf,size_t len)
 {
 	read_buf (f,buf,len);
 	buf[len] = 0;
@@ -375,9 +378,9 @@ unsigned char read_byte(int f)
 }
 
 /* write len bytes to fd */
-static void writefd_unbuffered(int fd,char *buf,int len)
+static void writefd_unbuffered(int fd,char *buf,size_t len)
 {
-	int total = 0;
+	size_t total = 0;
 	fd_set w_fds, r_fds;
 	int fd_count, count;
 	struct timeval tv;
@@ -424,7 +427,8 @@ static void writefd_unbuffered(int fd,char *buf,int len)
 		}
 
 		if (FD_ISSET(fd, &w_fds)) {
-			int ret, n = len-total;
+			int ret;
+			size_t n = len-total;
 			ret = write(fd,buf+total,n);
 
 			if (ret == -1 && errno == EINTR) {
@@ -483,10 +487,10 @@ void io_start_buffering(int fd)
 
 /* write an message to a multiplexed stream. If this fails then rsync
    exits */
-static void mplex_write(int fd, enum logcode code, char *buf, int len)
+static void mplex_write(int fd, enum logcode code, char *buf, size_t len)
 {
 	char buffer[4096];
-	int n = len;
+	size_t n = len;
 
 	SIVAL(buffer, 0, ((MPLEX_BASE + (int)code)<<24) + len);
 
@@ -533,7 +537,7 @@ void io_end_buffering(int fd)
 	}
 }
 
-static void writefd(int fd,char *buf,int len)
+static void writefd(int fd,char *buf,size_t len)
 {
 	stats.total_written += len;
 
@@ -587,7 +591,7 @@ void write_longint(int f, int64 x)
 	writefd(f,b,8);
 }
 
-void write_buf(int f,char *buf,int len)
+void write_buf(int f,char *buf,size_t len)
 {
 	writefd(f,buf,len);
 }
@@ -606,7 +610,7 @@ void write_byte(int f,unsigned char c)
 
 
 
-int read_line(int f, char *buf, int maxlen)
+int read_line(int f, char *buf, size_t maxlen)
 {
 	while (maxlen) {
 		buf[0] = 0;
@@ -664,7 +668,7 @@ void io_start_multiplex_in(int fd)
 }
 
 /* write an message to the multiplexed error stream */
-int io_multiplex_write(enum logcode code, char *buf, int len)
+int io_multiplex_write(enum logcode code, char *buf, size_t len)
 {
 	if (!io_multiplexing_out) return 0;
 

lib/compat.c

@@ -97,8 +97,10 @@
 
 
 #ifndef HAVE_STRLCPY
-/* like strncpy but does not 0 fill the buffer and always null 
-   terminates. bufsize is the size of the destination buffer */
+/* Like strncpy but does not 0 fill the buffer and always null 
+ * terminates. bufsize is the size of the destination buffer.
+ *
+ * Returns the index of the terminating byte. */
  size_t strlcpy(char *d, const char *s, size_t bufsize)
 {
 	size_t len = strlen(s);

loadparm.c

@@ -386,9 +386,16 @@ static void init_service(service *pservice)
 
 
 /**
- * Assign a copy of @p v to @p *s, freeing any existing values and
- * handling NULL strings.  @p *v must be initialized when this is
- * called, either to NULL or a malloc'd string.
+ * Assign a copy of @p v to @p *s.  Handles NULL strings.  @p *v must
+ * be initialized when this is called, either to NULL or a malloc'd
+ * string.
+ *
+ * @fixme There is a small leak here in that sometimes the existing
+ * value will be dynamically allocated, and the old copy is lost.
+ * However, we can't always deallocate the old value, because in the
+ * case of sDefault, it points to a static string.  It would be nice
+ * to have either all-strdup'd values, or to never need to free
+ * memory.
  **/
 static void string_set(char **s, const char *v)
 {
@@ -396,8 +403,6 @@ static void string_set(char **s, const char *v)
 		*s = NULL;
 		return;
 	}
-	if (*s)
-		free(*s);
 	*s = strdup(v);
 	if (!*s)
 		exit_cleanup(RERR_MALLOC);

log.c

@@ -286,13 +286,14 @@ void rprintf(enum logcode code, const char *format, ...)
 	int len;
 
 	va_start(ap, format);
+	/* Note: might return -1 */
 	len = vsnprintf(buf, sizeof(buf), format, ap);
 	va_end(ap);
 
 	/* Deal with buffer overruns.  Instead of panicking, just
 	 * truncate the resulting string.  Note that some vsnprintf()s
 	 * return -1 on truncation, e.g., glibc 2.0.6 and earlier. */
-	if (len > sizeof(buf)-1  ||  len < 0) {
+	if ((size_t) len > sizeof(buf)-1  ||  len < 0) {
 		const char ellipsis[] = "[...]";
 
 		/* Reset length, and zero-terminate the end of our buffer */
@@ -331,18 +332,21 @@ void rsyserr(enum logcode code, int errcode, const char *format, ...)
 {
 	va_list ap;  
 	char buf[1024];
-	int len, sys_len;
+	int len;
+	size_t sys_len;
         char *sysmsg;
 
 	va_start(ap, format);
+	/* Note: might return <0 */
 	len = vsnprintf(buf, sizeof(buf), format, ap);
 	va_end(ap);
 
-	if (len > sizeof(buf)-1) exit_cleanup(RERR_MESSAGEIO);
+	if ((size_t) len > sizeof(buf)-1)
+		exit_cleanup(RERR_MESSAGEIO);
 
         sysmsg = strerror(errcode);
         sys_len = strlen(sysmsg);
-        if (len + 3 + sys_len > sizeof(buf) - 1)
+        if ((size_t) len + 3 + sys_len > sizeof(buf) - 1)
                 exit_cleanup(RERR_MESSAGEIO);
 
         strcpy(buf + len, ": ");
@@ -399,7 +403,7 @@ static void log_formatted(enum logcode code,
 	char buf[1024];
 	char buf2[1024];
 	char *p, *s, *n;
-	int l;
+	size_t l;
 	extern struct stats stats;		
 	extern int am_sender;
 	extern int am_daemon;
@@ -466,7 +470,7 @@ static void log_formatted(enum logcode code,
 
 		l = strlen(n);
 
-		if ((l-1) + ((int)(s - &buf[0])) > sizeof(buf)) {
+		if (l + ((int)(s - &buf[0])) >= sizeof(buf)) {
 			rprintf(FERROR,"buffer overflow expanding %%%c - exiting\n",
 				p[0]);
 			exit_cleanup(RERR_MESSAGEIO);

main.c

@@ -2,7 +2,7 @@
    
    Copyright (C) 1996-2001 by Andrew Tridgell <tridge@samba.org>
    Copyright (C) Paul Mackerras 1996
-   Copyright (C) 2001 by Martin Pool <mbp@samba.org>
+   Copyright (C) 2001, 2002 by Martin Pool <mbp@samba.org>
    
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -27,6 +27,7 @@ struct stats stats;
 
 extern int verbose;
 
+static void show_malloc_stats(void);
 
 /****************************************************************************
 wait for a process to exit, calling io_flush while waiting
@@ -56,6 +57,11 @@ static void report(int f)
 	extern int remote_version;
 	int send_stats;
 
+	if (do_stats) {
+		/* These come out from every process */
+		show_malloc_stats();
+	}
+
 	if (am_daemon) {
 		log_exit(0, __FILE__, __LINE__);
 		if (f == -1 || !am_sender) return;
@@ -126,6 +132,38 @@ static void report(int f)
 }
 
 
+/**
+ * If our C library can get malloc statistics, then show them to FINFO
+ **/
+static void show_malloc_stats(void)
+{
+#ifdef HAVE_MALLINFO
+	struct mallinfo mi;
+	extern int am_server;
+	extern int am_sender;
+	extern int am_daemon;
+
+	mi = mallinfo();
+
+	rprintf(FINFO, RSYNC_NAME "[%d] (%s%s%s) heap statistics:\n",
+		getpid(),
+		am_server ? "server " : "",
+		am_daemon ? "daemon " : "",
+		am_sender ? "sender" : "receiver");
+	rprintf(FINFO, "  arena:     %10d   (bytes from sbrk)\n", mi.arena);
+	rprintf(FINFO, "  ordblks:   %10d   (chunks not in use)\n", mi.ordblks);
+	rprintf(FINFO, "  smblks:    %10d\n", mi.smblks);
+	rprintf(FINFO, "  hblks:     %10d   (chunks from mmap)\n", mi.hblks);
+	rprintf(FINFO, "  hblkhd:    %10d   (bytes from mmap)\n", mi.hblkhd);
+	rprintf(FINFO, "  usmblks:   %10d\n", mi.usmblks);
+	rprintf(FINFO, "  fsmblks:   %10d\n", mi.fsmblks);
+	rprintf(FINFO, "  uordblks:  %10d   (bytes used)\n", mi.uordblks);
+	rprintf(FINFO, "  fordblks:  %10d   (bytes free)\n", mi.fordblks);
+	rprintf(FINFO, "  keepcost:  %10d   (bytes in releasable chunk)\n", mi.keepcost);
+#endif /* HAVE_MALLINFO */
+}
+
+
 /* Start the remote shell.   cmd may be NULL to use the default. */
 static pid_t do_cmd(char *cmd,char *machine,char *user,char *path,int *f_in,int *f_out)
 {
@@ -765,8 +803,10 @@ int main(int argc,char *argv[])
 	extern int write_batch;  /*  dw */
 	extern char *batch_ext;   /*  dw */
 	int orig_argc;  /* dw */
+	char **orig_argv;
 
 	orig_argc = argc;   /* dw */
+	orig_argv = argv;
 
 	signal(SIGUSR1, sigusr1_handler);
 	signal(SIGUSR2, sigusr2_handler);
@@ -806,7 +846,7 @@ int main(int argc,char *argv[])
 
 	if (write_batch) { /* dw */
 	    create_batch_file_ext();
-	    write_batch_argvs_file(orig_argc, argc, argv);
+	    write_batch_argvs_file(orig_argc, orig_argv);
 	}
 
 	if (read_batch) { /* dw */

options.c

@@ -1,7 +1,7 @@
 /*  -*- c-file-style: "linux" -*-
     
     Copyright (C) 1998-2001 by Andrew Tridgell <tridge@samba.org>
-    Copyright (C) 2000-2001 by Martin Pool <mbp@samba.org>
+    Copyright (C) 2000, 2001, 2002 by Martin Pool <mbp@samba.org>
    
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -74,11 +74,21 @@ int modify_window=0;
 #endif
 int blocking_io=0;
 
-/** Global options set from command line. **/
-struct global_opts global_opts;
+/** Network address family. **/
+#ifdef INET6
+int default_af_hint = 0;	/* Any protocol */
+#else
+int default_af_hint = AF_INET;	/* Must use IPv4 */
+#endif
 
-int read_batch=0;  /* dw */
-int write_batch=0; /* dw */
+/** Do not go into the background when run as --daemon.  Good
+ * for debugging and required for running as a service on W32,
+ * or under Unix process-monitors. **/
+int no_detach = 0;
+
+
+int read_batch=0;
+int write_batch=0;
 
 char *backup_suffix = BACKUP_SUFFIX;
 char *tmpdir = NULL;
@@ -106,12 +116,13 @@ static int modify_window_set;
 char *bind_address;
 
 
-static void print_rsync_version(int f)
+static void print_rsync_version(enum logcode f)
 {
         char const *got_socketpair = "no ";
         char const *hardlinks = "no ";
         char const *links = "no ";
 	char const *ipv6 = "no ";
+	STRUCT_STAT *dumstat;
 
 #ifdef HAVE_SOCKETPAIR
         got_socketpair = "";
@@ -132,13 +143,20 @@ static void print_rsync_version(int f)
         rprintf(f, "%s  version %s  protocol version %d\n",
                 RSYNC_NAME, RSYNC_VERSION, PROTOCOL_VERSION);
         rprintf(f,
-                "Copyright (C) 1996-2001 by Andrew Tridgell and others\n");
+                "Copyright (C) 1996-2002 by Andrew Tridgell and others\n");
 	rprintf(f, "<http://rsync.samba.org/>\n");
         rprintf(f, "Capabilities: %d-bit files, %ssocketpairs, "
-                "%shard links, %ssymlinks, batchfiles, %sIPv6\n\n",
+                "%shard links, %ssymlinks, batchfiles, %sIPv6,\n",
                 (int) (sizeof(OFF_T) * 8),
                 got_socketpair, hardlinks, links, ipv6);
 
+	/* Note that this field may not have type ino_t.  It depends
+	 * on the complicated interaction between largefile feature
+	 * macros. */
+	rprintf(f, "              %d-bit system inums, %d-bit internal inums\n",
+		(int) (sizeof(dumstat->st_ino) * 8),
+		(int) (sizeof(INO64_T) * 8));
+
 #ifdef NO_INT64
         rprintf(f, "WARNING: no 64-bit integers on this platform!\n");
 #endif
@@ -171,8 +189,8 @@ void usage(enum logcode F)
   rprintf(F,"     --backup-dir            make backups into this directory\n");
   rprintf(F,"     --suffix=SUFFIX         override backup suffix\n");  
   rprintf(F," -u, --update                update only (don't overwrite newer files)\n");
-  rprintf(F," -l, --links                 preserve soft links\n");
-  rprintf(F," -L, --copy-links            treat soft links like regular files\n");
+  rprintf(F," -l, --links                 copy symlinks as symlinks\n");
+  rprintf(F," -L, --copy-links            copy the referent of symlinks\n");
   rprintf(F,"     --copy-unsafe-links     copy links outside the source tree\n");
   rprintf(F,"     --safe-links            ignore links outside the destination tree\n");
   rprintf(F," -H, --hard-links            preserve hard links\n");
@@ -212,7 +230,8 @@ void usage(enum logcode F)
   rprintf(F,"     --include-from=FILE     don't exclude patterns listed in FILE\n");
   rprintf(F,"     --version               print version number\n");  
   rprintf(F,"     --daemon                run as a rsync daemon\n");  
-  rprintf(F,"     --address               bind to the specified address\n");  
+  rprintf(F,"     --no-detach             do not detach from the parent\n");  
+  rprintf(F,"     --address=ADDRESS       bind to the specified address\n");  
   rprintf(F,"     --config=FILE           specify alternate rsyncd.conf file\n");  
   rprintf(F,"     --port=PORT             specify alternate rsyncd port number\n");
   rprintf(F,"     --blocking-io           use blocking IO for the remote shell\n");  
@@ -221,8 +240,8 @@ void usage(enum logcode F)
   rprintf(F,"     --log-format=FORMAT     log file transfers using specified format\n");  
   rprintf(F,"     --password-file=FILE    get password from FILE\n");
   rprintf(F,"     --bwlimit=KBPS          limit I/O bandwidth, KBytes per second\n");
-  rprintf(F," -f  --read-batch=EXT        read batch file\n");
-  rprintf(F," -F  --write-batch           write batch file\n");
+  rprintf(F,"     --read-batch=EXT        read batch file\n");
+  rprintf(F,"     --write-batch           write batch file\n");
   rprintf(F," -h, --help                  show this help screen\n");
 #ifdef INET6
   rprintf(F," -4                          prefer IPv4\n");
@@ -243,7 +262,7 @@ enum {OPT_VERSION = 1000, OPT_SUFFIX, OPT_SENDER, OPT_SERVER, OPT_EXCLUDE,
       OPT_LOG_FORMAT, OPT_PASSWORD_FILE, OPT_SIZE_ONLY, OPT_ADDRESS,
       OPT_DELETE_AFTER, OPT_EXISTING, OPT_MAX_DELETE, OPT_BACKUP_DIR, 
       OPT_IGNORE_ERRORS, OPT_BWLIMIT, OPT_BLOCKING_IO,
-      OPT_MODIFY_WINDOW};
+      OPT_MODIFY_WINDOW, OPT_READ_BATCH, OPT_WRITE_BATCH};
 
 static struct poptOption long_options[] = {
   /* longName, shortName, argInfo, argPtr, value, descrip, argDesc */
@@ -274,7 +293,7 @@ static struct poptOption long_options[] = {
   {"update",          'u', POPT_ARG_NONE,   &update_only},
   {"links",           'l', POPT_ARG_NONE,   &preserve_links},
   {"copy-links",      'L', POPT_ARG_NONE,   &copy_links},
-  {"whole",           'W', POPT_ARG_NONE,   &whole_file},
+  {"whole-file",      'W', POPT_ARG_NONE,   &whole_file},
   {"copy-unsafe-links", 0, POPT_ARG_NONE,   &copy_unsafe_links},
   {"perms",           'p', POPT_ARG_NONE,   &preserve_perms},
   {"owner",           'o', POPT_ARG_NONE,   &preserve_uid},
@@ -298,6 +317,7 @@ static struct poptOption long_options[] = {
   /* TODO: Should this take an optional int giving the compression level? */
   {"compress",        'z', POPT_ARG_NONE,   &do_compression},
   {"daemon",           0,  POPT_ARG_NONE,   &am_daemon},
+  {"no-detach",        0,  POPT_ARG_NONE,   &no_detach},
   {"stats",            0,  POPT_ARG_NONE,   &do_stats},
   {"progress",         0,  POPT_ARG_NONE,   &do_progress},
   {"partial",          0,  POPT_ARG_NONE,   &keep_partial},
@@ -311,11 +331,11 @@ static struct poptOption long_options[] = {
   {"address",          0,  POPT_ARG_STRING, &bind_address, 0},
   {"backup-dir",       0,  POPT_ARG_STRING, &backup_dir},
   {"hard-links",      'H', POPT_ARG_NONE,   &preserve_hard_links},
-  {"read-batch",      'f', POPT_ARG_STRING, &batch_ext, 'f'},
-  {"write-batch",     'F', POPT_ARG_NONE,   &write_batch, 0},
+  {"read-batch",       0,  POPT_ARG_STRING, &batch_ext, OPT_READ_BATCH},
+  {"write-batch",      0,  POPT_ARG_NONE,   &write_batch},
 #ifdef INET6
-  {0,		      '4', POPT_ARG_VAL,    &global_opts.af_hint,   AF_INET },
-  {0,		      '6', POPT_ARG_VAL,    &global_opts.af_hint,   AF_INET6 },
+  {0,		      '4', POPT_ARG_VAL,    &default_af_hint,   AF_INET },
+  {0,		      '6', POPT_ARG_VAL,    &default_af_hint,   AF_INET6 },
 #endif
   {0,0,0,0}
 };
@@ -491,9 +511,8 @@ int parse_arguments(int *argc, const char ***argv, int frommain)
 			keep_partial = 1;
 			break;
 
-
-		case 'f':
-			/* The filename is stored for us by popt */
+		case OPT_READ_BATCH:
+			/* The filename is stored in batch_ext for us by popt */
 			read_batch = 1;
 			break;
 
@@ -530,7 +549,8 @@ void server_options(char **args,int *argc)
 	static char mdelete[30];
 	static char mwindow[30];
 	static char bw[50];
-	static char fext[20]; /* dw */
+	static char fext[20];
+	static char wbatch[14];
 
 	int i, x;
 
@@ -585,8 +605,6 @@ void server_options(char **args,int *argc)
 		argstr[x++] = 'S';
 	if (do_compression)
 		argstr[x++] = 'z';
-	if (write_batch)
-	    argstr[x++] = 'F'; /* dw */
 
 	/* this is a complete hack - blame Rusty 
 
@@ -609,8 +627,13 @@ void server_options(char **args,int *argc)
 		args[ac++] = mdelete;
 	}    
 	
+	if (write_batch) {
+		snprintf(wbatch,sizeof(wbatch),"--write-batch");
+		args[ac++] = wbatch;
+	}
+
 	if (batch_ext != NULL) {
-		sprintf(fext,"-f%s",batch_ext);
+		snprintf(fext,sizeof(fext),"--read-batch=%s",batch_ext);
 		args[ac++] = fext;
 	}
 

packaging/lsb/rsync.spec

@@ -1,10 +1,10 @@
 Summary: Program for efficient remote updates of files.
 Name: rsync
-Version: 2.5.0
+Version: 2.5.1
 Release: 1
 Copyright: GPL
 Group: Applications/Networking
-Source:	ftp://samba.anu.edu.au/pub/rsync/rsync-2.5.0.tar.gz
+Source:	ftp://samba.anu.edu.au/pub/rsync/rsync-2.5.1.tar.gz
 URL: http://samba.anu.edu.au/rsync/
 Packager: Andrew Tridgell <tridge@samba.anu.edu.au>
 BuildRoot: /tmp/rsync

packaging/redhat/5.0/rsync.spec

@@ -1,10 +1,10 @@
 Summary: Program for efficient remote updates of files.
 Name: rsync
-Version: 2.5.0
+Version: 2.5.1
 Release: 1
 Copyright: GPL
 Group: Applications/Networking
-Source:	ftp://samba.anu.edu.au/pub/rsync/rsync-2.5.0.tar.gz
+Source:	ftp://samba.anu.edu.au/pub/rsync/rsync-2.5.1.tar.gz
 URL: http://samba.anu.edu.au/rsync/
 Packager: Andrew Tridgell <tridge@samba.anu.edu.au>
 BuildRoot: /tmp/rsync

packaging/redhat/7.1/rsync.spec

@@ -1,10 +1,10 @@
 Summary: Program for efficient remote updates of files.
 Name: rsync
-Version: 2.5.0
+Version: 2.5.1
 Release: 1
 Copyright: GPL
 Group: Applications/Networking
-Source:	ftp://samba.anu.edu.au/pub/rsync/rsync-2.5.0.tar.gz
+Source:	ftp://samba.anu.edu.au/pub/rsync/rsync-2.5.1.tar.gz
 URL: http://samba.anu.edu.au/rsync/
 Packager: Andrew Tridgell <tridge@samba.anu.edu.au>
 BuildRoot: /tmp/rsync

receiver.c

@@ -38,8 +38,8 @@ extern int make_backups;
 extern char *backup_suffix;
 
 static struct delete_list {
-	dev_t dev;
-	INO_T inode;
+	DEV64_T dev;
+	INO64_T inode;
 } *delete_list;
 static int dlist_len, dlist_alloc_len;
 
@@ -206,7 +206,8 @@ static int get_tmpname(char *fnametmp, char *fname)
 static int receive_data(int f_in,struct map_struct *buf,int fd,char *fname,
 			OFF_T total_size)
 {
-	int i,n,remainder,len,count;
+	int i;
+	unsigned int n,remainder,len,count;
 	OFF_T offset = 0;
 	OFF_T offset2;
 	char *data;
@@ -488,7 +489,7 @@ int recv_files(int f_in,struct file_list *flist,char *local_name,int f_gen)
 	}
 
 	if (preserve_hard_links)
-		do_hard_links(flist);
+		do_hard_links();
 
 	/* now we need to fix any directory permissions that were 
 	   modified during the transfer */

rsync.c

@@ -226,7 +226,6 @@ int set_perms(char *fname,struct file_struct *file,STRUCT_STAT *st,
 
 void sig_int(void)
 {
-	rprintf(FINFO,"\nrsync.c:sig_int() called.\n");
 	exit_cleanup(RERR_SIGNAL);
 }
 

rsync.h

@@ -1,7 +1,7 @@
 /* 
    Copyright (C) by Andrew Tridgell 1996, 2000
    Copyright (C) Paul Mackerras 1996
-   Copyright (C) 2001 by Martin Pool <mbp@samba.org>
+   Copyright (C) 2001, 2002 by Martin Pool <mbp@samba.org>
    
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -49,7 +49,7 @@
 #define SAME_TIME (1<<7)
 
 /* update this if you make incompatible changes */
-#define PROTOCOL_VERSION 25
+#define PROTOCOL_VERSION 26
 
 /* We refuse to interoperate with versions that are not in this range.
  * Note that we assume we'll work with later versions: the onus is on
@@ -185,6 +185,10 @@ enum logcode {FNONE=0, FERROR=1, FINFO=2, FLOG=3 };
 #include <glob.h>
 #endif
 
+#ifdef HAVE_MALLOC_H
+#  include <malloc.h>
+#endif
+
 /* these are needed for the uid/gid mapping code */
 #include <pwd.h>
 #include <grp.h>
@@ -263,17 +267,44 @@ enum logcode {FNONE=0, FERROR=1, FINFO=2, FLOG=3 };
 #elif HAVE_LONGLONG
 #define int64 long long
 #else
+/* As long as it gets... */
 #define int64 off_t
 #define NO_INT64
 #endif
 
-#if HAVE_SHORT_INO_T
-#  define INO_T uint32
-#elif HAVE_INO_T
-#  define INO_T ino_t
-#else
-#  define INO_T unsigned
-#endif
+/* Starting from protocol version 26, we always use 64-bit
+ * ino_t and dev_t internally, even if this platform does not
+ * allow files to have 64-bit inums.  That's because the
+ * receiver needs to find duplicate (dev,ino) tuples to detect
+ * hardlinks, and it might have files coming from a platform
+ * that has 64-bit inums.
+ *
+ * The only exception is if we're on a platform with no 64-bit type at
+ * all.
+ *
+ * Because we use read_longint() to get these off the wire, if you
+ * transfer devices or hardlinks with dev or inum > 2**32 to a machine
+ * with no 64-bit types then you will get an overflow error.  Probably
+ * not many people have that combination of machines, and you can
+ * avoid it by not preserving hardlinks or not transferring device
+ * nodes.  It's not clear that any other behaviour is better.
+ *
+ * Note that if you transfer devices from a 64-bit-devt machine (say,
+ * Solaris) to a 32-bit-devt machine (say, Linux-2.2/x86) then the
+ * device numbers will be truncated.  But it's a kind of silly thing
+ * to do anyhow.
+ *
+ * FIXME: In future, we should probable split the device number into
+ * major/minor, and transfer the two parts as 32-bit ints.  That gives
+ * you somewhat more of a chance that they'll come from a big machine
+ * to a little one in a useful way.
+ *
+ * FIXME: Really we need an unsigned type, and we perhaps ought to
+ * cope with platforms on which this is an unsigned int or even a
+ * struct.  Later.
+ */ 
+#define INO64_T int64
+#define DEV64_T int64
 
 #ifndef MIN
 #define MIN(a,b) ((a)<(b)?(a):(b))
@@ -304,9 +335,13 @@ struct file_struct {
 	time_t modtime;
 	OFF_T length;
 	mode_t mode;
-	INO_T inode;
-	dev_t dev;
-	dev_t rdev;
+
+	INO64_T inode;
+	/** Device this file lives upon */
+	DEV64_T dev;
+
+	/** If this is a device node, the device number. */
+	DEV64_T rdev;
 	uid_t uid;
 	gid_t gid;
 	char *basename;
@@ -342,11 +377,11 @@ struct sum_buf {
 };
 
 struct sum_struct {
-  OFF_T flength;		/* total file length */
-  int count;			/* how many chunks */
-  int remainder;		/* flength % block_length */
-  int n;			/* block_length */
-  struct sum_buf *sums;		/* points to info for each chunk */
+	OFF_T flength;		/* total file length */
+	size_t count;		/* how many chunks */
+	size_t remainder;	/* flength % block_length */
+	size_t n;		/* block_length */
+	struct sum_buf *sums;	/* points to info for each chunk */
 };
 
 struct map_struct {
@@ -390,6 +425,7 @@ static inline int flist_up(struct file_list *flist, int i)
 #include "proto.h"
 #include "lib/mdfour.h"
 #include "lib/permstring.h"
+#include "lib/addrinfo.h"
 
 /* We have replacement versions of these if they're missing. */
 #ifndef HAVE_ASPRINTF
@@ -563,12 +599,6 @@ size_t strlcat(char *d, const char *s, size_t bufsize);
 
 extern int verbose;
 
-extern struct global_opts {
-	/** Network address family. **/
-	int af_hint;
-} global_opts;
-
-
 #ifndef HAVE_INET_NTOP
 const char *                 
 inet_ntop(int af, const void *src, char *dst, size_t size);

rsync.yo

@@ -1,5 +1,5 @@
 mailto(rsync-bugs@samba.org)
-manpage(rsync)(1)(29 May 2001)()()
+manpage(rsync)(1)(14 Dec 2001)()()
 manpagename(rsync)(faster, flexible replacement for rcp)
 manpagesynopsis()
 
@@ -19,7 +19,7 @@ manpagedescription()
 
 rsync is a program that behaves in much the same way that rcp does,
 but has many more options and uses the rsync remote-update protocol to
-greatly speedup file transfers when the destination file already
+greatly speed up file transfers when the destination file already
 exists.
 
 The rsync remote-update protocol allows rsync to transfer just the
@@ -81,7 +81,7 @@ Once installed you can use rsync to any machine that you can use rsh
 to.  rsync uses rsh for its communications, unless both the source and
 destination are local.
 
-You can also specify an alternative to rsh, by either using the -e
+You can also specify an alternative to rsh, either by using the -e
 command line option, or by setting the RSYNC_RSH environment variable.
 
 One common substitute is to use ssh, which offers a high degree of
@@ -139,10 +139,10 @@ It is also possible to use rsync without using rsh or ssh as the
 transport. In this case you will connect to a remote rsync server
 running on TCP port 873. 
 
-You may establish the connetcion via a web proxy by setting the
+You may establish the connection via a web proxy by setting the
 environment variable RSYNC_PROXY to a hostname:port pair pointing to
-your web proxy. Note that your web proxy must allow proxying to port
-873, this must be configured in your proxy servers ruleset.
+your web proxy.  Note that your web proxy's configuration must allow
+proxying to port 873.
 
 Using rsync in this way is the same as using it with rsh or ssh except
 that:
@@ -226,8 +226,8 @@ verb(
      --backup-dir            make backups into this directory
      --suffix=SUFFIX         override backup suffix
  -u, --update                update only (don't overwrite newer files)
- -l, --links                 preserve soft links
- -L, --copy-links            treat soft links like regular files
+ -l, --links                 copy symlinks as symlinks
+ -L, --copy-links            copy the referent of symlinks
      --copy-unsafe-links     copy links outside the source tree
      --safe-links            ignore links outside the destination tree
  -H, --hard-links            preserve hard links
@@ -267,7 +267,8 @@ verb(
      --include-from=FILE     don't exclude patterns listed in FILE
      --version               print version number
      --daemon                run as a rsync daemon
-     --address               bind to the specified address
+     --no-detach             do not detach from the parent
+     --address=ADDRESS       bind to the specified address
      --config=FILE           specify alternate rsyncd.conf file
      --port=PORT             specify alternate rsyncd port number
      --blocking-io           use blocking IO for the remote shell
@@ -276,8 +277,8 @@ verb(
      --log-format=FORMAT     log file transfers using specified format
      --password-file=FILE    get password from FILE
      --bwlimit=KBPS          limit I/O bandwidth, KBytes per second
- -f, --read-batch=FILE       read batch file
- -F, --write-batch           write batch file
+     --read-batch=FILE       read batch file
+     --write-batch           write batch file
  -h, --help                  show this help screen
 
 
@@ -377,17 +378,16 @@ dit(bf(-u, --update)) This forces rsync to skip any files for which the
 destination file already exists and has a date later than the source
 file.
 
-dit(bf(-l, --links)) This tells rsync to recreate symbolic links on the
-remote system  to  be the same as the local system. Without this
-option, all symbolic links are skipped.
+dit(bf(-l, --links)) When symlinks are encountered, recreate the
+symlink on the destination.
 
-dit(bf(-L, --copy-links)) This tells rsync to treat symbolic links just
-like ordinary files.
+dit(bf(-L, --copy-links)) When symlinks are encountered, the file that
+they point to is copied, rather than the symlink.
 
-dit(bf(--copy-unsafe-links)) This tells rsync to treat symbolic links that
-point outside the source tree like ordinary files.  Absolute symlinks are
-also treated like ordinary files, and so are any symlinks in the source
-path itself when --relative is used.
+dit(bf(--copy-unsafe-links)) This tells rsync to copy the referent of
+symbolic links that point outside the source tree.  Absolute symlinks
+are also treated like ordinary files, and so are any symlinks in the
+source path itself when --relative is used.
 
 dit(bf(--safe-links)) This tells rsync to ignore any symbolic links
 which point outside the destination tree. All absolute symlinks are
@@ -413,16 +413,15 @@ the source and target are on the local machine.
 dit(bf(-p, --perms)) This option causes rsync to update the remote
 permissions to be the same as the local permissions.
 
-dit(bf(-o, --owner)) This option causes rsync to update the  remote  owner
-of the  file to be the same as the local owner. This is only available
-to the super-user.  Note that if the source system is a daemon using chroot,
-the --numeric-ids option is implied because the source system cannot get
-access to the usernames.
+dit(bf(-o, --owner)) This option causes rsync to set the owner of the
+destination file to be the same as the source file.  On most systems,
+only the super-user can set file ownership.  
 
-dit(bf(-g, --group)) This option causes rsync to update the  remote  group
-of the file to be the same as the local group.  If the receving system is
-not running as the super-user, only groups that the receiver is a member of
-will be preserved (by group name, not group id number).
+dit(bf(-g, --group)) This option causes rsync to set the group of the
+destination file to be the same as the source file.  If the receiving
+program is not running as the super-user, only groups that the
+receiver is a member of will be preserved (by group name, not group id
+number).
 
 dit(bf(-D, --devices)) This option causes rsync to transfer character and
 block device information to the remote system to recreate these
@@ -493,7 +492,7 @@ contains a directory of the same name.
 Since this option was added, deletions were reordered to be done depth-first
 so it is hardly ever needed anymore except in very obscure cases.
 
-dit(bf(-B , --block_size=BLOCKSIZE)) This controls the block size used in
+dit(bf(-B , --block-size=BLOCKSIZE)) This controls the block size used in
 the rsync algorithm. See the technical report for details.
 
 dit(bf(-e, --rsh=COMMAND)) This option allows you to choose an alternative
@@ -550,8 +549,9 @@ quote(RCS SCCS CVS CVS.adm RCSLOG cvslog.* tags TAGS .make.state
 then files listed in a $HOME/.cvsignore are added to the list and any
 files listed in the CVSIGNORE environment variable (space delimited).
 
-Finally in each directory any files listed in the .cvsignore file in
-that directory are added to the list.
+Finally, any file is ignored if it is in the same directory as a
+.cvsignore file and matches one of the patterns listed therein.  See
+the bf(cvs(1)) manual for more information.
 
 dit(bf(--csum-length=LENGTH)) By default the primary checksum used in
 rsync is a very strong 16 byte MD4 checksum. In most cases you will
@@ -608,21 +608,33 @@ what ownership to give files. The special uid 0 and the special group
 0 are never mapped via user/group names even if the --numeric-ids
 option is not specified.
 
-If the source system is a daemon using chroot, or if a user or group name
-does not exist on the destination system, then the numeric id from the
-source system is used instead.
+If the source system is a daemon using chroot, or if a user or group
+name does not exist on the destination system, then the numeric id
+from the source system is used instead.
 
 dit(bf(--timeout=TIMEOUT)) This option allows you to set a maximum IO
 timeout in seconds. If no data is transferred for the specified time
 then rsync will exit. The default is 0, which means no timeout.
 
-dit(bf(--daemon)) This tells rsync that it is to run as a rsync
-daemon. If standard input is a socket then rsync will assume that it
-is being run via inetd, otherwise it will detach from the current
-terminal and become a background daemon. The daemon will read the
-config file (/etc/rsyncd.conf) on each connect made by a client and
-respond to requests accordingly. See the rsyncd.conf(5) man page for more
-details. 
+dit(bf(--daemon)) This tells rsync that it is to run as a daemon.  The
+daemon may be accessed using the bf(host::module) or
+bf(rsync://host/module/) syntax.
+
+If standard input is a socket then rsync will assume that it is being
+run via inetd, otherwise it will detach from the current terminal and
+become a background daemon.  The daemon will read the config file
+(/etc/rsyncd.conf) on each connect made by a client and respond to
+requests accordingly.  See the rsyncd.conf(5) man page for more
+details.
+
+dit(bf(--no-detach)) When running as a daemon, this option instructs
+rsync to not detach itself and become a background process.  This
+option is required when running as a service on Cygwin, and may also
+be useful when rsync is supervised by a program such as
+bf(daemontools) or AIX's bf(System Resource Controller).
+bf(--no-detach) is also recommended when rsync is run under a
+debugger.  This option has no effect if rsync is run from inetd or
+sshd.
 
 dit(bf(--address)) By default rsync will bind to the wildcard address
 when run as a daemon with the --daemon option or when connecting to a
@@ -696,7 +708,7 @@ manpagesection(EXCLUDE PATTERNS)
 The exclude and include patterns specified to rsync allow for flexible
 selection of which files to transfer and which files to skip.
 
-rsync builds a ordered list of include/exclude options as specified on
+rsync builds an ordered list of include/exclude options as specified on
 the command line. When a filename is encountered, rsync checks the
 name against each exclude/include pattern in turn. The first matching
 pattern is acted on. If it is an exclude pattern, then that file is
@@ -749,7 +761,7 @@ itemize(
   part of an include option. The "- " part is discarded before matching.
 
   it() if the pattern is a single exclamation mark ! then the current
-  exclude list is reset, removing all previous exclude patterns.
+  include/exclude list is reset, removing all previously defined patterns.
 )
 
 The +/- rules are most useful in exclude lists, allowing you to have a
@@ -781,12 +793,15 @@ itemize(
 
 manpagesection(BATCH MODE)
 
+bf(Note:) Batch mode should be considered experimental in this version
+of rsync.  The interface or behaviour may change before it stabilizes.
+
 The following call generates 4 files that encapsulate the information
 for synchronizing the contents of bf(target_dir) with the updates found in
 bf(src_dir)
 
 quote(
-$ rsync -F [other rsync options here] \nl()
+$ rsync --write-batch [other rsync options here] \nl()
            /somewhere/src_dir /somewhere/target_dir
 )
 
@@ -802,6 +817,29 @@ it() bf(rsync_delta.<timestamp>) data blocks for file update & change
 See bf(http://www.ils.unc.edu/i2dsi/unc_rsync+.html) for papers and technical
 reports.
 
+manpagesection(SYMBOLIC LINKS)
+
+Three basic behaviours are possible when rsync encounters a symbolic
+link in the source directory.
+
+By default, symbolic links are not transferred at all.  A message
+"skipping non-regular" file is emitted for any symlinks that exist.
+
+If bf(--links) is specified, then symlinks are recreated with the same
+target on the destination.  Note that bf(--archive) implies
+bf(--links).
+
+If bf(--copy-links) is specified, then symlinks are "collapsed" by
+copying their referent, rather than the symlink.
+
+rsync also distinguishes "safe" and "unsafe" symbolic links.  An
+example where this might be used is a web site mirror that wishes
+ensure the rsync module they copy does not include symbolic links to
+bf(/etc/passwd) in the public section of the site.  Using
+bf(--copy-unsafe-links) will cause any links to be copied as the file
+they point to on the destination.  Using bf(--safe-links) will cause
+unsafe links to be ommitted altogether.
+
 manpagesection(DIAGNOSTICS)
 
 rsync occasionally produces error messages that may seem a little
@@ -927,16 +965,22 @@ Jean-loup Gailly and Mark Adler.
 manpagesection(THANKS)
 
 Thanks to Richard Brent, Brendan Mackay, Bill Waite, Stephen Rothwell
-and David Bell for helpful suggestions and testing of rsync. I've
-probably missed some people, my apologies if I have.
+and David Bell for helpful suggestions, patches and testing of rsync.
+I've probably missed some people, my apologies if I have.
+
+Especial thanks also to: David Dykstra, Jos Backus, Sebastian Krahmer.
 
 
 manpageauthor()
 
-rsync was written by Andrew Tridgell and Paul Mackerras.  They may be
-contacted via email at tridge@samba.org and
-Paul.Mackerras@cs.anu.edu.au
+rsync was written by Andrew Tridgell <tridge@samba.org> and Paul
+Mackerras.
 
-rsync is now also maintained by Martin Pool <mbp@samba.org>
+rsync is now maintained by Martin Pool <mbp@samba.org>.  
 
+Mailing lists for support and development are available at
+url(http://lists.samba.org)(lists.samba.org) 
 
+If you suspect you have found a security vulnerability in rsync,
+please send it directly to Martin Pool and Andrew Tridgell.  For other
+enquiries, please use the mailing list.

runtests.sh

@@ -1,6 +1,6 @@
 #! /bin/sh
 
-# Copyright (C) 2001 by Martin Pool <mbp@samba.org>
+# Copyright (C) 2001, 2002 by Martin Pool <mbp@samba.org>
 
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License version
@@ -160,16 +160,27 @@ missing=0
 passed=0
 failed=0
 
-scratchdir=./testtmp
-[ -d "$scratchdir" ] && rm -r "$scratchdir"
-mkdir "$scratchdir"
-scratchdir=`cd $scratchdir && pwd`
-echo "    scratchdir=$scratchdir"
+# Prefix for scratch directory.  We create separate directories for
+# each test case, so that they can be left behind in case of failure
+# to aid investigation.
+scratchbase="`pwd`"/testtmp
+echo "    scratchbase=$scratchbase"
 
 suitedir="$srcdir/testsuite"
 
 export scratchdir suitedir
 
+prep_scratch() {
+    [ -d "$scratchdir" ] && rm -rf "$scratchdir"
+    mkdir "$scratchdir"
+    return 0
+}
+
+discard_scratch() {
+    [ -d "$scratchdir" ] && rm -rf "$scratchdir"
+    return 0
+}
+
 if [ "x$whichtests" = x ]
 then
     whichtests="*.test"
@@ -177,29 +188,45 @@ fi
 
 for testscript in $suitedir/$whichtests
 do
-    testbase=`echo $testscript | sed 's!.*/!!'`
+    testbase=`echo $testscript | sed 's!.*/!!' | sed -e 's/.test\$//'`
+    scratchdir="$scratchbase.$testbase"
 
-    echo "----- $testbase starting"
+    prep_scratch
 
-    if sh $RUNSHFLAGS "$testscript"
-    then
-	echo "----- $testbase completed succesfully"
+    set +e
+    sh $RUNSHFLAGS "$testscript" >"$scratchdir/test.log" 2>&1
+    result=$?
+    set -e
+
+    case $result in
+    0)
+	echo "PASS    $testbase"
 	passed=`expr $passed + 1`
-    else 
-	case $? in
-	77)
-	    echo "----- $testbase skipped"
-	    skipped=`expr $skipped + 1`
-	    ;;
-	*)
-	    echo "----- $testbase failed!"
-	    failed=`expr $failed + 1`
-	    if [ "x$nopersist" = "xyes" ]
-	    then
-		exit 1
-	    fi
-	esac
-    fi
+	discard_scratch
+	;;
+    77)
+	echo "SKIP    $testbase"
+	skipped=`expr $skipped + 1`
+	discard_scratch
+	;;
+    78)
+        # It failed, but we expected that.  don't dump out error logs, 
+	# because most users won't want to see them.  But do leave
+	# the working directory around.
+	echo "XFAIL   $testbase"
+	failed=`expr $failed + 1`
+	;;
+    *)
+	echo "FAIL    $testbase"
+	echo "----- $testbase failed: log follows"
+	cat "$scratchdir/test.log"
+	echo "----- $testbase log ends"
+	failed=`expr $failed + 1`
+	if [ "x$nopersist" = "xyes" ]
+	then
+	    exit 1
+	fi
+    esac
 done
 
 echo '------------------------------------------------------------'

sender.c

@@ -159,13 +159,14 @@ void send_files(struct file_list *flist,int f_out,int f_in)
 		initial_stats = stats;
 
 		s = receive_sums(f_in);
-		if (write_batch) /* dw */
-		    write_batch_csum_info(&i,flist->count,s);
 		if (!s) {
 			io_error = 1;
 			rprintf(FERROR,"receive_sums failed\n");
 			return;
 		}
+
+		if (write_batch)
+		    write_batch_csum_info(&i,flist->count,s);
 	  
 		if (!read_batch) {
 			fd = do_open(fname, O_RDONLY, 0);

socket.c

@@ -1,7 +1,9 @@
 /* -*- c-file-style: "linux" -*-
    
+   rsync -- fast file replication program
+   
    Copyright (C) 1992-2001 by Andrew Tridgell <tridge@samba.org>
-   Copyright (C) 2001 by Martin Pool <mbp@samba.org>
+   Copyright (C) 2001, 2002 by Martin Pool <mbp@samba.org>
    
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -22,14 +24,15 @@
  * @file socket.c
  * 
  * Socket functions used in rsync.
+ *
+ * This file is now converted to use the new-style getaddrinfo()
+ * interface, which supports IPv6 but is also supported on recent
+ * IPv4-only machines.  On systems that don't have that interface, we
+ * emulate it using the KAME implementation.
  **/
 
 #include "rsync.h"
 
-#ifndef HAVE_GETADDRINFO
-#include "lib/addrinfo.h"
-#endif
-
 /* Establish a proxy connection on an open socket to a web roxy by
  * using the CONNECT method. */
 static int establish_proxy_connection(int fd, char *host, int port)
@@ -94,6 +97,39 @@ static int establish_proxy_connection(int fd, char *host, int port)
 }
 
 
+/**
+ * Try to set the local address for a newly-created socket.  Return -1
+ * if this fails.
+ **/
+int try_bind_local(int s,
+		   int ai_family, int ai_socktype,
+		   const char *bind_address)
+{
+	int error;
+	struct addrinfo bhints, *bres_all, *r;
+
+	memset(&bhints, 0, sizeof(bhints));
+	bhints.ai_family = ai_family;
+	bhints.ai_socktype = ai_socktype;
+	bhints.ai_flags = AI_PASSIVE;
+	if ((error = getaddrinfo(bind_address, NULL, &bhints, &bres_all))) {
+		rprintf(FERROR, RSYNC_NAME ": getaddrinfo %s: %s\n",
+			bind_address, gai_strerror(error));
+		return -1;
+	}
+
+	for (r = bres_all; r; r = r->ai_next) {
+		if (bind(s, r->ai_addr, r->ai_addrlen) == -1)
+			continue;
+		return s;
+	}
+
+	/* no error message; there might be some problem that allows
+	 * creation of the socket but not binding, perhaps if the
+	 * machine has no ipv6 address of this name. */
+	return -1;
+}
+
 
 /**
  * Open a socket to a tcp remote host with the specified port .
@@ -121,7 +157,6 @@ int open_socket_out(char *host, int port, const char *bind_address,
 	int type = SOCK_STREAM;
 	int error;
 	int s;
-	int result;
 	struct addrinfo hints, *res0, *res;
 	char portbuf[10];
 	char *h;
@@ -162,32 +197,22 @@ int open_socket_out(char *host, int port, const char *bind_address,
 	}
 
 	s = -1;
+	/* Try to connect to all addresses for this machine until we get
+	 * through.  It might e.g. be multi-homed, or have both IPv4 and IPv6
+	 * addresses.  We need to create a socket for each record, since the
+	 * address record tells us what protocol to use to try to connect. */
 	for (res = res0; res; res = res->ai_next) {
 		s = socket(res->ai_family, res->ai_socktype, res->ai_protocol);
 		if (s < 0)
 			continue;
 
-		if (bind_address) {
-			struct addrinfo bhints, *bres;
-
-			memset(&bhints, 0, sizeof(bhints));
-			bhints.ai_family = res->ai_family;
-			bhints.ai_socktype = type;
-			bhints.ai_flags = AI_PASSIVE;
-			error = getaddrinfo(bind_address, NULL, &bhints, &bres);
-			if (error) {
-				rprintf(FERROR, RSYNC_NAME ": getaddrinfo: bind address %s <noport>: %s\n",
-					bind_address, gai_strerror(error));
+		if (bind_address)
+			if (try_bind_local(s, res->ai_family, type,
+					   bind_address) == -1) {
+				close(s);
+				s = -1;
 				continue;
 			}
-			if (bres->ai_next) {
-				rprintf(FERROR, RSYNC_NAME ": getaddrinfo: bind address %s resolved to multiple hosts\n",
-					bind_address);
-				freeaddrinfo(bres);
-				continue;
-			}
-			bind(s, bres->ai_addr, bres->ai_addrlen);
-		}
 
 		if (connect(s, res->ai_addr, res->ai_addrlen) < 0) {
 			close(s);
@@ -263,7 +288,7 @@ static int open_socket_in(int type, int port, const char *bind_address,
 {
 	int one=1;
 	int s;
-	struct addrinfo hints, *res, *resp;
+	struct addrinfo hints, *all_ai, *resp;
 	char portbuf[10];
 	int error;
 
@@ -272,53 +297,43 @@ static int open_socket_in(int type, int port, const char *bind_address,
 	hints.ai_socktype = type;
 	hints.ai_flags = AI_PASSIVE;
 	snprintf(portbuf, sizeof(portbuf), "%d", port);
-	error = getaddrinfo(bind_address, portbuf, &hints, &res);
+	error = getaddrinfo(bind_address, portbuf, &hints, &all_ai);
 	if (error) {
 		rprintf(FERROR, RSYNC_NAME ": getaddrinfo: bind address %s: %s\n",
 			bind_address, gai_strerror(error));
 		return -1;
 	}
-	/* XXX: Do we need to care about getting multiple results
-	 * back?  I think probably not; if the user passed
-	 * bind_address == NULL and we set AI_PASSIVE then we ought to
-	 * get a wildcard result. */
 
-	resp = res;
-	while (1) {
-		s = socket(resp->ai_family, resp->ai_socktype, resp->ai_protocol);
+	/* We may not be able to create the socket, if for example the
+	 * machine knows about IPv6 in the C library, but not in the
+	 * kernel. */
+	for (resp = all_ai; resp; resp = resp->ai_next) {
+		s = socket(resp->ai_family, resp->ai_socktype,
+			   resp->ai_protocol);
 
-		if (s >= 0) {
-			break;	/* got a socket */
-		} else if ((resp = resp->ai_next)) {
-			switch (errno) {
-			case EPROTONOSUPPORT:
-			case EAFNOSUPPORT:
-			case EPFNOSUPPORT:
-				/* See if there's another address that will work... */
-				continue;
-			}
+		if (s == -1) 
+			/* See if there's another address that will work... */
+			continue;
+		
+		setsockopt(s, SOL_SOCKET, SO_REUSEADDR,
+			   (char *)&one, sizeof one);
+		
+		/* now we've got a socket - we need to bind it */
+		if (bind(s, all_ai->ai_addr, all_ai->ai_addrlen) < 0) {
+			/* Nope, try another */
+			close(s);
+			continue;
 		}
 		
-		rprintf(FERROR, RSYNC_NAME ": open inbound socket"
-			"(dom=%d, type=%d, proto=%d) failed: %s\n",
-			resp->ai_family, resp->ai_socktype, resp->ai_protocol,
-			strerror(errno));
-			goto fail;
+		return s;
 	}
 
-	setsockopt(s,SOL_SOCKET,SO_REUSEADDR,(char *)&one,sizeof(one));
+	rprintf(FERROR, RSYNC_NAME ": open inbound socket on port %d failed: "
+		"%s\n",
+		port, 
+		strerror(errno));
 
-	/* now we've got a socket - we need to bind it */
-	if (bind(s, res->ai_addr, res->ai_addrlen) < 0) { 
-		rprintf(FERROR, RSYNC_NAME ": bind failed on port %d\n", port);
-		close(s);
-		goto fail;
-	}
-
-	return s;
-
-fail:
-	freeaddrinfo(res);
+	freeaddrinfo(all_ai);
 	return -1; 
 }
 
@@ -353,10 +368,10 @@ void start_accept_loop(int port, int (*fn)(int ))
 {
 	int s;
 	extern char *bind_address;
+	extern int default_af_hint;
 
 	/* open an incoming socket */
-	s = open_socket_in(SOCK_STREAM, port, bind_address,
-			   global_opts.af_hint);
+	s = open_socket_in(SOCK_STREAM, port, bind_address, default_af_hint);
 	if (s == -1)
 		exit_cleanup(RERR_SOCKETIO);
 
@@ -372,8 +387,8 @@ void start_accept_loop(int port, int (*fn)(int ))
 	while (1) {
 		fd_set fds;
 		int fd;
-		struct sockaddr addr;
-		int in_addrlen = sizeof(addr);
+		struct sockaddr_storage addr;
+		socklen_t addrlen = sizeof addr;
 
 		/* close log file before the potentially very long select so
 		   file can be trimmed by another process instead of growing
@@ -389,7 +404,7 @@ void start_accept_loop(int port, int (*fn)(int ))
 
 		if(!FD_ISSET(s, &fds)) continue;
 
-		fd = accept(s,(struct sockaddr *)&addr,&in_addrlen);
+		fd = accept(s,(struct sockaddr *)&addr,&addrlen);
 
 		if (fd == -1) continue;
 
@@ -555,8 +570,8 @@ void become_daemon(void)
  **/
 char *client_addr(int fd)
 {
-	struct sockaddr ss;
-	int     length = sizeof(ss);
+	struct sockaddr_storage ss;
+	socklen_t length = sizeof ss;
 	static char addr_buf[100];
 	static int initialised;
 
@@ -564,23 +579,29 @@ char *client_addr(int fd)
 
 	initialised = 1;
 
-	if (getpeername(fd, &ss, &length)) {
+	if (getpeername(fd, (struct sockaddr *)&ss, &length)) {
 		exit_cleanup(RERR_SOCKETIO);
 	}
 
-	getnameinfo(&ss, length,
+	getnameinfo((struct sockaddr *)&ss, length,
 		addr_buf, sizeof(addr_buf), NULL, 0, NI_NUMERICHOST);
 	return addr_buf;
 }
 
 
+static int get_sockaddr_family(const struct sockaddr_storage *ss)
+{
+	return ((struct sockaddr *) ss)->sa_family;
+}
+
+
 /**
  * Return the DNS name of the client 
  **/
 char *client_name(int fd)
 {
-	struct sockaddr ss;
-	int     length = sizeof(ss);
+	struct sockaddr_storage ss;
+	socklen_t length = sizeof ss;
 	static char name_buf[100];
 	static char port_buf[100];
 	char *def = "UNKNOWN";
@@ -602,8 +623,14 @@ char *client_name(int fd)
 	}
 
 #ifdef INET6
-        if (ss.sa_family == AF_INET6 && 
+        if (get_sockaddr_family(&ss) == AF_INET6 && 
 	    IN6_IS_ADDR_V4MAPPED(&((struct sockaddr_in6 *)&ss)->sin6_addr)) {
+		/* OK, so ss is in the IPv6 family, but it is really
+		 * an IPv4 address: something like
+		 * "::ffff:10.130.1.2".  If we use it as-is, then the
+		 * reverse lookup might fail or perhaps something else
+		 * bad might happen.  So instead we convert it to an
+		 * equivalent address in the IPv4 address family.  */
 		struct sockaddr_in6 sin6;
 		struct sockaddr_in *sin;
 
@@ -616,6 +643,9 @@ char *client_name(int fd)
 		sin->sin_len = length;
 #endif
 		sin->sin_port = sin6.sin6_port;
+		/* FIXME: Isn't there a macro we can use here rather
+		 * than grovelling through the struct?  It might be
+		 * wrong on some systems. */
 		memcpy(&sin->sin_addr, &sin6.sin6_addr.s6_addr[12],
 			sizeof(sin->sin_addr));
         }
@@ -646,7 +676,7 @@ char *client_name(int fd)
 
 	/* XXX sin6_flowinfo and other fields */
 	for (res = res0; res; res = res->ai_next) {
-		if (res->ai_family != ss.sa_family)
+		if (res->ai_family != get_sockaddr_family(&ss))
 			continue;
 		if (res->ai_addrlen != length)
 			continue;
@@ -654,13 +684,11 @@ char *client_name(int fd)
 			break;
 	}
 
-	/* TODO: Do a  forward lookup as well to prevent spoofing */
-
 	if (res == NULL) {
 		strcpy(name_buf, def);
 		rprintf(FERROR, RSYNC_NAME ": "
-			"reverse name lookup mismatch on fd%d - spoofed address?\n",
-			fd);
+			"reverse name lookup for \"%s\" failed on fd%d - spoofed address? \n",
+			name_buf, fd);
 	}
 
 	freeaddrinfo(res0);

syscall.c

@@ -132,7 +132,7 @@ int do_mkstemp(char *template, mode_t perms)
 	}
 #else
 	if (!mktemp(template)) return -1;
-	return open(template, O_RDWR|O_EXCL|O_CREAT, perms);
+	return do_open(template, O_RDWR|O_EXCL|O_CREAT, perms);
 #endif
 }
 

testsuite/duplicates.test

@@ -0,0 +1,57 @@
+#! /bin/sh
+
+# Copyright (C) 2002 by Martin Pool <mbp@samba.org>
+
+# This program is distributable under the terms of the GNU GPL see
+# COPYING).
+
+# Test rsync handling of duplicate filenames.  
+
+# It's quite possible that the user might specify the same source file
+# more than once on the command line, perhaps through shell variables
+# or wildcard expansions.  It might cause problems for rsync if the
+# same name occurred more than once in the file list, because we might
+# be trying to update the first copy and generate checksums for the
+# second copy at the same time.  See clean_flist() for the implementation.
+
+# We don't need to worry about hardlinks or symlinks.  Because we
+# always rename-and-replace the new copy, they can't affect us.
+
+# This test is not great, because it is a timing-dependent bug.
+
+. $srcdir/testsuite/rsync.fns
+
+echo "SKIP THIS FOR NOW; It's a known bug"
+exit 77
+
+set -x
+
+# Build some hardlinks
+
+fromdir="$scratchdir/from"
+todir="$scratchdir/to"
+
+mkdir "$fromdir"
+name1="$fromdir/name1"
+name2="$fromdir/name2"
+echo "This is the file" > "$name1"
+ln -s "$name1" "$name2" || fail "can't create symlink"
+
+outfile="$scratchdir/rsync.out"
+
+checkit "rsync -avv \"$fromdir/\" \"$fromdir/\" \"$fromdir/\" \"$fromdir/\" \"$fromdir/\" \"$fromdir/\" \"$fromdir/\" \"$fromdir/\" \"$fromdir/\" \"$fromdir/\" \"$todir/\"" "$fromdir" "$todir" \
+    | tee "$outfile"
+
+# Make sure each file was only copied once...
+if [ `grep -c '^name1$' "$outfile"` != 1 ]
+then
+    test_xfail "name1 was not copied exactly once"
+fi
+if [ `grep -c '^name2$' "$outfile"` != 1 ]
+then
+    test_xfail "name2 was not copied exactly once"
+fi
+
+exit 0
+# last [] may have failed but if we get here then we've won
+

testsuite/hands.test

@@ -26,3 +26,5 @@ cp ${FROM}/text ${TO}/ThisShouldGo
 runtest " --delete" 'checkit "$RSYNC --delete -avH ${FROM}/ ${TO}" ${FROM}/ ${TO}'
 
 checkforlogs ${LOG}.?
+
+hands_cleanup

testsuite/hardlinks.test

@@ -0,0 +1,38 @@
+#! /bin/sh
+
+# Copyright (C) 2002 by Martin Pool <mbp@samba.org>
+
+# This program is distributable under the terms of the GNU GPL (see
+# COPYING).
+
+# Test rsync handling of hardlinks.  By default (in 2.5.1) rsync does
+# not detect symlinks and they get split into different files.  If you
+# specify -H, then hard links are detected and recreated as hardlinks
+# on the other end.
+
+. $srcdir/testsuite/rsync.fns
+
+set -x
+
+# Build some hardlinks
+
+fromdir="$scratchdir/from"
+todir="$scratchdir/to"
+
+# TODO: Need to test whether hardlinks are possible on this OS/filesystem
+
+mkdir "$fromdir"
+name1="$fromdir/name1"
+name2="$fromdir/name2"
+name3="$fromdir/name3"
+name4="$fromdir/name4"
+echo "This is the file" > "$name1"
+ln "$name1" "$name2" || fail "Can't create hardlink"
+ln "$name2" "$name3" || fail "Can't create hardlink"
+cp "$name2" "$name4" || fail "Can't copy file"
+
+checkit "rsync -aHvv \"$fromdir/\" \"$todir/\"" "$fromdir" "$todir"
+
+exit 0
+# last [] may have failed but if we get here then we've won
+

testsuite/longdir.test

@@ -7,7 +7,7 @@
 
 . "$suitedir/rsync.fns"
 
-set -x
+# set -x
 
 hands_setup
 

testsuite/rsync.fns

@@ -53,7 +53,8 @@ hands_setup() {
     # Clean before creation
     rm -rf $FROM
     rm -rf $TO
-    
+
+    [ -d $TMP ] || mkdir $TMP
     [ -d $FROM ] || mkdir $FROM
     [ -d $TO ] || mkdir $TO
 
@@ -72,7 +73,7 @@ hands_setup() {
     mkdir ${FROM}/emptydir
 
     # a hundred lines of text or so
-    rsync_ls_lR ${srcdir} > ${FROM}/filelist
+    rsync_ls_lR "${srcdir}" > ${FROM}/filelist
 
     # This might fail on systems that don't have -n
     echo $ECHO_N "This file has no trailing lf$ECHO_C" > ${FROM}/nolf
@@ -94,6 +95,11 @@ hands_setup() {
 }
 
 
+hands_cleanup() {
+    rm -r "$TMP"
+}
+
+
 
 
 ####################
@@ -129,35 +135,31 @@ makepath () {
 # there are any difference.  If there are, explain them.
 
 checkit() {
-    log=${LOG}
     failed=
-    # the log accumulates all output; we only display it if there 
-    # is a problem.
 
-    echo "Running: \"$1\""  >${log}
-    echo "">>${log}
-    eval "$1"  >>${log} 2>&1
+    # We can just write everything to stdout/stderr, because the
+    # wrapper hides it unless there is a problem.
+
+    echo "Running: \"$1\""  
+    eval "$1" 
     status=$?
     if [ $status != 0 ]; then
 	failed="YES";
     fi
 
-    echo "-------------">>${log}
-    echo "check how the files compare with diff:">>${log}
-    echo "">>${log}
-    diff -cr $2 $3 >>${log} 2>&1 || failed=YES
-    echo "-------------">>${log}
-    echo "check how the directory listings compare with diff:">>${log}
-    echo "">>${log}
-    ( cd "$2" && rsync_ls_lR ) > ${TMP}/ls-from 2>>${log}
-    ( cd "$3" && rsync_ls_lR ) > ${TMP}/ls-to  2>>${log}
-    diff -c ${TMP}/ls-from ${TMP}/ls-to >>${log} 2>&1 || failed=YES
+    echo "-------------"
+    echo "check how the files compare with diff:"
+    echo ""
+    diff -cr $2 $3 || failed=YES
+    echo "-------------"
+    echo "check how the directory listings compare with diff:"
+    echo ""
+    ( cd "$2" && rsync_ls_lR . ) > ${TMP}/ls-from 
+    ( cd "$3" && rsync_ls_lR . ) > ${TMP}/ls-to 
+    diff -c ${TMP}/ls-from ${TMP}/ls-to || failed=YES
     if [ -z "${failed}" ] ; then
-	rm $log
 	return 0
     else
-	cat ${log}
-	rm ${log}
 	return 1
     fi
 }
@@ -204,3 +206,29 @@ EOF
 }
 
 
+build_symlinks() {
+    fromdir="$scratchdir/from"
+    todir="$scratchdir/to"
+    mkdir "$fromdir"
+    date >"$fromdir/referent"
+    ln -s referent "$fromdir/relative"
+    ln -s "$fromdir/referent" "$fromdir/absolute"
+    ln -s nonexistent "$fromdir/dangling"
+    ln -s "$srcdir/rsync.c" "$fromdir/unsafe"
+}
+
+test_fail() {
+    echo "$@" >&2
+    exit 1
+}
+
+# It failed, but we expected that.  don't dump out error logs, 
+# because most users won't want to see them.  But do leave
+# the working directory around.
+test_xfail() {
+    echo "$@" >&2
+    exit 78
+}
+
+# be reproducible
+umask 077

testsuite/symlink-ignore.test

@@ -0,0 +1,29 @@
+#! /bin/sh
+
+# Copyright (C) 2001 by Martin Pool <mbp@samba.org>
+
+# This program is distributable under the terms of the GNU GPL (see
+# COPYING).
+
+# Test rsync's somewhat over-featured symlink control: the default
+# behaviour is that symlinks should not be copied at all.
+
+. $srcdir/testsuite/rsync.fns
+
+set -x
+
+build_symlinks || test_fail "failed to build symlinks"
+
+# Copy recursively, but without -l or -L or -a, and all the symlinks
+# should be missing.
+"$rsync_bin" -r "$fromdir/" "$todir" || test_fail "rsync returned $?"
+
+[ -f "$todir/referent" ] || test_fail "referent was not copied"
+[ -d "$todir/from" ] && test_fail "extra level of directories"
+[ -L "$todir/dangling" ] && test_fail "dangling symlink was copied"
+[ -L "$todir/relative" ] && test_fail "relative symlink was copied" 
+[ -L "$todir/absolute" ] && test_fail "absolute symlink was copied"
+
+exit 0
+# last [] may have failed but if we get here then we've one
+

tls.c

@@ -1,6 +1,6 @@
 /* -*- c-file-style: "linux" -*-
  *
- * Copyright (C) 2001 by Martin Pool <mbp@samba.org>
+ * Copyright (C) 2001, 2002 by Martin Pool <mbp@samba.org>
  * 
  * This program is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version
@@ -80,11 +80,18 @@ static void list_file (const char *fname)
 	 * undefined.  Also it tends not to be possible to reset a
 	 * symlink's mtime, so we have to ignore it too. */
 	if (S_ISLNK(buf.st_mode)) {
+		int len;
 		buf.st_mode &= ~0777;
 		buf.st_mtime = (time_t)0;
 		buf.st_uid = buf.st_gid = 0;
 		strcpy(linkbuf, " -> ");
-		readlink(fname, linkbuf+4, sizeof(linkbuf) - 4);
+		/* const-cast required for silly UNICOS headers */
+		len = readlink((char *) fname, linkbuf+4, sizeof(linkbuf) - 4);
+		if (len == -1) 
+			failed("readlink", fname);
+		else
+			/* it's not nul-terminated */
+			linkbuf[4+len] = 0;
 	} else {
 		linkbuf[0] = 0;
 	}
@@ -110,9 +117,10 @@ static void list_file (const char *fname)
 	
 	/* NB: need to pass size as a double because it might be be
 	 * too large for a long. */
-	printf("%s %12.0f %6d.%-6d %s %s%s\n",
+	printf("%s %12.0f %6d.%-6d %6d %s %s%s\n",
 	       permbuf, (double) buf.st_size,
 	       buf.st_uid, buf.st_gid,
+	       buf.st_nlink,
 	       datebuf, fname, linkbuf);
 }
 

util.c

@@ -2,7 +2,7 @@
     
     Copyright (C) 1996-2000 by Andrew Tridgell 
     Copyright (C) Paul Mackerras 1996
-    Copyright (C) 2001 by Martin Pool <mbp@samba.org>
+    Copyright (C) 2001, 2002 by Martin Pool <mbp@samba.org>
    
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -173,10 +173,7 @@ pid_t local_child(int argc, char **argv,int *f_in,int *f_out)
 		extern int am_sender;
 		extern int am_server;
 
-		if (read_batch)
-		    am_sender = 0;
-		else
-		    am_sender = !am_sender;
+		am_sender = read_batch ? 0 : !am_sender;
 		am_server = 1;		
 
 		if (dup2(to_child_pipe[0], STDIN_FILENO) < 0 ||
@@ -275,7 +272,7 @@ int create_directory_path(char *fname)
 
    derived from GNU C's cccp.c.
 */
-static int full_write(int desc, char *ptr, int len)
+static int full_write(int desc, char *ptr, size_t len)
 {
 	int total_written;
 	
@@ -301,11 +298,11 @@ static int full_write(int desc, char *ptr, int len)
    for an error.  
 
    derived from GNU C's cccp.c. */
-static int safe_read(int desc, char *ptr, int len)
+static int safe_read(int desc, char *ptr, size_t len)
 {
 	int n_chars;
  
-	if (len <= 0)
+	if (len == 0)
 		return len;
  
 #ifdef EINTR
@@ -811,28 +808,91 @@ int u_strcmp(const char *cs1, const char *cs2)
 	return (int)*s1 - (int)*s2;
 }
 
-static OFF_T last_ofs;
+static OFF_T  last_ofs;
+static struct timeval print_time;
+static struct timeval start_time;
+static OFF_T  start_ofs;
+
+static unsigned long msdiff(struct timeval *t1, struct timeval *t2)
+{
+    return (t2->tv_sec - t1->tv_sec) * 1000
+        + (t2->tv_usec - t1->tv_usec) / 1000;
+}
+
+
+/**
+ * @param ofs Current position in file
+ * @param size Total size of file
+ * @param is_last True if this is the last time progress will be
+ * printed for this file, so we should output a newline.  (Not
+ * necessarily the same as all bytes being received.)
+ **/
+static void rprint_progress(OFF_T ofs, OFF_T size, struct timeval *now,
+			    int is_last)
+{
+    int           pct  = (ofs == size) ? 100 : (int)((100.0*ofs)/size);
+    unsigned long diff = msdiff(&start_time, now);
+    double        rate = diff ? (double) (ofs-start_ofs) * 1000.0 / diff / 1024.0 : 0;
+    const char    *units;
+    double        remain = rate ? (double) (size-ofs) / rate / 1000.0: 0.0;
+    int 	  remain_h, remain_m, remain_s;
+
+    if (rate > 1024*1024) {
+	    rate /= 1024.0 * 1024.0;
+	    units = "GB/s";
+    } else if (rate > 1024) {
+	    rate /= 1024.0;
+	    units = "MB/s";
+    } else {
+	    units = "kB/s";
+    }
+
+    remain_s = (int) remain % 60;
+    remain_m = (int) (remain / 60.0) % 60;
+    remain_h = (int) (remain / 3600.0);
+    
+    rprintf(FINFO, "%12.0f %3d%% %7.2f%s %4d:%02d:%02d%s",
+	    (double) ofs, pct, rate, units,
+	    remain_h, remain_m, remain_s,
+	    is_last ? "\n" : "\r");
+}
 
 void end_progress(OFF_T size)
 {
 	extern int do_progress, am_server;
 
 	if (do_progress && !am_server) {
-		rprintf(FINFO,"%.0f (100%%)\n", (double)size);
+        	struct timeval now;
+                gettimeofday(&now, NULL);
+                rprint_progress(size, size, &now, True);
 	}
-	last_ofs = 0;
+	last_ofs   = 0;
+        start_ofs  = 0;
+        print_time.tv_sec  = print_time.tv_usec  = 0;
+        start_time.tv_sec  = start_time.tv_usec  = 0;
 }
 
 void show_progress(OFF_T ofs, OFF_T size)
 {
 	extern int do_progress, am_server;
+        struct timeval now;
 
-	if (do_progress && !am_server) {
-		if (ofs > last_ofs + 1000) {
-			int pct = (int)((100.0*ofs)/size);
-			rprintf(FINFO,"%.0f (%d%%)\r", (double)ofs, pct);
-			last_ofs = ofs;
-		}
+        gettimeofday(&now, NULL);
+
+        if (!start_time.tv_sec && !start_time.tv_usec) {
+        	start_time.tv_sec  = now.tv_sec;
+                start_time.tv_usec = now.tv_usec;
+                start_ofs          = ofs;
+        }
+
+	if (do_progress
+            && !am_server
+            && ofs > last_ofs + 1000
+            && msdiff(&print_time, &now) > 250) {
+        	rprint_progress(ofs, size, &now, False);
+                last_ofs = ofs;
+                print_time.tv_sec  = now.tv_sec;
+                print_time.tv_usec = now.tv_usec;
 	}
 }