Preparing for release of 3.2.4

The latest news.
Add debugging comment about read_buf_().
2026-05-26 15:55:42 -04:00 · 2022-04-15 13:31:16 -07:00 · 2022-04-11 09:50:58 -07:00 · 2022-04-11 09:50:31 -07:00 · 2022-04-11 09:08:11 -07:00 · 2022-04-11 08:57:19 -07:00
41 changed files with 713 additions and 367 deletions
--- a/.cirrus.yml
+++ b/.cirrus.yml
@@ -1,7 +1,7 @@
 freebsd_task:
  name: FreeBSD
  freebsd_instance:
-    image_family: freebsd-12-2
+    image_family: freebsd-13-0
  env:
    PATH: /usr/local/bin:$PATH
  prep_script:
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -86,16 +86,15 @@ jobs:
          rrsync

  cygwin-build:
-    runs-on: windows-latest
+    runs-on: windows-2022
    if: (github.event_name == 'schedule' || contains(github.event.head_commit.message, '[buildall]'))
    steps:
    - uses: actions/checkout@v2
-    - uses: crazy-max/ghaction-chocolatey@v1.2.2
-      with:
-        args: install -y --no-progress cygwin cyg-get
+    - name: cygwin
+      run: choco install -y --no-progress cygwin cyg-get
    - name: prep
      run: |
-        cyg-get make autoconf automake gcc-core attr libattr-devel python38 python38-pip libzstd-devel liblz4-devel libssl-devel libxxhash0 libxxhash-devel
+        cyg-get make autoconf automake gcc-core attr libattr-devel python39 python39-pip libzstd-devel liblz4-devel libssl-devel libxxhash0 libxxhash-devel
        curl.exe -o git-version.h https://gist.githubusercontent.com/WayneD/c11243fa374fc64d4e42f2855c8e3827/raw/rsync-git-version.h
        echo "C:/tools/cygwin/bin" >>$Env:GITHUB_PATH
    - name: commonmark
--- a/.gitignore
+++ b/.gitignore
@@ -16,6 +16,7 @@ aclocal.m4
 /proto.h
 /proto.h-tstamp
 /rsync*.[15]
+/rrsync
 /rrsync*.1
 /rsync*.html
 /rrsync*.html
--- a/INSTALL.md
+++ b/INSTALL.md
@@ -13,11 +13,11 @@ You need to have a C compiler installed and optionally a C++ compiler in order
 to try to build some hardware-accelerated checksum routines.  Rsync also needs
 a modern awk, which might be provided via gawk or nawk on some OSes.

-## Autoconf & man pages
+## Autoconf & manpages

 If you're installing from the git repo (instead of a release tar file) you'll
 also need the GNU autotools (autoconf & automake) and your choice of 2 python3
-markdown libraries: cmarkgfm or commonmark (needed to generate the man pages).
+markdown libraries: cmarkgfm or commonmark (needed to generate the manpages).
 If your OS doesn't provide a python3-cmarkgfm or python3-commonmark package,
 you can run the following to install the commonmark python library for your
 build user (after installing python3's pip package):
@@ -28,7 +28,7 @@ You can test if you've got it fixed by running (from the rsync checkout):

 >     ./md2man --test rsync-ssl.1.md

-Alternately, you can avoid generating the man pages by fetching the very latest
+Alternately, you can avoid generating the manpages by fetching the very latest
 versions (that match the latest git source) from the [generated-files][6] dir.
 One way to do that is to run:

--- a/Makefile.in
+++ b/Makefile.in
@@ -30,8 +30,9 @@ SHELL=/bin/sh
 .SUFFIXES:
 .SUFFIXES: .c .o

-SIMD_x86_64=simd-checksum-x86_64.o simd-checksum-avx2.o
-ASM_x86_64=lib/md5-asm-x86_64.o
+ROLL_SIMD_x86_64=simd-checksum-x86_64.o
+ROLL_ASM_x86_64=simd-checksum-avx2.o
+MD5_ASM_x86_64=lib/md5-asm-x86_64.o

 GENFILES=configure.sh aclocal.m4 config.h.in rsync.1 rsync.1.html \
 	 rsync-ssl.1 rsync-ssl.1.html rsyncd.conf.5 rsyncd.conf.5.html \
@@ -46,7 +47,7 @@ OBJS1=flist.o rsync.o generator.o receiver.o cleanup.o sender.o exclude.o \
 	util1.o util2.o main.o checksum.o match.o syscall.o log.o backup.o delete.o
 OBJS2=options.o io.o compat.o hlink.o token.o uidlist.o socket.o hashtable.o \
 	usage.o fileio.o batch.o clientname.o chmod.o acls.o xattrs.o
-OBJS3=progress.o pipe.o @ASM@ @SIMD@
+OBJS3=progress.o pipe.o @MD5_ASM@ @ROLL_SIMD@ @ROLL_ASM@
 DAEMON_OBJ = params.o loadparm.o clientserver.o access.o connection.o authenticate.o
 popt_OBJS=popt/findme.o  popt/popt.o  popt/poptconfig.o \
 	popt/popthelp.o popt/poptparse.o
@@ -147,13 +148,13 @@ git-version.h: ALWAYS_RUN
 ALWAYS_RUN:

 simd-checksum-x86_64.o: simd-checksum-x86_64.cpp
-	@$(srcdir)/cmd-or-msg disable-simd $(CXX) -I. $(CXXFLAGS) $(CPPFLAGS) -c -o $@ $(srcdir)/simd-checksum-x86_64.cpp
+	@$(srcdir)/cmd-or-msg disable-roll-simd $(CXX) -I. $(CXXFLAGS) $(CPPFLAGS) -c -o $@ $(srcdir)/simd-checksum-x86_64.cpp

 simd-checksum-avx2.o: simd-checksum-avx2.S
-	@$(srcdir)/cmd-or-msg disable-asm $(CC) $(CFLAGS) --include=$(srcdir)/rsync.h -DAVX2_ASM -I. @NOEXECSTACK@ -c -o $@ $(srcdir)/simd-checksum-avx2.S
+	@$(srcdir)/cmd-or-msg disable-roll-asm $(CC) $(CFLAGS) -I. @NOEXECSTACK@ -c -o $@ $(srcdir)/simd-checksum-avx2.S

-lib/md5-asm-x86_64.o: lib/md5-asm-x86_64.S config.h lib/md-defines.h
-	@$(srcdir)/cmd-or-msg disable-asm $(CC) -I. @NOEXECSTACK@ -c -o $@ $(srcdir)/lib/md5-asm-x86_64.S
+lib/md5-asm-x86_64.o: lib/md5-asm-x86_64.S lib/md-defines.h
+	@$(srcdir)/cmd-or-msg disable-md5-asm $(CC) -I. @NOEXECSTACK@ -c -o $@ $(srcdir)/lib/md5-asm-x86_64.S

 tls$(EXEEXT): $(TLS_OBJ)
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(TLS_OBJ) $(LIBS)
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,4 +1,4 @@
-# NEWS for rsync 3.2.4 (UNRELEASED)
+# NEWS for rsync 3.2.4 (15 Apr 2022)

 ## Changes in this version:

@@ -20,13 +20,13 @@
   If your rsync script depends on the old arg-splitting behavior, either run
   it with the [`--old-args`](rsync.1#opt) option or `export RSYNC_OLD_ARGS=1`
   in the script's environment.  See also the [ADVANCED USAGE](rsync.1#)
-   section of rsync's man page.
+   section of rsync's manpage for how to use a more modern arg style.

 - A long-standing bug was preventing rsync from figuring out the current
   locale's decimal point character, which made rsync always output numbers
   using the "C" locale.  Since this is now fixed in 3.2.4, a script that
-   parses rsync's decimal numbers (e.g. from the verbose footer) should be sure
-   to setup the environment in a way that the output continues to be in the C
+   parses rsync's decimal numbers (e.g. from the verbose footer) may want to
+   setup the environment in a way that the output continues to be in the C
   locale.  For instance, one of the following should work fine:

   ```shell
@@ -44,6 +44,10 @@
       export LC_NUMERIC=C.UTF-8
   ```

+### SECURITY FIXES:
+
+ - A fix for CVE-2018-25032 in the bundled zlib (memory corruption issue).
+
 ### BUG FIXES:

 - Fixed a bug with [`--inplace`](rsync.1#opt) + [`--sparse`](rsync.1#opt) (and
@@ -87,6 +91,12 @@
   per-file compression skipping has apparently never worked, so it is now
   documented as being ineffective.

+ - Fixed a truncate error when a `--write-devices` copy wrote a file onto a
+   device that was shorter than the device.
+
+ - Made `--write-devices` support both `--checksum` and `--no-whole-file` when
+   copying to a device.
+
 - Improved how the [`--stop-at`](rsync.1#opt), [`--stop-after`](rsync.1#opt),
   and (the deprecated) [`--time-limit`](rsync.1#opt) options check to see if
   the allowed time is over, which should make rsync exit more consistently.
@@ -97,6 +107,12 @@
 - Silence some chmod warnings about symlinks when it looks like we have a
   function to set their permissions but they can't really be set.

+ - Fixed a potential issue in git-set-file-times when handling commits with
+   high-bit characters in the description & when handling a description that
+   might mimick the git raw-commit deliniators.  (See the support dir.)
+
+ - The bundled systemd/rsync.service file now includes `Restart=on-failure`.
+
 ### ENHANCEMENTS:

 - Use openssl's `-verify_hostname` option in the rsync-ssl script.
@@ -110,6 +126,12 @@

 - Added the [`--fsync`](rsync.1#opt) option (promoted from the patches repo).

+ - Added the [`--copy-devices`](rsync.1#opt) option.  Compared to the
+   historical version from the rsync-patches repo, this version: properly
+   handles `--checksum`; fixes a truncation bug when doing an `--inplace` copy
+   onto a longer file; fixes several bugs in the `--itemize` output; and only
+   the sending side needs the enhanced rsync for the copy to work.
+
 - Reduced memory usage for an incremental transfer that has a bunch of small
   directories.

@@ -127,7 +149,7 @@
   that is being refused due to the Linux fs.protected_regular sysctl setting.

 - When [`--chown`](rsync.1#opt), [`--usermap`](rsync.1#opt), or
-   [`--groupmap`](rsync.1#opt), is specified, rsync now makes sure that the
+   [`--groupmap`](rsync.1#opt) is specified, rsync now makes sure that the
   appropriate [`--owner`](rsync.1#opt) and/or [`--group`](rsync.1#opt) options
   are enabled.

@@ -136,7 +158,11 @@
   (keeping the behavior the same as before), so specifying `--info=nonreg0`
   can be used to turn the warnings off.

- - More ASM optimizations from Shark64.
+ - An optional asm optimization for the rolling checksum from Shark64. Enable
+   it with `./configure --enable-roll-asm`.
+
+ - Using `--debug=FILTER` now outputs a caution message if a filter rule
+   has trailing whitespace.

 - Transformed rrsync into a python script with improvements:
   - Security has been beefed up.
@@ -154,7 +180,7 @@
     and `--delete*` rsync options on the server side.
   - The log format has been tweaked slightly to add seconds to the timestamp
     and to output the command executed as a tuple (making the args clearer).
-   - An rrsync.1 man page was added (in the support dir with rrsync).
+   - An rrsync.1 manpage was added (in the support dir with rrsync).

 - Added options to the lsh script to facilitate rrsync testing. (See the
   support dir.)
@@ -173,12 +199,12 @@
 - Try to support a client that sent a remote rsync a wacko stderr file handle
   (such as an older File::RsyncP perl library used by BackupPC).

- - Lots of man page improvements, including better html versions.
+ - Lots of manpage improvements, including better HTML versions.

 ### PACKAGING RELATED:

 - Give configure the `--with-rrsync` option if you want `make install` to
-   install the (now python3) rrsync script and its new man page.
+   install the (now python3) rrsync script and its new manpage.

 - If the rrsync script is installed, its package should be changed to depend
   on python3 and the (suggested but not mandatory) python3 braceexpand lib.
@@ -189,18 +215,32 @@
   using the output of `git describe` when building inside a non-shallow git
   checkout, though.)

- - Improved the IPv6 determination in configure.
+ - Renamed configure's `--enable-simd` option to `--enable-roll-simd` and added
+   the option `--enable-roll-asm` to use the new asm version of the code.  Both
+   are x86_64/amd64 only.

- - Made SIMD & ASM configure default to "no" on non-Linux hosts due to various
-   reports of problems on NetBSD & macOS hosts.  These tests were also tweaked
-   to allow enabling the feature on a host_cpu of amd64 (was only x86_64).
+ - Renamed configure's `--enable-asm` option to `--enable-md5-asm` to avoid
+   confusion with the asm option for the rolling checksum.  It is also honored
+   even when openssl crypto is in use.  This allows: normal MD4 & MD5, normal
+   MD4 + asm MD5, openssl MD4 & MD5, or openssl MD4 + asm MD5 depending on the
+   configure options selected.
+
+ - Made SIMD & asm configure checks default to "no" on non-Linux hosts due to
+   various reports of problems on NetBSD & macOS hosts.  These were also
+   tweaked to allow enabling the feature on a host_cpu of amd64 (was only
+   allowed on x86_64 before).

 - Fixed configure to not fail at the SIMD check when cross-compiling.

+ - Improved the IPv6 determination in configure.
+
 - Compile the C files with `-pedantic-errors` (when possible) so that we will
   get warned if a static initialization overflows in the future (among other
   things).

+ - When linking with an external zlib, rsync renames its `read_buf()` function
+   to `read_buf_()` to avoid a symbol clash on an unpatched zlib.
+
 - Added a SECURITY.md file.

 ### DEVELOPER RELATED:
@@ -383,7 +423,7 @@

 - Put optimizations into their own list in the `--version` output.

- - Improved the man page a bit more.
+ - Improved the manpage a bit more.

 ### PACKAGING RELATED:

@@ -610,7 +650,7 @@
 - The daemon now locks its pid file (when configured to use one) so that it
   will not fail to start when the file exists but no daemon is running.

- - Various man page improvements, including some html representations (that
+ - Various manpage improvements, including some html representations (that
   aren't installed by default).

 - Made `-V` the short option for `--version` and improved its information.
@@ -627,7 +667,7 @@

 - Add installed bash script: /usr/bin/rsync-ssl

- - Add installed man page: /usr/man/man1/rsync-ssl.1
+ - Add installed manpage: /usr/man/man1/rsync-ssl.1

 - Tweak auxiliary doc file names, such as: README.md, INSTALL.md, & NEWS.md.

@@ -649,8 +689,8 @@
   SIMD checksum optimizations.

 - Add _build_ dependency for _either_ python3-cmarkcfm or python3-commonmark
-   to allow for patching of man pages or building a git release.  This is not
-   required for a release-tar build, since it comes with pre-built man pages.
+   to allow for patching of manpages or building a git release.  This is not
+   required for a release-tar build, since it comes with pre-built manpages.
   Note that cmarkcfm is faster than commonmark, but they generate the same
   data.  The commonmark dependency is easiest to install since it's native
   python, and can even be installed via `pip3 install --user commonmark` if
@@ -663,7 +703,7 @@
 - Silenced some annoying warnings about major() & minor() by improving an
   autoconf include-file check.

- - Converted the man pages from yodl to markdown. They are now processed via a
+ - Converted the manpages from yodl to markdown. They are now processed via a
   simple python3 script using the cmarkgfm **or** commonmark library.  This
   should make it easier to package rsync, since yodl is rather obscure.

@@ -741,11 +781,11 @@

 ### DEVELOPER RELATED:

- - Tweak the `make` output when yodl isn't around to create the man pages.
+ - Tweak the `make` output when yodl isn't around to create the manpages.

 - Changed an obsolete autoconf compile macro.

- - Support newer yodl versions when converting man pages.
+ - Support newer yodl versions when converting manpages.

 ------------------------------------------------------------------------------

@@ -1234,7 +1274,7 @@

 - Avoid trying to reference `SO_BROADCAST` if the OS doesn't support it.

- - Fix some issues with the post-processing of the man pages.
+ - Fix some issues with the post-processing of the manpages.

 - Fixed the user home-dir handling in the lsh script. (See the support dir.)

@@ -1714,7 +1754,7 @@
   of files, and the ensuring that daemon excludes can't affect a dot-dir arg.

 - Improved some build rules for those that build in a separate directory from
-   the source, including better install rules for the man pages, and the fixing
+   the source, including better install rules for the manpages, and the fixing
   of a proto.h-tstamp rule that could make the binaries get rebuild without
   cause.

@@ -2322,7 +2362,7 @@

 - Added the `--log-file=FILE` and `--log-file-format=FORMAT` options. These
   can be used to tell any rsync to output what it is doing to a log file.
-   They work with a client rsync, a non-daemon server rsync (see the man page
+   They work with a client rsync, a non-daemon server rsync (see the manpage
   for instructions), and also allows the overriding of rsyncd.conf settings
   when starting a daemon.

@@ -2616,7 +2656,7 @@
 - Added two config items to the rsyncd.conf parsing: `pre-xfer exec` and
   `post-xfer exec`. These allow a command to be specified on a per-module
   basis that will be run before and/or after a daemon-mode transfer. (See the
-   man page for a list of the environment variables that are set with
+   manpage for a list of the environment variables that are set with
   information about the transfer.)

 - When using the `--relative` option, you can now insert a dot dir in the
@@ -2984,7 +3024,7 @@
   usually run with the `--no-detach` option that was necessary to see the
   error on stderr).

- - The man pages now consistently refer to an rsync daemon as a `daemon`
+ - The manpages now consistently refer to an rsync daemon as a `daemon`
   instead of a `server` (to distinguish it from the server process in a
   non-daemon transfer).

@@ -3551,12 +3591,12 @@
   without using a temporary file. The matching of existing data in the
   destination file can be severely limited by this, but there are also cases
   where this is more efficient (such as appending data).  Use only when needed
-   (see the man page for more details).
+   (see the manpage for more details).

 - Added the `write only` option for the daemon's config file.

 - Added long-option names for `-4` and `-6` (namely `--ipv4` and `--ipv6`) and
-   documented all these options in the man page.
+   documented all these options in the manpage.

 - Improved the handling of the `--bwlimit` option so that it's less bursty,
   more accurate, and works properly over a larger range of values.
@@ -3631,7 +3671,7 @@
 ### BUILD CHANGES:

 - Added a `gen` target to rebuild most of the generated files, including
-   configure, config.h.in, the man pages, and proto.h.
+   configure, config.h.in, the manpages, and proto.h.

 - If `make proto` doesn't find some changes in the prototypes, the proto.h
   file is left untouched (its time-stamp used to always be updated).
@@ -4229,7 +4269,7 @@
 - Added `--no-whole-file` and `--no-blocking-io` options (Dave Dykstra)

 - Made the `--write-batch` and `--read-batch` options actually work and added
-   documentation in the man page (Jos Backus)
+   documentation in the manpage (Jos Backus)

 - If the daemon is unable to fork a child to accept a connection, print an
   error message. (Colin Walters)
@@ -4442,7 +4482,7 @@

 | RELEASE DATE | VER.   | DATE OF COMMIT\* | PROTOCOL    |
 |--------------|--------|------------------|-------------|
-| ?? Jan 2022  | 3.2.4  |                  | 31          |
+| 15 Apr 2022  | 3.2.4  |                  | 31          |
 | 06 Aug 2020  | 3.2.3  |                  | 31          |
 | 04 Jul 2020  | 3.2.2  |                  | 31          |
 | 22 Jun 2020  | 3.2.1  |                  | 31          |
--- a/README.md
+++ b/README.md
@@ -65,8 +65,8 @@ RSYNC DAEMONS
 -------------

 Rsync can also talk to "rsync daemons" which can provide anonymous or
-authenticated rsync.  See the rsyncd.conf(5) man page for details on how
-to setup an rsync daemon.  See the rsync(1) man page for info on how to
+authenticated rsync.  See the rsyncd.conf(5) manpage for details on how
+to setup an rsync daemon.  See the rsync(1) manpage for info on how to
 connect to an rsync daemon.


--- a/checksum.c
+++ b/checksum.c
@@ -3,7 +3,7 @@
 *
 * Copyright (C) 1996 Andrew Tridgell
 * Copyright (C) 1996 Paul Mackerras
- * Copyright (C) 2004-2020 Wayne Davison
+ * Copyright (C) 2004-2022 Wayne Davison
 *
 * 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
@@ -179,7 +179,7 @@ int canonical_checksum(int csum_type)
 	return 0;
 }

-#ifndef HAVE_SIMD /* See simd-checksum-*.cpp. */
+#ifndef USE_ROLL_SIMD /* See simd-checksum-*.cpp. */
 /*
  a simple 32 bit checksum that can be updated from either end
  (inspired by Mark Adler's Adler-32 checksum)
@@ -222,23 +222,23 @@ void get_checksum2(char *buf, int32 len, char *sum)
 	  }
 #endif
 	  case CSUM_MD5: {
-		MD5_CTX m5;
+		md5_context m5;
 		uchar seedbuf[4];
-		MD5_Init(&m5);
+		md5_begin(&m5);
 		if (proper_seed_order) {
 			if (checksum_seed) {
 				SIVALu(seedbuf, 0, checksum_seed);
-				MD5_Update(&m5, seedbuf, 4);
+				md5_update(&m5, seedbuf, 4);
 			}
-			MD5_Update(&m5, (uchar *)buf, len);
+			md5_update(&m5, (uchar *)buf, len);
 		} else {
-			MD5_Update(&m5, (uchar *)buf, len);
+			md5_update(&m5, (uchar *)buf, len);
 			if (checksum_seed) {
 				SIVALu(seedbuf, 0, checksum_seed);
-				MD5_Update(&m5, seedbuf, 4);
+				md5_update(&m5, seedbuf, 4);
 			}
 		}
-		MD5_Final((uchar *)sum, &m5);
+		md5_result(&m5, (uchar *)sum);
 		break;
 	  }
 	  case CSUM_MD4:
@@ -374,18 +374,18 @@ void file_checksum(const char *fname, const STRUCT_STAT *st_p, char *sum)
 	  }
 #endif
 	  case CSUM_MD5: {
-		MD5_CTX m5;
+		md5_context m5;

-		MD5_Init(&m5);
+		md5_begin(&m5);

 		for (i = 0; i + CHUNK_SIZE <= len; i += CHUNK_SIZE)
-			MD5_Update(&m5, (uchar *)map_ptr(buf, i, CHUNK_SIZE), CHUNK_SIZE);
+			md5_update(&m5, (uchar *)map_ptr(buf, i, CHUNK_SIZE), CHUNK_SIZE);

 		remainder = (int32)(len - i);
 		if (remainder > 0)
-			MD5_Update(&m5, (uchar *)map_ptr(buf, i, remainder), remainder);
+			md5_update(&m5, (uchar *)map_ptr(buf, i, remainder), remainder);

-		MD5_Final((uchar *)sum, &m5);
+		md5_result(&m5, (uchar *)sum);
 		break;
 	  }
 	  case CSUM_MD4:
@@ -443,7 +443,7 @@ static union {
 #ifdef USE_OPENSSL
 	MD4_CTX m4;
 #endif
-	MD5_CTX m5;
+	md5_context m5;
 } ctx;
 #ifdef SUPPORT_XXHASH
 static XXH64_state_t* xxh64_state;
@@ -482,7 +482,7 @@ void sum_init(int csum_type, int seed)
 		break;
 #endif
 	  case CSUM_MD5:
-		MD5_Init(&ctx.m5);
+		md5_begin(&ctx.m5);
 		break;
 	  case CSUM_MD4:
 #ifdef USE_OPENSSL
@@ -532,7 +532,7 @@ void sum_update(const char *p, int32 len)
 		break;
 #endif
 	  case CSUM_MD5:
-		MD5_Update(&ctx.m5, (uchar *)p, len);
+		md5_update(&ctx.m5, (uchar *)p, len);
 		break;
 	  case CSUM_MD4:
 #ifdef USE_OPENSSL
@@ -597,7 +597,7 @@ int sum_end(char *sum)
 	  }
 #endif
 	  case CSUM_MD5:
-		MD5_Final((uchar *)sum, &ctx.m5);
+		md5_result(&ctx.m5, (uchar *)sum);
 		break;
 	  case CSUM_MD4:
 #ifdef USE_OPENSSL
--- a/compat.c
+++ b/compat.c
@@ -604,7 +604,7 @@ void setup_protocol(int f_out,int f_in)
 	if (remote_protocol < MIN_PROTOCOL_VERSION
 	 || remote_protocol > MAX_PROTOCOL_VERSION) {
 		rprintf(FERROR,"protocol version mismatch -- is your shell clean?\n");
-		rprintf(FERROR,"(see the rsync man page for an explanation)\n");
+		rprintf(FERROR,"(see the rsync manpage for an explanation)\n");
 		exit_cleanup(RERR_PROTOCOL);
 	}
 	if (remote_protocol < OLD_PROTOCOL_VERSION) {
--- a/configure.ac
+++ b/configure.ac
@@ -83,7 +83,7 @@ if test x"$enable_profile" = x"yes"; then
 	CFLAGS="$CFLAGS -pg"
 fi

-AC_MSG_CHECKING([if md2man can create man pages])
+AC_MSG_CHECKING([if md2man can create manpages])
 if test x"$ac_cv_path_PYTHON3" = x; then
    AC_MSG_RESULT(no - python3 not found)
    md2man_works=no
@@ -101,7 +101,7 @@ fi

 AC_MSG_CHECKING([if we require man-page building])
 AC_ARG_ENABLE([md2man],
-	AS_HELP_STRING([--disable-md2man],[disable to omit man page creation]))
+	AS_HELP_STRING([--disable-md2man],[disable to omit manpage creation]))
 if test x"$enable_md2man" != x"no"; then
    if test -f "$srcdir/rsync.1"; then
 	AC_MSG_RESULT(optional)
@@ -109,7 +109,7 @@ if test x"$enable_md2man" != x"no"; then
 	AC_MSG_RESULT(required)
 	if test x"$md2man_works" = x"no"; then
 	    err_msg="$err_msg$nl- You need python3 and either the cmarkgfm OR commonmark python3 lib in order"
-	    err_msg="$err_msg$nl  to build man pages based on the git source (man pages are included in the"
+	    err_msg="$err_msg$nl  to build manpages based on the git source (manpages are included in the"
 	    err_msg="$err_msg$nl  official release tar files)."
 	    no_lib="$no_lib md2man"
 	fi
@@ -136,7 +136,7 @@ if test x"$GCC" = x"yes"; then
 fi

 AC_ARG_WITH(rrsync,
-        AS_HELP_STRING([--with-rrsync],[also install the rrsync script and its man page]))
+        AS_HELP_STRING([--with-rrsync],[also install the rrsync script and its manpage]))
 if test x"$with_rrsync" != x"yes"; then
    with_rrsync=no
 else
@@ -229,12 +229,12 @@ fi
 AC_DEFINE_UNQUOTED(NOBODY_USER, "$NOBODY_USER", [unprivileged user--e.g. nobody])
 AC_DEFINE_UNQUOTED(NOBODY_GROUP, "$NOBODY_GROUP", [unprivileged group for unprivileged user])

-# SIMD optimizations
-SIMD=
+# rolling-checksum SIMD optimizations
+ROLL_SIMD=

-AC_MSG_CHECKING([whether to enable SIMD optimizations])
-AC_ARG_ENABLE(simd,
-    AS_HELP_STRING([--enable-simd],[enable/disable to control SIMD optimizations (requires c++)]))
+AC_MSG_CHECKING([whether to enable rolling-checksum SIMD optimizations])
+AC_ARG_ENABLE(roll-simd,
+    AS_HELP_STRING([--enable-roll-simd],[enable/disable to control rolling-checksum SIMD optimizations (requires c++)]))

 # Clag is crashing with -g -O2, so we'll get rid of -g for now.
 CXXFLAGS=`echo "$CXXFLAGS" | sed 's/-g //'`
@@ -263,14 +263,14 @@ __attribute__ ((target("ssse3"))) void more_testing(char* buf, int len)
 }
 ]])

-if test x"$enable_simd" = x""; then
+if test x"$enable_roll_simd" = x""; then
    case "$host_os" in
 	*linux*) ;;
-	*) enable_simd=no ;;
+	*) enable_roll_simd=no ;;
    esac
 fi

-if test x"$enable_simd" != x"no"; then
+if test x"$enable_roll_simd" != x"no"; then
    # For x86-64 SIMD, g++ >=5 or clang++ >=7 is required
    if test x"$host_cpu" = x"x86_64" || test x"$host_cpu" = x"amd64"; then
 	AC_LANG(C++)
@@ -283,23 +283,23 @@ if test x"$enable_simd" != x"no"; then
 	AC_LANG(C)
 	if test x"$CXX_OK" = x"yes"; then
 	    # AC_MSG_RESULT() is called below.
-	    SIMD="$host_cpu"
-	elif test x"$enable_simd" = x"yes"; then
+	    ROLL_SIMD="$host_cpu"
+	elif test x"$enable_roll_simd" = x"yes"; then
 	    AC_MSG_RESULT(error)
-	    AC_MSG_ERROR(The SIMD compilation test failed.
-Omit --enable-simd to continue without it.)
+	    AC_MSG_ERROR(The rolling-checksum SIMD compilation test failed.
+Omit --enable-roll-simd to continue without it.)
 	fi
-    elif test x"$enable_simd" = x"yes"; then
+    elif test x"$enable_roll_simd" = x"yes"; then
        AC_MSG_RESULT(unavailable)
-        AC_MSG_ERROR(The SIMD optimizations are currently x86_64|amd64 only.
-Omit --enable-simd to continue without it.)
+        AC_MSG_ERROR(The rolling-checksum SIMD optimizations are currently x86_64|amd64 only.
+Omit --enable-roll-simd to continue without it.)
    fi
 fi

-if test x"$SIMD" != x""; then
-    AC_MSG_RESULT([yes ($SIMD)])
-    AC_DEFINE(HAVE_SIMD, 1, [Define to 1 to enable SIMD optimizations])
-    SIMD='$(SIMD_'"$SIMD)"
+if test x"$ROLL_SIMD" != x""; then
+    AC_MSG_RESULT([yes ($ROLL_SIMD)])
+    AC_DEFINE(USE_ROLL_SIMD, 1, [Define to 1 to enable rolling-checksum SIMD optimizations])
+    ROLL_SIMD='$(ROLL_SIMD_'"$ROLL_SIMD)"
    # We only use c++ for its target attribute dispatching, disable unneeded bulky features
    CXXFLAGS="$CXXFLAGS -fno-exceptions -fno-rtti"
    # Apple often has "g++" as a symlink for clang. Try to find out the truth.
@@ -311,7 +311,7 @@ else
    AC_MSG_RESULT(no)
 fi

-AC_SUBST(SIMD)
+AC_SUBST(ROLL_SIMD)

 AC_MSG_CHECKING([if assembler accepts noexecstack])
 OLD_CFLAGS="$CFLAGS"
@@ -322,52 +322,19 @@ AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[ ]], [[return 0;]])],
 CFLAGS="$OLD_CFLAGS"
 AC_SUBST(NOEXECSTACK)

-ASM=
-
-AC_MSG_CHECKING([whether to enable ASM optimizations])
-AC_ARG_ENABLE(asm,
-    AS_HELP_STRING([--enable-asm],[enable/disable to control ASM optimizations]))
-
-if test x"$enable_asm" = x""; then
-    case "$host_os" in
-	*linux*) ;;
-	*) enable_asm=no ;;
-    esac
-fi
-
-if test x"$enable_asm" != x"no"; then
-    if test x"$host_cpu" = x"x86_64" || test x"$host_cpu" = x"amd64"; then
-	ASM="$host_cpu"
-    elif test x"$enable_asm" = x"yes"; then
-        AC_MSG_RESULT(unavailable)
-        AC_MSG_ERROR(The ASM optimizations are currently x86_64|amd64 only.
-Omit --enable-asm to continue without it.)
-    fi
-fi
-
-if test x"$ASM" != x""; then
-    AC_MSG_RESULT([yes ($ASM)])
-    AC_DEFINE(HAVE_ASM, 1, [Define to 1 to enable ASM optimizations])
-    ASM='$(ASM_'"$ASM)"
-else
-    AC_MSG_RESULT(no)
-fi
-
-AC_SUBST(ASM)
-
 # arrgh. libc in some old debian version screwed up the largefile
 # stuff, getting byte range locking wrong
 AC_CACHE_CHECK([for broken largefile support],rsync_cv_HAVE_BROKEN_LARGEFILE,[
 AC_RUN_IFELSE([AC_LANG_SOURCE([[
 #define _FILE_OFFSET_BITS 64
-#include <stdio.h>
-#include <fcntl.h>
-#ifdef HAVE_SYS_TYPES_H
-#include <sys/types.h>
+$ac_includes_default
+#ifdef HAVE_FCNTL_H
+# include <fcntl.h>
+#elif defined HAVE_SYS_FCNTL_H
+# include <sys/fcntl.h>
 #endif
+#ifdef HAVE_SYS_WAIT_H
 #include <sys/wait.h>
-#if HAVE_UNISTD_H
-#include <unistd.h>
 #endif

 int main(void)
@@ -457,7 +424,8 @@ if test x"$enable_openssl" != x"no"; then
    if test x"$ac_cv_header_openssl_md4_h" = x"yes" && test x"$ac_cv_header_openssl_md5_h" = x"yes"; then
      AC_MSG_RESULT(yes)
      AC_SEARCH_LIBS(MD5_Init, crypto,
-          [AC_DEFINE(USE_OPENSSL)],
+          [AC_DEFINE(USE_OPENSSL)
+	   enable_openssl=yes],
          [err_msg="$err_msg$nl- Failed to find MD5_Init function in openssl crypto lib.";
 	   no_lib="$no_lib openssl"])
    else
@@ -465,10 +433,67 @@ if test x"$enable_openssl" != x"no"; then
 	err_msg="$err_msg$nl- Failed to find openssl/md4.h and openssl/md5.h for openssl crypto lib support."
 	no_lib="$no_lib openssl"
    fi
+    if test x"$enable_md5_asm" != x"yes"; then
+	enable_md5_asm=no
+    fi
 else
    AC_MSG_RESULT(no)
 fi

+MD5_ASM=
+
+AC_MSG_CHECKING([whether to enable MD5 ASM optimizations])
+AC_ARG_ENABLE(md5-asm,
+    AS_HELP_STRING([--enable-md5-asm],[enable/disable to control MD5 ASM optimizations]))
+
+if test x"$enable_md5_asm" = x""; then
+    case "$host_os" in
+	*linux*) ;;
+	*) enable_md5_asm=no ;;
+    esac
+fi
+
+if test x"$enable_md5_asm" != x"no"; then
+    if test x"$host_cpu" = x"x86_64" || test x"$host_cpu" = x"amd64"; then
+	MD5_ASM="$host_cpu"
+    elif test x"$enable_md5_asm" = x"yes"; then
+        AC_MSG_RESULT(unavailable)
+        AC_MSG_ERROR(The ASM optimizations are currently x86_64|amd64 only.
+Omit --enable-md5-asm to continue without it.)
+    fi
+fi
+
+if test x"$MD5_ASM" != x""; then
+    AC_MSG_RESULT([yes ($MD5_ASM)])
+    AC_DEFINE(USE_MD5_ASM, 1, [Define to 1 to enable MD5 ASM optimizations])
+    MD5_ASM='$(MD5_ASM_'"$MD5_ASM)"
+else
+    AC_MSG_RESULT(no)
+fi
+
+AC_SUBST(MD5_ASM)
+
+ROLL_ASM=
+
+AC_MSG_CHECKING([whether to enable rolling-checksum ASM optimizations])
+AC_ARG_ENABLE(roll-asm,
+    AS_HELP_STRING([--enable-roll-asm],[enable/disable to control rolling-checksum ASM optimizations (requires --enable-roll-simd)]))
+
+if test x"$ROLL_SIMD" = x""; then
+    enable_roll_asm=no
+fi
+
+if test x"$enable_roll_asm" = x"yes"; then
+    ROLL_ASM="$host_cpu"
+    AC_MSG_RESULT([yes ($ROLL_ASM)])
+    AC_DEFINE(USE_ROLL_ASM, 1, [Define to 1 to enable rolling-checksum ASM optimizations (requires --enable-roll-simd)])
+    ROLL_ASM='$(ROLL_ASM_'"$ROLL_ASM)"
+else
+    AC_MSG_RESULT(no)
+fi
+
+AC_SUBST(ROLL_ASM)
+
 AC_MSG_CHECKING([whether to enable xxhash checksum support])
 AC_ARG_ENABLE([xxhash],
 	AS_HELP_STRING([--disable-xxhash],[disable to omit xxhash checksums]))
@@ -538,7 +563,7 @@ if test x"$no_lib" != x; then
    echo "$err_msg"
    echo ""
    echo "See the INSTALL file for hints on how to install the missing libraries and/or"
-    echo "how to generate (or fetch) man pages:"
+    echo "how to generate (or fetch) manpages:"
    echo "    https://github.com/WayneD/rsync/blob/master/INSTALL.md"
    echo ""
    echo "To disable one or more features, the relevant configure options are:"
--- a/exclude.c
+++ b/exclude.c
@@ -4,7 +4,7 @@
 * Copyright (C) 1996-2001 Andrew Tridgell <tridge@samba.org>
 * Copyright (C) 1996 Paul Mackerras
 * Copyright (C) 2002 Martin Pool
- * Copyright (C) 2003-2020 Wayne Davison
+ * Copyright (C) 2003-2022 Wayne Davison
 *
 * 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
@@ -152,13 +152,17 @@ static void add_rule(filter_rule_list *listp, const char *pat, unsigned int pat_
 {
 	const char *cp;
 	unsigned int pre_len, suf_len, slash_cnt = 0;
+	char *mention_rule_suffix;

-	if (DEBUG_GTE(FILTER, 2)) {
-		rprintf(FINFO, "[%s] add_rule(%s%.*s%s)%s\n",
+	if (DEBUG_GTE(FILTER, 1) && pat_len && (pat[pat_len-1] == ' ' || pat[pat_len-1] == '\t'))
+		mention_rule_suffix = " -- CAUTION: trailing whitespace!";
+	else
+		mention_rule_suffix = DEBUG_GTE(FILTER, 2) ? "" : NULL;
+	if (mention_rule_suffix) {
+		rprintf(FINFO, "[%s] add_rule(%s%.*s%s)%s%s\n",
 			who_am_i(), get_rule_prefix(rule, pat, 0, NULL),
-			(int)pat_len, pat,
-			(rule->rflags & FILTRULE_DIRECTORY) ? "/" : "",
-			listp->debug_type);
+			(int)pat_len, pat, (rule->rflags & FILTRULE_DIRECTORY) ? "/" : "",
+			listp->debug_type, mention_rule_suffix);
 	}

 	/* These flags also indicate that we're reading a list that
--- a/flist.c
+++ b/flist.c
@@ -43,6 +43,7 @@ extern int use_qsort;
 extern int xfer_dirs;
 extern int filesfrom_fd;
 extern int one_file_system;
+extern int copy_devices;
 extern int copy_dirlinks;
 extern int preserve_uid;
 extern int preserve_gid;
@@ -700,6 +701,7 @@ static struct file_struct *recv_file_entry(int f, struct file_list *flist, int x
 	int alloc_len, basename_len, linkname_len;
 	int extra_len = file_extra_cnt * EXTRA_LEN;
 	int first_hlink_ndx = -1;
+	char real_ISREG_entry;
 	int64 file_length;
 #ifdef CAN_SET_NSEC
 	uint32 modtime_nsec;
@@ -814,6 +816,7 @@ static struct file_struct *recv_file_entry(int f, struct file_list *flist, int x
 				linkname_len = strlen(F_SYMLINK(first)) + 1;
 			else
 				linkname_len = 0;
+			real_ISREG_entry = S_ISREG(mode) ? 1 : 0;
 			goto create_object;
 		}
 	}
@@ -941,10 +944,20 @@ static struct file_struct *recv_file_entry(int f, struct file_list *flist, int x
 #endif
 		linkname_len = 0;

+	if (copy_devices && IS_DEVICE(mode)) {
+		/* This is impossible in the official release, but some pre-release patches
+		 * didn't convert the device into a file before sending, so we'll do it here
+		 * (even though the length is typically 0 and any checksum data is zeros). */
+		mode = S_IFREG | (mode & ACCESSPERMS);
+		modtime = time(NULL); /* The mtime on the device is not up-to-date, so set it to "now". */
+		real_ISREG_entry = 0;
+	} else
+		real_ISREG_entry = S_ISREG(mode) ? 1 : 0;
+
 #ifdef SUPPORT_HARD_LINKS
  create_object:
 	if (preserve_hard_links) {
-		if (protocol_version < 28 && S_ISREG(mode))
+		if (protocol_version < 28 && real_ISREG_entry)
 			xflags |= XMIT_HLINKED;
 		if (xflags & XMIT_HLINKED)
 			extra_len += (inc_recurse+1) * EXTRA_LEN;
@@ -1160,8 +1173,8 @@ static struct file_struct *recv_file_entry(int f, struct file_list *flist, int x
 	}
 #endif

-	if (always_checksum && (S_ISREG(mode) || protocol_version < 28)) {
-		if (S_ISREG(mode))
+	if (always_checksum && (real_ISREG_entry || protocol_version < 28)) {
+		if (real_ISREG_entry)
 			bp = F_SUM(file);
 		else {
 			/* Prior to 28, we get a useless set of nulls. */
@@ -1360,6 +1373,18 @@ struct file_struct *make_file(const char *fname, struct file_list *flist,
 	linkname_len = 0;
 #endif

+	if (copy_devices && am_sender && IS_DEVICE(st.st_mode)) {
+		if (st.st_size == 0) {
+			int fd = do_open(fname, O_RDONLY, 0);
+			if (fd >= 0) {
+				st.st_size = get_device_size(fd, fname);
+				close(fd);
+			}
+		}
+		st.st_mode = S_IFREG | (st.st_mode & ACCESSPERMS);
+		st.st_mtime = time(NULL); /* The mtime on the device is not up-to-date, so set it to "now". */
+	}
+
 #ifdef ST_MTIME_NSEC
 	if (st.ST_MTIME_NSEC && protocol_version >= 31)
 		extra_len += EXTRA_LEN;
--- a/generator.c
+++ b/generator.c
@@ -35,11 +35,11 @@ extern int inc_recurse;
 extern int relative_paths;
 extern int implied_dirs;
 extern int keep_dirlinks;
+extern int write_devices;
 extern int preserve_acls;
 extern int preserve_xattrs;
 extern int preserve_links;
 extern int preserve_devices;
-extern int write_devices;
 extern int preserve_specials;
 extern int preserve_hard_links;
 extern int preserve_executability;
@@ -1793,6 +1793,12 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 		goto cleanup;
 	}

+	if (write_devices && IS_DEVICE(sx.st.st_mode) && sx.st.st_size == 0) {
+		/* This early open into fd skips the regular open below. */
+		if ((fd = do_open(fnamecmp, O_RDONLY, 0)) >= 0)
+			real_sx.st.st_size = sx.st.st_size = get_device_size(fd, fnamecmp);
+	}
+
 	if (fnamecmp_type <= FNAMECMP_BASIS_DIR_HIGH)
 		;
 	else if (fnamecmp_type >= FNAMECMP_FUZZY)
@@ -1858,7 +1864,7 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 	}

 	/* open the file */
-	if ((fd = do_open(fnamecmp, O_RDONLY, 0)) < 0) {
+	if (fd < 0 && (fd = do_open(fnamecmp, O_RDONLY, 0)) < 0) {
 		rsyserr(FERROR, errno, "failed to open %s, continuing",
 			full_fname(fnamecmp));
 	  pretend_missing:
@@ -1875,11 +1881,9 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,

 	if (inplace && make_backups > 0 && fnamecmp_type == FNAMECMP_FNAME) {
 		if (!(backupptr = get_backup_name(fname))) {
-			close(fd);
 			goto cleanup;
 		}
 		if (!(back_file = make_file(fname, NULL, NULL, 0, NO_FILTERS))) {
-			close(fd);
 			goto pretend_missing;
 		}
 		if (robust_unlink(backupptr) && errno != ENOENT) {
@@ -1887,14 +1891,12 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 				full_fname(backupptr));
 			unmake_file(back_file);
 			back_file = NULL;
-			close(fd);
 			goto cleanup;
 		}
 		if ((f_copy = do_open(backupptr, O_WRONLY | O_CREAT | O_TRUNC | O_EXCL, 0600)) < 0) {
 			rsyserr(FERROR_XFER, errno, "open %s", full_fname(backupptr));
 			unmake_file(back_file);
 			back_file = NULL;
-			close(fd);
 			goto cleanup;
 		}
 		fnamecmp_type = FNAMECMP_BACKUP;
@@ -1945,7 +1947,6 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 		write_sum_head(f_out, NULL);
 	else if (sx.st.st_size <= 0) {
 		write_sum_head(f_out, NULL);
-		close(fd);
 	} else {
 		if (generate_and_send_sums(fd, sx.st.st_size, f_out, f_copy) < 0) {
 			rprintf(FWARNING,
@@ -1953,10 +1954,11 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 				fnamecmp);
 			write_sum_head(f_out, NULL);
 		}
-		close(fd);
 	}

  cleanup:
+	if (fd >= 0)
+		close(fd);
 	if (back_file) {
 		int save_preserve_xattrs = preserve_xattrs;
 		if (f_copy >= 0)
--- a/io.c
+++ b/io.c
@@ -1843,6 +1843,7 @@ int64 read_longint(int f)
 #endif
 }

+/* Debugging note: this will be named read_buf_() when using an external zlib. */
 void read_buf(int f, char *buf, size_t len)
 {
 	if (f != iobuf.in_fd) {
--- a/lib/md5-asm-x86_64.S
+++ b/lib/md5-asm-x86_64.S
@@ -27,7 +27,7 @@
 #include "config.h"
 #include "md-defines.h"

-#if !defined USE_OPENSSL && CSUM_CHUNK == 64
+#ifdef USE_MD5_ASM /* { */

 #ifdef __APPLE__
 #define md5_process_asm _md5_process_asm
@@ -698,4 +698,4 @@ md5_process_asm:
 	pop	%rbp
 	ret

-#endif /* !USE_OPENSSL ... */
+#endif /* } USE_MD5_ASM */
--- a/lib/md5.c
+++ b/lib/md5.c
@@ -2,7 +2,7 @@
 * RFC 1321 compliant MD5 implementation
 *
 * Copyright (C) 2001-2003 Christophe Devine
- * Copyright (C) 2007-2020 Wayne Davison
+ * Copyright (C) 2007-2022 Wayne Davison
 *
 * 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
@@ -20,7 +20,7 @@

 #include "rsync.h"

-#ifndef USE_OPENSSL
+#if !defined USE_OPENSSL || USE_MD5_ASM /* { */
 void md5_begin(md_context *ctx)
 {
 	ctx->A = 0x67452301;
@@ -148,7 +148,10 @@ static void md5_process(md_context *ctx, const uchar data[CSUM_CHUNK])
 	ctx->D += D;
 }

-#if defined HAVE_ASM && CSUM_CHUNK == 64
+#ifdef USE_MD5_ASM
+#if CSUM_CHUNK != 64
+#error The MD5 ASM code does not support CSUM_CHUNK != 64
+#endif
 extern void md5_process_asm(md_context *ctx, const void *data, size_t num);
 #endif

@@ -176,20 +179,20 @@ void md5_update(md_context *ctx, const uchar *input, uint32 length)
 		left = 0;
 	}

-#if defined HAVE_ASM && CSUM_CHUNK == 64
+#ifdef USE_MD5_ASM /* { */
 	if (length >= CSUM_CHUNK) {
 		uint32 chunks = length / CSUM_CHUNK;
 		md5_process_asm(ctx, input, chunks);
 		length -= chunks * CSUM_CHUNK;
 		input += chunks * CSUM_CHUNK;
 	}
-#else
+#else /* } { */
 	while (length >= CSUM_CHUNK) {
 		md5_process(ctx, input);
 		length -= CSUM_CHUNK;
 		input  += CSUM_CHUNK;
 	}
-#endif
+#endif /* } */

 	if (length)
 		memcpy(ctx->buffer + left, input, length);
@@ -221,9 +224,9 @@ void md5_result(md_context *ctx, uchar digest[MD5_DIGEST_LEN])
 	SIVALu(digest, 8, ctx->C);
 	SIVALu(digest, 12, ctx->D);
 }
-#endif
+#endif /* } */

-#ifdef TEST_MD5
+#ifdef TEST_MD5 /* { */

 void get_md5(uchar *out, const uchar *input, int n)
 {
@@ -317,4 +320,4 @@ int main(int argc, char *argv[])
 	return 0;
 }

-#endif
+#endif /* } */
--- a/lib/mdigest.h
+++ b/lib/mdigest.h
@@ -17,12 +17,13 @@ void mdfour_begin(md_context *md);
 void mdfour_update(md_context *md, const uchar *in, uint32 length);
 void mdfour_result(md_context *md, uchar digest[MD4_DIGEST_LEN]);

-#ifndef USE_OPENSSL
-#define MD5_CTX md_context
-#define MD5_Init md5_begin
-#define MD5_Update md5_update
-#define MD5_Final(digest, cptr) md5_result(cptr, digest)
-
+#if defined USE_OPENSSL && !defined USE_MD5_ASM
+#define md5_context MD5_CTX
+#define md5_begin MD5_Init
+#define md5_update MD5_Update
+#define md5_result(cptr, digest) MD5_Final(digest, cptr)
+#else
+#define md5_context md_context
 void md5_begin(md_context *ctx);
 void md5_update(md_context *ctx, const uchar *input, uint32 length);
 void md5_result(md_context *ctx, uchar digest[MD5_DIGEST_LEN]);
--- a/6
+++ b/6
@@ -148,7 +148,7 @@ def parse_md_file(mdfn):
    fi = argparse.Namespace(**fi.groupdict())
    fi.want_manpage = not not fi.sect
    if fi.want_manpage:
-        fi.title = fi.prog + '(' + fi.sect + ') man page'
+        fi.title = fi.prog + '(' + fi.sect + ') manpage'
    else:
        fi.title = fi.prog + ' for rsync'

@@ -350,7 +350,7 @@ class TransformHtml(HTMLParser):
            st.txt += BOLD_FONT[0]
        elif tag == 'em' or  tag == 'i':
            if st.want_manpage:
-                tag = 'u' # Change it into underline to be more like the man page
+                tag = 'u' # Change it into underline to be more like the manpage
                st.txt += UNDR_FONT[0]
        elif tag == 'ol':
            start = 1
@@ -456,7 +456,7 @@ class TransformHtml(HTMLParser):
            add_to_txt = NORM_FONT[0]
        elif tag == 'em' or  tag == 'i':
            if st.want_manpage:
-                tag = 'u' # Change it into underline to be more like the man page
+                tag = 'u' # Change it into underline to be more like the manpage
                add_to_txt = NORM_FONT[0]
        elif tag == 'ol' or tag == 'ul':
            if st.list_state.pop() == 'dl':
--- a/options.c
+++ b/options.c
@@ -47,6 +47,7 @@ int append_mode = 0;
 int keep_dirlinks = 0;
 int copy_dirlinks = 0;
 int copy_links = 0;
+int copy_devices = 0;
 int write_devices = 0;
 int preserve_links = 0;
 int preserve_hard_links = 0;
@@ -656,6 +657,7 @@ static struct poptOption long_options[] = {
  {"no-D",             0,  POPT_ARG_NONE,   0, OPT_NO_D, 0, 0 },
  {"devices",          0,  POPT_ARG_VAL,    &preserve_devices, 1, 0, 0 },
  {"no-devices",       0,  POPT_ARG_VAL,    &preserve_devices, 0, 0, 0 },
+  {"copy-devices",     0,  POPT_ARG_NONE,   &copy_devices, 0, 0, 0 },
  {"write-devices",    0,  POPT_ARG_VAL,    &write_devices, 1, 0, 0 },
  {"no-write-devices", 0,  POPT_ARG_VAL,    &write_devices, 0, 0, 0 },
  {"specials",         0,  POPT_ARG_VAL,    &preserve_specials, 1, 0, 0 },
@@ -949,6 +951,7 @@ static void set_refuse_options(void)
 		 || strcmp("iconv", longName) == 0
 		 || strcmp("no-iconv", longName) == 0
 		 || strcmp("checksum-seed", longName) == 0
+		 || strcmp("copy-devices", longName) == 0 /* disable wild-match (it gets refused below) */
 		 || strcmp("write-devices", longName) == 0 /* disable wild-match (it gets refused below) */
 		 || strcmp("log-format", longName) == 0 /* aka out-format (NOT log-file-format) */
 		 || strcmp("sender", longName) == 0
@@ -960,6 +963,7 @@ static void set_refuse_options(void)
 	assert(list_end != NULL);

 	if (am_daemon) { /* Refused by default, but can be accepted via a negated exact match. */
+		parse_one_refuse_match(0, "copy-devices", list_end);
 		parse_one_refuse_match(0, "write-devices", list_end);
 	}

@@ -2917,6 +2921,9 @@ void server_options(char **args, int *argc_p)
 	else if (remove_source_files)
 		args[ac++] = "--remove-sent-files";

+	if (copy_devices && !am_sender)
+		args[ac++] = "--copy-devices";
+
 	if (preallocate_files && am_sender)
 		args[ac++] = "--preallocate";

--- a/packaging/lsb/rsync.spec
+++ b/packaging/lsb/rsync.spec
@@ -1,9 +1,9 @@
 Summary: A fast, versatile, remote (and local) file-copying tool
 Name: rsync
 Version: 3.2.4
-%define fullversion %{version}pre3
-Release: 0.1.pre3
-%define srcdir src-previews
+%define fullversion %{version}
+Release: 1
+%define srcdir src
 Group: Applications/Internet
 License: GPL
 Source0: https://rsync.samba.org/ftp/rsync/%{srcdir}/rsync-%{fullversion}.tar.gz
@@ -79,8 +79,8 @@ rm -rf $RPM_BUILD_ROOT
 %dir /etc/rsync-ssl/certs

 %changelog
-* Tue Jan 18 2022 Wayne Davison <wayne@opencoder.net>
-Released 3.2.4pre3.
+* Fri Apr 15 2022 Wayne Davison <wayne@opencoder.net>
+Released 3.2.4.

 * Fri Mar 21 2008 Wayne Davison <wayne@opencoder.net>
 Added installation of /etc/xinetd.d/rsync file and some commented-out
--- a/packaging/systemd/rsync.service
+++ b/packaging/systemd/rsync.service
@@ -7,6 +7,7 @@ Documentation=man:rsync(1) man:rsyncd.conf(5)
 [Service]
 ExecStart=/usr/bin/rsync --daemon --no-detach
 RestartSec=1
+Restart=on-failure

 # Citing README.md:
 #
--- a/2
+++ b/2
@@ -6,7 +6,7 @@
 #
 #   build     build the config files [the default w/no arg]
 #   fetch     fetch the latest dev autoconfig files
-#   fetchgen  fetch all the latest dev generated files (including man pages)
+#   fetchgen  fetch all the latest dev generated files (including manpages)
 #   fetchSRC  fetch the latest dev source files [NON-GENERATED FILES]
 #
 # The script stops after the first successful action.
--- a/receiver.c
+++ b/receiver.c
@@ -808,14 +808,16 @@ int recv_files(int f_in, int f_out, char *local_name)
 			continue;
 		}

-		if (fd1 != -1 && !(S_ISREG(st.st_mode) || (write_devices && IS_DEVICE(st.st_mode)))) {
+		if (write_devices && IS_DEVICE(st.st_mode)) {
+			if (fd1 != -1 && st.st_size == 0)
+				st.st_size = get_device_size(fd1, fname);
+			/* Mark the file entry as a device so that we don't try to truncate it later on. */
+			file->mode = S_IFBLK | (file->mode & ACCESSPERMS);
+		} else if (fd1 != -1 && !(S_ISREG(st.st_mode))) {
 			close(fd1);
 			fd1 = -1;
 		}

-		if (fd1 != -1 && IS_DEVICE(st.st_mode) && st.st_size == 0)
-			st.st_size = get_device_size(fd1, fname);
-
 		/* If we're not preserving permissions, change the file-list's
 		 * mode based on the local permissions and some heuristics. */
 		if (!preserve_perms) {
--- a/rsync-ssl.1.md
+++ b/rsync-ssl.1.md
@@ -8,7 +8,7 @@ rsync-ssl - a helper script for connecting to an ssl rsync daemon
 rsync-ssl [--type=SSL_TYPE] RSYNC_ARGS
 ```

-The online version of this man page (that includes cross-linking of topics)
+The online version of this manpage (that includes cross-linking of topics)
 is available at <https://download.samba.org/pub/rsync/rsync-ssl.1>.

 ## DESCRIPTION
@@ -117,7 +117,7 @@ Please report bugs! See the web site at <https://rsync.samba.org/>.

 ## VERSION

-This man page is current for version @VERSION@ of rsync.
+This manpage is current for version @VERSION@ of rsync.

 ## CREDITS

--- a/rsync.1.md
+++ b/rsync.1.md
@@ -26,7 +26,7 @@ Access via rsync daemon:
 Usages with just one SRC arg and no DEST arg will list the source files instead
 of copying.

-The online version of this man page (that includes cross-linking of topics)
+The online version of this manpage (that includes cross-linking of topics)
 is available at <https://download.samba.org/pub/rsync/rsync.1>.

 ## DESCRIPTION
@@ -276,7 +276,7 @@ In order to connect to an rsync daemon, the remote system needs to have a
 daemon already running (or it needs to have configured something like inetd to
 spawn an rsync daemon for incoming connections on a particular port).  For full
 information on how to start a daemon that will handling incoming socket
-connections, see the [**rsyncd.conf**(5)](rsyncd.conf.5) man page -- that is
+connections, see the [**rsyncd.conf**(5)](rsyncd.conf.5) manpage -- that is
 the config file for the daemon, and it contains the full details for how to run
 the daemon (including stand-alone and inetd configurations).

@@ -329,7 +329,7 @@ This is launched from cron every few hours.
 ## OPTION SUMMARY

 Here is a short summary of the options available in rsync.  Each option also
-has its own detailed description later in this man page.
+has its own detailed description later in this manpage.

 [comment]: # (help-rsync.h)
 [comment]: # (Keep these short enough that they'll be under 80 chars when indented by 7 chars.)
@@ -373,6 +373,8 @@ has its own detailed description later in this man page.
 --owner, -o              preserve owner (super-user only)
 --group, -g              preserve group
 --devices                preserve device files (super-user only)
+--copy-devices           copy device contents as a regular file
+--write-devices          write to devices as files (implies --inplace)
 --specials               preserve special files
 -D                       same as --devices --specials
 --times, -t              preserve modification times
@@ -385,7 +387,6 @@ has its own detailed description later in this man page.
 --fake-super             store/recover privileged attrs using xattrs
 --sparse, -S             turn sequences of nulls into sparse blocks
 --preallocate            allocate dest files before writing them
--write-devices          write to devices as files (implies --inplace)
 --dry-run, -n            perform a trial run with no changes made
 --whole-file, -W         copy files whole (w/o delta-xfer algorithm)
 --checksum-choice=STR    choose the checksum algorithm (aka --cc)
@@ -797,7 +798,7 @@ your home directory (remove the '=' for that).
    One side-effect of incremental recursion is that any missing
    sub-directories inside a recursively-scanned directory are (by default)
    created prior to recursing into the sub-dirs.  This earlier creation point
-    (commpared to a non-incremental recursion) allows rsync to then set the
+    (compared to a non-incremental recursion) allows rsync to then set the
    modify time of the finished directory right away (without having to delay
    that until a bunch of recursive copying has finished).  However, these
    early directories don't yet have their completed mode, mtime, or ownership
@@ -954,9 +955,9 @@ your home directory (remove the '=' for that).
    A caution for those that choose to combine [`--inplace`](#opt) with
    `--update`: an interrupted transfer will leave behind a partial file on the
    receiving side that has a very recent modified time, so re-running the
-    transfer will probably **not** continue the interrutped file.  As such, it
+    transfer will probably **not** continue the interrupted file.  As such, it
    is usually best to avoid combining this with[ `--inplace`](#opt) unless you
-    have implemented manual steps to handle any interrutped in-progress files.
+    have implemented manual steps to handle any interrupted in-progress files.

 0.  `--inplace`

@@ -1473,6 +1474,14 @@ your home directory (remove the '=' for that).
    The `-D` option is equivalent to "[`--devices`](#opt)
    [`--specials`](#opt)".

+0.  `--copy-devices`
+
+    This tells rsync to treat a device on the sending side as a regular file,
+    allowing it to be copied to a normal destination file (or another device
+    if `--write-devices` was also specifed).
+
+    This option is refused by default by an rsync daemon.
+
 0.  `--write-devices`

    This tells rsync to treat a device on the receiving side as a regular file,
@@ -1481,9 +1490,9 @@ your home directory (remove the '=' for that).
    This option implies the [`--inplace`](#opt) option.

    Be careful using this, as you should know what devices are present on the
-    receiving side of the transfer, especially if running rsync as root.
+    receiving side of the transfer, especially when running rsync as root.

-    This option is refused by an rsync daemon.
+    This option is refused by default by an rsync daemon.

 0.  `--times`, `-t`

@@ -1643,7 +1652,7 @@ your home directory (remove the '=' for that).

 0. `--no-whole-file`, `--no-W`

-    Disable whole-file updating when it is enaled by default for a local
+    Disable whole-file updating when it is enabled by default for a local
    transfer.  This usually slows rsync down, but it can be useful if you are
    trying to minimize the writes to the destination file (if combined with
    [`--inplace`](#opt)) or for testing the checksum-based update algorithm.
@@ -2871,7 +2880,7 @@ your home directory (remove the '=' for that).

    This option can provide endless fun for people who like to tune their
    systems to the utmost degree.  You can set all sorts of socket options
-    which may make transfers faster (or slower!).  Read the man page for the
+    which may make transfers faster (or slower!).  Read the manpage for the
    `setsockopt()` system call for details on some of the options you may be
    able to set.  By default no special socket options are set.  This only
    affects direct socket connections to a remote rsync daemon.
@@ -3565,7 +3574,7 @@ your home directory (remove the '=' for that).
    socket when directly contacting an rsync daemon, as well as the forwarding
    of the `-4` or `-6` option to ssh when rsync can deduce that ssh is being
    used as the remote shell.  For other remote shells you'll need to specify
-    the "`--rsh SHELL -4`" option directly (or whatever ipv4/ipv6 hint options
+    the "`--rsh SHELL -4`" option directly (or whatever IPv4/IPv6 hint options
    it uses).

    See also [the daemon version of these options](#dopt--ipv4).
@@ -3600,7 +3609,7 @@ The options allowed when starting an rsync daemon are as follows:
    background daemon.  The daemon will read the config file (rsyncd.conf) on
    each connect made by a client and respond to requests accordingly.

-    See the [**rsyncd.conf**(5)](rsyncd.conf.5) man page for more details.
+    See the [**rsyncd.conf**(5)](rsyncd.conf.5) manpage for more details.

 0.  `--address=ADDRESS`

@@ -3723,6 +3732,14 @@ first matching pattern is acted on: if it is an exclude pattern, then that file
 is skipped; if it is an include pattern then that filename is not skipped; if
 no matching pattern is found, then the filename is not skipped.

+Aside: because the interactions of filter rules can be complex, it is useful to
+use the `--debug=FILTER` option if things aren't working the way you expect.
+The level-1 output (the default if no level number is specified) mentions the
+filter rule that is first matched by each file in the transfer.  It also warns
+if a filter rule has trailing whitespace.  The level-2 output mentions a lot
+more filter events, including the definition of each rule and the handling of
+per-directory filter files.
+
 Rsync builds an ordered list of filter rules as specified on the command-line.
 Filter rules have the following syntax:

@@ -4429,7 +4446,7 @@ file is included or excluded.

 0. `RSYNC_PORT`

-    This environment variable does is not read by rsync, but is instead set in
+    This environment variable is not read by rsync, but is instead set in
    its sub-environment when rsync is running the remote shell in combination
    with a daemon connection.  This allows a script such as
    [`rsync-ssl`](rsync-ssl.1) to be able to know the port number that the user
@@ -4472,7 +4489,7 @@ Please report bugs! See the web site at <https://rsync.samba.org/>.

 ## VERSION

-This man page is current for version @VERSION@ of rsync.
+This manpage is current for version @VERSION@ of rsync.

 ## INTERNAL OPTIONS

--- a/rsync.h
+++ b/rsync.h
@@ -18,11 +18,6 @@
 * with this program; if not, visit the http://fsf.org website.
 */

-/* a non-zero CHAR_OFFSET makes the rolling sum stronger, but is
-   incompatible with older versions :-( */
-#define CHAR_OFFSET 0
-
-#ifndef AVX2_ASM /* do not include the rest of file for assembly */
 #define False 0
 #define True 1
 #define Unset (-1) /* Our BOOL values are always an int. */
@@ -43,6 +38,9 @@

 #define BACKUP_SUFFIX "~"

+/* a non-zero CHAR_OFFSET makes the rolling sum stronger, but is
+   incompatible with older versions :-( */
+#define CHAR_OFFSET 0

 /* These flags are only used during the flist transfer. */

@@ -1174,6 +1172,10 @@ struct name_num_obj {
 	struct name_num_item list[10]; /* we'll get a compile error/warning if this is ever too small */
 };

+#ifdef EXTERNAL_ZLIB
+#define read_buf read_buf_
+#endif
+
 #ifndef __cplusplus
 #include "proto.h"
 #endif
@@ -1477,7 +1479,6 @@ const char *get_panic_action(void);
    fprintf(stderr, "%s in %s at line %d\n", msg, __FILE__, __LINE__); \
    exit_cleanup(RERR_UNSUPPORTED); \
 } while (0)
-#endif  /* AVX2_ASM */

 #ifdef HAVE_MALLINFO2
 #define MEM_ALLOC_INFO mallinfo2
--- a/rsyncd.conf.5.md
+++ b/rsyncd.conf.5.md
@@ -6,7 +6,7 @@ rsyncd.conf - configuration file for rsync in daemon mode

 rsyncd.conf

-The online version of this man page (that includes cross-linking of topics)
+The online version of this manpage (that includes cross-linking of topics)
 is available at <https://download.samba.org/pub/rsync/rsyncd.conf.5>.

 ## DESCRIPTION
@@ -98,9 +98,9 @@ a literal % into a value is to use %%.

 0.  `motd file`

-    This parameter allows you to specify a "message of the day" to display to
-    clients on each connect. This usually contains site information and any
-    legal notices. The default is no motd file.  This can be overridden by the
+    This parameter allows you to specify a "message of the day" (MOTD) to display
+    to clients on each connect. This usually contains site information and any
+    legal notices. The default is no MOTD file.  This can be overridden by the
    `--dparam=motdfile=FILE` command-line option when starting the daemon.

 0.  `pid file`
@@ -128,7 +128,7 @@ a literal % into a value is to use %%.

    This parameter can provide endless fun for people who like to tune their
    systems to the utmost degree. You can set all sorts of socket options which
-    may make transfers faster (or slower!). Read the man page for the
+    may make transfers faster (or slower!). Read the manpage for the
    **setsockopt()** system call for details on some of the options you may be
    able to set. By default no special socket options are set.  These settings
    can also be specified via the `--sockopts` command-line option.
@@ -716,7 +716,7 @@ the values of parameters.  See the GLOBAL PARAMETERS section for more details.
      addresses which match the masked IP address will be allowed in.
    - a hostname pattern using wildcards. If the hostname of the connecting IP
      (as determined by a reverse lookup) matches the wildcarded name (using
-      the same rules as normal unix filename matching), the client is allowed
+      the same rules as normal Unix filename matching), the client is allowed
      in.  This only works if "[reverse lookup](#)" is enabled (the default).
    - a hostname. A plain hostname is matched against the reverse DNS of the
      connecting IP (if "[reverse lookup](#)" is enabled), and/or the IP of the
@@ -933,9 +933,10 @@ the values of parameters.  See the GLOBAL PARAMETERS section for more details.
    If you are un-refusing the compress option, you may want to match
    "`!compress*`" if you also want to allow the `--compress-level` option.

-    Note that the "write-devices" option is refused by default, but can be
-    explicitly accepted with "`!write-devices`".  The options "log-file" and
-    "log-file-format" are forcibly refused and cannot be accepted.
+    Note that the "copy-devices" & "write-devices" options are refused by
+    default, but they can be explicitly accepted with "`!copy-devices`" and/or
+    "`!write-devices`".  The options "log-file" and "log-file-format" are
+    forcibly refused and cannot be accepted.

    Here are all the options that are not matched by wild-cards:

@@ -1219,7 +1220,7 @@ Please report bugs! The rsync bug tracking system is online at

 ## VERSION

-This man page is current for version @VERSION@ of rsync.
+This manpage is current for version @VERSION@ of rsync.

 ## CREDITS

--- a/sender.c
+++ b/sender.c
@@ -37,6 +37,7 @@ extern int io_error;
 extern int flist_eof;
 extern int whole_file;
 extern int allowed_lull;
+extern int copy_devices;
 extern int preserve_xattrs;
 extern int protocol_version;
 extern int remove_source_files;
@@ -366,6 +367,15 @@ void send_files(int f_in, int f_out)
 			exit_cleanup(RERR_FILEIO);
 		}

+		if (IS_DEVICE(st.st_mode)) {
+			if (!copy_devices) {
+				rprintf(FERROR, "attempt to copy device contents without --copy-devices\n");
+				exit_cleanup(RERR_PROTOCOL);
+			}
+			if (st.st_size == 0)
+				st.st_size = get_device_size(fd, fname);
+		}
+
 		if (append_mode > 0 && st.st_size < F_LENGTH(file)) {
 			rprintf(FWARNING, "skipped diminished file: %s\n",
 				full_fname(fname));
--- a/simd-checksum-avx2.S
+++ b/simd-checksum-avx2.S
@@ -1,15 +1,21 @@
+#include "config.h"
+
+#ifdef USE_ROLL_ASM /* { */
+
+#define CHAR_OFFSET 0 /* Keep this the same as rsync.h, which isn't likely to change. */
+
 #ifdef __APPLE__
-#define get_checksum1_avx2  _get_checksum1_avx2
+#define get_checksum1_avx2_asm  _get_checksum1_avx2_asm
 #endif

 .intel_syntax noprefix
 .text

 	.p2align 5
-	.globl get_checksum1_avx2
+	.globl get_checksum1_avx2_asm

 # rdi=*buf, esi=len, edx=i, rcx= *ps1, r8= *ps2
-get_checksum1_avx2:
+get_checksum1_avx2_asm:
 	vmovd	xmm6,[rcx] # load *ps1
 	lea	eax, [rsi-128] # at least 128 bytes to process?
 	cmp	edx, eax
@@ -167,3 +173,5 @@ get_checksum1_avx2:
 	.byte 3
 	.byte 2
 	.byte 1
+
+#endif /* } USE_ROLL_ASM */
--- a/simd-checksum-x86_64.cpp
+++ b/simd-checksum-x86_64.cpp
@@ -51,12 +51,12 @@
 * GCC 4.x are not supported to ease configure.ac logic.
 */

-#ifdef __x86_64__
-#ifdef __cplusplus
+#ifdef __x86_64__ /* { */
+#ifdef __cplusplus /* { */

 #include "rsync.h"

-#ifdef HAVE_SIMD
+#ifdef USE_ROLL_SIMD /* { */

 #include <immintrin.h>

@@ -85,6 +85,9 @@ typedef long long __m256i_u __attribute__((__vector_size__(32), __may_alias__, _
 #define SSE2_HADDS_EPI16(a, b) _mm_adds_epi16(SSE2_INTERLEAVE_EVEN_EPI16(a, b), SSE2_INTERLEAVE_ODD_EPI16(a, b))
 #define SSE2_MADDUBS_EPI16(a, b) _mm_adds_epi16(SSE2_MULU_EVEN_EPI8(a, b), SSE2_MULU_ODD_EPI8(a, b))

+#ifndef USE_ROLL_ASM
+__attribute__ ((target("default"))) MVSTATIC int32 get_checksum1_avx2_64(schar* buf, int32 len, int32 i, uint32* ps1, uint32* ps2) { return i; }
+#endif
 __attribute__ ((target("default"))) MVSTATIC int32 get_checksum1_ssse3_32(schar* buf, int32 len, int32 i, uint32* ps1, uint32* ps2) { return i; }
 __attribute__ ((target("default"))) MVSTATIC int32 get_checksum1_sse2_32(schar* buf, int32 len, int32 i, uint32* ps1, uint32* ps2) { return i; }

@@ -245,7 +248,7 @@ __attribute__ ((target("sse2"))) MVSTATIC int32 get_checksum1_sse2_32(schar* buf

            // (4*buf[i] + 3*buf[i+1]), (2*buf[i+2], buf[i+3]), ... 2*[int16*8]
            __m128i mul_const = _mm_set1_epi32(4 + (3 << 8) + (2 << 16) + (1 << 24));
-            __m128i mul_add16_1 = SSE2_MADDUBS_EPI16(mul_const, in8_1);
+            __m128i mul_add16_1 = SSE2_MADDUBS_EPI16(mul_const, in8_1);
            __m128i mul_add16_2 = SSE2_MADDUBS_EPI16(mul_const, in8_2);

            // s2 += 32*s1
@@ -310,7 +313,127 @@ __attribute__ ((target("sse2"))) MVSTATIC int32 get_checksum1_sse2_32(schar* buf
    return i;
 }

-extern "C" __attribute__ ((target("avx2"))) int32 get_checksum1_avx2(schar* buf, int32 len, int32 i, uint32* ps1, uint32* ps2);
+#ifdef USE_ROLL_ASM /* { */
+
+extern "C" __attribute__ ((target("avx2"))) int32 get_checksum1_avx2_asm(schar* buf, int32 len, int32 i, uint32* ps1, uint32* ps2);
+
+#else /* } { */
+
+/*
+  AVX2 loop per 64 bytes:
+    int16 t1[16];
+    int16 t2[16];
+    for (int j = 0; j < 16; j++) {
+      t1[j] = buf[j*4 + i] + buf[j*4 + i+1] + buf[j*4 + i+2] + buf[j*4 + i+3];
+      t2[j] = 4*buf[j*4 + i] + 3*buf[j*4 + i+1] + 2*buf[j*4 + i+2] + buf[j*4 + i+3];
+    }
+    s2 += 64*s1 + (uint32)(
+              60*t1[0] + 56*t1[1] + 52*t1[2] + 48*t1[3] + 44*t1[4] + 40*t1[5] + 36*t1[6] + 32*t1[7] + 28*t1[8] + 24*t1[9] + 20*t1[10] + 16*t1[11] + 12*t1[12] + 8*t1[13] + 4*t1[14] +
+              t2[0] + t2[1] + t2[2] + t2[3] + t2[4] + t2[5] + t2[6] + t2[7] + t2[8] + t2[9] + t2[10] + t2[11] + t2[12] + t2[13] + t2[14] + t2[15]
+          ) + 2080*CHAR_OFFSET;
+    s1 += (uint32)(t1[0] + t1[1] + t1[2] + t1[3] + t1[4] + t1[5] + t1[6] + t1[7] + t1[8] + t1[9] + t1[10] + t1[11] + t1[12] + t1[13] + t1[14] + t1[15]) +
+          64*CHAR_OFFSET;
+ */
+
+__attribute__ ((target("avx2"))) MVSTATIC int32 get_checksum1_avx2_64(schar* buf, int32 len, int32 i, uint32* ps1, uint32* ps2)
+{
+    if (len > 64) {
+
+        uint32 x[4] = {0};
+        __m128i ss1 = _mm_cvtsi32_si128(*ps1);
+        __m128i ss2 = _mm_cvtsi32_si128(*ps2);
+
+        const char mul_t1_buf[16] = {60, 56, 52, 48, 44, 40, 36, 32, 28, 24, 20, 16, 12, 8, 4, 0};
+	__m128i tmp = _mm_load_si128((__m128i*) mul_t1_buf);
+        __m256i mul_t1 = _mm256_cvtepu8_epi16(tmp);
+	__m256i mul_const = _mm256_broadcastd_epi32(_mm_cvtsi32_si128(4 | (3 << 8) | (2 << 16) | (1 << 24)));
+        __m256i mul_one;
+       	    mul_one = _mm256_abs_epi8(_mm256_cmpeq_epi16(mul_one,mul_one)); // set all vector elements to 1
+
+        for (; i < (len-64); i+=64) {
+            // Load ... 4*[int8*16]
+            __m256i in8_1, in8_2;
+	    __m128i in8_1_low, in8_2_low, in8_1_high, in8_2_high;
+	    in8_1_low = _mm_loadu_si128((__m128i_u*)&buf[i]);
+	    in8_2_low = _mm_loadu_si128((__m128i_u*)&buf[i+16]);
+	    in8_1_high = _mm_loadu_si128((__m128i_u*)&buf[i+32]);
+	    in8_2_high = _mm_loadu_si128((__m128i_u*)&buf[i+48]);
+	    in8_1 = _mm256_inserti128_si256(_mm256_castsi128_si256(in8_1_low), in8_1_high,1);
+	    in8_2 = _mm256_inserti128_si256(_mm256_castsi128_si256(in8_2_low), in8_2_high,1);
+
+            // (1*buf[i] + 1*buf[i+1]), (1*buf[i+2], 1*buf[i+3]), ... 2*[int16*8]
+            // Fastest, even though multiply by 1
+            __m256i add16_1 = _mm256_maddubs_epi16(mul_one, in8_1);
+            __m256i add16_2 = _mm256_maddubs_epi16(mul_one, in8_2);
+
+            // (4*buf[i] + 3*buf[i+1]), (2*buf[i+2], buf[i+3]), ... 2*[int16*8]
+            __m256i mul_add16_1 = _mm256_maddubs_epi16(mul_const, in8_1);
+            __m256i mul_add16_2 = _mm256_maddubs_epi16(mul_const, in8_2);
+
+            // s2 += 64*s1
+            ss2 = _mm_add_epi32(ss2, _mm_slli_epi32(ss1, 6));
+
+            // [sum(t1[0]..t1[7]), X, X, X] [int32*4]; faster than multiple _mm_hadds_epi16
+            __m256i sum_add32 = _mm256_add_epi16(add16_1, add16_2);
+            sum_add32 = _mm256_add_epi16(sum_add32, _mm256_srli_epi32(sum_add32, 16));
+            sum_add32 = _mm256_add_epi16(sum_add32, _mm256_srli_si256(sum_add32, 4));
+            sum_add32 = _mm256_add_epi16(sum_add32, _mm256_srli_si256(sum_add32, 8));
+
+            // [sum(t2[0]..t2[7]), X, X, X] [int32*4]; faster than multiple _mm_hadds_epi16
+            __m256i sum_mul_add32 = _mm256_add_epi16(mul_add16_1, mul_add16_2);
+            sum_mul_add32 = _mm256_add_epi16(sum_mul_add32, _mm256_srli_epi32(sum_mul_add32, 16));
+            sum_mul_add32 = _mm256_add_epi16(sum_mul_add32, _mm256_srli_si256(sum_mul_add32, 4));
+            sum_mul_add32 = _mm256_add_epi16(sum_mul_add32, _mm256_srli_si256(sum_mul_add32, 8));
+
+            // s1 += t1[0] + t1[1] + t1[2] + t1[3] + t1[4] + t1[5] + t1[6] + t1[7]
+	    __m128i sum_add32_hi = _mm256_extracti128_si256(sum_add32, 0x1);
+            ss1 = _mm_add_epi32(ss1, _mm256_castsi256_si128(sum_add32));
+            ss1 = _mm_add_epi32(ss1, sum_add32_hi);
+
+            // s2 += t2[0] + t2[1] + t2[2] + t2[3] + t2[4] + t2[5] + t2[6] + t2[7]
+	    __m128i sum_mul_add32_hi = _mm256_extracti128_si256(sum_mul_add32, 0x1);
+            ss2 = _mm_add_epi32(ss2, _mm256_castsi256_si128(sum_mul_add32));
+            ss2 = _mm_add_epi32(ss2, sum_mul_add32_hi);
+
+            // [t1[0] + t1[1], t1[2] + t1[3] ...] [int16*8]
+            // We could've combined this with generating sum_add32 above and
+            // save an instruction but benchmarking shows that as being slower
+            __m256i add16 = _mm256_hadds_epi16(add16_1, add16_2);
+
+            // [t1[0], t1[1], ...] -> [t1[0]*28 + t1[1]*24, ...] [int32*4]
+            __m256i mul32 = _mm256_madd_epi16(add16, mul_t1);
+
+            // [sum(mul32), X, X, X] [int32*4]; faster than multiple _mm_hadd_epi32
+            mul32 = _mm256_add_epi32(mul32, _mm256_srli_si256(mul32, 4));
+            mul32 = _mm256_add_epi32(mul32, _mm256_srli_si256(mul32, 8));
+	    // prefetch 2 cacheline ahead
+            _mm_prefetch(&buf[i + 160], _MM_HINT_T0);
+
+            // s2 += 28*t1[0] + 24*t1[1] + 20*t1[2] + 16*t1[3] + 12*t1[4] + 8*t1[5] + 4*t1[6]
+	    __m128i mul32_hi = _mm256_extracti128_si256(mul32, 0x1);
+            ss2 = _mm_add_epi32(ss2, _mm256_castsi256_si128(mul32));
+            ss2 = _mm_add_epi32(ss2, mul32_hi);
+
+#if CHAR_OFFSET != 0
+            // s1 += 32*CHAR_OFFSET
+            __m128i char_offset_multiplier = _mm_set1_epi32(32 * CHAR_OFFSET);
+            ss1 = _mm_add_epi32(ss1, char_offset_multiplier);
+
+            // s2 += 528*CHAR_OFFSET
+            char_offset_multiplier = _mm_set1_epi32(528 * CHAR_OFFSET);
+            ss2 = _mm_add_epi32(ss2, char_offset_multiplier);
+#endif
+        }
+
+        _mm_store_si128((__m128i_u*)x, ss1);
+        *ps1 = x[0];
+        _mm_store_si128((__m128i_u*)x, ss2);
+        *ps2 = x[0];
+    }
+    return i;
+}
+
+#endif /* } !USE_ROLL_ASM */

 static int32 get_checksum1_default_1(schar* buf, int32 len, int32 i, uint32* ps1, uint32* ps2)
 {
@@ -338,7 +461,11 @@ static inline uint32 get_checksum1_cpp(char *buf1, int32 len)
    uint32 s2 = 0;

    // multiples of 64 bytes using AVX2 (if available)
-    i = get_checksum1_avx2((schar*)buf1, len, i, &s1, &s2);
+#ifdef USE_ROLL_ASM
+    i = get_checksum1_avx2_asm((schar*)buf1, len, i, &s1, &s2);
+#else
+    i = get_checksum1_avx2_64((schar*)buf1, len, i, &s1, &s2);
+#endif

    // multiples of 32 bytes using SSSE3 (if available)
    i = get_checksum1_ssse3_32((schar*)buf1, len, i, &s1, &s2);
@@ -407,7 +534,11 @@ int main() {
    benchmark("Raw-C", get_checksum1_default_1, (schar*)buf, BLOCK_LEN);
    benchmark("SSE2", get_checksum1_sse2_32, (schar*)buf, BLOCK_LEN);
    benchmark("SSSE3", get_checksum1_ssse3_32, (schar*)buf, BLOCK_LEN);
-    benchmark("AVX2", get_checksum1_avx2, (schar*)buf, BLOCK_LEN);
+#ifdef USE_ROLL_ASM
+    benchmark("AVX2-ASM", get_checksum1_avx2_asm, (schar*)buf, BLOCK_LEN);
+#else
+    benchmark("AVX2", get_checksum1_avx2_64, (schar*)buf, BLOCK_LEN);
+#endif

    free(buf);
    return 0;
@@ -417,6 +548,6 @@ int main() {
 #pragma clang optimize on
 #endif /* BENCHMARK_SIMD_CHECKSUM1 */

-#endif /* HAVE_SIMD */
-#endif /* __cplusplus */
-#endif /* __x86_64__ */
+#endif /* } USE_ROLL_SIMD */
+#endif /* } __cplusplus */
+#endif /* } __x86_64__ */
--- a/support/git-set-file-times
+++ b/support/git-set-file-times
@@ -38,7 +38,7 @@ def main():
                print_line(fn, mtime, mtime)
            ls.discard(fn)

-    cmd = git + 'log -r --name-only --no-color --pretty=raw --no-renames -z'.split()
+    cmd = git + 'log -r --name-only --format=%x00commit%x20%H%n%x00commit_time%x20%ct%n --no-renames -z'.split()
    if args.tree:
        cmd.append(args.tree)
    cmd += ['--'] + args.files
@@ -46,7 +46,7 @@ def main():
    proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, encoding='utf-8')
    for line in proc.stdout:
        line = line.strip()
-        m = re.match(r'^committer .*? (\d+) [-+]\d+$', line)
+        m = re.match(r'^\0commit_time (\d+)$', line)
        if m:
            commit_time = int(m[1])
        elif NULL_COMMIT_RE.search(line):
--- a/syscall.c
+++ b/syscall.c
@@ -247,7 +247,7 @@ int do_chmod(const char *path, mode_t mode)
 		else if (errno != ENOTSUP)
 			break;
 #endif
-
+		/* FALLTHROUGH */
 	default:
 		if (S_ISLNK(mode)) {
 # if defined HAVE_SETATTRLIST
--- a/testsuite/00-hello.test
+++ b/testsuite/00-hello.test
@@ -29,7 +29,7 @@ append_line test1
 checkit "$RSYNC -ai '$fromdir/' '$todir/'" "$fromdir" "$todir"

 copy_weird() {
-    checkit "$RSYNC $1 \"$2$fromdir/$weird_name/\" \"$3$todir/$weird_name\"" "$fromdir" "$todir"
+    checkit "$RSYNC $1 --rsync-path='$RSYNC' '$2$fromdir/$weird_name/' '$3$todir/$weird_name'" "$fromdir" "$todir"
 }

 append_line test2
@@ -47,7 +47,7 @@ copy_weird '-ais' '' 'lh:'
 echo test6

 touch "$fromdir/one" "$fromdir/two"
-(cd "$fromdir" && $RSYNC -ai --old-args lh:'one two' "$todir/")
+(cd "$fromdir" && $RSYNC -ai --old-args --rsync-path="$RSYNC" lh:'one two' "$todir/")
 if [ ! -f "$todir/one" ] || [ ! -f "$todir/two" ]; then
    test_fail "old-args copy of 'one two' failed"
 fi
@@ -55,7 +55,7 @@ fi
 echo test7

 rm "$todir/one" "$todir/two"
-(cd "$fromdir" && RSYNC_OLD_ARGS=1 $RSYNC -ai lh:'one two' "$todir/")
+(cd "$fromdir" && RSYNC_OLD_ARGS=1 $RSYNC -ai --rsync-path="$RSYNC" lh:'one two' "$todir/")

 # The script would have aborted on error, so getting here means we've won.
 exit 0
--- a/testsuite/alt-dest.test
+++ b/testsuite/alt-dest.test
@@ -0,0 +1,68 @@
+#!/bin/sh
+
+# Copyright (C) 2004-2022 Wayne Davison
+
+# This program is distributable under the terms of the GNU GPL (see
+# COPYING).
+
+# Test rsync handling of --compare-dest and similar options.
+
+. "$suitedir/rsync.fns"
+
+alt1dir="$tmpdir/alt1"
+alt2dir="$tmpdir/alt2"
+alt3dir="$tmpdir/alt3"
+
+SSH="$scratchdir/src/support/lsh.sh"
+
+# Build some files/dirs/links to copy
+
+hands_setup
+
+# Setup the alt and chk dirs
+$RSYNC -av --include=text --include='*/' --exclude='*' "$fromdir/" "$alt1dir/"
+$RSYNC -av --include=etc-ltr-list --include='*/' --exclude='*' "$fromdir/" "$alt2dir/"
+
+# Create a side dir where there is a candidate destfile of the same name as a sourcefile
+echo "This is a test file" >"$fromdir/likely"
+
+mkdir "$alt3dir"
+echo "This is a test file" >"$alt3dir/likely"
+
+sleep 1
+touch "$fromdir/dir/text" "$fromdir/likely"
+
+$RSYNC -av --exclude=/text --exclude=etc-ltr-list "$fromdir/" "$chkdir/"
+
+# Let's do it!
+checkit "$RSYNC -avv --no-whole-file \
+    --compare-dest='$alt1dir' --compare-dest='$alt2dir' \
+    '$fromdir/' '$todir/'" "$chkdir" "$todir"
+
+rm -rf "$todir"
+checkit "$RSYNC -avv --no-whole-file \
+    --copy-dest='$alt1dir' --copy-dest='$alt2dir' \
+    '$fromdir/' '$todir/'" "$fromdir" "$todir"
+
+# Test that copy_file() works correctly with tmpfiles
+for maybe_inplace in '' --inplace; do
+    rm -rf "$todir"
+    checkit "$RSYNC -av $maybe_inplace --copy-dest='$alt3dir' \
+	'$fromdir/' '$todir/'" "$fromdir" "$todir"
+
+    for srchost in '' 'localhost:'; do
+	if [ -z "$srchost" ]; then
+	    desthost='localhost:'
+	else
+	    desthost=''
+	fi
+
+	rm -rf "$todir"
+	checkit "$RSYNC -ave '$SSH' --rsync-path='$RSYNC' $maybe_inplace \
+	    --copy-dest='$alt3dir' '$srchost$fromdir/' '$desthost$todir/'" \
+	    "$fromdir" "$todir"
+    done
+done
+
+# The script would have aborted on error, so getting here means we've won.
+exit 0
--- a/testsuite/compare-dest.test
+++ b/testsuite/compare-dest.test
@@ -1,37 +0,0 @@
-#!/bin/sh
-
-# Copyright (C) 2004-2022 Wayne Davison
-
-# This program is distributable under the terms of the GNU GPL (see
-# COPYING).
-
-# Test rsync handling of the --compare-dest option.
-
-. "$suitedir/rsync.fns"
-
-alt1dir="$tmpdir/alt1"
-alt2dir="$tmpdir/alt2"
-
-# Build some files/dirs/links to copy
-
-hands_setup
-
-# Setup the alt and chk dirs
-$RSYNC -av --include=text --include='*/' --exclude='*' "$fromdir/" "$alt1dir/"
-$RSYNC -av --include=etc-ltr-list --include='*/' --exclude='*' "$fromdir/" "$alt2dir/"
-
-sleep 1
-touch "$fromdir/dir/text"
-
-$RSYNC -av --exclude=/text --exclude=etc-ltr-list "$fromdir/" "$chkdir/"
-
-# Let's do it!
-checkit "$RSYNC -avv --no-whole-file \
-    --compare-dest='$alt1dir' --compare-dest='$alt2dir' \
-    '$fromdir/' '$todir/'" "$chkdir" "$todir"
-checkit "$RSYNC -avv --no-whole-file \
-    --copy-dest='$alt1dir' --copy-dest='$alt2dir' \
-    '$fromdir/' '$todir/'" "$fromdir" "$todir"
-
-# The script would have aborted on error, so getting here means we've won.
-exit 0
--- a/usage.c
+++ b/usage.c
@@ -1,7 +1,7 @@
 /*
 * Some usage & version related functions.
 *
- * Copyright (C) 2002-2020 Wayne Davison
+ * Copyright (C) 2002-2022 Wayne Davison
 *
 * 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
@@ -139,21 +139,26 @@ static void print_info_flags(enum logcode f)

 	"*Optimizations",

-#ifndef HAVE_SIMD
+#ifndef USE_ROLL_SIMD
 		"no "
 #endif
-			"SIMD",
+			"SIMD-roll",

-#ifndef HAVE_ASM
+#ifndef USE_ROLL_ASM
 		"no "
 #endif
-			"asm",
+			"asm-roll",

 #ifndef USE_OPENSSL
 		"no "
 #endif
 			"openssl-crypto",

+#ifndef USE_MD5_ASM
+		"no "
+#endif
+			"asm-MD5",
+
 		NULL
 	};

@@ -240,7 +245,7 @@ void usage(enum logcode F)
 #include "help-rsync.h"
  rprintf(F,"\n");
  rprintf(F,"Use \"rsync --daemon --help\" to see the daemon-mode command-line options.\n");
-  rprintf(F,"Please see the rsync(1) and rsyncd.conf(5) man pages for full documentation.\n");
+  rprintf(F,"Please see the rsync(1) and rsyncd.conf(5) manpages for full documentation.\n");
  rprintf(F,"See https://rsync.samba.org/ for updates, bug reports, and answers\n");
 }

@@ -253,7 +258,7 @@ void daemon_usage(enum logcode F)
 #include "help-rsyncd.h"
  rprintf(F,"\n");
  rprintf(F,"If you were not trying to invoke rsync as a daemon, avoid using any of the\n");
-  rprintf(F,"daemon-specific rsync options.  See also the rsyncd.conf(5) man page.\n");
+  rprintf(F,"daemon-specific rsync options.  See also the rsyncd.conf(5) manpage.\n");
 }

 const char *rsync_version(void)
--- a/util1.c
+++ b/util1.c
@@ -320,16 +320,48 @@ static int safe_read(int desc, char *ptr, size_t len)
 	return n_chars;
 }

-/* Copy a file.  If ofd < 0, copy_file unlinks and opens the "dest" file.
- * Otherwise, it just writes to and closes the provided file descriptor.
+/* Remove existing file @dest and reopen, creating a new file with @mode */
+static int unlink_and_reopen(const char *dest, mode_t mode)
+{
+	int ofd;
+
+	if (robust_unlink(dest) && errno != ENOENT) {
+		int save_errno = errno;
+		rsyserr(FERROR_XFER, errno, "unlink %s", full_fname(dest));
+		errno = save_errno;
+		return -1;
+	}
+
+#ifdef SUPPORT_XATTRS
+	if (preserve_xattrs)
+		mode |= S_IWUSR;
+#endif
+	mode &= INITACCESSPERMS;
+	if ((ofd = do_open(dest, O_WRONLY | O_CREAT | O_TRUNC | O_EXCL, mode)) < 0) {
+		int save_errno = errno;
+		rsyserr(FERROR_XFER, save_errno, "open %s", full_fname(dest));
+		errno = save_errno;
+		return -1;
+	}
+	return ofd;
+}
+
+/* Copy contents of file @source to file @dest with mode @mode.
+ *
+ * If @tmpfilefd is < 0, copy_file unlinks @dest and then opens a new
+ * file with name @dest.
+ *
+ * Otherwise, copy_file writes to and closes the provided file
+ * descriptor.
+ *
 * In either case, if --xattrs are being preserved, the dest file will
 * have its xattrs set from the source file.
 *
 * This is used in conjunction with the --temp-dir, --backup, and
 * --copy-dest options. */
-int copy_file(const char *source, const char *dest, int ofd, mode_t mode)
+int copy_file(const char *source, const char *dest, int tmpfilefd, mode_t mode)
 {
-	int ifd;
+	int ifd, ofd;
 	char buf[1024 * 8];
 	int len;   /* Number of bytes read into `buf'. */
 	OFF_T prealloc_len = 0, offset = 0;
@@ -341,23 +373,12 @@ int copy_file(const char *source, const char *dest, int ofd, mode_t mode)
 		return -1;
 	}

-	if (ofd < 0) {
-		if (robust_unlink(dest) && errno != ENOENT) {
+	if (tmpfilefd >= 0) {
+		ofd = tmpfilefd;
+	} else {
+		ofd = unlink_and_reopen(dest, mode);
+		if (ofd < 0) {
 			int save_errno = errno;
-			rsyserr(FERROR_XFER, errno, "unlink %s", full_fname(dest));
-			close(ifd);
-			errno = save_errno;
-			return -1;
-		}
-
-#ifdef SUPPORT_XATTRS
-		if (preserve_xattrs)
-			mode |= S_IWUSR;
-#endif
-		mode &= INITACCESSPERMS;
-		if ((ofd = do_open(dest, O_WRONLY | O_CREAT | O_TRUNC | O_EXCL, mode)) < 0) {
-			int save_errno = errno;
-			rsyserr(FERROR_XFER, save_errno, "open %s", full_fname(dest));
 			close(ifd);
 			errno = save_errno;
 			return -1;
--- a/version.h
+++ b/version.h
@@ -1 +1 @@
-#define RSYNC_VERSION "3.2.4pre3"
+#define RSYNC_VERSION "3.2.4"
--- a/zlib/deflate.c
+++ b/zlib/deflate.c
@@ -227,11 +227,6 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy,
    int wrap = 1;
    static const char my_version[] = ZLIB_VERSION;

-    ushf *overlay;
-    /* We overlay pending_buf and d_buf+l_buf. This works since the average
-     * output size for (length,distance) codes is <= 24 bits.
-     */
-
    if (version == Z_NULL || version[0] != my_version[0] ||
        stream_size != sizeof(z_stream)) {
        return Z_VERSION_ERROR;
@@ -300,9 +295,47 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy,

    s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */

-    overlay = (ushf *) ZALLOC(strm, s->lit_bufsize, sizeof(ush)+2);
-    s->pending_buf = (uchf *) overlay;
-    s->pending_buf_size = (ulg)s->lit_bufsize * (sizeof(ush)+2L);
+    /* We overlay pending_buf and sym_buf. This works since the average size
+     * for length/distance pairs over any compressed block is assured to be 31
+     * bits or less.
+     *
+     * Analysis: The longest fixed codes are a length code of 8 bits plus 5
+     * extra bits, for lengths 131 to 257. The longest fixed distance codes are
+     * 5 bits plus 13 extra bits, for distances 16385 to 32768. The longest
+     * possible fixed-codes length/distance pair is then 31 bits total.
+     *
+     * sym_buf starts one-fourth of the way into pending_buf. So there are
+     * three bytes in sym_buf for every four bytes in pending_buf. Each symbol
+     * in sym_buf is three bytes -- two for the distance and one for the
+     * literal/length. As each symbol is consumed, the pointer to the next
+     * sym_buf value to read moves forward three bytes. From that symbol, up to
+     * 31 bits are written to pending_buf. The closest the written pending_buf
+     * bits gets to the next sym_buf symbol to read is just before the last
+     * code is written. At that time, 31*(n-2) bits have been written, just
+     * after 24*(n-2) bits have been consumed from sym_buf. sym_buf starts at
+     * 8*n bits into pending_buf. (Note that the symbol buffer fills when n-1
+     * symbols are written.) The closest the writing gets to what is unread is
+     * then n+14 bits. Here n is lit_bufsize, which is 16384 by default, and
+     * can range from 128 to 32768.
+     *
+     * Therefore, at a minimum, there are 142 bits of space between what is
+     * written and what is read in the overlain buffers, so the symbols cannot
+     * be overwritten by the compressed data. That space is actually 139 bits,
+     * due to the three-bit fixed-code block header.
+     *
+     * That covers the case where either Z_FIXED is specified, forcing fixed
+     * codes, or when the use of fixed codes is chosen, because that choice
+     * results in a smaller compressed block than dynamic codes. That latter
+     * condition then assures that the above analysis also covers all dynamic
+     * blocks. A dynamic-code block will only be chosen to be emitted if it has
+     * fewer bits than a fixed-code block would for the same set of symbols.
+     * Therefore its average symbol length is assured to be less than 31. So
+     * the compressed data for a dynamic block also cannot overwrite the
+     * symbols from which it is being constructed.
+     */
+
+    s->pending_buf = (uchf *) ZALLOC(strm, s->lit_bufsize, 4);
+    s->pending_buf_size = (ulg)s->lit_bufsize * 4;

    if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL ||
        s->pending_buf == Z_NULL) {
@@ -311,8 +344,12 @@ int ZEXPORT deflateInit2_(strm, level, method, windowBits, memLevel, strategy,
        deflateEnd (strm);
        return Z_MEM_ERROR;
    }
-    s->d_buf = overlay + s->lit_bufsize/sizeof(ush);
-    s->l_buf = s->pending_buf + (1+sizeof(ush))*s->lit_bufsize;
+    s->sym_buf = s->pending_buf + s->lit_bufsize;
+    s->sym_end = (s->lit_bufsize - 1) * 3;
+    /* We avoid equality with lit_bufsize*3 because of wraparound at 64K
+     * on 16 bit machines and because stored blocks are restricted to
+     * 64K-1 bytes.
+     */

    s->level = level;
    s->strategy = strategy;
@@ -473,7 +510,7 @@ int ZEXPORT deflatePrime (strm, bits, value)

    if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR;
    s = strm->state;
-    if ((Bytef *)(s->d_buf) < s->pending_out + ((Buf_size + 7) >> 3))
+    if (s->sym_buf < s->pending_out + ((Buf_size + 7) >> 3))
        return Z_BUF_ERROR;
    do {
        put = Buf_size - s->bi_valid;
@@ -1022,7 +1059,6 @@ int ZEXPORT deflateCopy (dest, source)
 #else
    deflate_state *ds;
    deflate_state *ss;
-    ushf *overlay;


    if (source == Z_NULL || dest == Z_NULL || source->state == Z_NULL) {
@@ -1042,8 +1078,7 @@ int ZEXPORT deflateCopy (dest, source)
    ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte));
    ds->prev   = (Posf *)  ZALLOC(dest, ds->w_size, sizeof(Pos));
    ds->head   = (Posf *)  ZALLOC(dest, ds->hash_size, sizeof(Pos));
-    overlay = (ushf *) ZALLOC(dest, ds->lit_bufsize, sizeof(ush)+2);
-    ds->pending_buf = (uchf *) overlay;
+    ds->pending_buf = (uchf *) ZALLOC(dest, ds->lit_bufsize, 4);

    if (ds->window == Z_NULL || ds->prev == Z_NULL || ds->head == Z_NULL ||
        ds->pending_buf == Z_NULL) {
@@ -1057,8 +1092,7 @@ int ZEXPORT deflateCopy (dest, source)
    zmemcpy(ds->pending_buf, ss->pending_buf, (uInt)ds->pending_buf_size);

    ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf);
-    ds->d_buf = overlay + ds->lit_bufsize/sizeof(ush);
-    ds->l_buf = ds->pending_buf + (1+sizeof(ush))*ds->lit_bufsize;
+    ds->sym_buf = ds->pending_buf + ds->lit_bufsize;

    ds->l_desc.dyn_tree = ds->dyn_ltree;
    ds->d_desc.dyn_tree = ds->dyn_dtree;
@@ -1737,7 +1771,7 @@ local block_state deflate_fast(s, flush)
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
-    if (s->last_lit)
+    if (s->sym_next)
        FLUSH_BLOCK(s, 0);
    return block_done;
 }
@@ -1878,7 +1912,7 @@ local block_state deflate_slow(s, flush)
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
-    if (s->last_lit)
+    if (s->sym_next)
        FLUSH_BLOCK(s, 0);
    return block_done;
 }
@@ -1953,7 +1987,7 @@ local block_state deflate_rle(s, flush)
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
-    if (s->last_lit)
+    if (s->sym_next)
        FLUSH_BLOCK(s, 0);
    return block_done;
 }
@@ -1992,7 +2026,7 @@ local block_state deflate_huff(s, flush)
        FLUSH_BLOCK(s, 1);
        return finish_done;
    }
-    if (s->last_lit)
+    if (s->sym_next)
        FLUSH_BLOCK(s, 0);
    return block_done;
 }
--- a/zlib/deflate.h
+++ b/zlib/deflate.h
@@ -214,7 +214,7 @@ typedef struct internal_state {
    /* Depth of each subtree used as tie breaker for trees of equal frequency
     */

-    uchf *l_buf;          /* buffer for literals or lengths */
+    uchf *sym_buf;        /* buffer for distances and literals/lengths */

    uInt  lit_bufsize;
    /* Size of match buffer for literals/lengths.  There are 4 reasons for
@@ -236,13 +236,8 @@ typedef struct internal_state {
     *   - I can't count above 4
     */

-    uInt last_lit;      /* running index in l_buf */
-
-    ushf *d_buf;
-    /* Buffer for distances. To simplify the code, d_buf and l_buf have
-     * the same number of elements. To use different lengths, an extra flag
-     * array would be necessary.
-     */
+    uInt sym_next;      /* running index in sym_buf */
+    uInt sym_end;       /* symbol table full when sym_next reaches this */

    ulg opt_len;        /* bit length of current block with optimal trees */
    ulg static_len;     /* bit length of current block with static trees */
@@ -322,20 +317,22 @@ void ZLIB_INTERNAL _tr_stored_block OF((deflate_state *s, charf *buf,

 # define _tr_tally_lit(s, c, flush) \
  { uch cc = (c); \
-    s->d_buf[s->last_lit] = 0; \
-    s->l_buf[s->last_lit++] = cc; \
+    s->sym_buf[s->sym_next++] = 0; \
+    s->sym_buf[s->sym_next++] = 0; \
+    s->sym_buf[s->sym_next++] = cc; \
    s->dyn_ltree[cc].Freq++; \
-    flush = (s->last_lit == s->lit_bufsize-1); \
+    flush = (s->sym_next == s->sym_end); \
   }
 # define _tr_tally_dist(s, distance, length, flush) \
  { uch len = (length); \
    ush dist = (distance); \
-    s->d_buf[s->last_lit] = dist; \
-    s->l_buf[s->last_lit++] = len; \
+    s->sym_buf[s->sym_next++] = dist; \
+    s->sym_buf[s->sym_next++] = dist >> 8; \
+    s->sym_buf[s->sym_next++] = len; \
    dist--; \
    s->dyn_ltree[_length_code[len]+LITERALS+1].Freq++; \
    s->dyn_dtree[d_code(dist)].Freq++; \
-    flush = (s->last_lit == s->lit_bufsize-1); \
+    flush = (s->sym_next == s->sym_end); \
  }
 #else
 # define _tr_tally_lit(s, c, flush) flush = _tr_tally(s, 0, c)
--- a/zlib/trees.c
+++ b/zlib/trees.c
@@ -418,7 +418,7 @@ local void init_block(s)

    s->dyn_ltree[END_BLOCK].Freq = 1;
    s->opt_len = s->static_len = 0L;
-    s->last_lit = s->matches = 0;
+    s->sym_next = s->matches = 0;
 }

 #define SMALLEST 1
@@ -943,7 +943,7 @@ void ZLIB_INTERNAL _tr_flush_block(s, buf, stored_len, last)

        Tracev((stderr, "\nopt %lu(%lu) stat %lu(%lu) stored %lu lit %u ",
                opt_lenb, s->opt_len, static_lenb, s->static_len, stored_len,
-                s->last_lit));
+                s->sym_next / 3));

        if (static_lenb <= opt_lenb) opt_lenb = static_lenb;

@@ -1012,8 +1012,9 @@ int ZLIB_INTERNAL _tr_tally (s, dist, lc)
    unsigned dist;  /* distance of matched string */
    unsigned lc;    /* match length-MIN_MATCH or unmatched char (if dist==0) */
 {
-    s->d_buf[s->last_lit] = (ush)dist;
-    s->l_buf[s->last_lit++] = (uch)lc;
+    s->sym_buf[s->sym_next++] = dist;
+    s->sym_buf[s->sym_next++] = dist >> 8;
+    s->sym_buf[s->sym_next++] = lc;
    if (dist == 0) {
        /* lc is the unmatched char */
        s->dyn_ltree[lc].Freq++;
@@ -1028,30 +1029,7 @@ int ZLIB_INTERNAL _tr_tally (s, dist, lc)
        s->dyn_ltree[_length_code[lc]+LITERALS+1].Freq++;
        s->dyn_dtree[d_code(dist)].Freq++;
    }
-
-#ifdef TRUNCATE_BLOCK
-    /* Try to guess if it is profitable to stop the current block here */
-    if ((s->last_lit & 0x1fff) == 0 && s->level > 2) {
-        /* Compute an upper bound for the compressed length */
-        ulg out_length = (ulg)s->last_lit*8L;
-        ulg in_length = (ulg)((long)s->strstart - s->block_start);
-        int dcode;
-        for (dcode = 0; dcode < D_CODES; dcode++) {
-            out_length += (ulg)s->dyn_dtree[dcode].Freq *
-                (5L+extra_dbits[dcode]);
-        }
-        out_length >>= 3;
-        Tracev((stderr,"\nlast_lit %u, in %ld, out ~%ld(%ld%%) ",
-               s->last_lit, in_length, out_length,
-               100L - out_length*100L/in_length));
-        if (s->matches < s->last_lit/2 && out_length < in_length/2) return 1;
-    }
-#endif
-    return (s->last_lit == s->lit_bufsize-1);
-    /* We avoid equality with lit_bufsize because of wraparound at 64K
-     * on 16 bit machines and because stored blocks are restricted to
-     * 64K-1 bytes.
-     */
+    return (s->sym_next == s->sym_end);
 }

 /* ===========================================================================
@@ -1064,13 +1042,14 @@ local void compress_block(s, ltree, dtree)
 {
    unsigned dist;      /* distance of matched string */
    int lc;             /* match length or unmatched char (if dist == 0) */
-    unsigned lx = 0;    /* running index in l_buf */
+    unsigned sx = 0;    /* running index in sym_buf */
    unsigned code;      /* the code to send */
    int extra;          /* number of extra bits to send */

-    if (s->last_lit != 0) do {
-        dist = s->d_buf[lx];
-        lc = s->l_buf[lx++];
+    if (s->sym_next != 0) do {
+        dist = s->sym_buf[sx++] & 0xff;
+        dist += (unsigned)(s->sym_buf[sx++] & 0xff) << 8;
+        lc = s->sym_buf[sx++];
        if (dist == 0) {
            send_code(s, lc, ltree); /* send a literal byte */
            Tracecv(isgraph(lc), (stderr," '%c' ", lc));
@@ -1095,11 +1074,10 @@ local void compress_block(s, ltree, dtree)
            }
        } /* literal or match pair ? */

-        /* Check that the overlay between pending_buf and d_buf+l_buf is ok: */
-        Assert((uInt)(s->pending) < s->lit_bufsize + 2*lx,
-               "pendingBuf overflow");
+        /* Check that the overlay between pending_buf and sym_buf is ok: */
+        Assert(s->pending < s->lit_bufsize + sx, "pendingBuf overflow");

-    } while (lx < s->last_lit);
+    } while (sx < s->sym_next);

    send_code(s, END_BLOCK, ltree);
 }
Author	SHA1	Message	Date
Wayne Davison	0ac7ebceef	Preparing for release of 3.2.4	2022-04-15 13:31:16 -07:00
Wayne Davison	85c56b2603	The latest news.	2022-04-11 09:50:58 -07:00
Wayne Davison	10aeb75cea	Add debugging comment about read_buf_().	2022-04-11 09:50:31 -07:00
Simon Deziel	d41bb98c09	systemd: restart daemon on-failure (#302 ) man 5 systemd.service: > Setting this to on-failure is the recommended choice for long-running services Partial fix for https://bugzilla.samba.org/show_bug.cgi?id=13463 Signed-off-by: Simon Deziel <simon@sdeziel.info>	2022-04-11 09:08:11 -07:00
Yoichi NAKAYAMA	2fda51692b	Specify log format to avoid malfunctions and unexpected errors. (#305 ) Solve the following problems: * mishandling of commit message lines similar to committer lines * UnicodeDecodeError with commit messages that cannot be interpreted as utf-8	2022-04-11 08:57:19 -07:00
Michal Ruprich	1de71e8a78	Fix for CVE-2018-25032 in zlib (#306 )	2022-04-11 08:50:50 -07:00
Wayne Davison	60dd42be60	Handle linking with a zlib with external read_buf.	2022-04-11 08:29:54 -07:00
Wayne Davison	d821e4cbfb	Preparing for release of 3.2.4pre4	2022-03-27 14:59:57 -07:00
Wayne Davison	8aa465117f	Add new & improved `--copy-devices` option.	2022-03-27 14:04:59 -07:00
Wayne Davison	8977815f5d	Some `--write-device` fixes.	2022-03-27 12:52:26 -07:00
Wayne Davison	a48c20c97c	Combine some alt-dest tests.	2022-03-26 10:01:12 -07:00
Wayne Davison	601f47436f	Rename compare-dest test.	2022-03-26 10:01:10 -07:00
Sam Mikes	ef76d6cfa5	Extract unlink_and_reopen from copy_file (#294 ) * add tests to exercise copy_file * Extract new function unlink_and_reopen from copy_file The argument `ofd` to copy_file is always set to -1 unless `open_tmpfile()` is called at generator.c:909 This change * removes assignment to a function argument * renames argument `ofd` to `tmpfilefd` in line with existing uses * extracts a new function `unlink_and_reopen` which is static to util1.c * rewrites header comments for copy_file	2022-03-26 09:14:10 -07:00
Wayne Davison	96ed4b47b9	Some word fixes.	2022-03-26 08:58:51 -07:00
Wayne Davison	13c4019e94	Also ignore a root-level rrsync file.	2022-03-13 10:45:09 -07:00
Wayne Davison	b7b387b1f7	Add FALLTHROUGH comment.	2022-03-13 09:31:44 -07:00
Wayne Davison	7569edfaef	Use ac_includes_default in largefile support test.	2022-03-09 18:38:03 -08:00
Wayne Davison	55b2a06812	Test newer FreeBSD.	2022-03-03 17:26:47 -08:00
Wayne Davison	b81a509556	Make asm use more selectable - Make the SIMD ASM code off by default. Use configure --enable-simd-asm to enable. - Allow MD5 ASM code to be requested even when OpenSSL is handling MD4 checksums. Use configure --enable-md5-asm to enable.	2022-03-03 17:00:57 -08:00
Wayne Davison	26f4dbe12c	Change usage (--version) output to note when ASM isn't really being used.	2022-02-21 16:41:50 -08:00
Sam Mikes	b3f1970f18	Fix wording in RSYNC_PORT section (#293 ) Fix wording from 'does is not read' -> 'is not read'	2022-02-21 14:00:45 -08:00
Wayne Davison	c51da9174f	Build Cygwin on windows-2022 with newer python. [buildall]	2022-02-09 14:00:13 -08:00
Wayne Davison	81f71f6f29	Add a CAUTION message to --debug=FILTER for trailing whitespace.	2022-01-27 08:53:41 -08:00
Wayne Davison	48e7005554	Add a couple more `--rsync-path` opts to the test. [buildall]	2022-01-20 10:51:13 -08:00
Wayne Davison	2b3e68814b	Specify the rsync that lsh.sh should run. [buildall]	2022-01-20 09:02:02 -08:00