Preparing for release of 3.2.0

xattr headers have been provided by glibc (at least on Linux/glibc)
for many years now. Reorder the inclusion of xattr headers to
attempt compatibility/legacy after the common case.
This prevents the warning without changing compatibility to
non-glibc systems.

* Add dependency on lib/sysxattrs.h header in Makefile

Co-authored-by: Wayne Davison <wayne@opencoder.net>

.cvsignore

@@ -1,25 +0,0 @@
-ID
-Makefile
-autom4te*.cache
-confdefs.h
-config.cache
-config.h
-config.log
-config.status
-conftest*
-dox
-getgroups
-gmon.out
-rsync
-shconfig
-testdir
-tests-dont-exist
-testtmp
-tls
-trimslash
-t_unsafe
-wildtest
-getfsdev
-.rsync-filter
-mkrounding
-rounding.h

.github/workflows/ccpp.yml

@@ -0,0 +1,31 @@
+name: C CI
+
+on:
+  push:
+    branches: [ master ]
+  pull_request:
+    branches: [ master ]
+
+jobs:
+  build:
+
+    runs-on: ubuntu-20.04
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: prepare-packages
+      run: sudo apt-get install fakeroot acl libacl1-dev attr libattr1-dev liblz4-dev libzstd-dev libxxhash-dev python3-cmarkgfm
+    - name: prepare-source
+      run: ./prepare-source
+    - name: configure
+      run: ./configure --with-included-popt --with-included-zlib
+    - name: make
+      run: make
+    - name: version-summary
+      run: ./rsync --version
+    - name: make check
+      run: make check
+    - name: make check30
+      run: make check30
+    - name: make check29
+      run: make check29

.gitignore

@@ -0,0 +1,52 @@
+*.[oa]
+*~
+dummy
+ID
+Makefile
+Makefile.old
+configure.sh
+configure.sh.old
+config.cache
+config.h
+config.h.in
+config.h.in.old
+config.log
+config.status
+aclocal.m4
+/proto.h
+/proto.h-tstamp
+/rsync*.1
+/rsync*.5
+/rsync*.html
+/help-rsync*.h
+/default-cvsignore.h
+/default-dont-compress.h
+/.md2man-works
+/autom4te*.cache
+/confdefs.h
+/conftest*
+/dox
+/getgroups
+/gmon.out
+/rsync
+/stunnel-rsyncd.conf
+/shconfig
+/testdir
+/tests-dont-exist
+/testtmp
+/tls
+/testrun
+/trimslash
+/t_unsafe
+/wildtest
+/getfsdev
+/rounding.h
+/doc/rsync.pdf
+/doc/rsync.ps
+/support/savetransfer
+/testsuite/chown-fake.test
+/testsuite/devices-fake.test
+/testsuite/xattrs-hlink.test
+/patches
+/SaVeDiR
+.deps

COPYING

@@ -672,3 +672,12 @@ may consider it more useful to permit linking proprietary applications with
 the library.  If this is what you want to do, use the GNU Lesser General
 Public License instead of this License.  But first, please read
 <http://www.gnu.org/philosophy/why-not-lgpl.html>.
+
+REGARDING OPENSSL AND XXHASH
+
+In addition, as a special exception, the copyright holders give
+permission to dynamically link rsync with the OpenSSL and xxhash
+libraries when those libraries are being distributed in compliance
+with their license terms, and to distribute a dynamically linked
+combination of rsync and these libraries.  This is also considered
+to be covered under the GPL's System Libraries exception.

INSTALL.md

@@ -1,13 +1,13 @@
 To build and install rsync:
 
-  $ ./configure
-  $ make
-  # make install
+    $ ./configure
+    $ make
+    # make install
 
 You may set the installation directory and other parameters by options
 to ./configure.  To see them, use:
 
-  $ ./configure --help
+    $ ./configure --help
 
 Configure tries to figure out if the local system uses group "nobody" or
 "nogroup" by looking in the /etc/group file.  (This is only used for the
@@ -17,7 +17,7 @@ for the daemon by editing the NOBODY_USER and NOBODY_GROUP defines in
 config.h, or just override them in your /etc/rsyncd.conf file.
 
 As of 2.4.7, rsync uses Eric Troan's popt option-parsing library.  A
-cut-down copy of release 1.6.4 is included in the rsync distribution,
+cut-down copy of a recent release is included in the rsync distribution,
 and will be used if there is no popt library on your build host, or if
 the --with-included-popt option is passed to ./configure.
 
@@ -25,6 +25,17 @@ If you configure using --enable-maintainer-mode, then rsync will try
 to pop up an xterm on DISPLAY=:0 if it crashes.  You might find this
 useful, but it should be turned off for production builds.
 
+MAKE COMPATIBILITY
+------------------
+
+Note that Makefile.in has a rule that uses a wildcard in a prerequisite.  If
+your make has a problem with this rule, you will see an error like this:
+
+    Don't know how to make ./*.c
+
+You can change the "proto.h-tstamp" target in Makefile.in to list all the *.c
+filenames explicitly in order to avoid this issue.
+
 RPM NOTES
 ---------
 

Makefile.in

@@ -5,52 +5,63 @@ prefix=@prefix@
 datarootdir=@datarootdir@
 exec_prefix=@exec_prefix@
 bindir=@bindir@
+libdir=@libdir@/rsync
 mandir=@mandir@
 
 LIBS=@LIBS@
 CC=@CC@
+AWK=@AWK@
 CFLAGS=@CFLAGS@
 CPPFLAGS=@CPPFLAGS@
+CXX=@CXX@
+CXXFLAGS=@CXXFLAGS@
 EXEEXT=@EXEEXT@
 LDFLAGS=@LDFLAGS@
+LIBOBJDIR=lib/
 
 INSTALLCMD=@INSTALL@
 INSTALLMAN=@INSTALL@
 
 srcdir=@srcdir@
+MKDIR_P=@MKDIR_P@
 VPATH=$(srcdir)
 SHELL=/bin/sh
 
-VERSION=@VERSION@
+VERSION=@RSYNC_VERSION@
 
 .SUFFIXES:
 .SUFFIXES: .c .o
 
-HEADERS=byteorder.h config.h errcode.h proto.h rsync.h ifuncs.h lib/pool_alloc.h
+SIMD_x86_64=simd-checksum-x86_64.o lib/md5-asm-x86_64.o
+
+GENFILES=configure.sh aclocal.m4 config.h.in proto.h proto.h-tstamp rsync.1 rsync.1.html \
+	 rsync-ssl.1 rsync-ssl.1.html rsyncd.conf.5 rsyncd.conf.5.html
+HEADERS=byteorder.h config.h errcode.h proto.h rsync.h ifuncs.h itypes.h inums.h \
+	lib/pool_alloc.h
 LIBOBJ=lib/wildmatch.o lib/compat.o lib/snprintf.o lib/mdfour.o lib/md5.o \
 	lib/permstring.o lib/pool_alloc.o lib/sysacls.o lib/sysxattrs.o @LIBOBJS@
-ZLIBOBJ=zlib/deflate.o zlib/inffast.o zlib/inflate.o zlib/inftrees.o \
+zlib_OBJS=zlib/deflate.o zlib/inffast.o zlib/inflate.o zlib/inftrees.o \
 	zlib/trees.o zlib/zutil.o zlib/adler32.o zlib/compress.o zlib/crc32.o
 OBJS1=flist.o rsync.o generator.o receiver.o cleanup.o sender.o exclude.o \
-	util.o main.o checksum.o match.o syscall.o log.o backup.o
+	util.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 \
 	fileio.o batch.o clientname.o chmod.o acls.o xattrs.o
 OBJS3=progress.o pipe.o
 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
-OBJS=$(OBJS1) $(OBJS2) $(OBJS3) $(DAEMON_OBJ) $(LIBOBJ) $(ZLIBOBJ) @BUILD_POPT@
+OBJS=$(OBJS1) $(OBJS2) $(OBJS3) @SIMD@ $(DAEMON_OBJ) $(LIBOBJ) @BUILD_ZLIB@ @BUILD_POPT@
 
-TLS_OBJ = tls.o syscall.o lib/compat.o lib/snprintf.o lib/permstring.o lib/sysxattrs.o @BUILD_POPT@
+TLS_OBJ = tls.o syscall.o t_stub.o lib/compat.o lib/snprintf.o lib/permstring.o lib/sysxattrs.o @BUILD_POPT@
 
 # Programs we must have to run the test cases
 CHECK_PROGS = rsync$(EXEEXT) tls$(EXEEXT) getgroups$(EXEEXT) getfsdev$(EXEEXT) \
-	trimslash$(EXEEXT) t_unsafe$(EXEEXT) wildtest$(EXEEXT)
+	testrun$(EXEEXT) trimslash$(EXEEXT) t_unsafe$(EXEEXT) wildtest$(EXEEXT)
 
-CHECK_SYMLINKS = testsuite/chown-fake.test testsuite/devices-fake.test
+CHECK_SYMLINKS = testsuite/chown-fake.test testsuite/devices-fake.test testsuite/xattrs-hlink.test
 
 # Objects for CHECK_PROGS to clean
-CHECK_OBJS=getgroups.o getfsdev.o t_stub.o t_unsafe.o trimslash.o wildtest.o
+CHECK_OBJS=tls.o testrun.o getgroups.o getfsdev.o t_stub.o t_unsafe.o trimslash.o wildtest.o
 
 # note that the -I. is needed to handle config.h when using VPATH
 .c.o:
@@ -58,15 +69,28 @@ CHECK_OBJS=getgroups.o getfsdev.o t_stub.o t_unsafe.o trimslash.o wildtest.o
 	$(CC) -I. -I$(srcdir) $(CFLAGS) $(CPPFLAGS) -c $< @CC_SHOBJ_FLAG@
 @OBJ_RESTORE@
 
-all: rsync$(EXEEXT)
+.PHONY: all
+all: Makefile rsync$(EXEEXT) stunnel-rsyncd.conf @MAKE_MAN@
 
+.PHONY: install
 install: all
-	-mkdir -p ${DESTDIR}${bindir}
+	-${MKDIR_P} ${DESTDIR}${bindir}
 	${INSTALLCMD} ${INSTALL_STRIP} -m 755 rsync$(EXEEXT) ${DESTDIR}${bindir}
-	-mkdir -p ${DESTDIR}${mandir}/man1
-	-mkdir -p ${DESTDIR}${mandir}/man5
-	${INSTALLMAN} -m 644 $(srcdir)/rsync.1 ${DESTDIR}${mandir}/man1
-	${INSTALLMAN} -m 644 $(srcdir)/rsyncd.conf.5 ${DESTDIR}${mandir}/man5
+	${INSTALLCMD} -m 755 rsync-ssl ${DESTDIR}${bindir}
+	-${MKDIR_P} ${DESTDIR}${mandir}/man1
+	-${MKDIR_P} ${DESTDIR}${mandir}/man5
+	if test -f rsync.1; then ${INSTALLMAN} -m 644 rsync.1 ${DESTDIR}${mandir}/man1; fi
+	if test -f rsync-ssl.1; then ${INSTALLMAN} -m 644 rsync-ssl.1 ${DESTDIR}${mandir}/man1; fi
+	if test -f rsyncd.conf.5; then ${INSTALLMAN} -m 644 rsyncd.conf.5 ${DESTDIR}${mandir}/man5; fi
+
+install-ssl-daemon: stunnel-rsyncd.conf
+	-${MKDIR_P} ${DESTDIR}/etc/stunnel
+	${INSTALLCMD} -m 644 stunnel-rsyncd.conf ${DESTDIR}/etc/stunnel/rsyncd.conf
+	@if ! ls /etc/rsync-ssl/certs/server.* >/dev/null 2>/dev/null; then \
+	    echo "Note that you'll need to install the certificate used by /etc/stunnel/rsyncd.conf"; \
+	fi
+
+install-all: install install-ssl-client install-ssl-daemon
 
 install-strip:
 	$(MAKE) INSTALL_STRIP='-s' install
@@ -75,73 +99,184 @@ rsync$(EXEEXT): $(OBJS)
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJS) $(LIBS)
 
 $(OBJS): $(HEADERS)
+$(CHECK_OBJS): $(HEADERS)
+tls.o xattrs.o: lib/sysxattrs.h
+options.o: latest-year.h help-rsync.h help-rsyncd.h
+exclude.o: default-cvsignore.h
+loadparm.o: default-dont-compress.h
 
 flist.o: rounding.h
 
-rounding.h: mkrounding$(EXEEXT)
-	./mkrounding$(EXEEXT) >rounding.h
+default-cvsignore.h default-dont-compress.h: rsync.1.md define-from-md.awk
+	$(AWK) -f $(srcdir)/define-from-md.awk -v hfile=$@ $(srcdir)/rsync.1.md
 
-mkrounding$(EXEEXT): mkrounding.c rsync.h
-	@sed '1,/^struct file_struct/d; /^}/,$$d' <$(srcdir)/rsync.h >mkrounding.h
-	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ -I. $(srcdir)/mkrounding.c
-	@rm mkrounding.h
+help-rsync.h help-rsyncd.h: rsync.1.md help-from-md.awk
+	$(AWK) -f $(srcdir)/help-from-md.awk -v hfile=$@ $(srcdir)/rsync.1.md
+
+rounding.h: rounding.c rsync.h proto.h
+	@for r in 0 1 3; do \
+	    if $(CC) $(CFLAGS) $(CPPFLAGS) $(LDFLAGS) -o rounding -DEXTRA_ROUNDING=$$r -I. $(srcdir)/rounding.c >rounding.out 2>&1; then \
+		echo "#define EXTRA_ROUNDING $$r" >rounding.h; \
+		if test -f "$$HOME/build_farm/build_test.fns"; then \
+		    echo "EXTRA_ROUNDING is $$r" >&2; \
+		fi; \
+		break; \
+	    fi; \
+	done
+	@rm -f rounding
+	@if test -f rounding.h; then : ; else \
+	    cat rounding.out 1>&2; \
+	    echo "Failed to create rounding.h!" 1>&2; \
+	    exit 1; \
+	fi
+	@rm -f rounding.out
+
+simd-checksum-x86_64.o: simd-checksum-x86_64.cpp
+	$(CXX) $(CXXFLAGS) $(CPPFLAGS) -c -o $@ $(srcdir)/simd-checksum-x86_64.cpp
+
+lib/md5-asm-x86_64.o: lib/md5-asm-x86_64.s
+	$(CC) -c -o $@ $(srcdir)/lib/md5-asm-x86_64.s
 
 tls$(EXEEXT): $(TLS_OBJ)
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(TLS_OBJ) $(LIBS)
 
+testrun$(EXEEXT): testrun.o
+	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ testrun.o
+
 getgroups$(EXEEXT): getgroups.o
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ getgroups.o $(LIBS)
 
 getfsdev$(EXEEXT): getfsdev.o
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ getfsdev.o $(LIBS)
 
-TRIMSLASH_OBJ = trimslash.o syscall.o lib/compat.o lib/snprintf.o
+TRIMSLASH_OBJ = trimslash.o syscall.o t_stub.o lib/compat.o lib/snprintf.o
 trimslash$(EXEEXT): $(TRIMSLASH_OBJ)
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(TRIMSLASH_OBJ) $(LIBS)
 
-T_UNSAFE_OBJ = t_unsafe.o syscall.o util.o t_stub.o lib/compat.o lib/snprintf.o
+T_UNSAFE_OBJ = t_unsafe.o syscall.o util.o util2.o t_stub.o lib/compat.o lib/snprintf.o lib/wildmatch.o
 t_unsafe$(EXEEXT): $(T_UNSAFE_OBJ)
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(T_UNSAFE_OBJ) $(LIBS)
 
-gen:
-	cd $(srcdir) && $(MAKE) -f prepare-source.mak gen
+.PHONY: conf
+conf: configure.sh config.h.in
 
-man:
-	cd $(srcdir) && $(MAKE) -f prepare-source.mak man
+.PHONY: gen
+gen: conf proto.h man
 
-proto:
-	cd $(srcdir) && $(MAKE) -f prepare-source.mak proto.h
+.PHONY: gensend
+gensend: gen
+	rsync -aic $(GENFILES) $${SAMBA_HOST-samba.org}:/home/ftp/pub/rsync/generated-files/
 
+aclocal.m4: $(srcdir)/m4/*.m4
+	aclocal -I $(srcdir)/m4
+
+configure.sh config.h.in: configure.ac aclocal.m4
+	@if test -f configure.sh; then cp -p configure.sh configure.sh.old; else touch configure.sh.old; fi
+	@if test -f config.h.in; then cp -p config.h.in config.h.in.old; else touch config.h.in.old; fi
+	autoconf -o configure.sh
+	autoheader && touch config.h.in
+	@if diff configure.sh configure.sh.old >/dev/null 2>&1; then \
+	    echo "configure.sh is unchanged."; \
+	    rm configure.sh.old; \
+	else \
+	    echo "configure.sh has CHANGED."; \
+	fi
+	@if diff config.h.in config.h.in.old >/dev/null 2>&1; then \
+	    echo "config.h.in is unchanged."; \
+	    rm config.h.in.old; \
+	else \
+	    echo "config.h.in has CHANGED."; \
+	fi
+	@if test -f configure.sh.old -o -f config.h.in.old; then \
+	    if test "$(MAKECMDGOALS)" = reconfigure; then \
+		echo 'Continuing with "make reconfigure".'; \
+	    else \
+		echo 'You may need to run:'; \
+		echo '  make reconfigure'; \
+		exit 1; \
+	    fi \
+	fi
+
+.PHONY: reconfigure
+reconfigure: configure.sh
+	./config.status --recheck
+	./config.status
+
+Makefile: Makefile.in config.status configure.sh config.h.in
+	@if test -f Makefile; then cp -p Makefile Makefile.old; else touch Makefile.old; fi
+	@./config.status
+	@if diff Makefile Makefile.old >/dev/null 2>&1; then \
+	    echo "Makefile is unchanged."; \
+	    rm Makefile.old; \
+	else \
+	    if test "$(MAKECMDGOALS)" = reconfigure; then \
+		echo 'Continuing with "make reconfigure".'; \
+	    else \
+		echo "Makefile updated -- rerun your make command."; \
+		exit 1; \
+	    fi \
+	fi
+
+stunnel-rsyncd.conf: $(srcdir)/stunnel-rsyncd.conf.in Makefile
+	sed 's;\@bindir\@;$(bindir);g' <$(srcdir)/stunnel-rsyncd.conf.in >stunnel-rsyncd.conf
+
+.PHONY: proto
+proto: proto.h-tstamp
+
+proto.h: proto.h-tstamp
+	@if test -f proto.h; then :; else cp -p $(srcdir)/proto.h .; fi
+
+proto.h-tstamp: $(srcdir)/*.c $(srcdir)/lib/compat.c config.h
+	$(AWK) -f $(srcdir)/mkproto.awk $(srcdir)/*.c $(srcdir)/lib/compat.c
+
+.PHONY: man
+man: rsync.1 rsync-ssl.1 rsyncd.conf.5
+
+rsync.1: rsync.1.md md2man NEWS.md Makefile
+	@$(srcdir)/maybe-make-man $(srcdir) rsync.1.md
+
+rsync-ssl.1: rsync-ssl.1.md md2man NEWS.md Makefile
+	@$(srcdir)/maybe-make-man $(srcdir) rsync-ssl.1.md
+
+rsyncd.conf.5: rsyncd.conf.5.md md2man NEWS.md Makefile
+	@$(srcdir)/maybe-make-man $(srcdir) rsyncd.conf.5.md
+
+.PHONY: clean
 clean: cleantests
-	rm -f *~ $(OBJS) $(TLS_OBJ) $(CHECK_PROGS) $(CHECK_OBJS) $(CHECK_SYMLINKS) \
-		mkrounding mkrounding.h rounding.h
+	rm -f *~ $(OBJS) $(CHECK_PROGS) $(CHECK_OBJS) $(CHECK_SYMLINKS) \
+		rounding rounding.h *.old rsync*.1 rsync*.5 rsync*.html
 
+.PHONY: cleantests
 cleantests:
 	rm -rf ./testtmp*
 
 # We try to delete built files from both the source and build
 # directories, just in case somebody previously configured things in
 # the source directory.
+.PHONY: distclean
 distclean: clean
 	rm -f Makefile config.h config.status
+	rm -f stunnel-rsyncd.conf
+	rm -f lib/dummy popt/dummy zlib/dummy
 	rm -f $(srcdir)/Makefile $(srcdir)/config.h $(srcdir)/config.status
-
+	rm -f $(srcdir)/lib/dummy $(srcdir)/popt/dummy $(srcdir)/zlib/dummy
 	rm -f config.cache config.log
 	rm -f $(srcdir)/config.cache $(srcdir)/config.log
-
 	rm -f shconfig $(srcdir)/shconfig
+	rm -f $(GENFILES)
+	rm -rf autom4te.cache
 
 # this target is really just for my use. It only works on a limited
 # range of machines and is used to produce a list of potentially
 # dead (ie. unused) functions in the code. (tridge)
+.PHONY: finddead
 finddead:
 	nm *.o */*.o |grep 'U ' | awk '{print $$2}' | sort -u > nmused.txt
 	nm *.o */*.o |grep 'T ' | awk '{print $$3}' | sort -u > nmfns.txt
 	comm -13 nmused.txt nmfns.txt
 
 # 'check' is the GNU name, 'test' is the name for everybody else :-)
-.PHONY: check test
-
+.PHONY: test
 test: check
 
 # There seems to be no standard way to specify some variables as
@@ -154,13 +289,19 @@ test: check
 # catch Bash-isms earlier even if we're running on GNU.  Of course, we
 # might lose in the future where POSIX diverges from old sh.
 
+.PHONY: check
 check: all $(CHECK_PROGS) $(CHECK_SYMLINKS)
 	rsync_bin=`pwd`/rsync$(EXEEXT) $(srcdir)/runtests.sh
 
+.PHONY: check29
 check29: all $(CHECK_PROGS) $(CHECK_SYMLINKS)
 	rsync_bin=`pwd`/rsync$(EXEEXT) $(srcdir)/runtests.sh --protocol=29
 
-wildtest.o: wildtest.c lib/wildmatch.c rsync.h
+.PHONY: check30
+check30: all $(CHECK_PROGS) $(CHECK_SYMLINKS)
+	rsync_bin=`pwd`/rsync$(EXEEXT) $(srcdir)/runtests.sh --protocol=30
+
+wildtest.o: wildtest.c t_stub.o lib/wildmatch.c rsync.h config.h
 wildtest$(EXEEXT): wildtest.o lib/compat.o lib/snprintf.o @BUILD_POPT@
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ wildtest.o lib/compat.o lib/snprintf.o @BUILD_POPT@ $(LIBS)
 
@@ -170,10 +311,14 @@ testsuite/chown-fake.test:
 testsuite/devices-fake.test:
 	ln -s devices.test $(srcdir)/testsuite/devices-fake.test
 
+testsuite/xattrs-hlink.test:
+	ln -s xattrs.test $(srcdir)/testsuite/xattrs-hlink.test
+
 # This does *not* depend on building or installing: you can use it to
 # check a version installed from a binary or some other source tree,
 # if you want.
 
+.PHONY: installcheck
 installcheck: $(CHECK_PROGS) $(CHECK_SYMLINKS)
 	POSIXLY_CORRECT=1 TOOLDIR=`pwd` rsync_bin="$(bindir)/rsync$(EXEEXT)" srcdir="$(srcdir)" $(srcdir)/runtests.sh
 
@@ -184,21 +329,12 @@ installcheck: $(CHECK_PROGS) $(CHECK_SYMLINKS)
 splint:
 	splint +unixlib +gnuextensions -weak rsync.c
 
-
-rsync.dvi: doc/rsync.texinfo
-	texi2dvi -o $@ $<
-
-rsync.ps: rsync.dvi
-	dvips -ta4 -o $@ $<
-
-rsync.pdf: doc/rsync.texinfo
-	texi2dvi -o $@ --pdf $<
-
-
+.PHONY: doxygen
 doxygen:
 	cd $(srcdir) && rm dox/html/* && doxygen
 
 # for maintainers only
+.PHONY: doxygen-upload
 doxygen-upload:
 	rsync -avzv $(srcdir)/dox/html/ --delete \
-	samba.org:/home/httpd/html/rsync/doxygen/head/
+	$${SAMBA_HOST-samba.org}:/home/httpd/html/rsync/doxygen/head/

NEWS

@@ -1,232 +0,0 @@
-NEWS for rsync 3.0.0 (UNRELEASED)
-Protocol: 30 (changed)
-Changes since 2.6.9:
-
-  NOTABLE CHANGES IN BEHAVIOR:
-
-    - The handling of implied directories when using --relative has changed to
-      send them as directories (e.g. no implied dir is ever sent as a symlink).
-      This avoids unexpected behavior and should not adversely affect most
-      people.  If you're one of those rare people who relied upon having an
-      implied dir be duplicated as a symlink, you should specify the transfer
-      of the symlink and the transfer of the referent directory as separate
-      args.  (See also --keep-dirlinks and --no-implied-dirs.)
-
-    - Requesting a remote file list without specifying -r (--recursive) now
-      sends the -d (--dirs) option to the remote rsync rather than sending -r
-      along with an extra exclude of /*/*.  If the remote rsync does not
-      understand the -d option (i.e. it is 2.6.3 or older), you will need to
-      either turn off -d (--no-d), or specify  -r --exclude='/*/*'  manually.
-
-  BUG FIXES:
-
-    - Fixed the output of -ii when combined with one of the --*-dest options:
-      it now itemizes all the items, not just the changed ones.
-
-    - Made the output of all file types consistent when using a --*-dest
-      option.  Prior versions used to output too many creation events for
-      matching items.
-
-    - The code that waits for a child pid now handles being interrupted by a
-      signal.  This fixes a problem with the pre-xfer exec function not being
-      able to get the exit status from the script.
-
-    - A negated filter rule (with a '!' modifier) no longer loses the negation
-      when sending the filter rules to the remote rsync.
-
-    - Fixed a problem with the --out-format (aka --log-format) option %f:  it
-      would output superfluous directory information for a non-daemon rsync.
-
-    - Fixed a problem with -vv (double --verbose) and --stats when "pushing"
-      files (which includes local copies).  Version 2.6.9 would complete the
-      copy, but exit with an error when the receiver output its memory stats.
-
-    - If --password-file is used on a non-daemon transfer, rsync now complains
-      and exits.  This should help users figure out that they can't use this
-      option to control a remote shell's password prompt.
-
-    - Make sure that directory permissions of a newly-created destination
-      directory are handled right when --perms is left off.
-
-    - The itemized output of a newly-created destination directory is now
-      output as a creation event, not a change event.
-
-    - Improved --hard-link so that more corner cases are handled correctly
-      when combined with options such as --link-dest and/or --ignore-existing.
-
-    - The --append option no longer updates a file that has the same size.
-
-    - Fixed a bug when combining --backup and --backup-dir with --inplace:
-      any missing backup directories are now created.
-
-    - Fixed a bug when using --backup and --inplace with --whole-file or
-      --read-batch: backup files are actually created now.
-
-    - Starting up an extra copy of an rsync daemon does not delete the pidfile
-      for the running daemon -- if the pidfile exists, the extra program will
-      exit with an error.
-
-    - The daemon pidfile is checked and created sooner in the startup sequence.
-
-    - If a daemon module's "path" value is not an absolute pathname, the code
-      now makes it absolute internally (making it work properly).
-
-  ENHANCEMENTS:
-
-    - A new incremental-recursion algorithm is now used when rsync is talking
-      to another 3.x version.  This starts the transfer going more quickly
-      (before all the files have been found), and requires much less memory.
-      See the --recursive option in the manpage for some restrictions.
-
-    - Lowered memory use in the non-incremental-recursion algorithm for typical
-      option values (usually saving from 21-29 bytes per file).
-
-    - The default --delete algorithm is now --delete-during when talking to a
-      3.x rsync.  This is a faster scan than using --delete-before (which is
-      the default when talking to older rsync versions), and is compatible with
-      the new incremental recursion mode.
-
-    - Rsync now allows multiple remote-source args to be specified rather than
-      having to rely on a special space-splitting side-effect of the remote-
-      shell.  Additional remote args must specify the same host or have an
-      empty hostname, as seen here:  :file1  ::module/file2.  This means that
-      local use of brace expansion now works:  rsync -av host:path/{f1,f2} .
-
-    - Added the --protect-args (-s) option, that tells rsync to send most of
-      the command-line args at the start of the transfer rather than as args
-      to the remote-shell command.  This protects them from space-splitting,
-      and only interprets basic wildcard special shell characters (*?[).
-
-    - Added the --delete-delay option, which is a more efficient way to delete
-      files at the end of the transfer without needing a separate delete pass.
-
-    - Added the --acls (-A) option to preserve Access Control Lists.  This is
-      an improved version of the prior patch that was available, and it even
-      supports OS X ACLs.  (If you need to have backward compatibility with
-      old, patched versions of rsync, apply the acls.diff file from the patches
-      dir.)
-
-    - Added the --xattrs (-X) option to preserver extended attributes.  This is
-      an improved version of the prior patch that was available.  (If you need
-      to have backward compatibility with old, patched versions of rsync, apply
-      the xattrs.diff file from the patches dir.)
-
-    - Added the --fake-super option that allows a non-super user to preserve
-      all attributes of a file by using a special extended-attribute idiom.
-      There is also an analogous "fake super" option for an rsync daemon.
-
-    - Added the --iconv option, which allows rsync to convert filenames from
-      one character-set to another during the transfer.  The default is to make
-      this feature available as long as your system has iconv_open().  If
-      compilation fails, specify --disable-iconv to configure, and then
-      rebuild.  If you want rsync to perform character-set conversions by
-      default, you can specify --enable-iconv=CONVERT_STRING with the default
-      value for the --iconv option that you wish to use.  For example,
-      "--enable-iconv=." is a good choice.  See the rsync manpage for an
-      explanation of the --iconv option's settings.
-
-    - Added the --skip-compress=LIST option to override the default list of
-      file suffixes that will not be compressed when using --compress.
-      
-    - The daemon's default for "dont compress" was extended to include:
-	  *.7z *.mp[34] *.mov *.avi *.ogg *.jpg *.jpeg
-      The matching routine was also optimized to run more quickly.
-
-    - The --max-delete option now outputs a warning if it skipped any file
-      deletions, including a count of how many deletions were skipped.  (Older
-      versions just silently stopped deleting things.)
-
-    - You may specify --max-delete=0 to a 3.0.0 client to request that it warn
-      about extraneous files without deleting anything.  If you're not sure
-      what version the client is, you can use the less-obvious --max-delete=-1,
-      as both old and new versions will treat that as the same request (though
-      older versions don't warn).
-
-    - The --hard-link option now uses less memory on both the sending and
-      receiving side for all protocol versions.  For protocol 30, the use of a
-      hashtable on the sending side allows us to more efficiently convey to the
-      receiver what files are linked together.  This reduces the amount of data
-      sent over the socket by a considerable margin (rather than adding more
-      data), and limits the in-memory storage of the device+inode information
-      to just the sending side for the new protocol 30, or to the receiving
-      side when speaking an older protocol (note that older rsync versions kept
-      the device+inode information on both sides).
-
-    - The filter rules now support a perishable ("p") modifier that marks rules
-      that should not have an effect in a directory that is being deleted.  e.g.
-      -f '-p .svn/' would only affect "live" .svn directories.
-
-    - If we get an error setting the time on a symlink, we don't complain about
-      it anymore (since some operating systems don't support that, and it's not
-      that important).
-
-    - Protocol 30 now uses MD5 checksums instead of MD4.
-
-    - Changed the --append option to not checksum the existing data in the
-      destination file, which speeds up file appending.
-
-    - Added the --append-verify option, which works like the older --append
-      option (verifying the existing data in the destination file).  For
-      compatibility with older rsync versions, any use of --append that is
-      talking protocol 29 or older will revert to the --append-verify method.
-
-    - Documented and extended the support for the RSYNC_CONNECT_PROG variable
-      that can be used to enhance the client side of a daemon connection.
-
-    - Improved the dashes and double-quotes in the nroff manpage output.
-
-    - We now support a lot more --no-OPTION override options.
-
-  INTERNAL:
-
-    - The file-list sorting algorithm now uses a sort that keeps any same-
-      named items in the same order as they were specified.  This allows
-      rsync to always ensure that the first of the duplicates is the one
-      that will be included in the copy.  The new sort was also faster
-      than the glibc version of qsort() and mergesort() in my testing.
-
-    - Rsync now supports the transfer of 64-bit timestamps (time_t values).
-
-    - Fixed a build problem with older (2.x) versions of gcc.
-
-    - Added some isType() functions that make dealing with signed characters
-      easier without forcing variables via casts.
-
-    - Changed strcat/strcpy/sprintf function calls to use safer versions.
-
-    - Upgraded the included popt version to 1.10.2 and improved its use of
-      string-handling functions.
-
-    - Added missing prototypes for compatibility functions from the lib dir.
-
-    - Configure determines if iconv() has a const arg, allowing us to avoid a
-      compiler warning.
-
-    - Made the sending of some numbers more efficient for protocol 30.
-
-    - Make sure that a daemon process doesn't mind if the client was weird and
-      omitted the --server option.
-
-    - Improved the use of "const" on pointers.
-
-    - Improved J.W.'s pool_alloc routines to add a way of freeing older
-      sections of a pool's memory.
-
-  DEVELOPER RELATED:
-
-    - Rsync is now licensed under the GPLv3 or later.
-
-    - When running the tests, we now put our per-test temp dirs into a sub-
-      directory named testtmp (which is created, if missing).  This allows
-      someone to symlink the testtmp directory to another filesystem (which is
-      useful if the build dir's filesystem does not support ACLs and xattrs,
-      and another file system does).
-
-    - Rsync now has a way of handling protocol-version changes during the
-      development of a new protocol version.  This causes any out-of-sync
-      versions to speak an older protocol rather than fail in a cryptic manner.
-      This addition makes it safe to deploy a pre-release version that may
-      interact with the public.  This new exchange of sub-version info does not
-      interfere with the {MIN,MAX}_PROTOCOL_VERSION checking algorithm (which
-      does not have enough range to allow the main protocol number to be
-      incremented for every minor tweak in that happens during development).

NEWS.md

@@ -0,0 +1,205 @@
+# NEWS for rsync 3.2.0 (19 Jun 2020)
+
+Protocol: 31 (unchanged)
+
+## Changes since 3.1.3:
+
+### BUG FIXES:
+
+ - Avoid a potential out-of-bounds read in daemon mode if argc can be made to
+   become 0.
+
+ - Fix the default list of skip-compress files for non-daemon transfers.
+
+ - Fix xattr filter rules losing an 'x' attribute in a non-local transfer.
+
+ - Avoid an error when a check for a potential fuzzy file happens to reference
+   a directory.
+
+ - Make the atomic-rsync helper script have a more consistent error-exit.
+
+ - Make sure that a signal handler calls `_exit()` instead of exit().
+
+ - Various zlib fixes, including security fixes for CVE-2016-9843,
+   CVE-2016-9842, CVE-2016-9841, and CVE-2016-9840.
+
+ - Fixed an issue with `--remove-source-files` not removing a source symlink
+   when combined with `--copy-links`.
+
+ - Fixed a bug where the daemon would fail to write early fatal error messages
+   to the client, such as refused or unknown command-line options.
+
+ - Fixed the block-size validation logic when dealing with older protocols.
+
+ - Some rrsync fixes and enhancements to handle the latest options.
+
+ - Fixed a problem with the `--link-dest`|`--copy-dest` code when `--xattrs`
+   was specified along with multiple alternate-destination directories (it
+   could possibly choose a bad file match while trying to find a better xattr
+   match).
+
+ - Fixed a couple bugs in the handling of files with the `--sparse` option.
+
+ - Fixed a bug in the writing of the batch.sh file (w/--write-batch) when the
+   source & destination args were not last on the command-line.
+
+ - Avoid a hang when an overabundance of messages clogs up all the I/O buffers.
+
+ - Fixed a mismatch in the RSYNC_PID values put into the environment of
+   `pre-xfer exec` and a `post-xfer exec`.
+
+ - Fixed a crash in the `--iconv` code.
+
+ - Fixed a rare crash in the popt_unalias() code.
+
+### ENHANCEMENTS:
+
+ - Various checksum enhancements, including the optional use of openssl's MD4 &
+   MD5 checksum algorithms, some x86-64 optimizations for the rolling checksum,
+   some x86-64 optimizations for the (non-openssl) MD5 checksum, the addition
+   of xxhash checksum support, and a negotiation heuristic that ensures that it
+   is easier to add new checksum algorithms in the future.  The environment
+   variable `RSYNC_CHECKSUM_LIST` can be used to customize the preference order
+   of the negotiation, or use `--checksum-choice` (`--cc`) to force a choice.
+
+ - Various compression enhancements, including the addition of zstd and lz4
+   compression algorithms and a negotiation heuristic that picks the best
+   compression option supported by both sides.  The environment variable
+   `RSYNC_COMPRESS_LIST` can be used to customize the preference order of the
+   negotiation, or use `--compress-choice` (`--zc`) to force a choice.
+
+ - Added a --debug=NSTR option that outputs details of the new negotiation
+   strings (for checksums and compression).  The first level just outputs the
+   result of each negotiation on the client, level 2 outputs the values of the
+   strings that were sent to and received from the server, and level 3 outputs
+   all those values on the server side too (when the server was given the debug
+   option).
+
+ - The --debug=OPTS command-line option is no longer auto-forwarded to the
+   remote rsync which allows for the client and server to have different levels
+   of debug specified. This also allows for newer debug options to be
+   specified, such as using --debug=NSTR to see the negotiated hash result,
+   without having the command fail if the server version is too old to handle
+   that debug item. Use -M--debug=OPTS to send the options to the remote side.
+
+ - Added the `--atimes` option based on the long-standing patch (just with some
+   fixes that the patch has been needing).
+
+ - Added `--open-noatime` option to open files using `O_NOATIME`.
+
+ - Added the `--write-devices` option based on the long-standing patch.
+
+ - Added openssl & preliminary gnutls support to the rsync-ssl script, which is
+   now installed by default.  This was unified with the old stunnel-rsync
+   helper script to simplify packaging.  Note that the script accepts the use
+   of --type=gnutls for gnutls testing, but does not look for gnutls-cli on the
+   path yet.  The use of type=gnutls will not work right until gnutls-cli no
+   longer drops data.
+
+ - Rsync was enhanced to set the `RSYNC_PORT` environment variable when running
+   a daemon-over-rsh script. Its value is the user-specified port number (set
+   via `--port` or an rsync:// URL) or 0 if the user didn't override the port.
+
+ - Added the `proxy protocol` daemon parameter that allows your rsyncd to know
+   the real remote IP when it is setup behind a proxy.
+
+ - Added negated matching to the daemon's `refuse options` setting by using
+   match strings that start with a `!` (such as `!compress*`).  This lets you
+   refuse all options except for a particular approved list, for example.  It
+   also lets rsync refuse certain options by default (such as `write-devices`)
+   while allowing the config to override that, as desired.
+
+ - Added the `early exec` daemon parameter that runs a script before the
+   transfer parameters are known, allowing some early setup based on module
+   name.
+
+ - Added status output in response to a signal (via both SIGINFO & SIGVTALRM).
+
+ - Added `--copy-as=USER` option to give some extra security to root-run rsync
+   commands into/from untrusted directories (such as backups and restores).
+
+ - When resuming the transfer of a file in the `--partial-dir`, rsync will now
+   update that partial file in-place instead of creating yet another tmp file
+   copy.  This requires both sender & receiver to be at least v3.2.0.
+
+ - Added support for `RSYNC_SHELL` & `RSYNC_NO_XFER_EXEC` environment variables
+   that affect the early, pre-xfer, and post-xfer exec rsync daemon parameters.
+
+ - Optimize the `--fuzzy --fuzzy` heuristic to avoid the fuzzy directory scan
+   until all other basis-file options are exhausted (such as `--link-dest`).
+
+ - Have the daemon log include the normal-exit sent/received stats when the
+   transfer exited with an error when possible (i.e. if it is the sender).
+
+ - 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
+   aren't installed by default).
+
+ - Made -V the short option for --version and improved its information.
+
+ - Pass the -4 or -6 option to the ssh command, making it easier to type than
+   `--rsh='ssh -4'` (or -6).
+
+ - Added example config for rsyncd SSL proxy configs to rsyncd.conf.
+
+ - More errors messages now mention if the error is coming from the sender or
+   the receiver.
+
+### PACKAGING RELATED:
+
+ - Add installed binary: /usr/bin/rsync-ssl
+
+ - Add installed man page: /usr/man/man1/rsync-ssl.1
+
+ - Tweak auxiliary doc file names, such as: README.md, INSTALL.md, NEWS.md, &
+   OLDNEWS.md.
+
+ - The rsync-ssl script wants to run openssl or stunnel4, so consider adding a
+   dependency for one of those options (though it's probably fine to just let
+   it complain about being unable to find the program and let the user decide
+   if they want to install one or the other).
+
+ - If you packaged rsync + rsync-ssl + rsync-ssl-daemon as separate packages,
+   the rsync-ssl package is now gone (rsync-ssl should be considered to be
+   mainstream now that Samba requires SSL for its rsync daemon).
+
+ - Add _build_ dependency for liblz4-dev, libxxhash-dev, libzstd-dev, and
+   libssl-dev.  These development libraries will give rsync extra compression
+   algorithms, extra checksum algorithms, and allow use of openssl's crypto
+   lib for (potentially) faster MD4/MD5 checksums.
+
+ - Add _build_ dependency for g++ or clang++ on x86_64 systems to enable the
+   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.
+   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 be installed via `pip3 install --user commonmark` if you
+   want to just install it for the build user (or omit `--user`).
+
+ - Remove yodl _build_ dependency (if it was even listed before).
+
+### DEVELOPER RELATED:
+
+ - 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
+   simple python3 script using the cmarkgfm **or** commonmark library.  This
+   should make it easier to package rsync, since yodl has gotten obscure.
+
+ - Improved some configure checks to work better with strict C99 compilers.
+
+ - Some perl building/packaging scripts were recoded into awk and python3.
+
+ - Some defines in byteorder.h were changed into static inline functions that
+   will help to ensure that the args don't get evaluated multiple times on
+   "careful alignment" hosts.
+
+ - Some code typos were fixed (as pointed out by a Fossies run).
+
+------------------------------------------------------------------------------

README

@@ -1,127 +0,0 @@
-WHAT IS RSYNC?
---------------
-
-rsync is a replacement for scp/rcp that has many more features.
-
-rsync uses the "rsync algorithm" which provides a very fast method for
-bringing remote files into sync. It does this by sending just the
-differences in the files across the link, without requiring that both
-sets of files are present at one of the ends of the link beforehand.
-At first glance this may seem impossible because the calculation of
-diffs between two files normally requires local access to both
-files. 
-
-A technical report describing the rsync algorithm is included with
-this package. 
-
-
-USAGE
------
-
-Basically you use rsync just like rcp, but rsync has many additional
-options.  To get a complete list of supported options type
-
-  rsync --help
-
-and see the manual for more information.
-
-
-SETUP
------
-
-Rsync normally uses ssh or rsh for communication.  It does not need to
-be setuid and requires no special privileges for installation.  You
-must, however, have a working ssh or rsh system.  Using ssh is
-recommended for its security features.
-
-Alternatively, rsync can run in `daemon' mode, listening on a socket.
-This is generally used for public file distribution, although
-authentication and access control are available.
-
-To install rsync, first run the "configure" script. This will create a
-Makefile and config.h appropriate for your system. Then type
-"make". 
-
-Note that on some systems you will have to force configure not to use
-gcc because gcc may not support some features (such as 64 bit file
-offsets) that your system may support. Set the environment variable CC
-to the name of your native compiler before running configure in this
-case. 
-
-Once built put a copy of rsync in your search path on the local and
-remote systems (or use "make install"). That's it!
-
-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
-connect to an rsync daemon.
-
-
-MAILING LIST
-------------
-
-There is a mailing list for the discussion of rsync and its
-applications. It is open to anyone to join. I will announce new
-versions on this list. 
-
-To join the mailing list see the web page at http://lists.samba.org/
-
-To send mail to everyone on the list send it to rsync@lists.samba.org
-
-
-BUG REPORTS
------------
-
-If you have web access then please look at
-
-  http://rsync.samba.org/
-
-That page contains links to the current bug list, and information on
-how to report a bug well.  You might also like to try searching the
-internet for the error message you've received, or looking in the
-mailing list archives at
-
-  http://mail-archive.com/rsync@lists.samba.org/
-
-To send a bug report, follow the instructions on the bug-tracking
-page of the web site.
-
-If you don't have web access, email your bug report to
-rsync@lists.samba.org.
-
-
-CVS TREE
---------
-
-If you want to get the very latest version of rsync direct from the
-source code repository then you can use anonymous cvs. You will need a
-recent version of cvs then use the following commands:
-
-	cvs -d :pserver:cvs@pserver.samba.org:/cvsroot login
-	Password: cvs
-
-	cvs -d :pserver:cvs@pserver.samba.org:/cvsroot co rsync
-
-Look at the cvs documentation for more details.
-
-
-COPYRIGHT
----------
-
-rsync was originally written by Andrew Tridgell and has been improved
-by many developers around the world.  rsync may be used, modified and
-redistributed only under the terms of the GNU General Public License,
-found in the file COPYING in this distribution, or at 
-
-  http://www.fsf.org/licenses/gpl.html
-
-
-AVAILABILITY
-------------
-
-The main web site for rsync is http://rsync.samba.org/
-The main ftp site is ftp://rsync.samba.org/pub/rsync/
-This is also available as rsync://rsync.samba.org/rsyncftp/

README.md

@@ -0,0 +1,149 @@
+WHAT IS RSYNC?
+--------------
+
+Rsync is a fast and extraordinarily versatile file copying tool for
+both remote and local files.
+
+Rsync uses a delta-transfer algorithm which provides a very fast method
+for bringing remote files into sync.  It does this by sending just the
+differences in the files across the link, without requiring that both
+sets of files are present at one of the ends of the link beforehand.  At
+first glance this may seem impossible because the calculation of diffs
+between two files normally requires local access to both files.
+
+A technical report describing the rsync algorithm is included with this
+package.
+
+
+USAGE
+-----
+
+Basically you use rsync just like scp, but rsync has many additional
+options.  To get a complete list of supported options type:
+
+    rsync --help
+
+See the manpage for more detailed information.
+
+
+SETUP
+-----
+
+Rsync normally uses ssh or rsh for communication with remote systems.
+It does not need to be setuid and requires no special privileges for
+installation.  You must, however, have a working ssh or rsh system.
+Using ssh is recommended for its security features.
+
+Alternatively, rsync can run in `daemon' mode, listening on a socket.
+This is generally used for public file distribution, although
+authentication and access control are available.
+
+To install rsync, first run the "configure" script.  This will create a
+Makefile and config.h appropriate for your system.  Then type "make".
+
+Note that on some systems you will have to force configure not to use
+gcc because gcc may not support some features (such as 64 bit file
+offsets) that your system may support.  Set the environment variable CC
+to the name of your native compiler before running configure in this
+case.
+
+Once built put a copy of rsync in your search path on the local and
+remote systems (or use "make install").  That's it!
+
+
+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
+connect to an rsync daemon.
+
+
+WEB SITE
+--------
+
+The main rsync web site is here:
+
+>  http://rsync.samba.org/
+
+You'll find a FAQ list, downloads, resources, HTML versions of the
+manpages, etc.
+
+
+MAILING LISTS
+-------------
+
+There is a mailing list for the discussion of rsync and its applications
+that is open to anyone to join.  New releases are announced on this
+list, and there is also an announcement-only mailing list for those that
+want official announcements.  See the mailing-list page for full
+details:
+
+>  http://rsync.samba.org/lists.html
+
+
+BUG REPORTS
+-----------
+
+To visit this web page for full the details on bug reporting:
+
+>  http://rsync.samba.org/bugzilla.html
+
+That page contains links to the current bug list, and information on how
+to report a bug well.  You might also like to try searching the Internet
+for the error message you've received, or looking in the mailing list
+archives at:
+
+>  http://mail-archive.com/rsync@lists.samba.org/
+
+To send a bug report, follow the instructions on the bug-tracking
+page of the web site.
+
+Alternately, email your bug report to <rsync@lists.samba.org>.
+
+
+GIT REPOSITORY
+--------------
+
+If you want to get the very latest version of rsync direct from the
+source code repository, then you will need to use git.  The git repo
+is hosted on github and on samba's site.  Feel free to access it here:
+
+>  https://github.com/WayneD/rsync
+
+or clone it from its samba repo:
+
+>     git clone git://git.samba.org/rsync.git
+
+See the download page for full details on all the ways to grab the
+source:
+
+>  http://rsync.samba.org/download.html
+
+
+COPYRIGHT
+---------
+
+Rsync was originally written by Andrew Tridgell and is currently
+maintained by Wayne Davison.   It has been improved by many developers
+from around the world.
+
+Rsync may be used, modified and redistributed only under the terms of
+the GNU General Public License, found in the file COPYING in this
+distribution, or at:
+
+>  http://www.fsf.org/licenses/gpl.html
+
+
+AVAILABILITY
+------------
+
+The main web site for rsync is http://rsync.samba.org/
+
+The main ftp site is ftp://rsync.samba.org/pub/rsync/
+
+This is also available as rsync://download.samba.org/rsyncftp/ if you
+connect via ssl. Use the `rsync-ssl` script if you have it, otherwise
+connect to the rsync server via a normal rsync command and it will
+output some instructions for how to connect.

TODO

@@ -94,7 +94,7 @@ Handling IPv6 on old machines
   platforms that have a half-working implementation, so redefining
   these functions clashes with system headers, and leaving them out
   breaks.  This affects at least OSF/1, RedHat 5, and Cobalt, which
-  are moderately improtant.
+  are moderately important.
 
   Perhaps the simplest solution would be to have two different files
   implementing the same interface, and choose either the new or the
@@ -236,7 +236,7 @@ Memory accounting
 
   At exit, show how much memory was used for the file list, etc.
 
-  Also we do a wierd exponential-growth allocation in flist.c.  I'm
+  We also do a weird exponential-growth allocation in flist.c.  I'm
   not sure this makes sense with modern mallocs.  At any rate it will
   make us allocate a huge amount of memory for large file lists.
 

access.c

@@ -2,7 +2,7 @@
  * Routines to authenticate access to a daemon (hosts allow/deny).
  *
  * Copyright (C) 1998 Andrew Tridgell
- * Copyright (C) 2004-2007 Wayne Davison
+ * Copyright (C) 2004-2020 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
@@ -20,14 +20,50 @@
 
 #include "rsync.h"
 
-static int match_hostname(char *host, char *tok)
+static int allow_forward_dns;
+
+extern const char undetermined_hostname[];
+
+static int match_hostname(const char **host_ptr, const char *addr, const char *tok)
 {
+	struct hostent *hp;
+	unsigned int i;
+	const char *host = *host_ptr;
+
 	if (!host || !*host)
 		return 0;
-	return wildmatch(tok, host);
+
+	/* First check if the reverse-DNS-determined hostname matches. */
+	if (iwildmatch(tok, host))
+		return 1;
+
+	if (!allow_forward_dns)
+		return 0;
+
+	/* Fail quietly if tok is an address or wildcarded entry, not a simple hostname. */
+	if (!tok[strspn(tok, ".0123456789")] || tok[strcspn(tok, ":/*?[")])
+		return 0;
+
+	/* Now try forward-DNS on the token (config-specified hostname) and see if the IP matches. */
+	if (!(hp = gethostbyname(tok)))
+		return 0;
+
+	for (i = 0; hp->h_addr_list[i] != NULL; i++) {
+		if (strcmp(addr, inet_ntoa(*(struct in_addr*)(hp->h_addr_list[i]))) == 0) {
+			/* If reverse lookups are off, we'll use the conf-specified
+			 * hostname in preference to UNDETERMINED. */
+			if (host == undetermined_hostname) {
+				if (!(*host_ptr = strdup(tok)))
+					*host_ptr = undetermined_hostname;
+			}
+			return 1;
+		}
+	}
+
+	return 0;
 }
 
-static int match_binary(char *b1, char *b2, char *mask, int addrlen)
+static int match_binary(const char *b1, const char *b2, const char *mask, int addrlen)
 {
 	int i;
 
@@ -56,7 +92,7 @@ static void make_mask(char *mask, int plen, int addrlen)
 	return;
 }
 
-static int match_address(char *addr, char *tok)
+static int match_address(const char *addr, const char *tok)
 {
 	char *p;
 	struct addrinfo hints, *resa, *rest;
@@ -70,24 +106,16 @@ static int match_address(char *addr, char *tok)
 #endif
 	char mask[16];
 	char *a = NULL, *t = NULL;
-	unsigned int len;
 
 	if (!addr || !*addr)
 		return 0;
 
 	p = strchr(tok,'/');
-	if (p) {
+	if (p)
 		*p = '\0';
-		len = p - tok;
-	} else
-		len = strlen(tok);
 
-	/* Fail quietly if tok is a hostname (not an address) */
-	if (strspn(tok, ".0123456789") != len
-#ifdef INET6
-	    && strchr(tok, ':') == NULL
-#endif
-	) {
+	/* Fail quietly if tok is a hostname, not an address. */
+	if (tok[strspn(tok, ".0123456789")] && strchr(tok, ':') == NULL) {
 		if (p)
 			*p = '/';
 		return 0;
@@ -130,8 +158,7 @@ static int match_address(char *addr, char *tok)
 		break;
 
 #ifdef INET6
-	case PF_INET6:
-	    {
+	case PF_INET6: {
 		struct sockaddr_in6 *sin6a, *sin6t;
 
 		sin6a = (struct sockaddr_in6 *)resa->ai_addr;
@@ -143,20 +170,19 @@ static int match_address(char *addr, char *tok)
 		addrlen = 16;
 
 #ifdef HAVE_SOCKADDR_IN6_SCOPE_ID
-		if (sin6t->sin6_scope_id &&
-		    sin6a->sin6_scope_id != sin6t->sin6_scope_id) {
+		if (sin6t->sin6_scope_id && sin6a->sin6_scope_id != sin6t->sin6_scope_id) {
 			ret = 0;
 			goto out;
 		}
 #endif
 
 		break;
-	    }
+	}
 #endif
 	default:
-	    rprintf(FLOG, "unknown family %u\n", rest->ai_family);
-	    ret = 0;
-	    goto out;
+		rprintf(FLOG, "unknown family %u\n", rest->ai_family);
+		ret = 0;
+		goto out;
 	}
 
 	bits = -1;
@@ -210,7 +236,7 @@ static int match_address(char *addr, char *tok)
 	return ret;
 }
 
-static int access_match(char *list, char *addr, char *host)
+static int access_match(const char *list, const char *addr, const char **host_ptr)
 {
 	char *tok;
 	char *list2 = strdup(list);
@@ -219,11 +245,9 @@ static int access_match(char *list, char *addr, char *host)
 		out_of_memory("access_match");
 
 	strlower(list2);
-	if (host)
-		strlower(host);
 
 	for (tok = strtok(list2, " ,\t"); tok; tok = strtok(NULL, " ,\t")) {
-		if (match_hostname(host, tok) || match_address(addr, tok)) {
+		if (match_hostname(host_ptr, addr, tok) || match_address(addr, tok)) {
 			free(list2);
 			return 1;
 		}
@@ -233,16 +257,21 @@ static int access_match(char *list, char *addr, char *host)
 	return 0;
 }
 
-int allow_access(char *addr, char *host, char *allow_list, char *deny_list)
+int allow_access(const char *addr, const char **host_ptr, int i)
 {
+	const char *allow_list = lp_hosts_allow(i);
+	const char *deny_list = lp_hosts_deny(i);
+
 	if (allow_list && !*allow_list)
 		allow_list = NULL;
 	if (deny_list && !*deny_list)
 		deny_list = NULL;
 
+	allow_forward_dns = lp_forward_lookup(i);
+
 	/* If we match an allow-list item, we always allow access. */
 	if (allow_list) {
-		if (access_match(allow_list, addr, host))
+		if (access_match(allow_list, addr, host_ptr))
 			return 1;
 		/* For an allow-list w/o a deny-list, disallow non-matches. */
 		if (!deny_list)
@@ -251,7 +280,7 @@ int allow_access(char *addr, char *host, char *allow_list, char *deny_list)
 
 	/* If we match a deny-list item (and got past any allow-list
 	 * items), we always disallow access. */
-	if (deny_list && access_match(deny_list, addr, host))
+	if (deny_list && access_match(deny_list, addr, host_ptr))
 		return 0;
 
 	/* Allow all other access. */

acls.c

@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1996 Andrew Tridgell
  * Copyright (C) 1996 Paul Mackerras
- * Copyright (C) 2006 Wayne Davison
+ * Copyright (C) 2006-2020 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
@@ -31,6 +31,8 @@ extern int list_only;
 extern int orig_umask;
 extern int numeric_ids;
 extern int inc_recurse;
+extern int preserve_devices;
+extern int preserve_specials;
 
 /* Flags used to indicate what items are being transmitted for an entry. */
 #define XMIT_USER_OBJ (1<<0)
@@ -46,7 +48,7 @@ extern int inc_recurse;
 /* When we send the access bits over the wire, we shift them 2 bits to the
  * left and use the lower 2 bits as flags (relevant only to a name entry).
  * This makes the protocol more efficient than sending a value that would
- * be likely to have its hightest bits set. */
+ * be likely to have its highest bits set. */
 #define XFLAG_NAME_FOLLOWS 0x0001u
 #define XFLAG_NAME_IS_USER 0x0002u
 
@@ -88,13 +90,26 @@ static const rsync_acl empty_rsync_acl = {
 static item_list access_acl_list = EMPTY_ITEM_LIST;
 static item_list default_acl_list = EMPTY_ITEM_LIST;
 
+static size_t prior_access_count = (size_t)-1;
+static size_t prior_default_count = (size_t)-1;
+
 /* === Calculations on ACL types === */
 
 static const char *str_acl_type(SMB_ACL_TYPE_T type)
 {
-	return type == SMB_ACL_TYPE_ACCESS ? "SMB_ACL_TYPE_ACCESS"
-	     : type == SMB_ACL_TYPE_DEFAULT ? "SMB_ACL_TYPE_DEFAULT"
-	     : "unknown SMB_ACL_TYPE_T";
+	switch (type) {
+	case SMB_ACL_TYPE_ACCESS:
+#ifdef HAVE_OSX_ACLS
+		return "ACL_TYPE_EXTENDED";
+#else
+		return "ACL_TYPE_ACCESS";
+#endif
+	case SMB_ACL_TYPE_DEFAULT:
+		return "ACL_TYPE_DEFAULT";
+	default:
+		break;
+	}
+	return "unknown ACL type!";
 }
 
 static int calc_sacl_entries(const rsync_acl *racl)
@@ -102,10 +117,11 @@ static int calc_sacl_entries(const rsync_acl *racl)
 	/* A System ACL always gets user/group/other permission entries. */
 	return racl->names.count
 #ifdef ACLS_NEED_MASK
-	     + 4;
+	     + 1
 #else
-	     + (racl->mask_obj != NO_ENTRY) + 3;
+	     + (racl->mask_obj != NO_ENTRY)
 #endif
+	     + 3;
 }
 
 /* Extracts and returns the permission bits from the ACL.  This cannot be
@@ -119,15 +135,21 @@ static int rsync_acl_get_perms(const rsync_acl *racl)
 
 /* Removes the permission-bit entries from the ACL because these
  * can be reconstructed from the file's mode. */
-static void rsync_acl_strip_perms(rsync_acl *racl)
+static void rsync_acl_strip_perms(stat_x *sxp)
 {
+	rsync_acl *racl = sxp->acc_acl;
+
 	racl->user_obj = NO_ENTRY;
 	if (racl->mask_obj == NO_ENTRY)
 		racl->group_obj = NO_ENTRY;
 	else {
-		if (racl->group_obj == racl->mask_obj)
+		int group_perms = (sxp->st.st_mode >> 3) & 7;
+		if (racl->group_obj == group_perms)
 			racl->group_obj = NO_ENTRY;
-		racl->mask_obj = NO_ENTRY;
+#ifndef HAVE_SOLARIS_ACLS
+		if (racl->names.count != 0 && racl->mask_obj == group_perms)
+			racl->mask_obj = NO_ENTRY;
+#endif
 	}
 	racl->other_obj = NO_ENTRY;
 }
@@ -300,7 +322,7 @@ static BOOL unpack_smb_acl(SMB_ACL_T sacl, rsync_acl *racl)
 		ida->access = access;
 	}
 	if (rc) {
-		rsyserr(FERROR, errno, "unpack_smb_acl: %s()", errfun);
+		rsyserr(FERROR_XFER, errno, "unpack_smb_acl: %s()", errfun);
 		rsync_acl_free(racl);
 		return False;
 	}
@@ -310,14 +332,12 @@ static BOOL unpack_smb_acl(SMB_ACL_T sacl, rsync_acl *racl)
 	if (temp_ida_list.count) {
 #ifdef SMB_ACL_NEED_SORT
 		if (temp_ida_list.count > 1) {
-			qsort(temp_ida_list.items, temp_ida_list.count,
-			      sizeof (id_access), id_access_sorter);
+			qsort(temp_ida_list.items, temp_ida_list.count, sizeof (id_access), id_access_sorter);
 		}
 #endif
 		if (!(racl->names.idas = new_array(id_access, temp_ida_list.count)))
 			out_of_memory("unpack_smb_acl");
-		memcpy(racl->names.idas, temp_ida_list.items,
-		       temp_ida_list.count * sizeof (id_access));
+		memcpy(racl->names.idas, temp_ida_list.items, temp_ida_list.count * sizeof (id_access));
 	} else
 		racl->names.idas = NULL;
 
@@ -326,15 +346,6 @@ static BOOL unpack_smb_acl(SMB_ACL_T sacl, rsync_acl *racl)
 	/* Truncate the temporary list now that its idas have been saved. */
 	temp_ida_list.count = 0;
 
-#ifdef ACLS_NEED_MASK
-	if (!racl->names.count && racl->mask_obj != NO_ENTRY) {
-		/* Throw away a superfluous mask, but mask off the
-		 * group perms with it first. */
-		racl->group_obj &= racl->mask_obj;
-		racl->mask_obj = NO_ENTRY;
-	}
-#endif
-
 	return True;
 }
 
@@ -356,7 +367,7 @@ static BOOL unpack_smb_acl(SMB_ACL_T sacl, rsync_acl *racl)
 static int store_access_in_entry(uint32 access, SMB_ACL_ENTRY_T entry)
 {
 	if (sys_acl_set_access_bits(entry, access)) {
-		rsyserr(FERROR, errno, "store_access_in_entry sys_acl_set_access_bits()");
+		rsyserr(FERROR_XFER, errno, "store_access_in_entry sys_acl_set_access_bits()");
 		return -1;
 	}
 	return 0;
@@ -375,7 +386,7 @@ static BOOL pack_smb_acl(SMB_ACL_T *smb_acl, const rsync_acl *racl)
 	SMB_ACL_ENTRY_T entry;
 
 	if (!(*smb_acl = sys_acl_init(calc_sacl_entries(racl)))) {
-		rsyserr(FERROR, errno, "pack_smb_acl: sys_acl_init()");
+		rsyserr(FERROR_XFER, errno, "pack_smb_acl: sys_acl_init()");
 		return False;
 	}
 
@@ -410,7 +421,7 @@ static BOOL pack_smb_acl(SMB_ACL_T *smb_acl, const rsync_acl *racl)
 #ifdef ACLS_NEED_MASK
 	mask_bits = racl->mask_obj == NO_ENTRY ? racl->group_obj & ~NO_ENTRY : racl->mask_obj;
 	COE( sys_acl_create_entry,(smb_acl, &entry) );
-	COE( sys_acl_set_info,(entry, SMB_ACL_MASK, mask_bits, NULL) );
+	COE( sys_acl_set_info,(entry, SMB_ACL_MASK, mask_bits, 0) );
 #else
 	if (racl->mask_obj != NO_ENTRY) {
 		COE( sys_acl_create_entry,(smb_acl, &entry) );
@@ -424,14 +435,14 @@ static BOOL pack_smb_acl(SMB_ACL_T *smb_acl, const rsync_acl *racl)
 
 #ifdef DEBUG
 	if (sys_acl_valid(*smb_acl) < 0)
-		rprintf(FERROR, "pack_smb_acl: warning: system says the ACL I packed is invalid\n");
+		rprintf(FERROR_XFER, "pack_smb_acl: warning: system says the ACL I packed is invalid\n");
 #endif
 
 	return True;
 
   error_exit:
 	if (errfun) {
-		rsyserr(FERROR, errno, "pack_smb_acl %s()", errfun);
+		rsyserr(FERROR_XFER, errno, "pack_smb_acl %s()", errfun);
 	}
 	sys_acl_free_acl(*smb_acl);
 	return False;
@@ -476,13 +487,21 @@ static int get_rsync_acl(const char *fname, rsync_acl *racl,
 		if ((buf = get_xattr_acl(fname, type == SMB_ACL_TYPE_ACCESS, &len)) == NULL)
 			return 0;
 		cnt = (len - 4*4) / (4+4);
-		if (len < 4*4 || len != (size_t)cnt*(4+4) + 4*4)
+		if (len < 4*4 || len != (size_t)cnt*(4+4) + 4*4) {
+			free(buf);
 			return -1;
+		}
 
 		racl->user_obj = IVAL(buf, 0);
+		if (racl->user_obj == NO_ENTRY)
+			racl->user_obj = (mode >> 6) & 7;
 		racl->group_obj = IVAL(buf, 4);
+		if (racl->group_obj == NO_ENTRY)
+			racl->group_obj = (mode >> 3) & 7;
 		racl->mask_obj = IVAL(buf, 8);
 		racl->other_obj = IVAL(buf, 12);
+		if (racl->other_obj == NO_ENTRY)
+			racl->other_obj = mode & 7;
 
 		if (cnt) {
 			char *bp = buf + 4*4;
@@ -512,7 +531,7 @@ static int get_rsync_acl(const char *fname, rsync_acl *racl,
 		if (type == SMB_ACL_TYPE_ACCESS)
 			rsync_acl_fake_perms(racl, mode);
 	} else {
-		rsyserr(FERROR, errno, "get_acl: sys_acl_get_file(%s, %s)",
+		rsyserr(FERROR_XFER, errno, "get_acl: sys_acl_get_file(%s, %s)",
 			fname, str_acl_type(type));
 		return -1;
 	}
@@ -524,6 +543,24 @@ static int get_rsync_acl(const char *fname, rsync_acl *racl,
 int get_acl(const char *fname, stat_x *sxp)
 {
 	sxp->acc_acl = create_racl();
+
+	if (S_ISREG(sxp->st.st_mode) || S_ISDIR(sxp->st.st_mode)) {
+		/* Everyone supports this. */
+	} else if (S_ISLNK(sxp->st.st_mode)) {
+		return 0;
+	} else if (IS_SPECIAL(sxp->st.st_mode)) {
+#ifndef NO_SPECIAL_ACLS
+		if (!preserve_specials)
+#endif
+			return 0;
+	} else if (IS_DEVICE(sxp->st.st_mode)) {
+#ifndef NO_DEVICE_ACLS
+		if (!preserve_devices)
+#endif
+			return 0;
+	} else if (IS_MISSING_FILE(sxp->st))
+		return 0;
+
 	if (get_rsync_acl(fname, sxp->acc_acl, SMB_ACL_TYPE_ACCESS,
 			  sxp->st.st_mode) < 0) {
 		free_acl(sxp);
@@ -545,7 +582,7 @@ int get_acl(const char *fname, stat_x *sxp)
 /* === Send functions === */
 
 /* Send the ida list over the file descriptor. */
-static void send_ida_entries(const ida_entries *idal, int f)
+static void send_ida_entries(int f, const ida_entries *idal)
 {
 	id_access *ida;
 	size_t count = idal->count;
@@ -554,12 +591,12 @@ static void send_ida_entries(const ida_entries *idal, int f)
 
 	for (ida = idal->idas; count--; ida++) {
 		uint32 xbits = ida->access << 2;
-		char *name;
+		const char *name;
 		if (ida->access & NAME_IS_USER) {
 			xbits |= XFLAG_NAME_IS_USER;
-			name = add_uid(ida->id);
+			name = numeric_ids ? NULL : add_uid(ida->id);
 		} else
-			name = add_gid(ida->id);
+			name = numeric_ids ? NULL : add_gid(ida->id);
 		write_varint(f, ida->id);
 		if (inc_recurse && name) {
 			int len = strlen(name);
@@ -571,8 +608,8 @@ static void send_ida_entries(const ida_entries *idal, int f)
 	}
 }
 
-static void send_rsync_acl(rsync_acl *racl, SMB_ACL_TYPE_T type,
-			   item_list *racl_list, int f)
+static void send_rsync_acl(int f, rsync_acl *racl, SMB_ACL_TYPE_T type,
+			   item_list *racl_list)
 {
 	int ndx = find_matching_rsync_acl(racl, type, racl_list);
 
@@ -605,7 +642,7 @@ static void send_rsync_acl(rsync_acl *racl, SMB_ACL_TYPE_T type,
 		if (flags & XMIT_OTHER_OBJ)
 			write_varint(f, racl->other_obj);
 		if (flags & XMIT_NAME_LIST)
-			send_ida_entries(&racl->names, f);
+			send_ida_entries(f, &racl->names);
 
 		/* Give the allocated data to the new list object. */
 		*new_racl = *racl;
@@ -615,28 +652,28 @@ static void send_rsync_acl(rsync_acl *racl, SMB_ACL_TYPE_T type,
 
 /* Send the ACL from the stat_x structure down the indicated file descriptor.
  * This also frees the ACL data. */
-void send_acl(stat_x *sxp, int f)
+void send_acl(int f, stat_x *sxp)
 {
 	if (!sxp->acc_acl) {
 		sxp->acc_acl = create_racl();
 		rsync_acl_fake_perms(sxp->acc_acl, sxp->st.st_mode);
 	}
 	/* Avoid sending values that can be inferred from other data. */
-	rsync_acl_strip_perms(sxp->acc_acl);
+	rsync_acl_strip_perms(sxp);
 
-	send_rsync_acl(sxp->acc_acl, SMB_ACL_TYPE_ACCESS, &access_acl_list, f);
+	send_rsync_acl(f, sxp->acc_acl, SMB_ACL_TYPE_ACCESS, &access_acl_list);
 
 	if (S_ISDIR(sxp->st.st_mode)) {
 		if (!sxp->def_acl)
 			sxp->def_acl = create_racl();
 
-		send_rsync_acl(sxp->def_acl, SMB_ACL_TYPE_DEFAULT, &default_acl_list, f);
+		send_rsync_acl(f, sxp->def_acl, SMB_ACL_TYPE_DEFAULT, &default_acl_list);
 	}
 }
 
 /* === Receive functions === */
 
-static uint32 recv_acl_access(uchar *name_follows_ptr, int f)
+static uint32 recv_acl_access(int f, uchar *name_follows_ptr)
 {
 	uint32 access = read_varint(f);
 
@@ -653,7 +690,7 @@ static uint32 recv_acl_access(uchar *name_follows_ptr, int f)
 			access |= NAME_IS_USER;
 	} else if (am_root >= 0 && access & ~SMB_ACL_VALID_OBJ_BITS) {
 	  value_error:
-		rprintf(FERROR, "recv_acl_access: value out of range: %x\n",
+		rprintf(FERROR_XFER, "recv_acl_access: value out of range: %x\n",
 			access);
 		exit_cleanup(RERR_STREAMIO);
 	}
@@ -661,7 +698,7 @@ static uint32 recv_acl_access(uchar *name_follows_ptr, int f)
 	return access;
 }
 
-static uchar recv_ida_entries(ida_entries *ent, int f)
+static uchar recv_ida_entries(int f, ida_entries *ent)
 {
 	uchar computed_mask_bits = 0;
 	int i, count = read_varint(f);
@@ -677,7 +714,7 @@ static uchar recv_ida_entries(ida_entries *ent, int f)
 	for (i = 0; i < count; i++) {
 		uchar has_name;
 		id_t id = read_varint(f);
-		uint32 access = recv_acl_access(&has_name, f);
+		uint32 access = recv_acl_access(f, &has_name);
 
 		if (has_name) {
 			if (access & NAME_IS_USER)
@@ -700,7 +737,7 @@ static uchar recv_ida_entries(ida_entries *ent, int f)
 	return computed_mask_bits & ~NO_ENTRY;
 }
 
-static int recv_rsync_acl(item_list *racl_list, SMB_ACL_TYPE_T type, int f)
+static int recv_rsync_acl(int f, item_list *racl_list, SMB_ACL_TYPE_T type, mode_t mode)
 {
 	uchar computed_mask_bits = 0;
 	acl_duo *duo_item;
@@ -708,14 +745,14 @@ static int recv_rsync_acl(item_list *racl_list, SMB_ACL_TYPE_T type, int f)
 	int ndx = read_varint(f);
 
 	if (ndx < 0 || (size_t)ndx > racl_list->count) {
-		rprintf(FERROR, "recv_acl_index: %s ACL index %d > %d\n",
+		rprintf(FERROR_XFER, "recv_acl_index: %s ACL index %d > %d\n",
 			str_acl_type(type), ndx, (int)racl_list->count);
 		exit_cleanup(RERR_STREAMIO);
 	}
 
 	if (ndx != 0)
 		return ndx - 1;
-	
+
 	ndx = racl_list->count;
 	duo_item = EXPAND_ITEM_LIST(racl_list, acl_duo, 1000);
 	duo_item->racl = empty_rsync_acl;
@@ -723,29 +760,28 @@ static int recv_rsync_acl(item_list *racl_list, SMB_ACL_TYPE_T type, int f)
 	flags = read_byte(f);
 
 	if (flags & XMIT_USER_OBJ)
-		duo_item->racl.user_obj = recv_acl_access(NULL, f);
+		duo_item->racl.user_obj = recv_acl_access(f, NULL);
 	if (flags & XMIT_GROUP_OBJ)
-		duo_item->racl.group_obj = recv_acl_access(NULL, f);
+		duo_item->racl.group_obj = recv_acl_access(f, NULL);
 	if (flags & XMIT_MASK_OBJ)
-		duo_item->racl.mask_obj = recv_acl_access(NULL, f);
+		duo_item->racl.mask_obj = recv_acl_access(f, NULL);
 	if (flags & XMIT_OTHER_OBJ)
-		duo_item->racl.other_obj = recv_acl_access(NULL, f);
+		duo_item->racl.other_obj = recv_acl_access(f, NULL);
 	if (flags & XMIT_NAME_LIST)
-		computed_mask_bits |= recv_ida_entries(&duo_item->racl.names, f);
+		computed_mask_bits |= recv_ida_entries(f, &duo_item->racl.names);
 
 #ifdef HAVE_OSX_ACLS
 	/* If we received a superfluous mask, throw it away. */
 	duo_item->racl.mask_obj = NO_ENTRY;
 #else
-	if (!duo_item->racl.names.count) {
-		/* If we received a superfluous mask, throw it away. */
-		if (duo_item->racl.mask_obj != NO_ENTRY) {
-			/* Mask off the group perms with it first. */
-			duo_item->racl.group_obj &= duo_item->racl.mask_obj | NO_ENTRY;
-			duo_item->racl.mask_obj = NO_ENTRY;
-		}
-	} else if (duo_item->racl.mask_obj == NO_ENTRY) /* Must be non-empty with lists. */
-		duo_item->racl.mask_obj = (computed_mask_bits | duo_item->racl.group_obj) & ~NO_ENTRY;
+	if (duo_item->racl.names.count && duo_item->racl.mask_obj == NO_ENTRY) {
+		/* Mask must be non-empty with lists. */
+		if (type == SMB_ACL_TYPE_ACCESS)
+			computed_mask_bits = (mode >> 3) & 7;
+		else
+			computed_mask_bits |= duo_item->racl.group_obj & ~NO_ENTRY;
+		duo_item->racl.mask_obj = computed_mask_bits;
+	}
 #endif
 
 	duo_item->sacl = NULL;
@@ -754,12 +790,12 @@ static int recv_rsync_acl(item_list *racl_list, SMB_ACL_TYPE_T type, int f)
 }
 
 /* Receive the ACL info the sender has included for this file-list entry. */
-void receive_acl(struct file_struct *file, int f)
+void receive_acl(int f, struct file_struct *file)
 {
-	F_ACL(file) = recv_rsync_acl(&access_acl_list, SMB_ACL_TYPE_ACCESS, f);
+	F_ACL(file) = recv_rsync_acl(f, &access_acl_list, SMB_ACL_TYPE_ACCESS, file->mode);
 
 	if (S_ISDIR(file->mode))
-		F_DIR_DEFACL(file) = recv_rsync_acl(&default_acl_list, SMB_ACL_TYPE_DEFAULT, f);
+		F_DIR_DEFACL(file) = recv_rsync_acl(f, &default_acl_list, SMB_ACL_TYPE_DEFAULT, 0);
 }
 
 static int cache_rsync_acl(rsync_acl *racl, SMB_ACL_TYPE_T type, item_list *racl_list)
@@ -782,14 +818,45 @@ static int cache_rsync_acl(rsync_acl *racl, SMB_ACL_TYPE_T type, item_list *racl
 
 /* Turn the ACL data in stat_x into cached ACL data, setting the index
  * values in the file struct. */
-void cache_acl(struct file_struct *file, stat_x *sxp)
+void cache_tmp_acl(struct file_struct *file, stat_x *sxp)
 {
-	F_ACL(file) = cache_rsync_acl(sxp->acc_acl,
-				      SMB_ACL_TYPE_ACCESS, &access_acl_list);
+	if (prior_access_count == (size_t)-1)
+		prior_access_count = access_acl_list.count;
+
+	F_ACL(file) = cache_rsync_acl(sxp->acc_acl, SMB_ACL_TYPE_ACCESS, &access_acl_list);
 
 	if (S_ISDIR(sxp->st.st_mode)) {
-		F_DIR_DEFACL(file) = cache_rsync_acl(sxp->def_acl,
-				      SMB_ACL_TYPE_DEFAULT, &default_acl_list);
+		if (prior_default_count == (size_t)-1)
+			prior_default_count = default_acl_list.count;
+		F_DIR_DEFACL(file) = cache_rsync_acl(sxp->def_acl, SMB_ACL_TYPE_DEFAULT, &default_acl_list);
+	}
+}
+
+static void uncache_duo_acls(item_list *duo_list, size_t start)
+{
+	acl_duo *duo_item = duo_list->items;
+	acl_duo *duo_start = duo_item + start;
+
+	duo_item += duo_list->count;
+	duo_list->count = start;
+
+	while (duo_item-- > duo_start) {
+		rsync_acl_free(&duo_item->racl);
+		if (duo_item->sacl)
+			sys_acl_free_acl(duo_item->sacl);
+	}
+}
+
+void uncache_tmp_acls(void)
+{
+	if (prior_access_count != (size_t)-1) {
+		uncache_duo_acls(&access_acl_list, prior_access_count);
+		prior_access_count = (size_t)-1;
+	}
+
+	if (prior_default_count != (size_t)-1) {
+		uncache_duo_acls(&default_acl_list, prior_default_count);
+		prior_default_count = (size_t)-1;
 	}
 }
 
@@ -838,12 +905,14 @@ static mode_t change_sacl_perms(SMB_ACL_T sacl, rsync_acl *racl, mode_t old_mode
 			COE2( store_access_in_entry,((mode >> 3) & 7, entry) );
 			break;
 		case SMB_ACL_MASK:
+#ifndef HAVE_SOLARIS_ACLS
 #ifndef ACLS_NEED_MASK
 			/* mask is only empty when we don't need it. */
 			if (racl->mask_obj == NO_ENTRY)
 				break;
 #endif
 			COE2( store_access_in_entry,((mode >> 3) & 7, entry) );
+#endif
 			break;
 		case SMB_ACL_OTHER:
 			COE2( store_access_in_entry,(mode & 7, entry) );
@@ -853,10 +922,10 @@ static mode_t change_sacl_perms(SMB_ACL_T sacl, rsync_acl *racl, mode_t old_mode
 	if (rc) {
 	  error_exit:
 		if (errfun) {
-			rsyserr(FERROR, errno, "change_sacl_perms: %s()",
+			rsyserr(FERROR_XFER, errno, "change_sacl_perms: %s()",
 				errfun);
 		}
-		return (mode_t)~0;
+		return (mode_t)-1;
 	}
 
 #ifdef SMB_ACL_LOSES_SPECIAL_MODE_BITS
@@ -885,7 +954,7 @@ static int set_rsync_acl(const char *fname, acl_duo *duo_item,
 #endif
 			rc = sys_acl_delete_def_file(fname);
 		if (rc < 0) {
-			rsyserr(FERROR, errno, "set_acl: sys_acl_delete_def_file(%s)",
+			rsyserr(FERROR_XFER, errno, "set_acl: sys_acl_delete_def_file(%s)",
 				fname);
 			return -1;
 		}
@@ -923,15 +992,14 @@ static int set_rsync_acl(const char *fname, acl_duo *duo_item,
 		mode = 0; /* eliminate compiler warning */
 #else
 		if (type == SMB_ACL_TYPE_ACCESS) {
-			cur_mode = change_sacl_perms(duo_item->sacl, &duo_item->racl,
-						     cur_mode, mode);
-			if (cur_mode == (mode_t)~0)
+			cur_mode = change_sacl_perms(duo_item->sacl, &duo_item->racl, cur_mode, mode);
+			if (cur_mode == (mode_t)-1)
 				return 0;
 		}
 #endif
 		if (sys_acl_set_file(fname, type, duo_item->sacl) < 0) {
-			rsyserr(FERROR, errno, "set_acl: sys_acl_set_file(%s, %s)",
-			fname, str_acl_type(type));
+			rsyserr(FERROR_XFER, errno, "set_acl: sys_acl_set_file(%s, %s)",
+				fname, str_acl_type(type));
 			return -1;
 		}
 		if (type == SMB_ACL_TYPE_ACCESS)
@@ -941,17 +1009,17 @@ static int set_rsync_acl(const char *fname, acl_duo *duo_item,
 	return 0;
 }
 
-/* Set ACL on indicated filename.
+/* Given a fname, this sets extended access ACL entries, the default ACL (for a
+ * dir), and the regular mode bits on the file.  Call this with fname set to
+ * NULL to just check if the ACL is different.
  *
- * This sets extended access ACL entries and default ACL.  If convenient,
- * it sets permission bits along with the access ACL and signals having
- * done so by modifying sxp->st.st_mode.
+ * If the ACL operation has a side-effect of changing the file's mode, the
+ * sxp->st.st_mode value will be changed to match.
  *
- * Returns 1 for unchanged, 0 for changed, -1 for failed.  Call this
- * with fname set to NULL to just check if the ACL is unchanged. */
-int set_acl(const char *fname, const struct file_struct *file, stat_x *sxp)
+ * Returns 0 for an unchanged ACL, 1 for changed, -1 for failed. */
+int set_acl(const char *fname, const struct file_struct *file, stat_x *sxp, mode_t new_mode)
 {
-	int unchanged = 1;
+	int changed = 0;
 	int32 ndx;
 	BOOL eq;
 
@@ -965,18 +1033,18 @@ int set_acl(const char *fname, const struct file_struct *file, stat_x *sxp)
 		acl_duo *duo_item = access_acl_list.items;
 		duo_item += ndx;
 		eq = sxp->acc_acl
-		  && rsync_acl_equal_enough(sxp->acc_acl, &duo_item->racl, file->mode);
+		  && rsync_acl_equal_enough(sxp->acc_acl, &duo_item->racl, new_mode);
 		if (!eq) {
-			unchanged = 0;
+			changed = 1;
 			if (!dry_run && fname
 			 && set_rsync_acl(fname, duo_item, SMB_ACL_TYPE_ACCESS,
-					  sxp, file->mode) < 0)
-				unchanged = -1;
+					  sxp, new_mode) < 0)
+				return -1;
 		}
 	}
 
-	if (!S_ISDIR(sxp->st.st_mode))
-		return unchanged;
+	if (!S_ISDIR(new_mode))
+		return changed;
 
 	ndx = F_DIR_DEFACL(file);
 	if (ndx >= 0 && (size_t)ndx < default_acl_list.count) {
@@ -984,16 +1052,15 @@ int set_acl(const char *fname, const struct file_struct *file, stat_x *sxp)
 		duo_item += ndx;
 		eq = sxp->def_acl && rsync_acl_equal(sxp->def_acl, &duo_item->racl);
 		if (!eq) {
-			if (unchanged > 0)
-				unchanged = 0;
+			changed = 1;
 			if (!dry_run && fname
 			 && set_rsync_acl(fname, duo_item, SMB_ACL_TYPE_DEFAULT,
-					  sxp, file->mode) < 0)
-				unchanged = -1;
+					  sxp, new_mode) < 0)
+				return -1;
 		}
 	}
 
-	return unchanged;
+	return changed;
 }
 
 /* Non-incremental recursion needs to convert all the received IDs.
@@ -1036,21 +1103,23 @@ int default_perms_for_dir(const char *dir)
 	if (sacl == NULL) {
 		/* Couldn't get an ACL.  Darn. */
 		switch (errno) {
+		case EINVAL:
+			/* If SMB_ACL_TYPE_DEFAULT isn't valid, then the ACLs must be non-POSIX. */
+			break;
 #ifdef ENOTSUP
 		case ENOTSUP:
 #endif
 		case ENOSYS:
 			/* No ACLs are available. */
 			break;
-		case ENOENT:
-			if (dry_run) {
+		default:
+			if (dry_run && errno == ENOENT) {
 				/* We're doing a dry run, so the containing directory
 				 * wasn't actually created.  Don't worry about it. */
 				break;
 			}
-			/* Otherwise fall through. */
-		default:
-			rprintf(FERROR, "default_perms_for_dir: sys_acl_get_file(%s, %s): %s, falling back on umask\n",
+			rprintf(FWARNING,
+				"default_perms_for_dir: sys_acl_get_file(%s, %s): %s, falling back on umask\n",
 				dir, str_acl_type(SMB_ACL_TYPE_DEFAULT), strerror(errno));
 		}
 		return perms;
@@ -1061,14 +1130,14 @@ int default_perms_for_dir(const char *dir)
 	ok = unpack_smb_acl(sacl, &racl);
 	sys_acl_free_acl(sacl);
 	if (!ok) {
-		rprintf(FERROR, "default_perms_for_dir: unpack_smb_acl failed, falling back on umask\n");
+		rprintf(FWARNING, "default_perms_for_dir: unpack_smb_acl failed, falling back on umask\n");
 		return perms;
 	}
 
 	/* Apply the permission-bit entries of the default ACL, if any. */
 	if (racl.user_obj != NO_ENTRY) {
 		perms = rsync_acl_get_perms(&racl);
-		if (verbose > 2)
+		if (DEBUG_GTE(ACL, 1))
 			rprintf(FINFO, "got ACL-based default perms %o for directory %s\n", perms, dir);
 	}
 

authenticate.c

@@ -2,7 +2,7 @@
  * Support rsync daemon authentication.
  *
  * Copyright (C) 1998-2000 Andrew Tridgell
- * Copyright (C) 2002-2007 Wayne Davison
+ * Copyright (C) 2002-2020 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
@@ -19,7 +19,9 @@
  */
 
 #include "rsync.h"
+#include "itypes.h"
 
+extern int read_only;
 extern char *password_file;
 
 /***************************************************************************
@@ -69,32 +71,48 @@ static void gen_challenge(const char *addr, char *challenge)
 	SIVAL(input, 20, tv.tv_usec);
 	SIVAL(input, 24, getpid());
 
-	sum_init(0);
+	sum_init(-1, 0);
 	sum_update(input, sizeof input);
 	len = sum_end(digest);
 
 	base64_encode(digest, len, challenge, 0);
 }
 
+/* Generate an MD4 hash created from the combination of the password
+ * and the challenge string and return it base64-encoded. */
+static void generate_hash(const char *in, const char *challenge, char *out)
+{
+	char buf[MAX_DIGEST_LEN];
+	int len;
+
+	sum_init(-1, 0);
+	sum_update(in, strlen(in));
+	sum_update(challenge, strlen(challenge));
+	len = sum_end(buf);
+
+	base64_encode(buf, len, out, 0);
+}
 
 /* Return the secret for a user from the secret file, null terminated.
  * Maximum length is len (not counting the null). */
-static int get_secret(int module, const char *user, char *secret, int len)
+static const char *check_secret(int module, const char *user, const char *group,
+				const char *challenge, const char *pass)
 {
+	char line[1024];
+	char pass2[MAX_DIGEST_LEN*2];
 	const char *fname = lp_secrets_file(module);
 	STRUCT_STAT st;
-	int fd, ok = 1;
-	const char *p;
-	char ch, *s;
+	int ok = 1;
+	int user_len = strlen(user);
+	int group_len = group ? strlen(group) : 0;
+	char *err;
+	FILE *fh;
 
-	if (!fname || !*fname)
-		return 0;
+	if (!fname || !*fname || (fh = fopen(fname, "r")) == NULL)
+		return "no secrets file";
 
-	if ((fd = open(fname, O_RDONLY)) < 0)
-		return 0;
-
-	if (do_stat(fname, &st) == -1) {
-		rsyserr(FLOG, errno, "stat(%s)", fname);
+	if (do_fstat(fileno(fh), &st) == -1) {
+		rsyserr(FLOG, errno, "fstat(%s)", fname);
 		ok = 0;
 	} else if (lp_strict_modes(module)) {
 		if ((st.st_mode & 06) != 0) {
@@ -106,115 +124,94 @@ static int get_secret(int module, const char *user, char *secret, int len)
 		}
 	}
 	if (!ok) {
-		rprintf(FLOG, "continuing without secrets file\n");
-		close(fd);
-		return 0;
+		fclose(fh);
+		return "ignoring secrets file";
 	}
 
 	if (*user == '#') {
 		/* Reject attempt to match a comment. */
-		close(fd);
-		return 0;
+		fclose(fh);
+		return "invalid username";
 	}
 
-	/* Try to find a line that starts with the user name and a ':'. */
-	p = user;
-	while (1) {
-		if (read(fd, &ch, 1) != 1) {
-			close(fd);
-			return 0;
-		}
-		if (ch == '\n')
-			p = user;
-		else if (p) {
-			if (*p == ch)
-				p++;
-			else if (!*p && ch == ':')
-				break;
-			else
-				p = NULL;
-		}
-	}
-
-	/* Slurp the secret into the "secret" buffer. */
-	s = secret;
-	while (len > 0) {
-		if (read(fd, s, 1) != 1 || *s == '\n')
-			break;
-		if (*s == '\r')
+	/* Try to find a line that starts with the user (or @group) name and a ':'. */
+	err = "secret not found";
+	while ((user || group) && fgets(line, sizeof line, fh) != NULL) {
+		const char **ptr, *s = strtok(line, "\n\r");
+		int len;
+		if (!s)
 			continue;
-		s++;
-		len--;
+		if (*s == '@') {
+			ptr = &group;
+			len = group_len;
+			s++;
+		} else {
+			ptr = &user;
+			len = user_len;
+		}
+		if (!*ptr || strncmp(s, *ptr, len) != 0 || s[len] != ':')
+			continue;
+		generate_hash(s+len+1, challenge, pass2);
+		if (strcmp(pass, pass2) == 0) {
+			err = NULL;
+			break;
+		}
+		err = "password mismatch";
+		*ptr = NULL; /* Don't look for name again. */
 	}
-	*s = '\0';
-	close(fd);
 
-	return 1;
+	fclose(fh);
+
+	force_memzero(line, sizeof line);
+	force_memzero(pass2, sizeof pass2);
+
+	return err;
 }
 
 static const char *getpassf(const char *filename)
 {
 	STRUCT_STAT st;
 	char buffer[512], *p;
-	int fd, n, ok = 1;
-	const char *envpw = getenv("RSYNC_PASSWORD");
+	int n;
 
 	if (!filename)
 		return NULL;
 
-	if ((fd = open(filename,O_RDONLY)) < 0) {
-		rsyserr(FERROR, errno, "could not open password file \"%s\"",
-			filename);
-		if (envpw)
-			rprintf(FERROR, "falling back to RSYNC_PASSWORD environment variable.\n");
-		return NULL;
-	}
+	if (strcmp(filename, "-") == 0) {
+		n = fgets(buffer, sizeof buffer, stdin) == NULL ? -1 : (int)strlen(buffer);
+	} else {
+		int fd;
 
-	if (do_stat(filename, &st) == -1) {
-		rsyserr(FERROR, errno, "stat(%s)", filename);
-		ok = 0;
-	} else if ((st.st_mode & 06) != 0) {
-		rprintf(FERROR,"password file must not be other-accessible\n");
-		ok = 0;
-	} else if (MY_UID() == 0 && st.st_uid != 0) {
-		rprintf(FERROR,"password file must be owned by root when running as root\n");
-		ok = 0;
-	}
-	if (!ok) {
-		rprintf(FERROR,"continuing without password file\n");
-		if (envpw)
-			rprintf(FERROR, "using RSYNC_PASSWORD environment variable.\n");
+		if ((fd = open(filename,O_RDONLY)) < 0) {
+			rsyserr(FERROR, errno, "could not open password file %s", filename);
+			exit_cleanup(RERR_SYNTAX);
+		}
+
+		if (do_stat(filename, &st) == -1) {
+			rsyserr(FERROR, errno, "stat(%s)", filename);
+			exit_cleanup(RERR_SYNTAX);
+		}
+		if ((st.st_mode & 06) != 0) {
+			rprintf(FERROR, "ERROR: password file must not be other-accessible\n");
+			exit_cleanup(RERR_SYNTAX);
+		}
+		if (MY_UID() == 0 && st.st_uid != 0) {
+			rprintf(FERROR, "ERROR: password file must be owned by root when running as root\n");
+			exit_cleanup(RERR_SYNTAX);
+		}
+
+		n = read(fd, buffer, sizeof buffer - 1);
 		close(fd);
-		return NULL;
 	}
 
-	if (envpw)
-		rprintf(FERROR, "RSYNC_PASSWORD environment variable ignored\n");
-
-	n = read(fd, buffer, sizeof buffer - 1);
-	close(fd);
 	if (n > 0) {
 		buffer[n] = '\0';
 		if ((p = strtok(buffer, "\n\r")) != NULL)
 			return strdup(p);
 	}
 
-	return NULL;
-}
-
-/* Generate an MD4 hash created from the combination of the password
- * and the challenge string and return it base64-encoded. */
-static void generate_hash(const char *in, const char *challenge, char *out)
-{
-	char buf[MAX_DIGEST_LEN];
-	int len;
-
-	sum_init(0);
-	sum_update(in, strlen(in));
-	sum_update(challenge, strlen(challenge));
-	len = sum_end(buf);
-
-	base64_encode(buf, len, out, 0);
+	rprintf(FERROR, "ERROR: failed to read a password from %s\n", filename);
+	exit_cleanup(RERR_SYNTAX);
 }
 
 /* Possibly negotiate authentication with the client.  Use "leader" to
@@ -229,9 +226,12 @@ char *auth_server(int f_in, int f_out, int module, const char *host,
 	char *users = lp_auth_users(module);
 	char challenge[MAX_DIGEST_LEN*2];
 	char line[BIGPATHBUFLEN];
-	char secret[512];
-	char pass2[MAX_DIGEST_LEN*2];
+	char **auth_uid_groups = NULL;
+	int auth_uid_groups_cnt = -1;
+	const char *err = NULL;
+	int group_match = -1;
 	char *tok, *pass;
+	char opt_ch = '\0';
 
 	/* if no auth list then allow anyone in! */
 	if (!users || !*users)
@@ -241,7 +241,7 @@ char *auth_server(int f_in, int f_out, int module, const char *host,
 
 	io_printf(f_out, "%s%s\n", leader, challenge);
 
-	if (!read_line_old(f_in, line, sizeof line)
+	if (!read_line_old(f_in, line, sizeof line, 0)
 	 || (pass = strchr(line, ' ')) == NULL) {
 		rprintf(FLOG, "auth failed on module %s from %s (%s): "
 			"invalid challenge response\n",
@@ -254,36 +254,92 @@ char *auth_server(int f_in, int f_out, int module, const char *host,
 		out_of_memory("auth_server");
 
 	for (tok = strtok(users, " ,\t"); tok; tok = strtok(NULL, " ,\t")) {
-		if (wildmatch(tok, line))
-			break;
+		char *opts;
+		/* See if the user appended :deny, :ro, or :rw. */
+		if ((opts = strchr(tok, ':')) != NULL) {
+			*opts++ = '\0';
+			opt_ch = isUpper(opts) ? toLower(opts) : *opts;
+			if (opt_ch == 'r') { /* handle ro and rw */
+				opt_ch = isUpper(opts+1) ? toLower(opts+1) : opts[1];
+				if (opt_ch == 'o')
+					opt_ch = 'r';
+				else if (opt_ch != 'w')
+					opt_ch = '\0';
+			} else if (opt_ch != 'd') /* if it's not deny, ignore it */
+				opt_ch = '\0';
+		} else
+			opt_ch = '\0';
+		if (*tok != '@') {
+			/* Match the username */
+			if (wildmatch(tok, line))
+				break;
+		} else {
+#ifdef HAVE_GETGROUPLIST
+			int j;
+			/* See if authorizing user is a real user, and if so, see
+			 * if it is in a group that matches tok+1 wildmat. */
+			if (auth_uid_groups_cnt < 0) {
+				item_list gid_list = EMPTY_ITEM_LIST;
+				uid_t auth_uid;
+				if (!user_to_uid(line, &auth_uid, False)
+				 || getallgroups(auth_uid, &gid_list) != NULL)
+					auth_uid_groups_cnt = 0;
+				else {
+					gid_t *gid_array = gid_list.items;
+					auth_uid_groups_cnt = gid_list.count;
+					if ((auth_uid_groups = new_array(char *, auth_uid_groups_cnt)) == NULL)
+						out_of_memory("auth_server");
+					for (j = 0; j < auth_uid_groups_cnt; j++)
+						auth_uid_groups[j] = gid_to_group(gid_array[j]);
+				}
+			}
+			for (j = 0; j < auth_uid_groups_cnt; j++) {
+				if (auth_uid_groups[j] && wildmatch(tok+1, auth_uid_groups[j])) {
+					group_match = j;
+					break;
+				}
+			}
+			if (group_match >= 0)
+				break;
+#else
+			rprintf(FLOG, "your computer doesn't support getgrouplist(), so no @group authorization is possible.\n");
+#endif
+		}
 	}
+
 	free(users);
 
-	if (!tok) {
-		rprintf(FLOG, "auth failed on module %s from %s (%s): "
-			"unauthorized user\n",
-			lp_name(module), host, addr);
+	if (!tok)
+		err = "no matching rule";
+	else if (opt_ch == 'd')
+		err = "denied by rule";
+	else {
+		char *group = group_match >= 0 ? auth_uid_groups[group_match] : NULL;
+		err = check_secret(module, line, group, challenge, pass);
+	}
+
+	force_memzero(challenge, sizeof challenge);
+	force_memzero(pass, strlen(pass));
+
+	if (auth_uid_groups) {
+		int j;
+		for (j = 0; j < auth_uid_groups_cnt; j++) {
+			if (auth_uid_groups[j])
+				free(auth_uid_groups[j]);
+		}
+		free(auth_uid_groups);
+	}
+
+	if (err) {
+		rprintf(FLOG, "auth failed on module %s from %s (%s) for %s: %s\n",
+			lp_name(module), host, addr, line, err);
 		return NULL;
 	}
 
-	memset(secret, 0, sizeof secret);
-	if (!get_secret(module, line, secret, sizeof secret - 1)) {
-		memset(secret, 0, sizeof secret);
-		rprintf(FLOG, "auth failed on module %s from %s (%s): "
-			"missing secret for user \"%s\"\n",
-			lp_name(module), host, addr, line);
-		return NULL;
-	}
-
-	generate_hash(secret, challenge, pass2);
-	memset(secret, 0, sizeof secret);
-
-	if (strcmp(pass, pass2) != 0) {
-		rprintf(FLOG, "auth failed on module %s from %s (%s): "
-			"password mismatch\n",
-			lp_name(module), host, addr);
-		return NULL;
-	}
+	if (opt_ch == 'r')
+		read_only = 1;
+	else if (opt_ch == 'w')
+		read_only = 0;
 
 	return strdup(line);
 }
@@ -301,12 +357,12 @@ void auth_client(int fd, const char *user, const char *challenge)
 		/* XXX: cyeoh says that getpass is deprecated, because
 		 * it may return a truncated password on some systems,
 		 * and it is not in the LSB.
-                 *
-                 * Andrew Klein says that getpassphrase() is present
-                 * on Solaris and reads up to 256 characters.
-                 *
-                 * OpenBSD has a readpassphrase() that might be more suitable.
-                 */
+		 *
+		 * Andrew Klein says that getpassphrase() is present
+		 * on Solaris and reads up to 256 characters.
+		 *
+		 * OpenBSD has a readpassphrase() that might be more suitable.
+		 */
 		pass = getpass("Password: ");
 	}
 

backup.c

@@ -2,7 +2,7 @@
  * Backup handling code.
  *
  * Copyright (C) 1999 Andrew Tridgell
- * Copyright (C) 2003-2007 Wayne Davison
+ * Copyright (C) 2003-2020 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
@@ -19,8 +19,8 @@
  */
 
 #include "rsync.h"
+#include "ifuncs.h"
 
-extern int verbose;
 extern int am_root;
 extern int preserve_acls;
 extern int preserve_xattrs;
@@ -34,199 +34,242 @@ extern char backup_dir_buf[MAXPATHLEN];
 extern char *backup_suffix;
 extern char *backup_dir;
 
-/* make a complete pathname for backup file */
+/* Returns -1 on error, 0 on missing dir, and 1 on present dir. */
+static int validate_backup_dir(void)
+{
+	STRUCT_STAT st;
+
+	if (do_lstat(backup_dir_buf, &st) < 0) {
+		if (errno == ENOENT)
+			return 0;
+		rsyserr(FERROR, errno, "backup lstat %s failed", backup_dir_buf);
+		return -1;
+	}
+	if (!S_ISDIR(st.st_mode)) {
+		int flags = get_del_for_flag(st.st_mode) | DEL_FOR_BACKUP | DEL_RECURSE;
+		if (delete_item(backup_dir_buf, st.st_mode, flags) == 0)
+			return 0;
+		return -1;
+	}
+	return 1;
+}
+
+/* Create a backup path from the given fname, putting the result into
+ * backup_dir_buf.  Any new directories (compared to the prior backup
+ * path) are ensured to exist as directories, replacing anything else
+ * that may be in the way (e.g. a symlink). */
+static BOOL copy_valid_path(const char *fname)
+{
+	const char *f;
+	int val;
+	BOOL ret = True;
+	stat_x sx;
+	char *b, *rel = backup_dir_buf + backup_dir_len, *name = rel;
+
+	for (f = fname, b = rel; *f && *f == *b; f++, b++) {
+		if (*b == '/')
+			name = b + 1;
+	}
+
+	if (stringjoin(rel, backup_dir_remainder, fname, backup_suffix, NULL) >= backup_dir_remainder) {
+		rprintf(FERROR, "backup filename too long\n");
+		*name = '\0';
+		return False;
+	}
+
+	for ( ; ; name = b + 1) {
+		if ((b = strchr(name, '/')) == NULL)
+			return True;
+		*b = '\0';
+
+		val = validate_backup_dir();
+		if (val == 0)
+			break;
+		if (val < 0) {
+			*name = '\0';
+			return False;
+		}
+
+		*b = '/';
+	}
+
+	init_stat_x(&sx);
+
+	for ( ; b; name = b + 1, b = strchr(name, '/')) {
+		*b = '\0';
+
+		while (do_mkdir(backup_dir_buf, ACCESSPERMS) < 0) {
+			if (errno == EEXIST) {
+				val = validate_backup_dir();
+				if (val > 0)
+					break;
+				if (val == 0)
+					continue;
+			} else
+				rsyserr(FERROR, errno, "backup mkdir %s failed", backup_dir_buf);
+			*name = '\0';
+			ret = False;
+			goto cleanup;
+		}
+
+		/* Try to transfer the directory settings of the actual dir
+		 * that the files are coming from. */
+		if (x_stat(rel, &sx.st, NULL) < 0)
+			rsyserr(FERROR, errno, "backup stat %s failed", full_fname(rel));
+		else {
+			struct file_struct *file;
+			if (!(file = make_file(rel, NULL, NULL, 0, NO_FILTERS)))
+				continue;
+#ifdef SUPPORT_ACLS
+			if (preserve_acls && !S_ISLNK(file->mode)) {
+				get_acl(rel, &sx);
+				cache_tmp_acl(file, &sx);
+				free_acl(&sx);
+			}
+#endif
+#ifdef SUPPORT_XATTRS
+			if (preserve_xattrs) {
+				get_xattr(rel, &sx);
+				cache_tmp_xattr(file, &sx);
+				free_xattr(&sx);
+			}
+#endif
+			set_file_attrs(backup_dir_buf, file, NULL, NULL, 0);
+			unmake_file(file);
+		}
+
+		*b = '/';
+	}
+
+  cleanup:
+
+#ifdef SUPPORT_ACLS
+	uncache_tmp_acls();
+#endif
+#ifdef SUPPORT_XATTRS
+	uncache_tmp_xattrs();
+#endif
+
+	return ret;
+}
+
+/* Make a complete pathname for backup file and verify any new path elements. */
 char *get_backup_name(const char *fname)
 {
 	if (backup_dir) {
-		if (stringjoin(backup_dir_buf + backup_dir_len, backup_dir_remainder,
-			       fname, backup_suffix, NULL) < backup_dir_remainder)
-			return backup_dir_buf;
-	} else {
-		if (stringjoin(backup_dir_buf, MAXPATHLEN,
-			       fname, backup_suffix, NULL) < MAXPATHLEN)
+		static int initialized = 0;
+		if (!initialized) {
+			int ret;
+			if (backup_dir_len > 1)
+				backup_dir_buf[backup_dir_len-1] = '\0';
+			ret = make_path(backup_dir_buf, 0);
+			if (backup_dir_len > 1)
+				backup_dir_buf[backup_dir_len-1] = '/';
+			if (ret < 0)
+				return NULL;
+			initialized = 1;
+		}
+		/* copy fname into backup_dir_buf while validating the dirs. */
+		if (copy_valid_path(fname))
 			return backup_dir_buf;
+		/* copy_valid_path() has printed an error message. */
+		return NULL;
 	}
 
+	if (stringjoin(backup_dir_buf, MAXPATHLEN, fname, backup_suffix, NULL) < MAXPATHLEN)
+		return backup_dir_buf;
+
 	rprintf(FERROR, "backup filename too long\n");
 	return NULL;
 }
 
-/* simple backup creates a backup with a suffix in the same directory */
-static int make_simple_backup(const char *fname)
+/* Has same return codes as make_backup(). */
+static inline int link_or_rename(const char *from, const char *to,
+				 BOOL prefer_rename, STRUCT_STAT *stp)
 {
-	int rename_errno;
-	const char *fnamebak = get_backup_name(fname);
-
-	if (!fnamebak)
-		return 0;
-
-	while (1) {
-		if (do_rename(fname, fnamebak) == 0) {
-			if (verbose > 1) {
-				rprintf(FINFO, "backed up %s to %s\n",
-					fname, fnamebak);
-			}
-			break;
-		}
-		/* cygwin (at least version b19) reports EINVAL */
-		if (errno == ENOENT || errno == EINVAL)
-			break;
-
-		rename_errno = errno;
-		if (errno == EISDIR && do_rmdir(fnamebak) == 0)
-			continue;
-		if (errno == ENOTDIR && do_unlink(fnamebak) == 0)
-			continue;
-
-		rsyserr(FERROR, rename_errno, "rename %s to backup %s",
-			fname, fnamebak);
-		errno = rename_errno;
-		return 0;
-	}
-
-	return 1;
-}
-
-
-/****************************************************************************
-Create a directory given an absolute path, perms based upon another directory
-path
-****************************************************************************/
-int make_bak_dir(const char *fullpath)
-{
-	char fbuf[MAXPATHLEN], *rel, *end, *p;
-	struct file_struct *file;
-	int len = backup_dir_len;
-	stat_x sx;
-
-	while (*fullpath == '.' && fullpath[1] == '/') {
-		fullpath += 2;
-		len -= 2;
-	}
-
-	if (strlcpy(fbuf, fullpath, sizeof fbuf) >= sizeof fbuf)
-		return -1;
-
-	rel = fbuf + len;
-	end = p = rel + strlen(rel);
-
-	/* Try to find an existing dir, starting from the deepest dir. */
-	while (1) {
-		if (--p == fbuf)
-			return -1;
-		if (*p == '/') {
-			*p = '\0';
-			if (mkdir_defmode(fbuf) == 0)
-				break;
-			if (errno != ENOENT) {
-				rsyserr(FERROR, errno,
-					"make_bak_dir mkdir %s failed",
-					full_fname(fbuf));
-				return -1;
-			}
-		}
-	}
-
-	/* Make all the dirs that we didn't find on the way here. */
-	while (1) {
-		if (p >= rel) {
-			/* Try to transfer the directory settings of the
-			 * actual dir that the files are coming from. */
-			if (x_stat(rel, &sx.st, NULL) < 0) {
-				rsyserr(FERROR, errno,
-					"make_bak_dir stat %s failed",
-					full_fname(rel));
-			} else {
-#ifdef SUPPORT_ACLS
-				sx.acc_acl = sx.def_acl = NULL;
+#ifdef SUPPORT_HARD_LINKS
+	if (!prefer_rename) {
+#ifndef CAN_HARDLINK_SYMLINK
+		if (S_ISLNK(stp->st_mode))
+			return 0; /* Use copy code. */
 #endif
-#ifdef SUPPORT_XATTRS
-				sx.xattr = NULL;
+#ifndef CAN_HARDLINK_SPECIAL
+		if (IS_SPECIAL(stp->st_mode) || IS_DEVICE(stp->st_mode))
+			return 0; /* Use copy code. */
 #endif
-				if (!(file = make_file(rel, NULL, NULL, 0, NO_FILTERS)))
-					continue;
-#ifdef SUPPORT_ACLS
-				if (preserve_acls && !S_ISLNK(file->mode)) {
-					get_acl(rel, &sx);
-					cache_acl(file, &sx);
-					free_acl(&sx);
-				}
-#endif
-#ifdef SUPPORT_XATTRS
-				if (preserve_xattrs) {
-					get_xattr(rel, &sx);
-					cache_xattr(file, &sx);
-					free_xattr(&sx);
-				}
-#endif
-				set_file_attrs(fbuf, file, NULL, NULL, 0);
-				unmake_file(file);
-			}
-		}
-		*p = '/';
-		p += strlen(p);
-		if (p == end)
-			break;
-		if (mkdir_defmode(fbuf) < 0) {
-			rsyserr(FERROR, errno, "make_bak_dir mkdir %s failed",
-				full_fname(fbuf));
-			return -1;
+		if (do_link(from, to) == 0) {
+			if (DEBUG_GTE(BACKUP, 1))
+				rprintf(FINFO, "make_backup: HLINK %s successful.\n", from);
+			return 2;
 		}
+		/* We prefer to rename a regular file rather than copy it. */
+		if (!S_ISREG(stp->st_mode) || errno == EEXIST || errno == EISDIR)
+			return 0;
 	}
-
-	return 0;
-}
-
-/* robustly move a file, creating new directory structures if necessary */
-static int robust_move(const char *src, char *dst)
-{
-	if (robust_rename(src, dst, NULL, 0755) < 0
-	 && (errno != ENOENT || make_bak_dir(dst) < 0
-	  || robust_rename(src, dst, NULL, 0755) < 0))
-		return -1;
-	return 0;
-}
-
-
-/* If we have a --backup-dir, then we get here from make_backup().
- * We will move the file to be deleted into a parallel directory tree. */
-static int keep_backup(const char *fname)
-{
-	stat_x sx;
-	struct file_struct *file;
-	char *buf;
-	int kept = 0;
-	int ret_code;
-
-	/* return if no file to keep */
-	if (x_lstat(fname, &sx.st, NULL) < 0)
+#endif
+	if (do_rename(from, to) == 0) {
+		if (stp->st_nlink > 1 && !S_ISDIR(stp->st_mode)) {
+			/* If someone has hard-linked the file into the backup
+			 * dir, rename() might return success but do nothing! */
+			robust_unlink(from); /* Just in case... */
+		}
+		if (DEBUG_GTE(BACKUP, 1))
+			rprintf(FINFO, "make_backup: RENAME %s successful.\n", from);
 		return 1;
-#ifdef SUPPORT_ACLS
-	sx.acc_acl = sx.def_acl = NULL;
-#endif
-#ifdef SUPPORT_XATTRS
-	sx.xattr = NULL;
-#endif
-
-	if (!(file = make_file(fname, NULL, NULL, 0, NO_FILTERS)))
-		return 1; /* the file could have disappeared */
-
-	if (!(buf = get_backup_name(fname))) {
-		unmake_file(file);
-		return 0;
 	}
+	return 0;
+}
+
+/* Hard-link, rename, or copy an item to the backup name.  Returns 0 for
+ * failure, 1 if item was moved, 2 if item was duplicated or hard linked
+ * into backup area, or 3 if item doesn't exist or isn't a regular file. */
+int make_backup(const char *fname, BOOL prefer_rename)
+{
+	stat_x sx;
+	struct file_struct *file;
+	int save_preserve_xattrs;
+	char *buf;
+	int ret = 0;
+
+	init_stat_x(&sx);
+	/* Return success if no file to keep. */
+	if (x_lstat(fname, &sx.st, NULL) < 0)
+		return 3;
+
+	if (!(buf = get_backup_name(fname)))
+		return 0;
+
+	/* Try a hard-link or a rename first.  Using rename is not atomic, but
+	 * is more efficient than forcing a copy for larger files when no hard-
+	 * linking is possible. */
+	if ((ret = link_or_rename(fname, buf, prefer_rename, &sx.st)) != 0)
+		goto success;
+	if (errno == EEXIST || errno == EISDIR) {
+		STRUCT_STAT bakst;
+		if (do_lstat(buf, &bakst) == 0) {
+			int flags = get_del_for_flag(bakst.st_mode) | DEL_FOR_BACKUP | DEL_RECURSE;
+			if (delete_item(buf, bakst.st_mode, flags) != 0)
+				return 0;
+		}
+		if ((ret = link_or_rename(fname, buf, prefer_rename, &sx.st)) != 0)
+			goto success;
+	}
+
+	/* Fall back to making a copy. */
+	if (!(file = make_file(fname, NULL, &sx.st, 0, NO_FILTERS)))
+		return 3; /* the file could have disappeared */
 
 #ifdef SUPPORT_ACLS
 	if (preserve_acls && !S_ISLNK(file->mode)) {
 		get_acl(fname, &sx);
-		cache_acl(file, &sx);
+		cache_tmp_acl(file, &sx);
 		free_acl(&sx);
 	}
 #endif
 #ifdef SUPPORT_XATTRS
 	if (preserve_xattrs) {
 		get_xattr(fname, &sx);
-		cache_xattr(file, &sx);
+		cache_tmp_xattr(file, &sx);
 		free_xattr(&sx);
 	}
 #endif
@@ -234,95 +277,78 @@ static int keep_backup(const char *fname)
 	/* Check to see if this is a device file, or link */
 	if ((am_root && preserve_devices && IS_DEVICE(file->mode))
 	 || (preserve_specials && IS_SPECIAL(file->mode))) {
-		uint32 *devp = F_RDEV_P(file);
-		dev_t rdev = MAKEDEV(DEV_MAJOR(devp), DEV_MINOR(devp));
-		do_unlink(buf);
-		if (do_mknod(buf, file->mode, rdev) < 0
-		    && (errno != ENOENT || make_bak_dir(buf) < 0
-		     || do_mknod(buf, file->mode, rdev) < 0)) {
-			rsyserr(FERROR, errno, "mknod %s failed",
-				full_fname(buf));
-		} else if (verbose > 2) {
-			rprintf(FINFO, "make_backup: DEVICE %s successful.\n",
-				fname);
-		}
-		kept = 1;
-		do_unlink(fname);
-	}
-
-	if (!kept && S_ISDIR(file->mode)) {
-		/* make an empty directory */
-		if (do_mkdir(buf, file->mode) < 0
-		    && (errno != ENOENT || make_bak_dir(buf) < 0
-		     || do_mkdir(buf, file->mode) < 0)) {
-			rsyserr(FINFO, errno, "mkdir %s failed",
-				full_fname(buf));
-		}
-
-		ret_code = do_rmdir(fname);
-		if (verbose > 2) {
-			rprintf(FINFO, "make_backup: RMDIR %s returns %i\n",
-				full_fname(fname), ret_code);
-		}
-		kept = 1;
+		if (do_mknod(buf, file->mode, sx.st.st_rdev) < 0)
+			rsyserr(FERROR, errno, "mknod %s failed", full_fname(buf));
+		else if (DEBUG_GTE(BACKUP, 1))
+			rprintf(FINFO, "make_backup: DEVICE %s successful.\n", fname);
+		ret = 2;
 	}
 
 #ifdef SUPPORT_LINKS
-	if (!kept && preserve_links && S_ISLNK(file->mode)) {
+	if (!ret && preserve_links && S_ISLNK(file->mode)) {
 		const char *sl = F_SYMLINK(file);
-		if (safe_symlinks && unsafe_symlink(sl, buf)) {
-			if (verbose) {
-				rprintf(FINFO, "ignoring unsafe symlink %s -> %s\n",
-					full_fname(buf), sl);
+		if (safe_symlinks && unsafe_symlink(sl, fname)) {
+			if (INFO_GTE(SYMSAFE, 1)) {
+				rprintf(FINFO, "not backing up unsafe symlink \"%s\" -> \"%s\"\n",
+					fname, sl);
 			}
-			kept = 1;
+			ret = 2;
 		} else {
-			do_unlink(buf);
-			if (do_symlink(sl, buf) < 0
-			    && (errno != ENOENT || make_bak_dir(buf) < 0
-			     || do_symlink(sl, buf) < 0)) {
-				rsyserr(FERROR, errno, "link %s -> \"%s\"",
-					full_fname(buf), sl);
-			}
-			do_unlink(fname);
-			kept = 1;
+			if (do_symlink(sl, buf) < 0)
+				rsyserr(FERROR, errno, "link %s -> \"%s\"", full_fname(buf), sl);
+			else if (DEBUG_GTE(BACKUP, 1))
+				rprintf(FINFO, "make_backup: SYMLINK %s successful.\n", fname);
+			ret = 2;
 		}
 	}
 #endif
 
-	if (!kept && !S_ISREG(file->mode)) {
-		rprintf(FINFO, "make_bak: skipping non-regular file %s\n",
-			fname);
+	if (!ret && !S_ISREG(file->mode)) {
+		rprintf(FINFO, "make_bak: skipping non-regular file %s\n", fname);
 		unmake_file(file);
-		return 1;
+#ifdef SUPPORT_ACLS
+		uncache_tmp_acls();
+#endif
+#ifdef SUPPORT_XATTRS
+		uncache_tmp_xattrs();
+#endif
+		return 3;
 	}
 
-	/* move to keep tree if a file */
-	if (!kept) {
-		if (robust_move(fname, buf) != 0) {
+	/* Copy to backup tree if a file. */
+	if (!ret) {
+		if (copy_file(fname, buf, -1, file->mode) < 0) {
 			rsyserr(FERROR, errno, "keep_backup failed: %s -> \"%s\"",
 				full_fname(fname), buf);
-		} else if (sx.st.st_nlink > 1) {
-			/* If someone has hard-linked the file into the backup
-			 * dir, rename() might return success but do nothing! */
-			robust_unlink(fname); /* Just in case... */
+			unmake_file(file);
+#ifdef SUPPORT_ACLS
+			uncache_tmp_acls();
+#endif
+#ifdef SUPPORT_XATTRS
+			uncache_tmp_xattrs();
+#endif
+			return 0;
 		}
+		if (DEBUG_GTE(BACKUP, 1))
+			rprintf(FINFO, "make_backup: COPY %s successful.\n", fname);
+		ret = 2;
 	}
-	set_file_attrs(buf, file, NULL, fname, 0);
+
+	save_preserve_xattrs = preserve_xattrs;
+	preserve_xattrs = 0;
+	set_file_attrs(buf, file, NULL, fname, ATTRS_ACCURATE_TIME);
+	preserve_xattrs = save_preserve_xattrs;
+
 	unmake_file(file);
+#ifdef SUPPORT_ACLS
+	uncache_tmp_acls();
+#endif
+#ifdef SUPPORT_XATTRS
+	uncache_tmp_xattrs();
+#endif
 
-	if (verbose > 1) {
-		rprintf(FINFO, "backed up %s to %s\n",
-			fname, buf);
-	}
-	return 1;
-}
-
-
-/* main backup switch routine */
-int make_backup(const char *fname)
-{
-	if (backup_dir)
-		return keep_backup(fname);
-	return make_simple_backup(fname);
+  success:
+	if (INFO_GTE(BACKUP, 1))
+		rprintf(FINFO, "backed up %s to %s\n", fname, buf);
+	return ret;
 }

batch.c

@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1999 Weiss
  * Copyright (C) 2004 Chris Shoemaker
- * Copyright (C) 2004-2007 Wayne Davison
+ * Copyright (C) 2004-2020 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"
-#include "zlib/zlib.h"
+#include <zlib.h>
 #include <time.h>
 
 extern int eol_nulls;
@@ -31,15 +31,33 @@ extern int preserve_hard_links;
 extern int preserve_devices;
 extern int preserve_uid;
 extern int preserve_gid;
+extern int preserve_acls;
+extern int preserve_xattrs;
 extern int always_checksum;
 extern int do_compression;
-extern int def_compress_level;
+extern int inplace;
+extern int append_mode;
+extern int write_batch;
+extern int xfersum_type;
 extern int protocol_version;
+extern int raw_argc, cooked_argc;
+extern char **raw_argv, **cooked_argv;
 extern char *batch_name;
+extern const char *checksum_choice;
+extern const char *compress_choice;
+#ifdef ICONV_OPTION
+extern char *iconv_opt;
+#endif
 
-extern struct filter_list_struct filter_list;
+extern filter_rule_list filter_list;
 
-static int tweaked_compress_level;
+int batch_fd = -1;
+int batch_sh_fd = -1;
+int batch_stream_flags;
+
+static int tweaked_append;
+static int tweaked_append_verify;
+static int tweaked_iconv;
 
 static int *flag_ptr[] = {
 	&recurse,		/* 0 */
@@ -50,7 +68,13 @@ static int *flag_ptr[] = {
 	&preserve_hard_links,	/* 5 */
 	&always_checksum,	/* 6 */
 	&xfer_dirs,		/* 7 (protocol 29) */
-	&tweaked_compress_level,/* 8 (protocol 29) */
+	&do_compression,	/* 8 (protocol 29) */
+	&tweaked_iconv,		/* 9  (protocol 30) */
+	&preserve_acls,		/* 10 (protocol 30) */
+	&preserve_xattrs,	/* 11 (protocol 30) */
+	&inplace,		/* 12 (protocol 30) */
+	&tweaked_append,	/* 13 (protocol 30) */
+	&tweaked_append_verify,	/* 14 (protocol 30) */
 	NULL
 };
 
@@ -64,6 +88,12 @@ static char *flag_name[] = {
 	"--checksum (-c)",
 	"--dirs (-d)",
 	"--compress (-z)",
+	"--iconv",
+	"--acls (-A)",
+	"--xattrs (-X)",
+	"--inplace",
+	"--append",
+	"--append-verify",
 	NULL
 };
 
@@ -71,16 +101,14 @@ void write_stream_flags(int fd)
 {
 	int i, flags;
 
-#if Z_DEFAULT_COMPRESSION == -1
-	tweaked_compress_level = do_compression ? def_compress_level + 2 : 0;
-#else
-#error internal logic error!  Fix def_compress_level logic above and below too!
+	tweaked_append = append_mode == 1;
+	tweaked_append_verify = append_mode == 2;
+#ifdef ICONV_OPTION
+	tweaked_iconv = iconv_opt != NULL;
 #endif
 
 	/* Start the batch file with a bitmap of data-stream-affecting
 	 * flags. */
-	if (protocol_version < 29)
-		flag_ptr[7] = NULL;
 	for (i = 0, flags = 0; flag_ptr[i]; i++) {
 		if (*flag_ptr[i])
 			flags |= 1 << i;
@@ -90,14 +118,32 @@ void write_stream_flags(int fd)
 
 void read_stream_flags(int fd)
 {
-	int i, flags;
+	batch_stream_flags = read_int(fd);
+}
+
+void check_batch_flags(void)
+{
+	int i;
 
 	if (protocol_version < 29)
 		flag_ptr[7] = NULL;
-	for (i = 0, flags = read_int(fd); flag_ptr[i]; i++) {
-		int set = flags & (1 << i) ? 1 : 0;
+	else if (protocol_version < 30)
+		flag_ptr[9] = NULL;
+	tweaked_append = append_mode == 1;
+	tweaked_append_verify = append_mode == 2;
+#ifdef ICONV_OPTION
+	tweaked_iconv = iconv_opt != NULL;
+#endif
+	for (i = 0; flag_ptr[i]; i++) {
+		int set = batch_stream_flags & (1 << i) ? 1 : 0;
 		if (*flag_ptr[i] != set) {
-			if (verbose) {
+			if (i == 9) {
+				rprintf(FERROR,
+					"%s specify the --iconv option to use this batch file.\n",
+					set ? "Please" : "Do not");
+				exit_cleanup(RERR_SYNTAX);
+			}
+			if (INFO_GTE(MISC, 1)) {
 				rprintf(FINFO,
 					"%sing the %s option to match the batchfile.\n",
 					set ? "Sett" : "Clear", flag_name[i]);
@@ -112,48 +158,64 @@ void read_stream_flags(int fd)
 			xfer_dirs = 0;
 	}
 
-	if (tweaked_compress_level == 0 || tweaked_compress_level == 2)
-		do_compression = 0;
-	else {
-		do_compression = 1;
-		def_compress_level = tweaked_compress_level - 2;
-	}
+	if (tweaked_append)
+		append_mode = 1;
+	else if (tweaked_append_verify)
+		append_mode = 2;
 }
 
-static void write_arg(int fd, char *arg)
+static int write_arg(const char *arg)
 {
-	char *x, *s;
+	const char *x, *s;
+	int len, err = 0;
 
 	if (*arg == '-' && (x = strchr(arg, '=')) != NULL) {
-		write(fd, arg, x - arg + 1);
+		err |= write(batch_sh_fd, arg, x - arg + 1) != x - arg + 1;
 		arg += x - arg + 1;
 	}
 
 	if (strpbrk(arg, " \"'&;|[]()$#!*?^\\") != NULL) {
-		write(fd, "'", 1);
+		err |= write(batch_sh_fd, "'", 1) != 1;
 		for (s = arg; (x = strchr(s, '\'')) != NULL; s = x + 1) {
-			write(fd, s, x - s + 1);
-			write(fd, "'", 1);
+			err |= write(batch_sh_fd, s, x - s + 1) != x - s + 1;
+			err |= write(batch_sh_fd, "'", 1) != 1;
 		}
-		write(fd, s, strlen(s));
-		write(fd, "'", 1);
-		return;
+		len = strlen(s);
+		err |= write(batch_sh_fd, s, len) != len;
+		err |= write(batch_sh_fd, "'", 1) != 1;
+		return err;
 	}
 
-	write(fd, arg, strlen(arg));
+	len = strlen(arg);
+	err |= write(batch_sh_fd, arg, len) != len;
+
+	return err;
+}
+
+/* Writes out a space and then an option (or other string) with an optional "=" + arg suffix. */
+static int write_opt(const char *opt, const char *arg)
+{
+	int len = strlen(opt);
+	int err = write(batch_sh_fd, " ", 1) != 1;
+	err = write(batch_sh_fd, opt, len) != len ? 1 : 0; 
+	if (arg) {
+		err |= write(batch_sh_fd, "=", 1) != 1;
+		err |= write_arg(arg);
+	}
+	return err;
 }
 
 static void write_filter_rules(int fd)
 {
-	struct filter_struct *ent;
+	filter_rule *ent;
 
 	write_sbuf(fd, " <<'#E#'\n");
 	for (ent = filter_list.head; ent; ent = ent->next) {
 		unsigned int plen;
-		char *p = get_rule_prefix(ent->match_flags, "- ", 0, &plen);
+		char *p = get_rule_prefix(ent, "- ", 0, &plen);
 		write_buf(fd, p, plen);
 		write_sbuf(fd, ent->pattern);
-		if (ent->match_flags & MATCHFLG_DIRECTORY)
+		if (ent->rflags & FILTRULE_DIRECTORY)
 			write_byte(fd, '/');
 		write_byte(fd, eol_nulls ? 0 : '\n');
 	}
@@ -162,40 +224,74 @@ static void write_filter_rules(int fd)
 	write_sbuf(fd, "#E#");
 }
 
+/* This sets batch_fd and (for --write-batch) batch_sh_fd. */
+void open_batch_files(void)
+{
+	if (write_batch) {
+		char filename[MAXPATHLEN];
+
+		stringjoin(filename, sizeof filename, batch_name, ".sh", NULL);
+
+		batch_sh_fd = do_open(filename, O_WRONLY | O_CREAT | O_TRUNC, S_IRUSR | S_IWUSR | S_IXUSR);
+		if (batch_sh_fd < 0) {
+			rsyserr(FERROR, errno, "Batch file %s open error", full_fname(filename));
+			exit_cleanup(RERR_FILESELECT);
+		}
+
+		batch_fd = do_open(batch_name, O_WRONLY | O_CREAT | O_TRUNC, S_IRUSR | S_IWUSR);
+	} else if (strcmp(batch_name, "-") == 0)
+		batch_fd = STDIN_FILENO;
+	else
+		batch_fd = do_open(batch_name, O_RDONLY, S_IRUSR | S_IWUSR);
+
+	if (batch_fd < 0) {
+		rsyserr(FERROR, errno, "Batch file %s open error", full_fname(batch_name));
+		exit_cleanup(RERR_FILEIO);
+	}
+}
+
 /* This routine tries to write out an equivalent --read-batch command
  * given the user's --write-batch args.  However, it doesn't really
  * understand most of the options, so it uses some overly simple
  * heuristics to munge the command line into something that will
  * (hopefully) work. */
-void write_batch_shell_file(int argc, char *argv[], int file_arg_cnt)
+void write_batch_shell_file(void)
 {
-	int fd, i, len;
-	char *p, filename[MAXPATHLEN];
-
-	stringjoin(filename, sizeof filename,
-		   batch_name, ".sh", NULL);
-	fd = do_open(filename, O_WRONLY | O_CREAT | O_TRUNC,
-		     S_IRUSR | S_IWUSR | S_IEXEC);
-	if (fd < 0) {
-		rsyserr(FERROR, errno, "Batch file %s open error",
-			filename);
-		exit_cleanup(1);
-	}
+	int i, j, len, err = 0;
+	char *p, *p2;
 
 	/* Write argvs info to BATCH.sh file */
-	write_arg(fd, argv[0]);
+	err |= write_arg(raw_argv[0]);
 	if (filter_list.head) {
 		if (protocol_version >= 29)
-			write_sbuf(fd, " --filter=._-");
+			err |= write_opt("--filter", "._-");
 		else
-			write_sbuf(fd, " --exclude-from=-");
+			err |= write_opt("--exclude-from", "-");
 	}
-	for (i = 1; i < argc - file_arg_cnt; i++) {
-		p = argv[i];
+
+	/* We need to make sure that any protocol-based or negotiated choices get accurately
+	 * reflected in the options we save AND that we avoid any need for --read-batch to
+	 * do a string-based negotiation (since we don't write them into the file). */
+	if (do_compression)
+		err |= write_opt("--compress-choice", compress_choice);
+	if (strchr(checksum_choice, ',') || xfersum_type != parse_csum_name(NULL, -1))
+		err |= write_opt("--checksum-choice", checksum_choice);
+
+	/* Elide the filename args from the option list, but scan for them in reverse. */
+	for (i = raw_argc-1, j = cooked_argc-1; i > 0 && j >= 0; i--) {
+		if (strcmp(raw_argv[i], cooked_argv[j]) == 0) {
+			raw_argv[i] = NULL;
+			j--;
+		}
+	}
+
+	for (i = 1; i < raw_argc; i++) {
+		if (!(p = raw_argv[i]))
+			continue;
 		if (strncmp(p, "--files-from", 12) == 0
-		    || strncmp(p, "--filter", 8) == 0
-		    || strncmp(p, "--include", 9) == 0
-		    || strncmp(p, "--exclude", 9) == 0) {
+		 || strncmp(p, "--filter", 8) == 0
+		 || strncmp(p, "--include", 9) == 0
+		 || strncmp(p, "--exclude", 9) == 0) {
 			if (strchr(p, '=') == NULL)
 				i++;
 			continue;
@@ -204,27 +300,24 @@ void write_batch_shell_file(int argc, char *argv[], int file_arg_cnt)
 			i++;
 			continue;
 		}
-		write(fd, " ", 1);
 		if (strncmp(p, "--write-batch", len = 13) == 0
-		 || strncmp(p, "--only-write-batch", len = 18) == 0) {
-			write(fd, "--read-batch", 12);
-			if (p[len] == '=') {
-				write(fd, "=", 1);
-				write_arg(fd, p + len + 1);
-			}
-		} else
-			write_arg(fd, p);
+		 || strncmp(p, "--only-write-batch", len = 18) == 0)
+			err |= write_opt("--read-batch", p[len] == '=' ? p + len + 1 : NULL);
+		else {
+			err |= write(batch_sh_fd, " ", 1) != 1;
+			err |= write_arg(p);
+		}
 	}
-	if (!(p = check_for_hostspec(argv[argc - 1], &p, &i)))
-		p = argv[argc - 1];
-	write(fd, " ${1:-", 6);
-	write_arg(fd, p);
-	write_byte(fd, '}');
+	if (!(p = check_for_hostspec(cooked_argv[cooked_argc - 1], &p2, &i)))
+		p = cooked_argv[cooked_argc - 1];
+	err |= write_opt("${1:-", NULL);
+	err |= write_arg(p);
+	err |= write(batch_sh_fd, "}", 1) != 1;
 	if (filter_list.head)
-		write_filter_rules(fd);
-	if (write(fd, "\n", 1) != 1 || close(fd) < 0) {
-		rsyserr(FERROR, errno, "Batch file %s write error",
-			filename);
-		exit_cleanup(1);
+		write_filter_rules(batch_sh_fd);
+	if (write(batch_sh_fd, "\n", 1) != 1 || close(batch_sh_fd) < 0 || err) {
+		rsyserr(FERROR, errno, "Batch file %s.sh write error", batch_name);
+		exit_cleanup(RERR_FILEIO);
 	}
+	batch_sh_fd = -1;
 }

byteorder.h

@@ -2,7 +2,7 @@
  * Simple byteorder handling.
  *
  * Copyright (C) 1992-1995 Andrew Tridgell
- * Copyright (C) 2007 Wayne Davison
+ * Copyright (C) 2007-2020 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
@@ -22,7 +22,7 @@
 
 /* We know that the x86 can handle misalignment and has the same
  * byte order (LSB-first) as the 32-bit numbers we transmit. */
-#ifdef __i386__
+#if defined __i386__ || defined __i486__ || defined __i586__ || defined __i686__ || __amd64
 #define CAREFUL_ALIGNMENT 0
 #endif
 
@@ -32,21 +32,101 @@
 
 #define CVAL(buf,pos) (((unsigned char *)(buf))[pos])
 #define UVAL(buf,pos) ((uint32)CVAL(buf,pos))
-#define SCVAL(buf,pos,val) (CVAL(buf,pos) = (val))
 
 #if CAREFUL_ALIGNMENT
-#define PVAL(buf,pos) (UVAL(buf,pos)|UVAL(buf,(pos)+1)<<8)
-#define IVAL(buf,pos) (PVAL(buf,pos)|PVAL(buf,(pos)+2)<<16)
-#define SSVALX(buf,pos,val) (CVAL(buf,pos)=(val)&0xFF,CVAL(buf,pos+1)=(val)>>8)
-#define SIVALX(buf,pos,val) (SSVALX(buf,pos,val&0xFFFF),SSVALX(buf,pos+2,val>>16))
-#define SIVAL(buf,pos,val) SIVALX((buf),(pos),((uint32)(val)))
-#else
-/* this handles things for architectures like the 386 that can handle
-   alignment errors */
-/*
-   WARNING: This section is dependent on the length of int32
-   being correct. set CAREFUL_ALIGNMENT if it is not.
-*/
-#define IVAL(buf,pos) (*(uint32 *)((char *)(buf) + (pos)))
-#define SIVAL(buf,pos,val) IVAL(buf,pos)=((uint32)(val))
-#endif
+
+static inline uint32
+IVALu(const uchar *buf, int pos)
+{
+	return UVAL(buf, pos)
+	     | UVAL(buf, pos + 1) << 8
+	     | UVAL(buf, pos + 2) << 16
+	     | UVAL(buf, pos + 3) << 24;
+}
+
+static inline void
+SIVALu(uchar *buf, int pos, uint32 val)
+{
+	CVAL(buf, pos)     = val;
+	CVAL(buf, pos + 1) = val >> 8;
+	CVAL(buf, pos + 2) = val >> 16;
+	CVAL(buf, pos + 3) = val >> 24;
+}
+
+static inline int64
+IVAL64(const char *buf, int pos)
+{
+	return IVALu((uchar*)buf, pos) | (int64)IVALu((uchar*)buf, pos + 4) << 32;
+}
+
+static inline void
+SIVAL64(char *buf, int pos, int64 val)
+{
+	SIVALu((uchar*)buf, pos, val);
+	SIVALu((uchar*)buf, pos + 4, val >> 32);
+}
+
+#else /* !CAREFUL_ALIGNMENT */
+
+/* This handles things for architectures like the 386 that can handle alignment errors.
+ * WARNING: This section is dependent on the length of an int32 (and thus a uint32)
+ * being correct (4 bytes)!  Set CAREFUL_ALIGNMENT if it is not. */
+
+static inline uint32
+IVALu(const uchar *buf, int pos)
+{
+	union {
+		const uchar *b;
+		const uint32 *num;
+	} u;
+	u.b = buf + pos;
+	return *u.num;
+}
+
+static inline void
+SIVALu(uchar *buf, int pos, uint32 val)
+{
+	union {
+		uchar *b;
+		uint32 *num;
+	} u;
+	u.b = buf + pos;
+	*u.num = val;
+}
+
+static inline int64
+IVAL64(const char *buf, int pos)
+{
+	union {
+		const char *b;
+		const int64 *num;
+	} u;
+	u.b = buf + pos;
+	return *u.num;
+}
+
+static inline void
+SIVAL64(char *buf, int pos, int64 val)
+{
+	union {
+		char *b;
+		int64 *num;
+	} u;
+	u.b = buf + pos;
+	*u.num = val;
+}
+
+#endif /* !CAREFUL_ALIGNMENT */
+
+static inline uint32
+IVAL(const char *buf, int pos)
+{
+	return IVALu((uchar*)buf, pos);
+}
+
+static inline void
+SIVAL(char *buf, int pos, uint32 val)
+{
+	SIVALu((uchar*)buf, pos, val);
+}
+

case_N.h

@@ -1,7 +1,7 @@
 /*
- * End-of-run cleanup helper code used by cleanup.c.
+ * Allow an arbitrary sequence of case labels.
  *
- * Copyright (C) 2006-2007 Wayne Davison
+ * Copyright (C) 2006-2020 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
@@ -17,63 +17,76 @@
  * with this program; if not, visit the http://fsf.org website.
  */
 
-/* This is included by cleanup.c multiple times, once for every segement in
- * the _exit_cleanup() code.  This produces the next "case N:" statement in
- * sequence and increments the cleanup_step variable by 1.  This ensures that
- * our case statements never get out of whack due to added/removed steps. */
+/* This is included multiple times, once for every segment in a switch statement.
+ * This produces the next "case N:" statement in sequence. */
 
-#if !defined EXIT_CLEANUP_CASE_0
-#define EXIT_CLEANUP_CASE_0
+#if !defined CASE_N_STATE_0
+#define CASE_N_STATE_0
 	case 0:
-#elif !defined EXIT_CLEANUP_CASE_1
-#define EXIT_CLEANUP_CASE_1
+#elif !defined CASE_N_STATE_1
+#define CASE_N_STATE_1
+	/* FALLTHROUGH */
 	case 1:
-#elif !defined EXIT_CLEANUP_CASE_2
-#define EXIT_CLEANUP_CASE_2
+#elif !defined CASE_N_STATE_2
+#define CASE_N_STATE_2
+	/* FALLTHROUGH */
 	case 2:
-#elif !defined EXIT_CLEANUP_CASE_3
-#define EXIT_CLEANUP_CASE_3
+#elif !defined CASE_N_STATE_3
+#define CASE_N_STATE_3
+	/* FALLTHROUGH */
 	case 3:
-#elif !defined EXIT_CLEANUP_CASE_4
-#define EXIT_CLEANUP_CASE_4
+#elif !defined CASE_N_STATE_4
+#define CASE_N_STATE_4
+	/* FALLTHROUGH */
 	case 4:
-#elif !defined EXIT_CLEANUP_CASE_5
-#define EXIT_CLEANUP_CASE_5
+#elif !defined CASE_N_STATE_5
+#define CASE_N_STATE_5
+	/* FALLTHROUGH */
 	case 5:
-#elif !defined EXIT_CLEANUP_CASE_6
-#define EXIT_CLEANUP_CASE_6
+#elif !defined CASE_N_STATE_6
+#define CASE_N_STATE_6
+	/* FALLTHROUGH */
 	case 6:
-#elif !defined EXIT_CLEANUP_CASE_7
-#define EXIT_CLEANUP_CASE_7
+#elif !defined CASE_N_STATE_7
+#define CASE_N_STATE_7
+	/* FALLTHROUGH */
 	case 7:
-#elif !defined EXIT_CLEANUP_CASE_8
-#define EXIT_CLEANUP_CASE_8
+#elif !defined CASE_N_STATE_8
+#define CASE_N_STATE_8
+	/* FALLTHROUGH */
 	case 8:
-#elif !defined EXIT_CLEANUP_CASE_9
-#define EXIT_CLEANUP_CASE_9
+#elif !defined CASE_N_STATE_9
+#define CASE_N_STATE_9
+	/* FALLTHROUGH */
 	case 9:
-#elif !defined EXIT_CLEANUP_CASE_10
-#define EXIT_CLEANUP_CASE_10
+#elif !defined CASE_N_STATE_10
+#define CASE_N_STATE_10
+	/* FALLTHROUGH */
 	case 10:
-#elif !defined EXIT_CLEANUP_CASE_11
-#define EXIT_CLEANUP_CASE_11
+#elif !defined CASE_N_STATE_11
+#define CASE_N_STATE_11
+	/* FALLTHROUGH */
 	case 11:
-#elif !defined EXIT_CLEANUP_CASE_12
-#define EXIT_CLEANUP_CASE_12
+#elif !defined CASE_N_STATE_12
+#define CASE_N_STATE_12
+	/* FALLTHROUGH */
 	case 12:
-#elif !defined EXIT_CLEANUP_CASE_13
-#define EXIT_CLEANUP_CASE_13
+#elif !defined CASE_N_STATE_13
+#define CASE_N_STATE_13
+	/* FALLTHROUGH */
 	case 13:
-#elif !defined EXIT_CLEANUP_CASE_14
-#define EXIT_CLEANUP_CASE_14
+#elif !defined CASE_N_STATE_14
+#define CASE_N_STATE_14
+	/* FALLTHROUGH */
 	case 14:
-#elif !defined EXIT_CLEANUP_CASE_15
-#define EXIT_CLEANUP_CASE_15
+#elif !defined CASE_N_STATE_15
+#define CASE_N_STATE_15
+	/* FALLTHROUGH */
 	case 15:
-#elif !defined EXIT_CLEANUP_CASE_16
-#define EXIT_CLEANUP_CASE_16
+#elif !defined CASE_N_STATE_16
+#define CASE_N_STATE_16
+	/* FALLTHROUGH */
 	case 16:
 #else
 #error Need to add more case statements!
 #endif
-		cleanup_step++;

checksum.c

@@ -3,13 +3,20 @@
  *
  * Copyright (C) 1996 Andrew Tridgell
  * Copyright (C) 1996 Paul Mackerras
- * Copyright (C) 2004-2007 Wayne Davison
+ * Copyright (C) 2004-2020 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
  * the Free Software Foundation; either version 3 of the License, or
  * (at your option) any later version.
  *
+ * In addition, as a special exception, the copyright holders give
+ * permission to dynamically link rsync with the OpenSSL and xxhash
+ * libraries when those libraries are being distributed in compliance
+ * with their license terms, and to distribute a dynamically linked
+ * combination of rsync and these libraries.  This is also considered
+ * to be covered under the GPL's System Libraries exception.
+ *
  * This program is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
@@ -20,49 +27,217 @@
  */
 
 #include "rsync.h"
+#ifdef SUPPORT_XXHASH
+#include "xxhash.h"
+#endif
 
+extern int am_server;
+extern int whole_file;
 extern int checksum_seed;
 extern int protocol_version;
+extern int proper_seed_order;
+extern const char *checksum_choice;
 
-int csum_length = SHORT_SUM_LENGTH; /* initial value */
+struct name_num_obj valid_checksums = {
+	"checksum", NULL, NULL, 0, 0, {
+#ifdef SUPPORT_XXHASH
+		{ CSUM_XXH64, "xxh64", NULL },
+		{ CSUM_XXH64, "xxhash", NULL },
+#endif
+		{ CSUM_MD5, "md5", NULL },
+		{ CSUM_MD4, "md4", NULL },
+		{ CSUM_NONE, "none", NULL },
+		{ 0, NULL, NULL }
+	}
+};
 
+int xfersum_type = 0; /* used for the file transfer checksums */
+int checksum_type = 0; /* used for the pre-transfer (--checksum) checksums */
+
+int parse_csum_name(const char *name, int len)
+{
+	struct name_num_item *nni;
+
+	if (len < 0 && name)
+		len = strlen(name);
+
+	if (!name || (len == 4 && strncasecmp(name, "auto", 4) == 0)) {
+		if (protocol_version >= 30)
+			return CSUM_MD5;
+		if (protocol_version >= 27)
+			return CSUM_MD4_OLD;
+		if (protocol_version >= 21)
+			return CSUM_MD4_BUSTED;
+		return CSUM_MD4_ARCHAIC;
+	}
+
+	nni = get_nni_by_name(&valid_checksums, name, len);
+
+	if (!nni) {
+		rprintf(FERROR, "unknown checksum name: %s\n", name);
+		exit_cleanup(RERR_UNSUPPORTED);
+	}
+
+	return nni->num;
+}
+
+static const char *checksum_name(int num)
+{
+	struct name_num_item *nni = get_nni_by_num(&valid_checksums, num);
+
+	return nni ? nni->name : num < CSUM_MD4 ? "MD4" : "UNKNOWN";
+}
+
+void parse_checksum_choice(int final_call)
+{
+	if (valid_checksums.negotiated_name)
+		xfersum_type = checksum_type = valid_checksums.negotiated_num;
+	else {
+		char *cp = checksum_choice ? strchr(checksum_choice, ',') : NULL;
+		if (cp) {
+			xfersum_type = parse_csum_name(checksum_choice, cp - checksum_choice);
+			checksum_type = parse_csum_name(cp+1, -1);
+		} else
+			xfersum_type = checksum_type = parse_csum_name(checksum_choice, -1);
+	}
+
+	if (xfersum_type == CSUM_NONE)
+		whole_file = 1;
+
+	/* Snag the checksum name for both write_batch's option output & the following debug output. */
+	if (valid_checksums.negotiated_name)
+		checksum_choice = valid_checksums.negotiated_name;
+	else if (checksum_choice == NULL)
+		checksum_choice = checksum_name(xfersum_type);
+
+	if (final_call && DEBUG_GTE(NSTR, am_server ? 3 : 1)) {
+		rprintf(FINFO, "%s%s checksum: %s\n",
+			am_server ? "Server" : "Client",
+			valid_checksums.negotiated_name ? " negotiated" : "",
+			checksum_choice);
+	}
+}
+
+int csum_len_for_type(int cst, BOOL flist_csum)
+{
+	switch (cst) {
+	  case CSUM_NONE:
+		return 1;
+	  case CSUM_MD4_ARCHAIC:
+		/* The oldest checksum code is rather weird: the file-list code only sent
+		 * 2-byte checksums, but all other checksums were full MD4 length. */
+		return flist_csum ? 2 : MD4_DIGEST_LEN;
+	  case CSUM_MD4:
+	  case CSUM_MD4_OLD:
+	  case CSUM_MD4_BUSTED:
+		return MD4_DIGEST_LEN;
+	  case CSUM_MD5:
+		return MD5_DIGEST_LEN;
+#ifdef SUPPORT_XXHASH
+	  case CSUM_XXH64:
+		return 64/8;
+#endif
+	  default: /* paranoia to prevent missing case values */
+		exit_cleanup(RERR_UNSUPPORTED);
+	}
+	return 0;
+}
+
+/* Returns 0 if the checksum is not canonical (i.e. it includes a seed value).
+ * Returns 1 if the public sum order matches our internal sum order.
+ * Returns -1 if the public sum order is the reverse of our internal sum order.
+ */
+int canonical_checksum(int csum_type)
+{
+	switch (csum_type) {
+	  case CSUM_NONE:
+	  case CSUM_MD4_ARCHAIC:
+	  case CSUM_MD4_OLD:
+	  case CSUM_MD4_BUSTED:
+		break;
+	  case CSUM_MD4:
+	  case CSUM_MD5:
+		return -1;
+#ifdef SUPPORT_XXHASH
+	  case CSUM_XXH64:
+		return 1;
+#endif
+	  default: /* paranoia to prevent missing case values */
+		exit_cleanup(RERR_UNSUPPORTED);
+	}
+	return 0;
+}
+
+#ifndef HAVE_SIMD /* See simd-checksum-*.cpp. */
 /*
-  a simple 32 bit checksum that can be upadted from either end
+  a simple 32 bit checksum that can be updated from either end
   (inspired by Mark Adler's Adler-32 checksum)
   */
 uint32 get_checksum1(char *buf1, int32 len)
 {
-    int32 i;
-    uint32 s1, s2;
-    schar *buf = (schar *)buf1;
+	int32 i;
+	uint32 s1, s2;
+	schar *buf = (schar *)buf1;
 
-    s1 = s2 = 0;
-    for (i = 0; i < (len-4); i+=4) {
-	s2 += 4*(s1 + buf[i]) + 3*buf[i+1] + 2*buf[i+2] + buf[i+3] +
-	  10*CHAR_OFFSET;
-	s1 += (buf[i+0] + buf[i+1] + buf[i+2] + buf[i+3] + 4*CHAR_OFFSET);
-    }
-    for (; i < len; i++) {
-	s1 += (buf[i]+CHAR_OFFSET); s2 += s1;
-    }
-    return (s1 & 0xffff) + (s2 << 16);
+	s1 = s2 = 0;
+	for (i = 0; i < (len-4); i+=4) {
+		s2 += 4*(s1 + buf[i]) + 3*buf[i+1] + 2*buf[i+2] + buf[i+3] + 10*CHAR_OFFSET;
+		s1 += (buf[i+0] + buf[i+1] + buf[i+2] + buf[i+3] + 4*CHAR_OFFSET);
+	}
+	for (; i < len; i++) {
+		s1 += (buf[i]+CHAR_OFFSET); s2 += s1;
+	}
+	return (s1 & 0xffff) + (s2 << 16);
 }
-
+#endif
 
 void get_checksum2(char *buf, int32 len, char *sum)
 {
-	md_context m;
-
-	if (protocol_version >= 30) {
+	switch (xfersum_type) {
+#ifdef SUPPORT_XXHASH
+	  case CSUM_XXH64:
+		SIVAL64(sum, 0, XXH64(buf, len, checksum_seed));
+		break;
+#endif
+	  case CSUM_MD5: {
+		MD5_CTX m5;
 		uchar seedbuf[4];
-		md5_begin(&m);
-		md5_update(&m, (uchar *)buf, len);
-		if (checksum_seed) {
-			SIVAL(seedbuf, 0, checksum_seed);
-			md5_update(&m, seedbuf, 4);
+		MD5_Init(&m5);
+		if (proper_seed_order) {
+			if (checksum_seed) {
+				SIVALu(seedbuf, 0, checksum_seed);
+				MD5_Update(&m5, seedbuf, 4);
+			}
+			MD5_Update(&m5, (uchar *)buf, len);
+		} else {
+			MD5_Update(&m5, (uchar *)buf, len);
+			if (checksum_seed) {
+				SIVALu(seedbuf, 0, checksum_seed);
+				MD5_Update(&m5, seedbuf, 4);
+			}
 		}
-		md5_result(&m, (uchar *)sum);
-	} else {
+		MD5_Final((uchar *)sum, &m5);
+		break;
+	  }
+	  case CSUM_MD4:
+#ifdef USE_OPENSSL
+	  {
+		MD4_CTX m4;
+		MD4_Init(&m4);
+		MD4_Update(&m4, (uchar *)buf, len);
+		if (checksum_seed) {
+			uchar seedbuf[4];
+			SIVALu(seedbuf, 0, checksum_seed);
+			MD4_Update(&m4, seedbuf, 4);
+		}
+		MD4_Final((uchar *)sum, &m4);
+		break;
+	  }
+#endif
+	  case CSUM_MD4_OLD:
+	  case CSUM_MD4_BUSTED:
+	  case CSUM_MD4_ARCHAIC: {
+		md_context m;
 		int32 i;
 		static char *buf1;
 		static int32 len1;
@@ -93,18 +268,21 @@ void get_checksum2(char *buf, int32 len, char *sum)
 		 * are multiples of 64.  This is fixed by calling mdfour_update()
 		 * even when there are no more bytes.
 		 */
-		if (len - i > 0 || protocol_version >= 27)
+		if (len - i > 0 || xfersum_type > CSUM_MD4_BUSTED)
 			mdfour_update(&m, (uchar *)(buf1+i), len-i);
 
 		mdfour_result(&m, (uchar *)sum);
+		break;
+	  }
+	  default: /* paranoia to prevent missing case values */
+		exit_cleanup(RERR_UNSUPPORTED);
 	}
 }
 
-void file_checksum(char *fname, char *sum, OFF_T size)
+void file_checksum(const char *fname, const STRUCT_STAT *st_p, char *sum)
 {
 	struct map_struct *buf;
-	OFF_T i, len = size;
-	md_context m;
+	OFF_T i, len = st_p->st_size;
 	int32 remainder;
 	int fd;
 
@@ -114,38 +292,86 @@ void file_checksum(char *fname, char *sum, OFF_T size)
 	if (fd == -1)
 		return;
 
-	buf = map_file(fd, size, MAX_MAP_SIZE, CSUM_CHUNK);
+	buf = map_file(fd, len, MAX_MAP_SIZE, CHUNK_SIZE);
 
-	if (protocol_version >= 30) {
-		md5_begin(&m);
+	switch (checksum_type) {
+#ifdef SUPPORT_XXHASH
+	  case CSUM_XXH64: {
+		static XXH64_state_t* state = NULL;
+		if (!state && !(state = XXH64_createState()))
+			out_of_memory("file_checksum");
 
-		for (i = 0; i + CSUM_CHUNK <= len; i += CSUM_CHUNK) {
-			md5_update(&m, (uchar *)map_ptr(buf, i, CSUM_CHUNK),
-				   CSUM_CHUNK);
-		}
+		XXH64_reset(state, 0);
+
+		for (i = 0; i + CHUNK_SIZE <= len; i += CHUNK_SIZE)
+			XXH64_update(state, (uchar *)map_ptr(buf, i, CHUNK_SIZE), CHUNK_SIZE);
 
 		remainder = (int32)(len - i);
 		if (remainder > 0)
-			md5_update(&m, (uchar *)map_ptr(buf, i, remainder), remainder);
+			XXH64_update(state, (uchar *)map_ptr(buf, i, remainder), remainder);
+
+		SIVAL64(sum, 0, XXH64_digest(state));
+		break;
+	  }
+#endif
+	  case CSUM_MD5: {
+		MD5_CTX m5;
+
+		MD5_Init(&m5);
+
+		for (i = 0; i + CHUNK_SIZE <= len; i += 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_Final((uchar *)sum, &m5);
+		break;
+	  }
+	  case CSUM_MD4:
+#ifdef USE_OPENSSL
+	  {
+		MD4_CTX m4;
+
+		MD4_Init(&m4);
+
+		for (i = 0; i + CHUNK_SIZE <= len; i += CHUNK_SIZE)
+			MD4_Update(&m4, (uchar *)map_ptr(buf, i, CHUNK_SIZE), CHUNK_SIZE);
+
+		remainder = (int32)(len - i);
+		if (remainder > 0)
+			MD4_Update(&m4, (uchar *)map_ptr(buf, i, remainder), remainder);
+
+		MD4_Final((uchar *)sum, &m4);
+		break;
+	  }
+#endif
+	  case CSUM_MD4_OLD:
+	  case CSUM_MD4_BUSTED:
+	  case CSUM_MD4_ARCHAIC: {
+		md_context m;
 
-		md5_result(&m, (uchar *)sum);
-	} else {
 		mdfour_begin(&m);
 
-		for (i = 0; i + CSUM_CHUNK <= len; i += CSUM_CHUNK) {
-			mdfour_update(&m, (uchar *)map_ptr(buf, i, CSUM_CHUNK),
-				      CSUM_CHUNK);
-		}
+		for (i = 0; i + CHUNK_SIZE <= len; i += CHUNK_SIZE)
+			mdfour_update(&m, (uchar *)map_ptr(buf, i, CHUNK_SIZE), CHUNK_SIZE);
 
 		/* Prior to version 27 an incorrect MD4 checksum was computed
 		 * by failing to call mdfour_tail() for block sizes that
 		 * are multiples of 64.  This is fixed by calling mdfour_update()
 		 * even when there are no more bytes. */
 		remainder = (int32)(len - i);
-		if (remainder > 0 || protocol_version >= 27)
+		if (remainder > 0 || checksum_type > CSUM_MD4_BUSTED)
 			mdfour_update(&m, (uchar *)map_ptr(buf, i, remainder), remainder);
 
 		mdfour_result(&m, (uchar *)sum);
+		break;
+	  }
+	  default:
+		rprintf(FERROR, "Invalid checksum-choice for --checksum: %s (%d)\n",
+			checksum_name(checksum_type), checksum_type);
+		exit_cleanup(RERR_UNSUPPORTED);
 	}
 
 	close(fd);
@@ -153,19 +379,57 @@ void file_checksum(char *fname, char *sum, OFF_T size)
 }
 
 static int32 sumresidue;
-static md_context md;
+static union {
+	md_context md;
+#ifdef USE_OPENSSL
+	MD4_CTX m4;
+#endif
+	MD5_CTX m5;
+} ctx;
+#ifdef SUPPORT_XXHASH
+static XXH64_state_t* xxh64_state;
+#endif
+static int cursum_type;
 
-void sum_init(int seed)
+void sum_init(int csum_type, int seed)
 {
 	char s[4];
 
-	if (protocol_version >= 30)
-		md5_begin(&md);
-	else {
-		mdfour_begin(&md);
+	if (csum_type < 0)
+		csum_type = parse_csum_name(NULL, 0);
+	cursum_type = csum_type;
+
+	switch (csum_type) {
+#ifdef SUPPORT_XXHASH
+	  case CSUM_XXH64:
+		if (!xxh64_state && !(xxh64_state = XXH64_createState()))
+			out_of_memory("sum_init");
+		XXH64_reset(xxh64_state, 0);
+		break;
+#endif
+	  case CSUM_MD5:
+		MD5_Init(&ctx.m5);
+		break;
+	  case CSUM_MD4:
+#ifdef USE_OPENSSL
+		MD4_Init(&ctx.m4);
+#else
+		mdfour_begin(&ctx.md);
+		sumresidue = 0;
+#endif
+		break;
+	  case CSUM_MD4_OLD:
+	  case CSUM_MD4_BUSTED:
+	  case CSUM_MD4_ARCHAIC:
+		mdfour_begin(&ctx.md);
 		sumresidue = 0;
 		SIVAL(s, 0, seed);
 		sum_update(s, 4);
+		break;
+	  case CSUM_NONE:
+		break;
+	  default: /* paranoia to prevent missing case values */
+		exit_cleanup(RERR_UNSUPPORTED);
 	}
 }
 
@@ -179,47 +443,90 @@ void sum_init(int seed)
  **/
 void sum_update(const char *p, int32 len)
 {
-	if (protocol_version >= 30) {
-		md5_update(&md, (uchar *)p, len);
-		return;
-	}
+	switch (cursum_type) {
+#ifdef SUPPORT_XXHASH
+	  case CSUM_XXH64:
+		XXH64_update(xxh64_state, p, len);
+		break;
+#endif
+	  case CSUM_MD5:
+		MD5_Update(&ctx.m5, (uchar *)p, len);
+		break;
+	  case CSUM_MD4:
+#ifdef USE_OPENSSL
+		MD4_Update(&ctx.m4, (uchar *)p, len);
+		break;
+#endif
+	  case CSUM_MD4_OLD:
+	  case CSUM_MD4_BUSTED:
+	  case CSUM_MD4_ARCHAIC:
+		if (len + sumresidue < CSUM_CHUNK) {
+			memcpy(ctx.md.buffer + sumresidue, p, len);
+			sumresidue += len;
+			break;
+		}
 
-	if (len + sumresidue < CSUM_CHUNK) {
-		memcpy(md.buffer + sumresidue, p, len);
-		sumresidue += len;
-		return;
-	}
+		if (sumresidue) {
+			int32 i = CSUM_CHUNK - sumresidue;
+			memcpy(ctx.md.buffer + sumresidue, p, i);
+			mdfour_update(&ctx.md, (uchar *)ctx.md.buffer, CSUM_CHUNK);
+			len -= i;
+			p += i;
+		}
 
-	if (sumresidue) {
-		int32 i = CSUM_CHUNK - sumresidue;
-		memcpy(md.buffer + sumresidue, p, i);
-		mdfour_update(&md, (uchar *)md.buffer, CSUM_CHUNK);
-		len -= i;
-		p += i;
-	}
+		while (len >= CSUM_CHUNK) {
+			mdfour_update(&ctx.md, (uchar *)p, CSUM_CHUNK);
+			len -= CSUM_CHUNK;
+			p += CSUM_CHUNK;
+		}
 
-	while (len >= CSUM_CHUNK) {
-		mdfour_update(&md, (uchar *)p, CSUM_CHUNK);
-		len -= CSUM_CHUNK;
-		p += CSUM_CHUNK;
+		sumresidue = len;
+		if (sumresidue)
+			memcpy(ctx.md.buffer, p, sumresidue);
+		break;
+	  case CSUM_NONE:
+		break;
+	  default: /* paranoia to prevent missing case values */
+		exit_cleanup(RERR_UNSUPPORTED);
 	}
-
-	sumresidue = len;
-	if (sumresidue)
-		memcpy(md.buffer, p, sumresidue);
 }
 
+/* NOTE: all the callers of sum_end() pass in a pointer to a buffer that is
+ * MAX_DIGEST_LEN in size, so even if the csum-len is shorter that that (i.e.
+ * CSUM_MD4_ARCHAIC), we don't have to worry about limiting the data we write
+ * into the "sum" buffer. */
 int sum_end(char *sum)
 {
-	if (protocol_version >= 30) {
-		md5_result(&md, (uchar *)sum);
-		return MD5_DIGEST_LEN;
+	switch (cursum_type) {
+#ifdef SUPPORT_XXHASH
+	  case CSUM_XXH64:
+		SIVAL64(sum, 0, XXH64_digest(xxh64_state));
+		break;
+#endif
+	  case CSUM_MD5:
+		MD5_Final((uchar *)sum, &ctx.m5);
+		break;
+	  case CSUM_MD4:
+#ifdef USE_OPENSSL
+		MD4_Final((uchar *)sum, &ctx.m4);
+		break;
+#endif
+	  case CSUM_MD4_OLD:
+		mdfour_update(&ctx.md, (uchar *)ctx.md.buffer, sumresidue);
+		mdfour_result(&ctx.md, (uchar *)sum);
+		break;
+	  case CSUM_MD4_BUSTED:
+	  case CSUM_MD4_ARCHAIC:
+		if (sumresidue)
+			mdfour_update(&ctx.md, (uchar *)ctx.md.buffer, sumresidue);
+		mdfour_result(&ctx.md, (uchar *)sum);
+		break;
+	  case CSUM_NONE:
+		*sum = '\0';
+		break;
+	  default: /* paranoia to prevent missing case values */
+		exit_cleanup(RERR_UNSUPPORTED);
 	}
 
-	if (sumresidue || protocol_version >= 27)
-		mdfour_update(&md, (uchar *)md.buffer, sumresidue);
-
-	mdfour_result(&md, (uchar *)sum);
-
-	return MD4_DIGEST_LEN;
+	return csum_len_for_type(cursum_type, 0);
 }

chmod.c

@@ -2,7 +2,7 @@
  * Implement the core of the --chmod option.
  *
  * Copyright (C) 2002 Scott Howard
- * Copyright (C) 2005-2007 Wayne Davison
+ * Copyright (C) 2005-2020 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
@@ -19,6 +19,7 @@
  */
 
 #include "rsync.h"
+#include "itypes.h"
 
 extern mode_t orig_umask;
 
@@ -35,13 +36,15 @@ struct chmod_mode_struct {
 #define CHMOD_ADD 1
 #define CHMOD_SUB 2
 #define CHMOD_EQ  3
+#define CHMOD_SET 4
 
 #define STATE_ERROR 0
 #define STATE_1ST_HALF 1
 #define STATE_2ND_HALF 2
+#define STATE_OCTAL_NUM 3
 
 /* Parse a chmod-style argument, and break it down into one or more AND/OR
- * pairs in a linked list.  We return a pointer to new items on succcess
+ * pairs in a linked list.  We return a pointer to new items on success
  * (appending the items to the specified list), or NULL on error. */
 struct chmod_mode_struct *parse_chmod(const char *modestr,
 				      struct chmod_mode_struct **root_mode_ptr)
@@ -87,6 +90,10 @@ struct chmod_mode_struct *parse_chmod(const char *modestr,
 				curr_mode->ModeAND = CHMOD_BITS - (where * 7) - (topoct ? topbits : 0);
 				curr_mode->ModeOR  = bits + topoct;
 				break;
+			case CHMOD_SET:
+				curr_mode->ModeAND = 0;
+				curr_mode->ModeOR  = bits;
+				break;
 			}
 
 			curr_mode->flags = flags;
@@ -99,7 +106,8 @@ struct chmod_mode_struct *parse_chmod(const char *modestr,
 			where = what = op = topoct = topbits = flags = 0;
 		}
 
-		if (state != STATE_2ND_HALF) {
+		switch (state) {
+		case STATE_1ST_HALF:
 			switch (*modestr) {
 			case 'D':
 				if (flags & FLAG_FILES_ONLY)
@@ -138,10 +146,17 @@ struct chmod_mode_struct *parse_chmod(const char *modestr,
 				state = STATE_2ND_HALF;
 				break;
 			default:
-				state = STATE_ERROR;
+				if (isDigit(modestr) && *modestr < '8' && !where) {
+					op = CHMOD_SET;
+					state =  STATE_OCTAL_NUM;
+					where = 1;
+					what = *modestr - '0';
+				} else
+					state = STATE_ERROR;
 				break;
 			}
-		} else {
+			break;
+		case STATE_2ND_HALF:
 			switch (*modestr) {
 			case 'r':
 				what |= 4;
@@ -168,6 +183,15 @@ struct chmod_mode_struct *parse_chmod(const char *modestr,
 				state = STATE_ERROR;
 				break;
 			}
+			break;
+		case STATE_OCTAL_NUM:
+			if (isDigit(modestr) && *modestr < '8') {
+				what = what*8 + *modestr - '0';
+				if (what > CHMOD_BITS)
+					state = STATE_ERROR;
+			} else
+				state = STATE_ERROR;
+			break;
 		}
 		modestr++;
 	}

cleanup.c

@@ -4,7 +4,7 @@
  * Copyright (C) 1996-2000 Andrew Tridgell
  * Copyright (C) 1996 Paul Mackerras
  * Copyright (C) 2002 Martin Pool
- * Copyright (C) 2003-2007 Wayne Davison
+ * Copyright (C) 2003-2020 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
@@ -22,14 +22,23 @@
 
 #include "rsync.h"
 
+extern int dry_run;
 extern int am_server;
 extern int am_daemon;
+extern int am_receiver;
+extern int am_sender;
 extern int io_error;
 extern int keep_partial;
-extern int log_got_error;
+extern int got_xfer_error;
+extern int protocol_version;
+extern int output_needs_newline;
 extern char *partial_dir;
 extern char *logfile_name;
 
+int called_from_signal_handler = 0;
+BOOL shutting_down = False;
+BOOL flush_ok_after_signal = False;
+
 #ifdef HAVE_SIGACTION
 static struct sigaction sigact;
 #endif
@@ -78,10 +87,10 @@ void close_all(void)
  **/
 int cleanup_got_literal = 0;
 
-static char *cleanup_fname;
-static char *cleanup_new_fname;
+static const char *cleanup_fname;
+static const char *cleanup_new_fname;
 static struct file_struct *cleanup_file;
-static int cleanup_fd_r, cleanup_fd_w;
+static int cleanup_fd_r = -1, cleanup_fd_w = -1;
 static pid_t cleanup_pid = 0;
 
 pid_t cleanup_child_pid = -1;
@@ -93,75 +102,104 @@ pid_t cleanup_child_pid = -1;
  **/
 NORETURN void _exit_cleanup(int code, const char *file, int line)
 {
-	static int cleanup_step = 0;
-	static int exit_code = 0;
-	static int unmodified_code = 0;
+	static int switch_step = 0;
+	static int exit_code = 0, exit_line = 0;
+	static const char *exit_file = NULL;
+	static int first_code = 0;
 
 	SIGACTION(SIGUSR1, SIG_IGN);
 	SIGACTION(SIGUSR2, SIG_IGN);
 
-	if (exit_code) /* Preserve first error code when recursing. */
-		code = exit_code;
+	if (!exit_code) { /* Preserve first error exit info when recursing. */
+		exit_code = code;
+		exit_file = file;
+		exit_line = line < 0 ? -line : line;
+	}
 
 	/* If this is the exit at the end of the run, the server side
-	 * should not attempt to output a message (see log.c). */
+	 * should not attempt to output a message (see log_exit()). */
 	if (am_server && code == 0)
 		am_server = 2;
 
 	/* Some of our actions might cause a recursive call back here, so we
 	 * keep track of where we are in the cleanup and never repeat a step. */
-	switch (cleanup_step) {
-#include "case_N.h" /* case 0: cleanup_step++; */
+	switch (switch_step) {
+#include "case_N.h" /* case 0: */
+		switch_step++;
 
-		exit_code = unmodified_code = code;
+		first_code = code;
 
-		if (verbose > 3) {
-			rprintf(FINFO,
-				"_exit_cleanup(code=%d, file=%s, line=%d): entered\n",
-				code, file, line);
+		if (output_needs_newline) {
+			fputc('\n', stdout);
+			output_needs_newline = 0;
+		}
+
+		if (DEBUG_GTE(EXIT, 2)) {
+			rprintf(FINFO,
+				"[%s] _exit_cleanup(code=%d, file=%s, line=%d): entered\n",
+				who_am_i(), code, file, line);
 		}
 
-		/* FALLTHROUGH */
 #include "case_N.h"
+		switch_step++;
 
 		if (cleanup_child_pid != -1) {
 			int status;
 			int pid = wait_process(cleanup_child_pid, &status, WNOHANG);
 			if (pid == cleanup_child_pid) {
 				status = WEXITSTATUS(status);
-				if (status > code)
-					code = exit_code = status;
+				if (status > exit_code)
+					exit_code = status;
 			}
 		}
 
-		/* FALLTHROUGH */
 #include "case_N.h"
+		switch_step++;
 
-		if (cleanup_got_literal && cleanup_fname && cleanup_new_fname
-		 && keep_partial && handle_partial_dir(cleanup_new_fname, PDIR_CREATE)) {
-			char *fname = cleanup_fname;
-			cleanup_fname = NULL;
-			if (cleanup_fd_r != -1)
+		if (cleanup_got_literal && (cleanup_fname || cleanup_fd_w != -1)) {
+			if (cleanup_fd_r != -1) {
 				close(cleanup_fd_r);
+				cleanup_fd_r = -1;
+			}
 			if (cleanup_fd_w != -1) {
 				flush_write_file(cleanup_fd_w);
 				close(cleanup_fd_w);
+				cleanup_fd_w = -1;
+			}
+			if (cleanup_fname && cleanup_new_fname && keep_partial
+			 && handle_partial_dir(cleanup_new_fname, PDIR_CREATE)) {
+				int tweak_modtime = 0;
+				const char *fname = cleanup_fname;
+				cleanup_fname = NULL;
+				if (!partial_dir) {
+					/* We don't want to leave a partial file with a modern time or it
+					 * could be skipped via --update.  Setting the time to something
+					 * really old also helps it to stand out as unfinished in an ls. */
+					tweak_modtime = 1;
+					cleanup_file->modtime = 0;
+				}
+				finish_transfer(cleanup_new_fname, fname, NULL, NULL,
+						cleanup_file, tweak_modtime, !partial_dir);
 			}
-			finish_transfer(cleanup_new_fname, fname, NULL, NULL,
-					cleanup_file, 0, !partial_dir);
 		}
 
-		/* FALLTHROUGH */
 #include "case_N.h"
+		switch_step++;
 
-		io_flush(FULL_FLUSH);
+		if (flush_ok_after_signal) {
+			flush_ok_after_signal = False;
+			if (code == RERR_SIGNAL)
+				io_flush(FULL_FLUSH);
+		}
+		if (!exit_code && !code)
+			io_flush(FULL_FLUSH);
 
-		/* FALLTHROUGH */
 #include "case_N.h"
+		switch_step++;
 
 		if (cleanup_fname)
 			do_unlink(cleanup_fname);
-		if (code)
+		if (exit_code)
 			kill_all(SIGUSR1);
 		if (cleanup_pid && cleanup_pid == getpid()) {
 			char *pidf = lp_pid_file();
@@ -169,32 +207,59 @@ NORETURN void _exit_cleanup(int code, const char *file, int line)
 				unlink(lp_pid_file());
 		}
 
-		if (code == 0) {
+		if (exit_code == 0) {
+			if (code)
+				exit_code = code;
 			if (io_error & IOERR_DEL_LIMIT)
-				code = exit_code = RERR_DEL_LIMIT;
+				exit_code = RERR_DEL_LIMIT;
 			if (io_error & IOERR_VANISHED)
-				code = exit_code = RERR_VANISHED;
-			if (io_error & IOERR_GENERAL || log_got_error)
-				code = exit_code = RERR_PARTIAL;
+				exit_code = RERR_VANISHED;
+			if (io_error & IOERR_GENERAL || got_xfer_error)
+				exit_code = RERR_PARTIAL;
 		}
 
-		if (code || am_daemon || (logfile_name && (am_server || !verbose)))
-			log_exit(code, file, line);
+		/* If line < 0, this exit is after a MSG_ERROR_EXIT event, so
+		 * we don't want to output a duplicate error. */
+		if ((exit_code && line > 0)
+		 || am_daemon || (logfile_name && (am_server || !INFO_GTE(STATS, 1))))
+			log_exit(exit_code, exit_file, exit_line);
 
-		/* FALLTHROUGH */
 #include "case_N.h"
+		switch_step++;
 
-		if (verbose > 2) {
+		if (DEBUG_GTE(EXIT, 1)) {
 			rprintf(FINFO,
-				"_exit_cleanup(code=%d, file=%s, line=%d): "
-				"about to call exit(%d)\n",
-				unmodified_code, file, line, code);
+				"[%s] _exit_cleanup(code=%d, file=%s, line=%d): "
+				"about to call exit(%d)%s\n",
+				who_am_i(), first_code, exit_file, exit_line, exit_code,
+				dry_run ? " (DRY RUN)" : "");
 		}
 
-		/* FALLTHROUGH */
 #include "case_N.h"
+		switch_step++;
 
-		if (am_server && code)
+		if (exit_code && exit_code != RERR_SOCKETIO && exit_code != RERR_STREAMIO && exit_code != RERR_SIGNAL1
+		 && exit_code != RERR_TIMEOUT && !shutting_down) {
+			if (protocol_version >= 31 || am_receiver) {
+				if (line > 0) {
+					if (DEBUG_GTE(EXIT, 3)) {
+						rprintf(FINFO, "[%s] sending MSG_ERROR_EXIT with exit_code %d\n",
+							who_am_i(), exit_code);
+					}
+					send_msg_int(MSG_ERROR_EXIT, exit_code);
+				}
+				if (!am_sender)
+					io_flush(MSG_FLUSH); /* Be sure to send all messages */
+				noop_io_until_death();
+			}
+			else if (!am_sender)
+				io_flush(MSG_FLUSH); /* Be sure to send all messages */
+		}
+
+#include "case_N.h"
+		switch_step++;
+
+		if (am_server && exit_code)
 			msleep(100);
 		close_all();
 
@@ -203,17 +268,20 @@ NORETURN void _exit_cleanup(int code, const char *file, int line)
 		break;
 	}
 
-	exit(code);
+	if (called_from_signal_handler)
+		_exit(exit_code);
+	exit(exit_code);
 }
 
 void cleanup_disable(void)
 {
 	cleanup_fname = cleanup_new_fname = NULL;
+	cleanup_fd_r = cleanup_fd_w = -1;
 	cleanup_got_literal = 0;
 }
 
 
-void cleanup_set(char *fnametmp, char *fname, struct file_struct *file,
+void cleanup_set(const char *fnametmp, const char *fname, struct file_struct *file,
 		 int fd_r, int fd_w)
 {
 	cleanup_fname = fnametmp;

clientname.c

@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1992-2001 Andrew Tridgell <tridge@samba.org>
  * Copyright (C) 2001, 2002 Martin Pool <mbp@samba.org>
- * Copyright (C) 2002-2007 Wayne Davison
+ * Copyright (C) 2002-2020 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
@@ -27,50 +27,60 @@
  */
 
 #include "rsync.h"
+#include "itypes.h"
+
+extern int am_daemon;
 
 static const char default_name[] = "UNKNOWN";
-extern int am_server;
+static const char proxyv2sig[] = "\r\n\r\n\0\r\nQUIT\n";
 
+static char ipaddr_buf[100];
 
-/**
- * Return the IP addr of the client as a string
- **/
+#define PROXY_V2_SIG_SIZE ((int)sizeof proxyv2sig - 1)
+#define PROXY_V2_HEADER_SIZE (PROXY_V2_SIG_SIZE + 1 + 1 + 2)
+
+#define CMD_LOCAL 0
+#define CMD_PROXY 1
+
+#define PROXY_FAM_TCPv4 0x11
+#define PROXY_FAM_TCPv6 0x21
+
+#define GET_SOCKADDR_FAMILY(ss) ((struct sockaddr*)ss)->sa_family
+
+static void client_sockaddr(int fd, struct sockaddr_storage *ss, socklen_t *ss_len);
+static int check_name(const char *ipaddr, const struct sockaddr_storage *ss, char *name_buf, size_t name_buf_size);
+static int valid_ipaddr(const char *s);
+
+/* Return the IP addr of the client as a string. */
 char *client_addr(int fd)
 {
-	static char addr_buf[100];
-	static int initialised;
 	struct sockaddr_storage ss;
 	socklen_t length = sizeof ss;
-	char *ssh_info, *p;
 
-	if (initialised)
-		return addr_buf;
+	if (*ipaddr_buf)
+		return ipaddr_buf;
 
-	initialised = 1;
-
-	if (am_server) {	/* daemon over --rsh mode */
-		strlcpy(addr_buf, "0.0.0.0", sizeof addr_buf);
-		if ((ssh_info = getenv("SSH_CONNECTION")) != NULL
-		    || (ssh_info = getenv("SSH_CLIENT")) != NULL
-		    || (ssh_info = getenv("SSH2_CLIENT")) != NULL) {
-			strlcpy(addr_buf, ssh_info, sizeof addr_buf);
+	if (am_daemon < 0) {	/* daemon over --rsh mode */
+		char *env_str;
+		strlcpy(ipaddr_buf, "0.0.0.0", sizeof ipaddr_buf);
+		if ((env_str = getenv("REMOTE_HOST")) != NULL
+		 || (env_str = getenv("SSH_CONNECTION")) != NULL
+		 || (env_str = getenv("SSH_CLIENT")) != NULL
+		 || (env_str = getenv("SSH2_CLIENT")) != NULL) {
+			char *p;
+			strlcpy(ipaddr_buf, env_str, sizeof ipaddr_buf);
 			/* Truncate the value to just the IP address. */
-			if ((p = strchr(addr_buf, ' ')) != NULL)
+			if ((p = strchr(ipaddr_buf, ' ')) != NULL)
 				*p = '\0';
 		}
-	} else {
-		client_sockaddr(fd, &ss, &length);
-		getnameinfo((struct sockaddr *)&ss, length,
-			    addr_buf, sizeof addr_buf, NULL, 0, NI_NUMERICHOST);
+		if (valid_ipaddr(ipaddr_buf))
+			return ipaddr_buf;
 	}
 
-	return addr_buf;
-}
+	client_sockaddr(fd, &ss, &length);
+	getnameinfo((struct sockaddr *)&ss, length, ipaddr_buf, sizeof ipaddr_buf, NULL, 0, NI_NUMERICHOST);
 
-
-static int get_sockaddr_family(const struct sockaddr_storage *ss)
-{
-	return ((struct sockaddr *) ss)->sa_family;
+	return ipaddr_buf;
 }
 
 
@@ -87,68 +97,216 @@ static int get_sockaddr_family(const struct sockaddr_storage *ss)
  * After translation from sockaddr to name we do a forward lookup to
  * make sure nobody is spoofing PTR records.
  **/
-char *client_name(int fd)
+char *client_name(const char *ipaddr)
 {
 	static char name_buf[100];
-	static char port_buf[100];
-	static int initialised;
+	char port_buf[100];
 	struct sockaddr_storage ss;
 	socklen_t ss_len;
+	struct addrinfo hint, *answer;
+	int err;
 
-	if (initialised)
+	if (*name_buf)
 		return name_buf;
 
 	strlcpy(name_buf, default_name, sizeof name_buf);
-	initialised = 1;
+
+	if (strcmp(ipaddr, "0.0.0.0") == 0)
+		return name_buf;
 
 	memset(&ss, 0, sizeof ss);
-
-	if (am_server) {	/* daemon over --rsh mode */
-		char *addr = client_addr(fd);
-		struct addrinfo hint, *answer;
-		int err;
-
-		memset(&hint, 0, sizeof hint);
+	memset(&hint, 0, sizeof hint);
 
 #ifdef AI_NUMERICHOST
-		hint.ai_flags = AI_NUMERICHOST;
+	hint.ai_flags = AI_NUMERICHOST;
 #endif
-		hint.ai_socktype = SOCK_STREAM;
+	hint.ai_socktype = SOCK_STREAM;
 
-		if ((err = getaddrinfo(addr, NULL, &hint, &answer)) != 0) {
-			rprintf(FLOG, "malformed address %s: %s\n",
-			        addr, gai_strerror(err));
-			return name_buf;
-		}
-
-		switch (answer->ai_family) {
-		case AF_INET:
-			ss_len = sizeof (struct sockaddr_in);
-			memcpy(&ss, answer->ai_addr, ss_len);
-			break;
-#ifdef INET6
-		case AF_INET6:
-			ss_len = sizeof (struct sockaddr_in6);
-			memcpy(&ss, answer->ai_addr, ss_len);
-			break;
-#endif
-		default:
-			exit_cleanup(RERR_SOCKETIO);
-		}
-		freeaddrinfo(answer);
-	} else {
-		ss_len = sizeof ss;
-		client_sockaddr(fd, &ss, &ss_len);
+	if ((err = getaddrinfo(ipaddr, NULL, &hint, &answer)) != 0) {
+		rprintf(FLOG, "malformed address %s: %s\n", ipaddr, gai_strerror(err));
+		return name_buf;
 	}
 
-	if (lookup_name(fd, &ss, ss_len, name_buf, sizeof name_buf,
-			port_buf, sizeof port_buf) == 0)
-		check_name(fd, &ss, name_buf, sizeof name_buf);
+	switch (answer->ai_family) {
+	case AF_INET:
+		ss_len = sizeof (struct sockaddr_in);
+		memcpy(&ss, answer->ai_addr, ss_len);
+		break;
+#ifdef INET6
+	case AF_INET6:
+		ss_len = sizeof (struct sockaddr_in6);
+		memcpy(&ss, answer->ai_addr, ss_len);
+		break;
+#endif
+	default:
+		assert(0);
+	}
+	freeaddrinfo(answer);
+
+	/* reverse lookup */
+	err = getnameinfo((struct sockaddr*)&ss, ss_len, name_buf, sizeof name_buf,
+			  port_buf, sizeof port_buf, NI_NAMEREQD | NI_NUMERICSERV);
+	if (err) {
+		strlcpy(name_buf, default_name, sizeof name_buf);
+		rprintf(FLOG, "name lookup failed for %s: %s\n", ipaddr, gai_strerror(err));
+	} else
+		check_name(ipaddr, &ss, name_buf, sizeof name_buf);
 
 	return name_buf;
 }
 
 
+/* Try to read an proxy protocol header (V1 or V2). Returns 1 on success or 0 on failure. */
+int read_proxy_protocol_header(int fd)
+{
+	union {
+		struct {
+			char line[108];
+		} v1;
+		struct {
+			char sig[PROXY_V2_SIG_SIZE];
+			char ver_cmd;
+			char fam;
+			char len[2];
+			union {
+				struct {
+					char src_addr[4];
+					char dst_addr[4];
+					char src_port[2];
+					char dst_port[2];
+				} ip4;
+				struct {
+					char src_addr[16];
+					char dst_addr[16];
+					char src_port[2];
+					char dst_port[2];
+				} ip6;
+				struct {
+					char src_addr[108];
+					char dst_addr[108];
+				} unx;
+			} addr;
+		} v2;
+	} hdr;
+
+	read_buf(fd, (char*)&hdr, PROXY_V2_SIG_SIZE);
+
+	if (memcmp(hdr.v2.sig, proxyv2sig, PROXY_V2_SIG_SIZE) == 0) { /* Proxy V2 */
+		int ver, cmd, size;
+
+		read_buf(fd, (char*)&hdr + PROXY_V2_SIG_SIZE, PROXY_V2_HEADER_SIZE - PROXY_V2_SIG_SIZE);
+
+		ver = (hdr.v2.ver_cmd & 0xf0) >> 4;
+		cmd = (hdr.v2.ver_cmd & 0x0f);
+		size = (hdr.v2.len[0] << 8) + hdr.v2.len[1];
+
+		if (ver != 2 || size + PROXY_V2_HEADER_SIZE > (int)sizeof hdr)
+			return 0;
+
+		/* Grab all the remaining data in the binary request. */
+		read_buf(fd, (char*)&hdr + PROXY_V2_HEADER_SIZE, size);
+
+		switch (cmd) {
+		case CMD_PROXY:
+			switch (hdr.v2.fam) {
+			case PROXY_FAM_TCPv4:
+				if (size != sizeof hdr.v2.addr.ip4)
+					return 0;
+				inet_ntop(AF_INET, hdr.v2.addr.ip4.src_addr, ipaddr_buf, sizeof ipaddr_buf);
+				return valid_ipaddr(ipaddr_buf);
+			case PROXY_FAM_TCPv6:
+				if (size != sizeof hdr.v2.addr.ip6)
+					return 0;
+				inet_ntop(AF_INET6, hdr.v2.addr.ip6.src_addr, ipaddr_buf, sizeof ipaddr_buf);
+				return valid_ipaddr(ipaddr_buf);
+			default:
+				break;
+			}
+			/* For an unsupported protocol we'll ignore the proxy data (leaving ipaddr_buf unset)
+			 * and accept the connection, which will get handled as a normal socket addr. */
+			return 1;
+		case CMD_LOCAL:
+			return 1;
+		default:
+			break;
+		}
+
+		return 0;
+	}
+
+	if (memcmp(hdr.v1.line, "PROXY", 5) == 0) { /* Proxy V1 */
+		char *endc, *sp, *p = hdr.v1.line + PROXY_V2_SIG_SIZE;
+		int port_chk;
+
+		*p = '\0';
+		if (!strchr(hdr.v1.line, '\n')) {
+			while (1) {
+				read_buf(fd, p, 1);
+				if (*p++ == '\n')
+					break;
+				if (p - hdr.v1.line >= (int)sizeof hdr.v1.line - 1)
+					return 0;
+			}
+			*p = '\0';
+		}
+
+		endc = strchr(hdr.v1.line, '\r');
+		if (!endc || endc[1] != '\n' || endc[2])
+			return 0;
+		*endc = '\0';
+
+		p = hdr.v1.line + 5;
+
+		if (!isSpace(p++))
+			return 0;
+		if (strncmp(p, "TCP4", 4) == 0)
+			p += 4;
+		else if (strncmp(p, "TCP6", 4) == 0)
+			p += 4;
+		else if (strncmp(p, "UNKNOWN", 7) == 0)
+			return 1;
+		else
+			return 0;
+
+		if (!isSpace(p++))
+			return 0;
+
+		if ((sp = strchr(p, ' ')) == NULL)
+			return 0;
+		*sp = '\0';
+		if (!valid_ipaddr(p))
+			return 0;
+		strlcpy(ipaddr_buf, p, sizeof ipaddr_buf); /* It will always fit when valid. */
+
+		p = sp + 1;
+		if ((sp = strchr(p, ' ')) == NULL)
+			return 0;
+		*sp = '\0';
+		if (!valid_ipaddr(p))
+			return 0;
+		/* Ignore destination address. */
+
+		p = sp + 1;
+		if ((sp = strchr(p, ' ')) == NULL)
+			return 0;
+		*sp = '\0';
+		port_chk = strtol(p, &endc, 10);
+		if (*endc || port_chk == 0)
+			return 0;
+		/* Ignore source port. */
+
+		p = sp + 1;
+		port_chk = strtol(p, &endc, 10);
+		if (*endc || port_chk == 0)
+			return 0;
+		/* Ignore destination port. */
+
+		return 1;
+	}
+
+	return 0;
+}
+
 
 /**
  * Get the sockaddr for the client.
@@ -156,9 +314,7 @@ char *client_name(int fd)
  * If it comes in as an ipv4 address mapped into IPv6 format then we
  * convert it back to a regular IPv4.
  **/
-void client_sockaddr(int fd,
-		     struct sockaddr_storage *ss,
-		     socklen_t *ss_len)
+static void client_sockaddr(int fd, struct sockaddr_storage *ss, socklen_t *ss_len)
 {
 	memset(ss, 0, sizeof *ss);
 
@@ -169,8 +325,8 @@ void client_sockaddr(int fd,
 	}
 
 #ifdef INET6
-	if (get_sockaddr_family(ss) == AF_INET6 &&
-	    IN6_IS_ADDR_V4MAPPED(&((struct sockaddr_in6 *)ss)->sin6_addr)) {
+	if (GET_SOCKADDR_FAMILY(ss) == AF_INET6
+	 && IN6_IS_ADDR_V4MAPPED(&((struct sockaddr_in6 *)ss)->sin6_addr)) {
 		/* OK, so ss is in the IPv6 family, but it is really
 		 * an IPv4 address: something like
 		 * "::ffff:10.130.1.2".  If we use it as-is, then the
@@ -193,51 +349,20 @@ void client_sockaddr(int fd,
 		/* There is a macro to extract the mapped part
 		 * (IN6_V4MAPPED_TO_SINADDR ?), but it does not seem
 		 * to be present in the Linux headers. */
-		memcpy(&sin->sin_addr, &sin6.sin6_addr.s6_addr[12],
-		    sizeof sin->sin_addr);
+		memcpy(&sin->sin_addr, &sin6.sin6_addr.s6_addr[12], sizeof sin->sin_addr);
 	}
 #endif
 }
 
 
-/**
- * Look up a name from @p ss into @p name_buf.
- *
- * @param fd file descriptor for client socket.
- **/
-int lookup_name(int fd, const struct sockaddr_storage *ss,
-		socklen_t ss_len,
-		char *name_buf, size_t name_buf_size,
-		char *port_buf, size_t port_buf_size)
-{
-	int name_err;
-
-	/* reverse lookup */
-	name_err = getnameinfo((struct sockaddr *) ss, ss_len,
-			       name_buf, name_buf_size,
-			       port_buf, port_buf_size,
-			       NI_NAMEREQD | NI_NUMERICSERV);
-	if (name_err != 0) {
-		strlcpy(name_buf, default_name, name_buf_size);
-		rprintf(FLOG, "name lookup failed for %s: %s\n",
-			client_addr(fd), gai_strerror(name_err));
-		return name_err;
-	}
-
-	return 0;
-}
-
-
-
 /**
  * Compare an addrinfo from the resolver to a sockinfo.
  *
  * Like strcmp, returns 0 for identical.
  **/
-int compare_addrinfo_sockaddr(const struct addrinfo *ai,
-			      const struct sockaddr_storage *ss)
+static int compare_addrinfo_sockaddr(const struct addrinfo *ai, const struct sockaddr_storage *ss)
 {
-	int ss_family = get_sockaddr_family(ss);
+	int ss_family = GET_SOCKADDR_FAMILY(ss);
 	const char fn[] = "compare_addrinfo_sockaddr";
 
 	if (ai->ai_family != ss_family) {
@@ -253,8 +378,7 @@ int compare_addrinfo_sockaddr(const struct addrinfo *ai,
 		sin1 = (const struct sockaddr_in *) ss;
 		sin2 = (const struct sockaddr_in *) ai->ai_addr;
 
-		return memcmp(&sin1->sin_addr, &sin2->sin_addr,
-			      sizeof sin1->sin_addr);
+		return memcmp(&sin1->sin_addr, &sin2->sin_addr, sizeof sin1->sin_addr);
 	}
 
 #ifdef INET6
@@ -264,14 +388,13 @@ int compare_addrinfo_sockaddr(const struct addrinfo *ai,
 		sin1 = (const struct sockaddr_in6 *) ss;
 		sin2 = (const struct sockaddr_in6 *) ai->ai_addr;
 
-		if (ai->ai_addrlen < sizeof (struct sockaddr_in6)) {
+		if (ai->ai_addrlen < (int)sizeof (struct sockaddr_in6)) {
 			rprintf(FLOG, "%s: too short sockaddr_in6; length=%d\n",
-				fn, ai->ai_addrlen);
+				fn, (int)ai->ai_addrlen);
 			return 1;
 		}
 
-		if (memcmp(&sin1->sin6_addr, &sin2->sin6_addr,
-			   sizeof sin1->sin6_addr))
+		if (memcmp(&sin1->sin6_addr, &sin2->sin6_addr, sizeof sin1->sin6_addr))
 			return 1;
 
 #ifdef HAVE_SOCKADDR_IN6_SCOPE_ID
@@ -297,13 +420,11 @@ int compare_addrinfo_sockaddr(const struct addrinfo *ai,
  * because it doesn't seem that it could be spoofed in any way, and
  * getaddrinfo on random service names seems to cause problems on AIX.
  **/
-int check_name(int fd,
-	       const struct sockaddr_storage *ss,
-	       char *name_buf, size_t name_buf_size)
+static int check_name(const char *ipaddr, const struct sockaddr_storage *ss, char *name_buf, size_t name_buf_size)
 {
 	struct addrinfo hints, *res, *res0;
 	int error;
-	int ss_family = get_sockaddr_family(ss);
+	int ss_family = GET_SOCKADDR_FAMILY(ss);
 
 	memset(&hints, 0, sizeof hints);
 	hints.ai_family = ss_family;
@@ -334,10 +455,74 @@ int check_name(int fd,
 		/* We hit the end of the list without finding an
 		 * address that was the same as ss. */
 		rprintf(FLOG, "%s is not a known address for \"%s\": "
-			"spoofed address?\n", client_addr(fd), name_buf);
+			"spoofed address?\n", ipaddr, name_buf);
 		strlcpy(name_buf, default_name, name_buf_size);
 	}
 
 	freeaddrinfo(res0);
 	return 0;
 }
+
+/* Returns 1 for a valid IPv4 or IPv6 addr, or 0 for a bad one. */
+static int valid_ipaddr(const char *s)
+{
+	int i;
+
+	if (strchr(s, ':') != NULL) { /* Only IPv6 has a colon. */
+		int count, saw_double_colon = 0;
+		int ipv4_at_end = 0;
+
+		if (*s == ':') { /* A colon at the start must be a :: */
+			if (*++s != ':')
+				return 0;
+			saw_double_colon = 1;
+			s++;
+		}
+
+		for (count = 0; count < 8; count++) {
+			if (!*s)
+				return saw_double_colon && count < 7;
+
+			if (strchr(s, ':') == NULL && strchr(s, '.') != NULL) {
+				if ((!saw_double_colon && count != 6) || (saw_double_colon && count > 6))
+					return 0;
+				ipv4_at_end = 1;
+				break;
+			}
+
+			if (!isHexDigit(s++)) /* Need 1-4 hex digits */
+				return 0;
+			if (isHexDigit(s) && isHexDigit(++s) && isHexDigit(++s) && isHexDigit(++s))
+				return 0;
+
+			if (*s == ':') {
+				if (!*++s)
+					return 0;
+				if (*s == ':') {
+					if (saw_double_colon)
+						return 0;
+					saw_double_colon = 1;
+					s++;
+				}
+			}
+		}
+
+		if (!ipv4_at_end)
+			return !*s;
+	}
+
+	/* IPv4 */
+	for (i = 0; i < 4; i++) {
+		long n;
+		char *end;
+
+		if (i && *s++ != '.')
+			return 0;
+		n = strtol(s, &end, 10);
+		if (n > 255 || n < 0 || end <= s || end > s+3)
+			return 0;
+		s = end;
+	}
+
+	return !*s;
+}

compat.c

@@ -3,7 +3,7 @@
  *
  * Copyright (C) Andrew Tridgell 1996
  * Copyright (C) Paul Mackerras 1996
- * Copyright (C) 2004-2007 Wayne Davison
+ * Copyright (C) 2004-2020 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
@@ -21,11 +21,6 @@
 
 #include "rsync.h"
 
-int remote_protocol = 0;
-int file_extra_cnt = 0; /* count of file-list extras that everyone gets */
-int inc_recurse = 0;
-
-extern int verbose;
 extern int am_server;
 extern int am_sender;
 extern int local_server;
@@ -33,9 +28,11 @@ extern int inplace;
 extern int recurse;
 extern int use_qsort;
 extern int allow_inc_recurse;
+extern int preallocate_files;
 extern int append_mode;
 extern int fuzzy_basis;
 extern int read_batch;
+extern int write_batch;
 extern int delay_updates;
 extern int checksum_seed;
 extern int basis_dir_cnt;
@@ -44,29 +41,76 @@ extern int protocol_version;
 extern int protect_args;
 extern int preserve_uid;
 extern int preserve_gid;
+extern int preserve_atimes;
 extern int preserve_acls;
 extern int preserve_xattrs;
+extern int xfer_flags_as_varint;
 extern int need_messages_from_generator;
 extern int delete_mode, delete_before, delete_during, delete_after;
-extern char *shell_cmd; /* contains VER.SUB string if client is a pre-release */
+extern int do_compression;
+extern int do_compression_level;
+extern char *shell_cmd;
 extern char *partial_dir;
-extern char *dest_option;
 extern char *files_from;
 extern char *filesfrom_host;
-extern struct filter_list_struct filter_list;
+extern const char *checksum_choice;
+extern const char *compress_choice;
+extern filter_rule_list filter_list;
+extern int need_unsorted_flist;
 #ifdef ICONV_OPTION
-extern char *iconv_opt;
 extern iconv_t ic_send, ic_recv;
+extern char *iconv_opt;
 #endif
+extern struct name_num_obj valid_checksums;
+
+int remote_protocol = 0;
+int file_extra_cnt = 0; /* count of file-list extras that everyone gets */
+int inc_recurse = 0;
+int compat_flags = 0;
+int use_safe_inc_flist = 0;
+int want_xattr_optim = 0;
+int proper_seed_order = 0;
+int inplace_partial = 0;
+int do_negotiated_strings = 0;
 
 /* These index values are for the file-list's extra-attribute array. */
-int uid_ndx, gid_ndx, acls_ndx, xattrs_ndx;
-#ifdef ICONV_OPTION
-int ic_ndx;
+int pathname_ndx, depth_ndx, atimes_ndx, uid_ndx, gid_ndx, acls_ndx, xattrs_ndx, unsort_ndx;
 
+int receiver_symlink_times = 0; /* receiver can set the time on a symlink */
+int sender_symlink_iconv = 0;	/* sender should convert symlink content */
+
+#ifdef ICONV_OPTION
 int filesfrom_convert = 0;
 #endif
 
+#define MAX_NSTR_STRLEN 256
+
+struct name_num_obj valid_compressions = {
+	"compress", NULL, NULL, 0, 0, {
+#ifdef SUPPORT_ZSTD
+		{ CPRES_ZSTD, "zstd", NULL },
+#endif
+#ifdef SUPPORT_LZ4
+		{ CPRES_LZ4, "lz4", NULL },
+#endif
+		{ CPRES_ZLIBX, "zlibx", NULL },
+		{ CPRES_ZLIB, "zlib", NULL },
+		{ CPRES_NONE, "none", NULL },
+		{ 0, NULL, NULL }
+	}
+};
+
+#define CF_INC_RECURSE	 (1<<0)
+#define CF_SYMLINK_TIMES (1<<1)
+#define CF_SYMLINK_ICONV (1<<2)
+#define CF_SAFE_FLIST	 (1<<3)
+#define CF_AVOID_XATTR_OPTIM (1<<4)
+#define CF_CHKSUM_SEED_FIX (1<<5)
+#define CF_INPLACE_PARTIAL_DIR (1<<6)
+#define CF_VARINT_FLIST_FLAGS (1<<7)
+
+static const char *client_info;
+
 /* The server makes sure that if either side only supports a pre-release
  * version of a protocol, that both sides must speak a compatible version
  * of that protocol for it to be advertised as available. */
@@ -80,8 +124,9 @@ static void check_sub_protocol(void)
 	int our_sub = 0;
 #endif
 
-	if (!shell_cmd || !(dot = strchr(shell_cmd, '.'))
-	 || !(their_protocol = atoi(shell_cmd))
+	/* client_info starts with VER.SUB string if client is a pre-release. */
+	if (!(their_protocol = atoi(client_info))
+	 || !(dot = strchr(client_info, '.'))
 	 || !(their_sub = atoi(dot+1))) {
 #if SUBPROTOCOL_VERSION != 0
 		if (our_sub)
@@ -102,12 +147,323 @@ static void check_sub_protocol(void)
 		protocol_version--;
 }
 
+void set_allow_inc_recurse(void)
+{
+	client_info = shell_cmd ? shell_cmd : "";
+
+	if (!recurse || use_qsort)
+		allow_inc_recurse = 0;
+	else if (!am_sender
+	 && (delete_before || delete_after
+	  || delay_updates || prune_empty_dirs))
+		allow_inc_recurse = 0;
+	else if (am_server && !local_server
+	 && (strchr(client_info, 'i') == NULL))
+		allow_inc_recurse = 0;
+}
+
+void parse_compress_choice(int final_call)
+{
+	if (valid_compressions.negotiated_name)
+		do_compression = valid_compressions.negotiated_num;
+	else if (compress_choice) {
+		struct name_num_item *nni = get_nni_by_name(&valid_compressions, compress_choice, -1);
+		if (!nni) {
+			rprintf(FERROR, "unknown compress name: %s\n", compress_choice);
+			exit_cleanup(RERR_UNSUPPORTED);
+		}
+		do_compression = nni->num;
+	} else if (do_compression)
+		do_compression = CPRES_ZLIB;
+	else
+		do_compression = CPRES_NONE;
+
+	if (do_compression != CPRES_NONE && final_call)
+		init_compression_level(); /* There's a chance this might turn compression off! */
+
+	if (do_compression == CPRES_NONE)
+		compress_choice = NULL;
+
+	/* Snag the compression name for both write_batch's option output & the following debug output. */
+	if (valid_compressions.negotiated_name)
+		compress_choice = valid_compressions.negotiated_name;
+	else if (compress_choice == NULL) {
+		struct name_num_item *nni = get_nni_by_num(&valid_compressions, do_compression);
+		compress_choice = nni ? nni->name : "UNKNOWN";
+	}
+
+	if (final_call && DEBUG_GTE(NSTR, am_server ? 3 : 1)
+	 && (do_compression != CPRES_NONE || do_compression_level != CLVL_NOT_SPECIFIED)) {
+		rprintf(FINFO, "%s%s compress: %s (level %d)\n",
+			am_server ? "Server" : "Client",
+			valid_compressions.negotiated_name ? " negotiated" : "",
+			compress_choice, do_compression_level);
+	}
+}
+
+struct name_num_item *get_nni_by_name(struct name_num_obj *nno, const char *name, int len)
+{
+	struct name_num_item *nni;
+
+	if (len < 0)
+		len = strlen(name);
+
+	for (nni = nno->list; nni->name; nni++) {
+		if (strncasecmp(name, nni->name, len) == 0 && nni->name[len] == '\0')
+			return nni;
+	}
+
+	return NULL;
+}
+
+struct name_num_item *get_nni_by_num(struct name_num_obj *nno, int num)
+{
+	struct name_num_item *nni;
+
+	for (nni = nno->list; nni->name; nni++) {
+		if (num == nni->num)
+			return nni;
+	}
+
+	return NULL;
+}
+
+static void init_nno_saw(struct name_num_obj *nno, int val)
+{
+	struct name_num_item *nni;
+	int cnt;
+
+	if (!nno->saw_len) {
+		for (nni = nno->list; nni->name; nni++) {
+			if (nni->num >= nno->saw_len)
+				nno->saw_len = nni->num + 1;
+		}
+	}
+
+	if (!nno->saw) {
+		if (!(nno->saw = new_array0(uchar, nno->saw_len)))
+			out_of_memory("init_nno_saw");
+
+		/* We'll take this opportunity to make sure that the main_name values are set right. */
+		for (cnt = 1, nni = nno->list; nni->name; nni++, cnt++) {
+			if (nno->saw[nni->num])
+				nni->main_name = nno->list[nno->saw[nni->num]-1].name;
+			else
+				nno->saw[nni->num] = cnt;
+		}
+	}
+
+	memset(nno->saw, val, nno->saw_len);
+}
+
+/* Simplify the user-provided string so that it contains valid names without any duplicates.
+ * It also sets the "saw" flags to a 1-relative count of which name was seen first. */
+static int parse_nni_str(struct name_num_obj *nno, const char *from, char *tobuf, int tobuf_len)
+{
+	char *to = tobuf, *tok = NULL;
+	int cnt = 0;
+
+	while (1) {
+		if (*from == ' ' || !*from) {
+			if (tok) {
+				struct name_num_item *nni = get_nni_by_name(nno, tok, to - tok);
+				if (nni && !nno->saw[nni->num]) {
+					nno->saw[nni->num] = ++cnt;
+					if (nni->main_name) {
+						to = tok + strlcpy(tok, nni->main_name, tobuf_len - (tok - tobuf));
+						if (to - tobuf >= tobuf_len) {
+							to = tok - 1;
+							break;
+						}
+					}
+				} else
+					to = tok - (tok != tobuf);
+				tok = NULL;
+			}
+			if (!*from++)
+				break;
+			continue;
+		}
+		if (!tok) {
+			if (to != tobuf)
+				*to++ = ' ';
+			tok = to;
+		}
+		if (to - tobuf >= tobuf_len - 1) {
+			to = tok - (tok != tobuf);
+			break;
+		}
+		*to++ = *from++;
+	}
+	*to = '\0';
+
+	return to - tobuf;
+}
+
+static void recv_negotiate_str(int f_in, struct name_num_obj *nno, char *tmpbuf, int len)
+{
+	struct name_num_item *ret = NULL;
+
+	if (len < 0)
+		len = read_vstring(f_in, tmpbuf, MAX_NSTR_STRLEN);
+
+	if (DEBUG_GTE(NSTR, am_server ? 3 : 2)) {
+		if (am_server)
+			rprintf(FINFO, "Client %s list (on server): %s\n", nno->type, tmpbuf);
+		else
+			rprintf(FINFO, "Server %s list (on client): %s\n", nno->type, tmpbuf);
+	}
+
+	if (len > 0) {
+		int best = nno->saw_len; /* We want best == 1 from the client list, so start with a big number. */
+		char *tok;
+		if (am_server)
+			init_nno_saw(nno, 1); /* Since we're parsing client names, anything we parse first is #1. */
+		for (tok = strtok(tmpbuf, " \t"); tok; tok = strtok(NULL, " \t")) {
+			struct name_num_item *nni = get_nni_by_name(nno, tok, -1);
+			if (!nni || !nno->saw[nni->num] || best <= nno->saw[nni->num])
+				continue;
+			ret = nni;
+			best = nno->saw[nni->num];
+			if (best == 1)
+				break;
+		}
+		if (ret) {
+			free(nno->saw);
+			nno->saw = NULL;
+			nno->negotiated_name = ret->main_name ? ret->main_name : ret->name;
+			nno->negotiated_num = ret->num;
+			return;
+		}
+	}
+
+	if (!am_server)
+		rprintf(FERROR, "Failed to negotiate a common %s\n", nno->type);
+	exit_cleanup(RERR_UNSUPPORTED);
+}
+
+/* The saw buffer is initialized and used to store ordinal values from 1 to N
+ * for the order of the args in the array.  If dup_markup == '\0', duplicates
+ * are removed otherwise the char is prefixed to the duplicate term and, if it
+ * is an opening paren/bracket/brace, the matching closing char is suffixed. */
+int get_default_nno_list(struct name_num_obj *nno, char *to_buf, int to_buf_len, char dup_markup)
+{
+	struct name_num_item *nni;
+	int len = 0, cnt = 0;
+	char delim = '\0', post_delim;
+
+	switch (dup_markup) {
+	case '(': post_delim = ')'; break;
+	case '[': post_delim = ']'; break;
+	case '{': post_delim = '}'; break;
+	default: post_delim = '\0'; break;
+	}
+
+	init_nno_saw(nno, 0);
+
+	for (nni = nno->list, len = 0; nni->name; nni++) {
+		if (nni->main_name) {
+			if (!dup_markup)
+				continue;
+			delim = dup_markup;
+		}
+		if (len)
+			to_buf[len++]= ' ';
+		if (delim) {
+			to_buf[len++]= delim;
+			delim = post_delim;
+		}
+		len += strlcpy(to_buf+len, nni->name, to_buf_len - len);
+		if (len >= to_buf_len - 3)
+			exit_cleanup(RERR_UNSUPPORTED); /* IMPOSSIBLE... */
+		if (delim) {
+			to_buf[len++]= delim;
+			delim = '\0';
+		}
+		nno->saw[nni->num] = ++cnt;
+	}
+
+	return len;
+}
+
+static void send_negotiate_str(int f_out, struct name_num_obj *nno, const char *env_name)
+{
+	char tmpbuf[MAX_NSTR_STRLEN];
+	const char *list_str = getenv(env_name);
+	int len, fail_if_empty = list_str && strstr(list_str, "FAIL");
+
+	if (!do_negotiated_strings) {
+		if (!am_server && fail_if_empty) {
+			rprintf(FERROR, "Remote rsync is too old for %s negotiation\n", nno->type);
+			exit_cleanup(RERR_UNSUPPORTED);
+		}
+		return;
+	}
+
+	if (list_str && *list_str && (!am_server || local_server)) {
+		init_nno_saw(nno, 0);
+		len = parse_nni_str(nno, list_str, tmpbuf, MAX_NSTR_STRLEN);
+		if (fail_if_empty && !len)
+			len = strlcpy(tmpbuf, "FAIL", MAX_NSTR_STRLEN);
+		list_str = tmpbuf;
+	} else
+		list_str = NULL;
+
+	if (!list_str || !*list_str)
+		len = get_default_nno_list(nno, tmpbuf, MAX_NSTR_STRLEN, '\0');
+
+	if (DEBUG_GTE(NSTR, am_server ? 3 : 2)) {
+		if (am_server)
+			rprintf(FINFO, "Server %s list (on server): %s\n", nno->type, tmpbuf);
+		else
+			rprintf(FINFO, "Client %s list (on client): %s\n", nno->type, tmpbuf);
+	}
+
+	if (local_server) {
+		/* A local server doesn't bother to send/recv the strings, it just constructs
+		 * and parses the same string on both sides. */
+		recv_negotiate_str(-1, nno, tmpbuf, len);
+	} else {
+		/* Each side sends their list of valid names to the other side and then both sides
+		 * pick the first name in the client's list that is also in the server's list. */
+		write_vstring(f_out, tmpbuf, len);
+	}
+}
+
+static void negotiate_the_strings(int f_in, int f_out)
+{
+	/* We send all the negotiation strings before we start to read them to help avoid a slow startup. */
+
+	if (!checksum_choice)
+		send_negotiate_str(f_out, &valid_checksums, "RSYNC_CHECKSUM_LIST");
+
+	if (do_compression && !compress_choice)
+		send_negotiate_str(f_out, &valid_compressions, "RSYNC_COMPRESS_LIST");
+
+	if (valid_checksums.saw) {
+		char tmpbuf[MAX_NSTR_STRLEN];
+		recv_negotiate_str(f_in, &valid_checksums, tmpbuf, -1);
+	}
+
+	if (valid_compressions.saw) {
+		char tmpbuf[MAX_NSTR_STRLEN];
+		recv_negotiate_str(f_in, &valid_compressions, tmpbuf, -1);
+	}
+}
+
 void setup_protocol(int f_out,int f_in)
 {
-	if (am_sender)
-		file_extra_cnt += PTR_EXTRA_CNT;
+	assert(file_extra_cnt == 0);
+	assert(EXTRA64_CNT == 2 || EXTRA64_CNT == 1);
+
+	/* All int64 values must be set first so that they are guaranteed to be
+	 * aligned for direct int64-pointer memory access. */
+	if (preserve_atimes)
+		atimes_ndx = (file_extra_cnt += EXTRA64_CNT);
+	if (am_sender) /* This is most likely in the in64 union as well. */
+		pathname_ndx = (file_extra_cnt += PTR_EXTRA_CNT);
 	else
-		file_extra_cnt++;
+		depth_ndx = ++file_extra_cnt;
 	if (preserve_uid)
 		uid_ndx = ++file_extra_cnt;
 	if (preserve_gid)
@@ -117,6 +473,9 @@ void setup_protocol(int f_out,int f_in)
 	if (preserve_xattrs)
 		xattrs_ndx = ++file_extra_cnt;
 
+	if (am_server)
+		set_allow_inc_recurse();
+
 	if (remote_protocol == 0) {
 		if (am_server && !local_server)
 			check_sub_protocol();
@@ -132,7 +491,7 @@ void setup_protocol(int f_out,int f_in)
 		exit_cleanup(RERR_PROTOCOL);
 	}
 
-	if (verbose > 3) {
+	if (DEBUG_GTE(PROTO, 1)) {
 		rprintf(FINFO, "(%s) Protocol versions: remote=%d, negotiated=%d\n",
 			am_server? "Server" : "Client", remote_protocol, protocol_version);
 	}
@@ -156,22 +515,32 @@ void setup_protocol(int f_out,int f_in)
 			PROTOCOL_VERSION, am_server? "Server" : "Client");
 		exit_cleanup(RERR_PROTOCOL);
 	}
+	if (read_batch)
+		check_batch_flags();
+
+#ifndef SUPPORT_PREALLOCATION
+	if (preallocate_files && !am_sender) {
+		rprintf(FERROR, "preallocation is not supported on this %s\n",
+			am_server ? "Server" : "Client");
+		exit_cleanup(RERR_SYNTAX);
+	}
+#endif
 
 	if (protocol_version < 30) {
 		if (append_mode == 1)
 			append_mode = 2;
 		if (preserve_acls && !local_server) {
 			rprintf(FERROR,
-			    "--acls requires protocol 30 or higher"
-			    " (negotiated %d).\n",
-			    protocol_version);
+				"--acls requires protocol 30 or higher"
+				" (negotiated %d).\n",
+				protocol_version);
 			exit_cleanup(RERR_PROTOCOL);
 		}
 		if (preserve_xattrs && !local_server) {
 			rprintf(FERROR,
-			    "--xattrs requires protocol 30 or higher"
-			    " (negotiated %d).\n",
-			    protocol_version);
+				"--xattrs requires protocol 30 or higher"
+				" (negotiated %d).\n",
+				protocol_version);
 			exit_cleanup(RERR_PROTOCOL);
 		}
 	}
@@ -186,53 +555,113 @@ void setup_protocol(int f_out,int f_in)
 	if (protocol_version < 29) {
 		if (fuzzy_basis) {
 			rprintf(FERROR,
-			    "--fuzzy requires protocol 29 or higher"
-			    " (negotiated %d).\n",
-			    protocol_version);
+				"--fuzzy requires protocol 29 or higher"
+				" (negotiated %d).\n",
+				protocol_version);
 			exit_cleanup(RERR_PROTOCOL);
 		}
 
 		if (basis_dir_cnt && inplace) {
 			rprintf(FERROR,
-			    "%s with --inplace requires protocol 29 or higher"
-			    " (negotiated %d).\n",
-			    dest_option, protocol_version);
+				"%s with --inplace requires protocol 29 or higher"
+				" (negotiated %d).\n",
+				alt_dest_opt(0), protocol_version);
 			exit_cleanup(RERR_PROTOCOL);
 		}
 
 		if (basis_dir_cnt > 1) {
 			rprintf(FERROR,
-			    "Using more than one %s option requires protocol"
-			    " 29 or higher (negotiated %d).\n",
-			    dest_option, protocol_version);
+				"Using more than one %s option requires protocol"
+				" 29 or higher (negotiated %d).\n",
+				alt_dest_opt(0), protocol_version);
 			exit_cleanup(RERR_PROTOCOL);
 		}
 
 		if (prune_empty_dirs) {
 			rprintf(FERROR,
-			    "--prune-empty-dirs requires protocol 29 or higher"
-			    " (negotiated %d).\n",
-			    protocol_version);
+				"--prune-empty-dirs requires protocol 29 or higher"
+				" (negotiated %d).\n",
+				protocol_version);
 			exit_cleanup(RERR_PROTOCOL);
 		}
 	} else if (protocol_version >= 30) {
-		if (recurse && allow_inc_recurse
-		 && !delete_before && !delete_after && !delay_updates
-		 && !use_qsort && !prune_empty_dirs)
-			inc_recurse = 1;
+		if (am_server) {
+			compat_flags = allow_inc_recurse ? CF_INC_RECURSE : 0;
+#ifdef CAN_SET_SYMLINK_TIMES
+			compat_flags |= CF_SYMLINK_TIMES;
+#endif
+#ifdef ICONV_OPTION
+			compat_flags |= CF_SYMLINK_ICONV;
+#endif
+			if (local_server || strchr(client_info, 'f') != NULL)
+				compat_flags |= CF_SAFE_FLIST;
+			if (local_server || strchr(client_info, 'x') != NULL)
+				compat_flags |= CF_AVOID_XATTR_OPTIM;
+			if (local_server || strchr(client_info, 'C') != NULL)
+				compat_flags |= CF_CHKSUM_SEED_FIX;
+			if (local_server || strchr(client_info, 'I') != NULL)
+				compat_flags |= CF_INPLACE_PARTIAL_DIR;
+			if (local_server || strchr(client_info, 'v') != NULL) {
+				if (!write_batch || protocol_version >= 30) {
+					do_negotiated_strings = 1;
+					compat_flags |= CF_VARINT_FLIST_FLAGS;
+				}
+			}
+			if (strchr(client_info, 'V') != NULL) { /* Support a pre-release 'V' that got superseded */
+				if (!write_batch)
+					compat_flags |= CF_VARINT_FLIST_FLAGS;
+				write_byte(f_out, compat_flags);
+			} else
+				write_varint(f_out, compat_flags);
+		} else { /* read_varint() is compatible with the older write_byte() when the 0x80 bit isn't on. */
+			compat_flags = read_varint(f_in);
+			if  (compat_flags & CF_VARINT_FLIST_FLAGS)
+				do_negotiated_strings = 1;
+		}
+		/* The inc_recurse var MUST be set to 0 or 1. */
+		inc_recurse = compat_flags & CF_INC_RECURSE ? 1 : 0;
+		want_xattr_optim = protocol_version >= 31 && !(compat_flags & CF_AVOID_XATTR_OPTIM);
+		proper_seed_order = compat_flags & CF_CHKSUM_SEED_FIX ? 1 : 0;
+		xfer_flags_as_varint = compat_flags & CF_VARINT_FLIST_FLAGS ? 1 : 0;
+		if (am_sender) {
+			receiver_symlink_times = am_server
+			    ? strchr(client_info, 'L') != NULL
+			    : !!(compat_flags & CF_SYMLINK_TIMES);
+		}
+#ifdef CAN_SET_SYMLINK_TIMES
+		else
+			receiver_symlink_times = 1;
+#endif
+#ifdef ICONV_OPTION
+		sender_symlink_iconv = iconv_opt && (am_server
+		    ? local_server || strchr(client_info, 's') != NULL
+		    : !!(compat_flags & CF_SYMLINK_ICONV));
+#endif
+		if (inc_recurse && !allow_inc_recurse) {
+			/* This should only be able to happen in a batch. */
+			fprintf(stderr,
+				"Incompatible options specified for inc-recursive %s.\n",
+				read_batch ? "batch file" : "connection");
+			exit_cleanup(RERR_SYNTAX);
+		}
+		use_safe_inc_flist = (compat_flags & CF_SAFE_FLIST) || protocol_version >= 31;
 		need_messages_from_generator = 1;
+		if (compat_flags & CF_INPLACE_PARTIAL_DIR)
+			inplace_partial = 1;
+#ifdef CAN_SET_SYMLINK_TIMES
+	} else if (!am_sender) {
+		receiver_symlink_times = 1;
+#endif
 	}
 
-#ifdef ICONV_OPTION
-	if (iconv_opt && (!am_sender || inc_recurse))
-		ic_ndx = ++file_extra_cnt;
-#endif
+	if (need_unsorted_flist && (!am_sender || inc_recurse))
+		unsort_ndx = ++file_extra_cnt;
 
 	if (partial_dir && *partial_dir != '/' && (!am_server || local_server)) {
-		int flags = MATCHFLG_NO_PREFIXES | MATCHFLG_DIRECTORY;
+		int rflags = FILTRULE_NO_PREFIXES | FILTRULE_DIRECTORY;
 		if (!am_sender || protocol_version >= 30)
-			flags |= MATCHFLG_PERISHABLE;
-		parse_rule(&filter_list, partial_dir, flags, 0);
+			rflags |= FILTRULE_PERISHABLE;
+		parse_filter_str(&filter_list, partial_dir, rule_template(rflags), 0);
 	}
 
 
@@ -245,11 +674,21 @@ void setup_protocol(int f_out,int f_in)
 	}
 #endif
 
+	negotiate_the_strings(f_in, f_out);
+
 	if (am_server) {
 		if (!checksum_seed)
-			checksum_seed = time(NULL);
+			checksum_seed = time(NULL) ^ (getpid() << 6);
 		write_int(f_out, checksum_seed);
 	} else {
 		checksum_seed = read_int(f_in);
 	}
+
+	parse_checksum_choice(1); /* Sets checksum_type & xfersum_type */
+	parse_compress_choice(1); /* Sets do_compression */
+
+	if (write_batch && !am_server)
+		write_batch_shell_file();
+
+	init_flist();
 }

configure

@@ -0,0 +1,27 @@
+#!/bin/sh -e
+# This configure script ensures that the configure.sh script exists, and
+# if not, it tries to fetch rsync's generated files or build them.  We
+# then transfer control to the configure.sh script to do the real work.
+
+dir=`dirname $0`
+realconfigure="$dir/configure.sh"
+
+if test ! -f "$realconfigure"; then
+    if test -f "$HOME/build_farm/build_test.fns"; then
+	# Test the included popt
+	set -- --with-included-popt "${@}"
+	# Allow the build farm to grab latest files via rsync.
+	actions='build fetch'
+    else
+	actions='build'
+    fi
+    if "$dir/prepare-source" $actions; then
+	:
+    else
+	echo 'Failed to build configure.sh and/or config.h.in -- giving up.' >&2
+	rm -f "$realconfigure"
+	exit 1
+    fi
+fi
+
+exec "$realconfigure" "${@}"

connection.c

@@ -2,6 +2,7 @@
  * Support the max connections option.
  *
  * Copyright (C) 1998 Andrew Tridgell
+ * Copyright (C) 2006-2020 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

csprotocol.txt

@@ -3,21 +3,22 @@ basically a summary of clientserver.c and authenticate.c.
 
  -- Martin Pool <mbp@samba.org>
 
-$Id$
-
-
-
 
 This is the protocol used for rsync --daemon; i.e. connections to port
 873 rather than invocations over a remote shell.
 
 When the server accepts a connection, it prints a greeting
 
-  @RSYNCD: <version>
+  @RSYNCD: <version>.<subprotocol>
 
-where <version> is the numeric version; currently 24.  It follows this
-with a free text message-of-the-day.  It expects to see a similar
-greeting back from the client.
+where <version> is the numeric version (see PROTOCOL_VERSION in rsync.h)
+'.' is a literal period, and <subprotocol> is the numeric subprotocol
+version (see SUBPROTOCOL_VERSION -- it will be 0 for final releases).
+Protocols prior to 30 only output <version> alone.  The daemon expects
+to see a similar greeting back from the client.  For protocols prior to
+30, an absent ".<subprotocol>" value is assumed to be 0.  For protocol
+30, an absent value is a fatal error.  The daemon then follows this line
+with a free-format text message-of-the-day (if any is defined).
 
 The server is now in the connected state.  The client can either send
 the command
@@ -75,8 +76,13 @@ stay tuned (or write it yourself!).
 ------------
 Protocol version changes
 
-25	 (2001-08-20, 2.4.7pre2) 
+30	(2007-10-04, 3.0.0pre1)
 
-	 Send an explicit "@RSYNC EXIT" command at the end of the
-	 module listing.  We never intentionally end the transmission
-	 by just closing the socket anymore.
+	The use of a ".<subprotocol>" number was added to
+	@RSYNCD: <version>.<subprotocol>
+
+25	(2001-08-20, 2.4.7pre2) 
+
+	Send an explicit "@RSYNC EXIT" command at the end of the
+	module listing.  We never intentionally end the transmission
+	by just closing the socket anymore.

define-from-md.awk

@@ -0,0 +1,41 @@
+#!/usr/bin/awk -f
+
+# The caller must pass args: -v hfile=NAME rsync.1.md
+
+BEGIN {
+    heading = "/* DO NOT EDIT THIS FILE!  It is auto-generated from a list of values in " ARGV[1] "! */"
+    if (hfile ~ /compress/) {
+	define = "#define DEFAULT_DONT_COMPRESS"
+	prefix = "*."
+    } else {
+	define = "#define DEFAULT_CVSIGNORE"
+	prefix = ""
+    }
+    value_list = ""
+}
+
+/^    > [^ ]+$/ {
+    gsub(/`/, "")
+    if (value_list != "") value_list = value_list " "
+    value_list = value_list prefix $2
+    next
+}
+
+value_list ~ /\.gz / && hfile ~ /compress/ {
+    exit
+}
+
+value_list ~ /SCCS / && hfile ~ /cvsignore/ {
+    exit
+}
+
+value_list = ""
+
+END {
+    if (value_list != "")
+	print heading "\n\n" define " \"" value_list "\"" > hfile
+    else {
+	print "Failed to find a value list in " ARGV[1] " for " hfile
+	exit 1
+    }
+}

delete.c

@@ -0,0 +1,240 @@
+/*
+ * Deletion routines used in rsync.
+ *
+ * Copyright (C) 1996-2000 Andrew Tridgell
+ * Copyright (C) 1996 Paul Mackerras
+ * Copyright (C) 2002 Martin Pool <mbp@samba.org>
+ * Copyright (C) 2003-2020 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
+ * the Free Software Foundation; either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, visit the http://fsf.org website.
+ */
+
+#include "rsync.h"
+
+extern int am_root;
+extern int make_backups;
+extern int max_delete;
+extern char *backup_dir;
+extern char *backup_suffix;
+extern int backup_suffix_len;
+extern struct stats stats;
+
+int ignore_perishable = 0;
+int non_perishable_cnt = 0;
+int skipped_deletes = 0;
+
+static inline int is_backup_file(char *fn)
+{
+	int k = strlen(fn) - backup_suffix_len;
+	return k > 0 && strcmp(fn+k, backup_suffix) == 0;
+}
+
+/* The directory is about to be deleted: if DEL_RECURSE is given, delete all
+ * its contents, otherwise just checks for content.  Returns DR_SUCCESS or
+ * DR_NOT_EMPTY.  Note that fname must point to a MAXPATHLEN buffer!  (The
+ * buffer is used for recursion, but returned unchanged.)
+ */
+static enum delret delete_dir_contents(char *fname, uint16 flags)
+{
+	struct file_list *dirlist;
+	enum delret ret;
+	unsigned remainder;
+	void *save_filters;
+	int j, dlen;
+	char *p;
+
+	if (DEBUG_GTE(DEL, 3)) {
+		rprintf(FINFO, "delete_dir_contents(%s) flags=%d\n",
+			fname, flags);
+	}
+
+	dlen = strlen(fname);
+	save_filters = push_local_filters(fname, dlen);
+
+	non_perishable_cnt = 0;
+	dirlist = get_dirlist(fname, dlen, 0);
+	ret = non_perishable_cnt ? DR_NOT_EMPTY : DR_SUCCESS;
+
+	if (!dirlist->used)
+		goto done;
+
+	if (!(flags & DEL_RECURSE)) {
+		ret = DR_NOT_EMPTY;
+		goto done;
+	}
+
+	p = fname + dlen;
+	if (dlen != 1 || *fname != '/')
+		*p++ = '/';
+	remainder = MAXPATHLEN - (p - fname);
+
+	/* We do our own recursion, so make delete_item() non-recursive. */
+	flags = (flags & ~(DEL_RECURSE|DEL_MAKE_ROOM|DEL_NO_UID_WRITE))
+	      | DEL_DIR_IS_EMPTY;
+
+	for (j = dirlist->used; j--; ) {
+		struct file_struct *fp = dirlist->files[j];
+
+		if (fp->flags & FLAG_MOUNT_DIR && S_ISDIR(fp->mode)) {
+			if (DEBUG_GTE(DEL, 1)) {
+				rprintf(FINFO,
+					"mount point, %s, pins parent directory\n",
+					f_name(fp, NULL));
+			}
+			ret = DR_NOT_EMPTY;
+			continue;
+		}
+
+		strlcpy(p, fp->basename, remainder);
+		if (!(fp->mode & S_IWUSR) && !am_root && fp->flags & FLAG_OWNED_BY_US)
+			do_chmod(fname, fp->mode | S_IWUSR);
+		/* Save stack by recursing to ourself directly. */
+		if (S_ISDIR(fp->mode)) {
+			if (delete_dir_contents(fname, flags | DEL_RECURSE) != DR_SUCCESS)
+				ret = DR_NOT_EMPTY;
+		}
+		if (delete_item(fname, fp->mode, flags) != DR_SUCCESS)
+			ret = DR_NOT_EMPTY;
+	}
+
+	fname[dlen] = '\0';
+
+  done:
+	flist_free(dirlist);
+	pop_local_filters(save_filters);
+
+	if (ret == DR_NOT_EMPTY) {
+		rprintf(FINFO, "cannot delete non-empty directory: %s\n",
+			fname);
+	}
+	return ret;
+}
+
+/* Delete a file or directory.  If DEL_RECURSE is set in the flags, this will
+ * delete recursively.
+ *
+ * Note that fbuf must point to a MAXPATHLEN buffer if the mode indicates it's
+ * a directory! (The buffer is used for recursion, but returned unchanged.)
+ */
+enum delret delete_item(char *fbuf, uint16 mode, uint16 flags)
+{
+	enum delret ret;
+	char *what;
+	int ok;
+
+	if (DEBUG_GTE(DEL, 2)) {
+		rprintf(FINFO, "delete_item(%s) mode=%o flags=%d\n",
+			fbuf, (int)mode, (int)flags);
+	}
+
+	if (flags & DEL_NO_UID_WRITE)
+		do_chmod(fbuf, mode | S_IWUSR);
+
+	if (S_ISDIR(mode) && !(flags & DEL_DIR_IS_EMPTY)) {
+		/* This only happens on the first call to delete_item() since
+		 * delete_dir_contents() always calls us w/DEL_DIR_IS_EMPTY. */
+		ignore_perishable = 1;
+		/* If DEL_RECURSE is not set, this just reports emptiness. */
+		ret = delete_dir_contents(fbuf, flags);
+		ignore_perishable = 0;
+		if (ret == DR_NOT_EMPTY || ret == DR_AT_LIMIT)
+			goto check_ret;
+		/* OK: try to delete the directory. */
+	}
+
+	if (!(flags & DEL_MAKE_ROOM) && max_delete >= 0 && stats.deleted_files >= max_delete) {
+		skipped_deletes++;
+		return DR_AT_LIMIT;
+	}
+
+	if (S_ISDIR(mode)) {
+		what = "rmdir";
+		ok = do_rmdir(fbuf) == 0;
+	} else {
+		if (make_backups > 0 && !(flags & DEL_FOR_BACKUP) && (backup_dir || !is_backup_file(fbuf))) {
+			what = "make_backup";
+			ok = make_backup(fbuf, True);
+			if (ok == 2) {
+				what = "unlink";
+				ok = robust_unlink(fbuf) == 0;
+			}
+		} else {
+			what = "unlink";
+			ok = robust_unlink(fbuf) == 0;
+		}
+	}
+
+	if (ok) {
+		if (!(flags & DEL_MAKE_ROOM)) {
+			log_delete(fbuf, mode);
+			stats.deleted_files++;
+			if (S_ISREG(mode)) {
+				/* Nothing more to count */
+			} else if (S_ISDIR(mode))
+				stats.deleted_dirs++;
+#ifdef SUPPORT_LINKS
+			else if (S_ISLNK(mode))
+				stats.deleted_symlinks++;
+#endif
+			else if (IS_DEVICE(mode))
+				stats.deleted_symlinks++;
+			else
+				stats.deleted_specials++;
+		}
+		ret = DR_SUCCESS;
+	} else {
+		if (S_ISDIR(mode) && errno == ENOTEMPTY) {
+			rprintf(FINFO, "cannot delete non-empty directory: %s\n",
+				fbuf);
+			ret = DR_NOT_EMPTY;
+		} else if (errno != ENOENT) {
+			rsyserr(FERROR_XFER, errno, "delete_file: %s(%s) failed",
+				what, fbuf);
+			ret = DR_FAILURE;
+		} else
+			ret = DR_SUCCESS;
+	}
+
+  check_ret:
+	if (ret != DR_SUCCESS && flags & DEL_MAKE_ROOM) {
+		const char *desc;
+		switch (flags & DEL_MAKE_ROOM) {
+		case DEL_FOR_FILE: desc = "regular file"; break;
+		case DEL_FOR_DIR: desc = "directory"; break;
+		case DEL_FOR_SYMLINK: desc = "symlink"; break;
+		case DEL_FOR_DEVICE: desc = "device file"; break;
+		case DEL_FOR_SPECIAL: desc = "special file"; break;
+		default: exit_cleanup(RERR_UNSUPPORTED); /* IMPOSSIBLE */
+		}
+		rprintf(FERROR_XFER, "could not make way for %s %s: %s\n",
+			flags & DEL_FOR_BACKUP ? "backup" : "new",
+			desc, fbuf);
+	}
+	return ret;
+}
+
+uint16 get_del_for_flag(uint16 mode)
+{
+	if (S_ISREG(mode))
+		return DEL_FOR_FILE;
+	if (S_ISDIR(mode))
+		return DEL_FOR_DIR;
+	if (S_ISLNK(mode))
+		return DEL_FOR_SYMLINK;
+	if (IS_DEVICE(mode))
+		return DEL_FOR_DEVICE;
+	if (IS_SPECIAL(mode))
+		return DEL_FOR_SPECIAL;
+	exit_cleanup(RERR_UNSUPPORTED); /* IMPOSSIBLE */
+}

doc/.cvsignore

@@ -1,2 +0,0 @@
-rsync.pdf
-rsync.ps

doc/rsync.sgml

@@ -126,7 +126,7 @@
       
       <para><option>-P</option>
 	Display a progress indicator while files are transferred.  This should
-	normally be ommitted if rsync is not run on a terminal.
+	normally be omitted if rsync is not run on a terminal.
       </para>
     </section>
 
@@ -348,4 +348,4 @@ running rsync giving the network directory.
     
     <para><ulink url="http://www.ccp14.ac.uk/ccp14admin/rsync/"></ulink></para>
   </appendix>
-</book>
+</book>

errcode.h

@@ -2,7 +2,7 @@
  * Error codes returned by rsync.
  *
  * Copyright (C) 1998-2000 Andrew Tridgell
- * Copyright (C) 2003-2007 Wayne Davison
+ * Copyright (C) 2003-2019 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
@@ -45,6 +45,7 @@
 #define RERR_DEL_LIMIT  25      /* skipped some deletes due to --max-delete */
 
 #define RERR_TIMEOUT    30      /* timeout in data send/receive */
+#define RERR_CONTIMEOUT 35      /* timeout waiting for daemon connection */
 
 /* Although it doesn't seem to be specified anywhere,
  * ssh and the shell seem to return these values:

fileio.c

@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1998 Andrew Tridgell
  * Copyright (C) 2002 Martin Pool
- * Copyright (C) 2004-2007 Wayne Davison
+ * Copyright (C) 2004-2020 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
@@ -20,68 +20,104 @@
  */
 
 #include "rsync.h"
+#include "inums.h"
 
 #ifndef ENODATA
 #define ENODATA EAGAIN
 #endif
 
+/* We want all reads to be aligned on 1K boundaries. */
+#define ALIGN_BOUNDARY 1024
+/* How far past the boundary is an offset? */
+#define ALIGNED_OVERSHOOT(oft) ((oft) & (ALIGN_BOUNDARY-1))
+/* Round up a length to the next boundary */
+#define ALIGNED_LENGTH(len) ((((len) - 1) | (ALIGN_BOUNDARY-1)) + 1)
+
 extern int sparse_files;
 
-static char last_byte;
-static size_t sparse_seek = 0;
+OFF_T preallocated_len = 0;
 
-int sparse_end(int f)
+static OFF_T sparse_seek = 0;
+static OFF_T sparse_past_write = 0;
+
+int sparse_end(int f, OFF_T size)
 {
 	int ret;
 
+	sparse_past_write = 0;
+
 	if (!sparse_seek)
 		return 0;
 
-	do_lseek(f, sparse_seek-1, SEEK_CUR);
+#ifdef HAVE_FTRUNCATE
+	ret = do_ftruncate(f, size);
+#else
+	if (do_lseek(f, sparse_seek-1, SEEK_CUR) != size-1)
+		ret = -1;
+	else {
+		do {
+			ret = write(f, "", 1);
+		} while (ret < 0 && errno == EINTR);
+
+		ret = ret <= 0 ? -1 : 0;
+	}
+#endif
+
 	sparse_seek = 0;
 
-	do {
-		ret = write(f, "", 1);
-	} while (ret < 0 && errno == EINTR);
-
-	return ret <= 0 ? -1 : 0;
+	return ret;
 }
 
-
-static int write_sparse(int f, char *buf, size_t len)
+/* Note that the offset is just the caller letting us know where
+ * the current file position is in the file. The use_seek arg tells
+ * us that we should seek over matching data instead of writing it. */
+static int write_sparse(int f, int use_seek, OFF_T offset, const char *buf, int len)
 {
-	size_t l1 = 0, l2 = 0;
+	int l1 = 0, l2 = 0;
 	int ret;
 
 	for (l1 = 0; l1 < len && buf[l1] == 0; l1++) {}
 	for (l2 = 0; l2 < len-l1 && buf[len-(l2+1)] == 0; l2++) {}
 
-	/* XXX Riddle me this: why does this function SLOW DOWN when I
-	 * remove the following (unneeded) line?? Core Duo weirdness? */
-	last_byte = buf[len-1];
-
 	sparse_seek += l1;
 
 	if (l1 == len)
 		return len;
 
-	if (sparse_seek)
-		do_lseek(f, sparse_seek, SEEK_CUR);
+	if (sparse_seek) {
+		if (sparse_past_write >= preallocated_len) {
+			if (do_lseek(f, sparse_seek, SEEK_CUR) < 0)
+				return -1;
+		} else if (do_punch_hole(f, sparse_past_write, sparse_seek) < 0) {
+			sparse_seek = 0;
+			return -1;
+		}
+	}
 	sparse_seek = l2;
+	sparse_past_write = offset + len - l2;
+
+	if (use_seek) {
+		/* The in-place data already matches. */
+		if (do_lseek(f, len - (l1+l2), SEEK_CUR) < 0)
+			return -1;
+		return len;
+	}
 
 	while ((ret = write(f, buf + l1, len - (l1+l2))) <= 0) {
 		if (ret < 0 && errno == EINTR)
 			continue;
+		sparse_seek = 0;
 		return ret;
 	}
 
-	if (ret != (int)(len - (l1+l2)))
+	if (ret != (int)(len - (l1+l2))) {
+		sparse_seek = 0;
 		return l1+ret;
+	}
 
 	return len;
 }
 
-
 static char *wf_writeBuf;
 static size_t wf_writeBufSize;
 static size_t wf_writeBufCnt;
@@ -103,12 +139,10 @@ int flush_write_file(int f)
 	return ret;
 }
 
-
-/*
- * write_file does not allow incomplete writes.  It loops internally
- * until len bytes are written or errno is set.
- */
-int write_file(int f,char *buf,size_t len)
+/* write_file does not allow incomplete writes.  It loops internally
+ * until len bytes are written or errno is set.  Note that use_seek and
+ * offset are only used in sparse processing (see write_sparse()). */
+int write_file(int f, int use_seek, OFF_T offset, const char *buf, int len)
 {
 	int ret = 0;
 
@@ -116,7 +150,8 @@ int write_file(int f,char *buf,size_t len)
 		int r1;
 		if (sparse_files > 0) {
 			int len1 = MIN(len, SPARSE_WRITE_SIZE);
-			r1 = write_sparse(f, buf, len1);
+			r1 = write_sparse(f, use_seek, offset, buf, len1);
+			offset += r1;
 		} else {
 			if (!wf_writeBuf) {
 				wf_writeBufSize = WRITE_SIZE * 8;
@@ -125,7 +160,7 @@ int write_file(int f,char *buf,size_t len)
 				if (!wf_writeBuf)
 					out_of_memory("write_file");
 			}
-			r1 = MIN(len, wf_writeBufSize - wf_writeBufCnt);
+			r1 = (int)MIN((size_t)len, wf_writeBufSize - wf_writeBufCnt);
 			if (r1) {
 				memcpy(wf_writeBuf + wf_writeBufCnt, buf, r1);
 				wf_writeBufCnt += r1;
@@ -149,13 +184,36 @@ int write_file(int f,char *buf,size_t len)
 	return ret;
 }
 
+/* An in-place update found identical data at an identical location. We either
+ * just seek past it, or (for an in-place sparse update), we give the data to
+ * the sparse processor with the use_seek flag set. */
+int skip_matched(int fd, OFF_T offset, const char *buf, int len)
+{
+	OFF_T pos;
+
+	if (sparse_files > 0) {
+		if (write_file(fd, 1, offset, buf, len) != len)
+			return -1;
+		return 0;
+	}
+
+	if (flush_write_file(fd) < 0)
+		return -1;
+
+	if ((pos = do_lseek(fd, len, SEEK_CUR)) != offset + len) {
+		rsyserr(FERROR_XFER, errno, "lseek returned %s, not %s",
+			big_num(pos), big_num(offset));
+		return -1;
+	}
+
+	return 0;
+}
 
 /* This provides functionality somewhat similar to mmap() but using read().
  * It gives sliding window access to a file.  mmap() is not used because of
  * the possibility of another program (such as a mailer) truncating the
  * file thus giving us a SIGBUS. */
-struct map_struct *map_file(int fd, OFF_T len, int32 read_size,
-			    int32 blk_size)
+struct map_struct *map_file(int fd, OFF_T len, int32 read_size, int32 blk_size)
 {
 	struct map_struct *map;
 
@@ -167,7 +225,7 @@ struct map_struct *map_file(int fd, OFF_T len, int32 read_size,
 
 	map->fd = fd;
 	map->file_size = len;
-	map->def_window_size = read_size;
+	map->def_window_size = ALIGNED_LENGTH(read_size);
 
 	return map;
 }
@@ -176,9 +234,8 @@ struct map_struct *map_file(int fd, OFF_T len, int32 read_size,
 /* slide the read window in the file */
 char *map_ptr(struct map_struct *map, OFF_T offset, int32 len)
 {
-	int32 nread;
 	OFF_T window_start, read_start;
-	int32 window_size, read_size, read_offset;
+	int32 window_size, read_size, read_offset, align_fudge;
 
 	if (len == 0)
 		return NULL;
@@ -193,12 +250,13 @@ char *map_ptr(struct map_struct *map, OFF_T offset, int32 len)
 		return map->p + (offset - map->p_offset);
 
 	/* nope, we are going to have to do a read. Work out our desired window */
-	window_start = offset;
+	align_fudge = (int32)ALIGNED_OVERSHOOT(offset);
+	window_start = offset - align_fudge;
 	window_size = map->def_window_size;
 	if (window_start + window_size > map->file_size)
 		window_size = (int32)(map->file_size - window_start);
-	if (len > window_size)
-		window_size = len;
+	if (window_size < len + align_fudge)
+		window_size = ALIGNED_LENGTH(len + align_fudge);
 
 	/* make sure we have allocated enough memory for the window */
 	if (window_size > map->p_size) {
@@ -208,11 +266,9 @@ char *map_ptr(struct map_struct *map, OFF_T offset, int32 len)
 		map->p_size = window_size;
 	}
 
-	/* Now try to avoid re-reading any bytes by reusing any bytes
-	 * from the previous buffer. */
-	if (window_start >= map->p_offset &&
-	    window_start < map->p_offset + map->p_len &&
-	    window_start + window_size >= map->p_offset + map->p_len) {
+	/* Now try to avoid re-reading any bytes by reusing any bytes from the previous buffer. */
+	if (window_start >= map->p_offset && window_start < map->p_offset + map->p_len
+	 && window_start + window_size >= map->p_offset + map->p_len) {
 		read_start = map->p_offset + map->p_len;
 		read_offset = (int32)(read_start - window_start);
 		read_size = window_size - read_offset;
@@ -232,8 +288,8 @@ char *map_ptr(struct map_struct *map, OFF_T offset, int32 len)
 	if (map->p_fd_offset != read_start) {
 		OFF_T ret = do_lseek(map->fd, read_start, SEEK_SET);
 		if (ret != read_start) {
-			rsyserr(FERROR, errno, "lseek returned %.0f, not %.0f",
-				(double)ret, (double)read_start);
+			rsyserr(FERROR, errno, "lseek returned %s, not %s",
+				big_num(ret), big_num(read_start));
 			exit_cleanup(RERR_FILEIO);
 		}
 		map->p_fd_offset = read_start;
@@ -242,7 +298,7 @@ char *map_ptr(struct map_struct *map, OFF_T offset, int32 len)
 	map->p_len = window_size;
 
 	while (read_size > 0) {
-		nread = read(map->fd, map->p + read_offset, read_size);
+		int32 nread = read(map->fd, map->p + read_offset, read_size);
 		if (nread <= 0) {
 			if (!map->status)
 				map->status = nread ? errno : ENODATA;
@@ -256,10 +312,9 @@ char *map_ptr(struct map_struct *map, OFF_T offset, int32 len)
 		read_size -= nread;
 	}
 
-	return map->p;
+	return map->p + align_fudge;
 }
 
-
 int unmap_file(struct map_struct *map)
 {
 	int	ret;
@@ -269,7 +324,9 @@ int unmap_file(struct map_struct *map)
 		map->p = NULL;
 	}
 	ret = map->status;
-	memset(map, 0, sizeof map[0]);
+#if 0 /* I don't think we really need this. */
+	force_memzero(map, sizeof map[0]);
+#endif
 	free(map);
 
 	return ret;

getfsdev.c

@@ -15,8 +15,7 @@
 			fprintf(stderr, "Unable to stat `%s'\n", *argv);
 			exit(1);
 		}
-		printf("%ld/%ld\n", (long)major(st.st_dev),
-				    (long)minor(st.st_dev));
+		printf("%ld/%ld\n", (long)major(st.st_dev), (long)minor(st.st_dev));
 	}
 
 	return 0;

getgroups.c

@@ -3,7 +3,7 @@
  * `id -G` on Linux, but it's too hard to find a portable equivalent.
  *
  * Copyright (C) 2002 Martin Pool
- * Copyright (C) 2003-2007 Wayne Davison
+ * Copyright (C) 2003-2019 Wayne Davison
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 3 as

hashtable.c

@@ -1,7 +1,7 @@
 /*
  * Routines to provide a memory-efficient hashtable.
  *
- * Copyright (C) 2007 Wayne Davison
+ * Copyright (C) 2007-2020 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
@@ -23,13 +23,13 @@
 
 struct hashtable *hashtable_create(int size, int key64)
 {
+	int req = size;
 	struct hashtable *tbl;
-	int node_size = key64 ? sizeof (struct ht_int64_node )
+	int node_size = key64 ? sizeof (struct ht_int64_node)
 			      : sizeof (struct ht_int32_node);
 
 	/* Pick a power of 2 that can hold the requested size. */
 	if (size & (size-1) || size < 16) {
-		int req = size;
 		size = 16;
 		while (size < req)
 			size *= 2;
@@ -41,25 +41,55 @@ struct hashtable *hashtable_create(int size, int key64)
 	tbl->size = size;
 	tbl->entries = 0;
 	tbl->node_size = node_size;
+	tbl->key64 = key64 ? 1 : 0;
+
+	if (DEBUG_GTE(HASH, 1)) {
+		char buf[32];
+		if (req != size)
+			snprintf(buf, sizeof buf, "req: %d, ", req);
+		else
+			*buf = '\0';
+		rprintf(FINFO, "[%s] created hashtable %lx (%ssize: %d, keys: %d-bit)\n",
+			who_am_i(), (long)tbl, buf, size, key64 ? 64 : 32);
+	}
 
 	return tbl;
 }
 
 void hashtable_destroy(struct hashtable *tbl)
 {
+	if (DEBUG_GTE(HASH, 1)) {
+		rprintf(FINFO, "[%s] destroyed hashtable %lx (size: %d, keys: %d-bit)\n",
+			who_am_i(), (long)tbl, tbl->size, tbl->key64 ? 64 : 32);
+	}
 	free(tbl->nodes);
 	free(tbl);
 }
 
-/* This returns the node for the indicated key, either newly created or
- * already existing.  Returns NULL if not allocating and not found. */
-void *hashtable_find(struct hashtable *tbl, int64 key, int allocate_if_missing)
+/* Returns the node that holds the indicated key if it exists. When it does not
+ * exist, it returns either NULL (when data_when_new is NULL), or it returns a
+ * new node with its node->data set to the indicated value.
+ *
+ * If your code doesn't know the data value for a new node in advance (usually
+ * because it doesn't know if a node is new or not) you should pass in a unique
+ * (non-0) value that you can use to check if the returned node is new. You can
+ * then overwrite the data with any value you want (even 0) since it only needs
+ * to be different than whatever data_when_new value you use later on.
+ *
+ * This return is a void* just because it might be pointing at a ht_int32_node
+ * or a ht_int64_node, and that makes the caller's assignment a little easier. */
+void *hashtable_find(struct hashtable *tbl, int64 key, void *data_when_new)
 {
-	int key64 = (tbl->node_size > sizeof (struct ht_int32_node));
+	int key64 = tbl->key64;
 	struct ht_int32_node *node;
 	uint32 ndx;
 
-	if (allocate_if_missing && tbl->entries > HASH_LOAD_LIMIT(tbl->size)) {
+	if (key64 ? key == 0 : (int32)key == 0) {
+		rprintf(FERROR, "Internal hashtable error: illegal key supplied!\n");
+		exit_cleanup(RERR_MESSAGEIO);
+	}
+
+	if (data_when_new && tbl->entries > HASH_LOAD_LIMIT(tbl->size)) {
 		void *old_nodes = tbl->nodes;
 		int size = tbl->size * 2;
 		int i;
@@ -69,13 +99,22 @@ void *hashtable_find(struct hashtable *tbl, int64 key, int allocate_if_missing)
 		tbl->size = size;
 		tbl->entries = 0;
 
+		if (DEBUG_GTE(HASH, 1)) {
+			rprintf(FINFO, "[%s] growing hashtable %lx (size: %d, keys: %d-bit)\n",
+				who_am_i(), (long)tbl, size, key64 ? 64 : 32);
+		}
+
 		for (i = size / 2; i-- > 0; ) {
 			struct ht_int32_node *move_node = HT_NODE(tbl, old_nodes, i);
 			int64 move_key = HT_KEY(move_node, key64);
 			if (move_key == 0)
 				continue;
-			node = hashtable_find(tbl, move_key, 1);
-			node->data = move_node->data;
+			if (move_node->data)
+				hashtable_find(tbl, move_key, move_node->data);
+			else {
+				node = hashtable_find(tbl, move_key, "");
+				node->data = 0;
+			}
 		}
 
 		free(old_nodes);
@@ -86,7 +125,7 @@ void *hashtable_find(struct hashtable *tbl, int64 key, int allocate_if_missing)
 		uchar buf[4], *keyp = buf;
 		int i;
 
-		SIVAL(buf, 0, key);
+		SIVALu(buf, 0, key);
 		for (ndx = 0, i = 0; i < 4; i++) {
 			ndx += keyp[i];
 			ndx += (ndx << 10);
@@ -103,7 +142,9 @@ void *hashtable_find(struct hashtable *tbl, int64 key, int allocate_if_missing)
 		a = b = c = 0xdeadbeef + (8 << 2);
 
 #define rot(x,k) (((x)<<(k)) ^ ((x)>>(32-(k))))
+#if SIZEOF_INT64 >= 8
 		b += (uint32)(key >> 32);
+#endif
 		a += (uint32)key;
 		c ^= b; c -= rot(b, 14);
 		a ^= c; a -= rot(c, 11);
@@ -128,7 +169,7 @@ void *hashtable_find(struct hashtable *tbl, int64 key, int allocate_if_missing)
 		if (nkey == key)
 			return node;
 		if (nkey == 0) {
-			if (!allocate_if_missing)
+			if (!data_when_new)
 				return NULL;
 			break;
 		}
@@ -139,7 +180,321 @@ void *hashtable_find(struct hashtable *tbl, int64 key, int allocate_if_missing)
 	if (key64)
 		((struct ht_int64_node*)node)->key = key;
 	else
-		node->key = key;
+		node->key = (int32)key;
+	node->data = data_when_new;
 	tbl->entries++;
 	return node;
 }
+
+#ifndef WORDS_BIGENDIAN
+# define HASH_LITTLE_ENDIAN 1
+# define HASH_BIG_ENDIAN 0
+#else
+# define HASH_LITTLE_ENDIAN 0
+# define HASH_BIG_ENDIAN 1
+#endif
+
+/*
+ -------------------------------------------------------------------------------
+ lookup3.c, by Bob Jenkins, May 2006, Public Domain.
+
+ These are functions for producing 32-bit hashes for hash table lookup.
+ hash_word(), hashlittle(), hashlittle2(), hashbig(), mix(), and final()
+ are externally useful functions.  Routines to test the hash are included
+ if SELF_TEST is defined.  You can use this free for any purpose.  It's in
+ the public domain.  It has no warranty.
+
+ You probably want to use hashlittle().  hashlittle() and hashbig()
+ hash byte arrays.  hashlittle() is is faster than hashbig() on
+ little-endian machines.  Intel and AMD are little-endian machines.
+ On second thought, you probably want hashlittle2(), which is identical to
+ hashlittle() except it returns two 32-bit hashes for the price of one.
+ You could implement hashbig2() if you wanted but I haven't bothered here.
+
+ If you want to find a hash of, say, exactly 7 integers, do
+   a = i1;  b = i2;  c = i3;
+   mix(a,b,c);
+   a += i4; b += i5; c += i6;
+   mix(a,b,c);
+   a += i7;
+   final(a,b,c);
+ then use c as the hash value.  If you have a variable length array of
+ 4-byte integers to hash, use hash_word().  If you have a byte array (like
+ a character string), use hashlittle().  If you have several byte arrays, or
+ a mix of things, see the comments above hashlittle().
+
+ Why is this so big?  I read 12 bytes at a time into 3 4-byte integers,
+ then mix those integers.  This is fast (you can do a lot more thorough
+ mixing with 12*3 instructions on 3 integers than you can with 3 instructions
+ on 1 byte), but shoehorning those bytes into integers efficiently is messy.
+*/
+
+#define hashsize(n) ((uint32_t)1<<(n))
+#define hashmask(n) (hashsize(n)-1)
+#define rot(x,k) (((x)<<(k)) | ((x)>>(32-(k))))
+
+/*
+ -------------------------------------------------------------------------------
+ mix -- mix 3 32-bit values reversibly.
+
+ This is reversible, so any information in (a,b,c) before mix() is
+ still in (a,b,c) after mix().
+
+ If four pairs of (a,b,c) inputs are run through mix(), or through
+ mix() in reverse, there are at least 32 bits of the output that
+ are sometimes the same for one pair and different for another pair.
+ This was tested for:
+ * pairs that differed by one bit, by two bits, in any combination
+   of top bits of (a,b,c), or in any combination of bottom bits of
+   (a,b,c).
+ * "differ" is defined as +, -, ^, or ~^.  For + and -, I transformed
+   the output delta to a Gray code (a^(a>>1)) so a string of 1's (as
+   is commonly produced by subtraction) look like a single 1-bit
+   difference.
+ * the base values were pseudorandom, all zero but one bit set, or
+   all zero plus a counter that starts at zero.
+
+ Some k values for my "a-=c; a^=rot(c,k); c+=b;" arrangement that
+ satisfy this are
+     4  6  8 16 19  4
+     9 15  3 18 27 15
+    14  9  3  7 17  3
+ Well, "9 15 3 18 27 15" didn't quite get 32 bits diffing
+ for "differ" defined as + with a one-bit base and a two-bit delta.  I
+ used http://burtleburtle.net/bob/hash/avalanche.html to choose
+ the operations, constants, and arrangements of the variables.
+
+ This does not achieve avalanche.  There are input bits of (a,b,c)
+ that fail to affect some output bits of (a,b,c), especially of a.  The
+ most thoroughly mixed value is c, but it doesn't really even achieve
+ avalanche in c.
+
+ This allows some parallelism.  Read-after-writes are good at doubling
+ the number of bits affected, so the goal of mixing pulls in the opposite
+ direction as the goal of parallelism.  I did what I could.  Rotates
+ seem to cost as much as shifts on every machine I could lay my hands
+ on, and rotates are much kinder to the top and bottom bits, so I used
+ rotates.
+ -------------------------------------------------------------------------------
+*/
+#define mix(a,b,c) \
+{ \
+  a -= c;  a ^= rot(c, 4);  c += b; \
+  b -= a;  b ^= rot(a, 6);  a += c; \
+  c -= b;  c ^= rot(b, 8);  b += a; \
+  a -= c;  a ^= rot(c,16);  c += b; \
+  b -= a;  b ^= rot(a,19);  a += c; \
+  c -= b;  c ^= rot(b, 4);  b += a; \
+}
+
+/*
+ -------------------------------------------------------------------------------
+ final -- final mixing of 3 32-bit values (a,b,c) into c
+
+ Pairs of (a,b,c) values differing in only a few bits will usually
+ produce values of c that look totally different.  This was tested for
+ * pairs that differed by one bit, by two bits, in any combination
+   of top bits of (a,b,c), or in any combination of bottom bits of
+   (a,b,c).
+ * "differ" is defined as +, -, ^, or ~^.  For + and -, I transformed
+   the output delta to a Gray code (a^(a>>1)) so a string of 1's (as
+   is commonly produced by subtraction) look like a single 1-bit
+   difference.
+ * the base values were pseudorandom, all zero but one bit set, or
+   all zero plus a counter that starts at zero.
+
+ These constants passed:
+  14 11 25 16 4 14 24
+  12 14 25 16 4 14 24
+ and these came close:
+   4  8 15 26 3 22 24
+  10  8 15 26 3 22 24
+  11  8 15 26 3 22 24
+ -------------------------------------------------------------------------------
+*/
+#define final(a,b,c) \
+{ \
+  c ^= b; c -= rot(b,14); \
+  a ^= c; a -= rot(c,11); \
+  b ^= a; b -= rot(a,25); \
+  c ^= b; c -= rot(b,16); \
+  a ^= c; a -= rot(c,4);  \
+  b ^= a; b -= rot(a,14); \
+  c ^= b; c -= rot(b,24); \
+}
+
+
+/*
+ -------------------------------------------------------------------------------
+ hashlittle() -- hash a variable-length key into a 32-bit value
+   k       : the key (the unaligned variable-length array of bytes)
+   length  : the length of the key, counting by bytes
+   val2    : IN: can be any 4-byte value OUT: second 32 bit hash.
+ Returns a 32-bit value.  Every bit of the key affects every bit of
+ the return value.  Two keys differing by one or two bits will have
+ totally different hash values.  Note that the return value is better
+ mixed than val2, so use that first.
+
+ The best hash table sizes are powers of 2.  There is no need to do
+ mod a prime (mod is sooo slow!).  If you need less than 32 bits,
+ use a bitmask.  For example, if you need only 10 bits, do
+   h = (h & hashmask(10));
+ In which case, the hash table should have hashsize(10) elements.
+
+ If you are hashing n strings (uint8_t **)k, do it like this:
+   for (i=0, h=0; i<n; ++i) h = hashlittle( k[i], len[i], h);
+
+ By Bob Jenkins, 2006.  bob_jenkins@burtleburtle.net.  You may use this
+ code any way you wish, private, educational, or commercial.  It's free.
+
+ Use for hash table lookup, or anything where one collision in 2^^32 is
+ acceptable.  Do NOT use for cryptographic purposes.
+ -------------------------------------------------------------------------------
+*/
+
+uint32_t hashlittle(const void *key, size_t length)
+{
+  uint32_t a,b,c;                                          /* internal state */
+  union { const void *ptr; size_t i; } u;     /* needed for Mac Powerbook G4 */
+
+  /* Set up the internal state */
+  a = b = c = 0xdeadbeef + ((uint32_t)length);
+
+  u.ptr = key;
+  if (HASH_LITTLE_ENDIAN && ((u.i & 0x3) == 0)) {
+    const uint32_t *k = (const uint32_t *)key;         /* read 32-bit chunks */
+    const uint8_t  *k8;
+
+    /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */
+    while (length > 12)
+    {
+      a += k[0];
+      b += k[1];
+      c += k[2];
+      mix(a,b,c);
+      length -= 12;
+      k += 3;
+    }
+
+    /*----------------------------- handle the last (probably partial) block */
+    k8 = (const uint8_t *)k;
+    switch(length)
+    {
+    case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
+    case 11: c+=((uint32_t)k8[10])<<16;  /* fall through */
+    case 10: c+=((uint32_t)k8[9])<<8;    /* fall through */
+    case 9 : c+=k8[8];                   /* fall through */
+    case 8 : b+=k[1]; a+=k[0]; break;
+    case 7 : b+=((uint32_t)k8[6])<<16;   /* fall through */
+    case 6 : b+=((uint32_t)k8[5])<<8;    /* fall through */
+    case 5 : b+=k8[4];                   /* fall through */
+    case 4 : a+=k[0]; break;
+    case 3 : a+=((uint32_t)k8[2])<<16;   /* fall through */
+    case 2 : a+=((uint32_t)k8[1])<<8;    /* fall through */
+    case 1 : a+=k8[0]; break;
+    case 0 : return c;
+    }
+  } else if (HASH_LITTLE_ENDIAN && ((u.i & 0x1) == 0)) {
+    const uint16_t *k = (const uint16_t *)key;         /* read 16-bit chunks */
+    const uint8_t  *k8;
+
+    /*--------------- all but last block: aligned reads and different mixing */
+    while (length > 12)
+    {
+      a += k[0] + (((uint32_t)k[1])<<16);
+      b += k[2] + (((uint32_t)k[3])<<16);
+      c += k[4] + (((uint32_t)k[5])<<16);
+      mix(a,b,c);
+      length -= 12;
+      k += 6;
+    }
+
+    /*----------------------------- handle the last (probably partial) block */
+    k8 = (const uint8_t *)k;
+    switch(length)
+    {
+    case 12: c+=k[4]+(((uint32_t)k[5])<<16);
+             b+=k[2]+(((uint32_t)k[3])<<16);
+             a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 11: c+=((uint32_t)k8[10])<<16;     /* fall through */
+    case 10: c+=k[4];
+             b+=k[2]+(((uint32_t)k[3])<<16);
+             a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 9 : c+=k8[8];                      /* fall through */
+    case 8 : b+=k[2]+(((uint32_t)k[3])<<16);
+             a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 7 : b+=((uint32_t)k8[6])<<16;      /* fall through */
+    case 6 : b+=k[2];
+             a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 5 : b+=k8[4];                      /* fall through */
+    case 4 : a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 3 : a+=((uint32_t)k8[2])<<16;      /* fall through */
+    case 2 : a+=k[0];
+             break;
+    case 1 : a+=k8[0];
+             break;
+    case 0 : return c;                     /* zero length requires no mixing */
+    }
+
+  } else {                        /* need to read the key one byte at a time */
+    const uint8_t *k = (const uint8_t *)key;
+
+    /*--------------- all but the last block: affect some 32 bits of (a,b,c) */
+    while (length > 12)
+    {
+      a += k[0];
+      a += ((uint32_t)k[1])<<8;
+      a += ((uint32_t)k[2])<<16;
+      a += ((uint32_t)k[3])<<24;
+      b += k[4];
+      b += ((uint32_t)k[5])<<8;
+      b += ((uint32_t)k[6])<<16;
+      b += ((uint32_t)k[7])<<24;
+      c += k[8];
+      c += ((uint32_t)k[9])<<8;
+      c += ((uint32_t)k[10])<<16;
+      c += ((uint32_t)k[11])<<24;
+      mix(a,b,c);
+      length -= 12;
+      k += 12;
+    }
+
+    /*-------------------------------- last block: affect all 32 bits of (c) */
+    switch(length)                   /* all the case statements fall through */
+    {
+    case 12: c+=((uint32_t)k[11])<<24;
+	     /* FALLTHROUGH */
+    case 11: c+=((uint32_t)k[10])<<16;
+	     /* FALLTHROUGH */
+    case 10: c+=((uint32_t)k[9])<<8;
+	     /* FALLTHROUGH */
+    case 9 : c+=k[8];
+	     /* FALLTHROUGH */
+    case 8 : b+=((uint32_t)k[7])<<24;
+	     /* FALLTHROUGH */
+    case 7 : b+=((uint32_t)k[6])<<16;
+	     /* FALLTHROUGH */
+    case 6 : b+=((uint32_t)k[5])<<8;
+	     /* FALLTHROUGH */
+    case 5 : b+=k[4];
+	     /* FALLTHROUGH */
+    case 4 : a+=((uint32_t)k[3])<<24;
+	     /* FALLTHROUGH */
+    case 3 : a+=((uint32_t)k[2])<<16;
+	     /* FALLTHROUGH */
+    case 2 : a+=((uint32_t)k[1])<<8;
+	     /* FALLTHROUGH */
+    case 1 : a+=k[0];
+             break;
+    case 0 : return c;
+    }
+  }
+
+  final(a,b,c);
+  return c;
+}

help-from-md.awk

@@ -0,0 +1,40 @@
+#!/usr/bin/awk -f
+
+# The caller must pass args: -v hfile=help-NAME.h NAME.NUM.md
+
+BEGIN {
+    heading = "/* DO NOT EDIT THIS FILE!  It is auto-generated from the option list in " ARGV[1] "! */"
+    findcomment = hfile
+    sub("\\.", "\\.", findcomment)
+    findcomment = "\\[comment\\].*" findcomment
+    backtick_cnt = 0
+    prints = ""
+}
+
+/^```/ {
+    backtick_cnt++
+    next
+}
+
+foundcomment {
+    if (backtick_cnt > 1) exit
+    if (backtick_cnt == 1) {
+	gsub(/"/, "\\\"")
+	prints = prints "\n  rprintf(F,\"" $0 "\\n\");"
+    }
+    next
+}
+
+$0 ~ findcomment {
+    foundcomment = 1
+    backtick_cnt = 0
+}
+
+END {
+    if (foundcomment && backtick_cnt > 1)
+	print heading "\n" prints > hfile
+    else {
+	print "Failed to find " hfile " section in " ARGV[1]
+	exit 1
+    }
+}

hlink.c

@@ -4,7 +4,7 @@
  * Copyright (C) 1996 Andrew Tridgell
  * Copyright (C) 1996 Paul Mackerras
  * Copyright (C) 2002 Martin Pool <mbp@samba.org>
- * Copyright (C) 2004-2007 Wayne Davison
+ * Copyright (C) 2004-2020 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
@@ -21,24 +21,24 @@
  */
 
 #include "rsync.h"
+#include "inums.h"
+#include "ifuncs.h"
 
-extern int verbose;
 extern int dry_run;
+extern int list_only;
 extern int am_sender;
 extern int inc_recurse;
 extern int do_xfers;
-extern int link_dest;
+extern int alt_dest_type;
 extern int preserve_acls;
-extern int make_backups;
+extern int preserve_xattrs;
 extern int protocol_version;
 extern int remove_source_files;
 extern int stdout_format_has_i;
 extern int maybe_ATTRS_REPORT;
-extern char *basis_dir[];
+extern int unsort_ndx;
+extern char *basis_dir[MAX_BASIS_DIRS+1];
 extern struct file_list *cur_flist;
-#ifdef ICONV_OPTION
-extern int ic_ndx;
-#endif
 
 #ifdef SUPPORT_HARD_LINKS
 
@@ -48,6 +48,8 @@ extern int ic_ndx;
  * we can avoid the pool of dev+inode data.  For incremental recursion mode,
  * the receiver will use a ndx hash to remember old pathnames. */
 
+static void *data_when_new = "";
+
 static struct hashtable *dev_tbl;
 
 static struct hashtable *prior_hlinks;
@@ -57,25 +59,29 @@ static struct file_list *hlink_flist;
 void init_hard_links(void)
 {
 	if (am_sender || protocol_version < 30)
-		dev_tbl = hashtable_create(16, SIZEOF_INT64 == 8);
+		dev_tbl = hashtable_create(16, HT_KEY64);
 	else if (inc_recurse)
-		prior_hlinks = hashtable_create(1024, 0);
+		prior_hlinks = hashtable_create(1024, HT_KEY32);
 }
 
 struct ht_int64_node *idev_find(int64 dev, int64 ino)
 {
 	static struct ht_int64_node *dev_node = NULL;
-	struct hashtable *tbl;
 
-	if (!dev_node || dev_node->key != dev) {
+	/* Note that some OSes have a dev == 0, so increment to avoid storing a 0. */
+	if (!dev_node || dev_node->key != dev+1) {
 		/* We keep a separate hash table of inodes for every device. */
-		dev_node = hashtable_find(dev_tbl, dev, 1);
-		if (!(tbl = dev_node->data))
-			tbl = dev_node->data = hashtable_create(512, SIZEOF_INT64 == 8);
-	} else
-		tbl = dev_node->data;
+		dev_node = hashtable_find(dev_tbl, dev+1, data_when_new);
+		if (dev_node->data == data_when_new) {
+			dev_node->data = hashtable_create(512, HT_KEY64);
+			if (DEBUG_GTE(HLINK, 3)) {
+				rprintf(FINFO, "[%s] created hashtable for dev %s\n",
+					who_am_i(), big_num(dev));
+			}
+		}
+	}
 
-	return hashtable_find(tbl, ino, 1);
+	return hashtable_find(dev_node->data, ino, (void*)-1L);
 }
 
 void idev_destroy(void)
@@ -111,27 +117,31 @@ static void match_gnums(int32 *ndx_list, int ndx_count)
 	struct ht_int32_node *node = NULL;
 	int32 gnum, gnum_next;
 
-	qsort(ndx_list, ndx_count, sizeof ndx_list[0],
-	     (int (*)()) hlink_compare_gnum);
+	qsort(ndx_list, ndx_count, sizeof ndx_list[0], (int (*)()) hlink_compare_gnum);
 
 	for (from = 0; from < ndx_count; from++) {
 		file = hlink_flist->sorted[ndx_list[from]];
 		gnum = F_HL_GNUM(file);
 		if (inc_recurse) {
-			node = hashtable_find(prior_hlinks, gnum, 1);
-			if (!node->data) {
-				node->data = new_array0(char, 5);
+			node = hashtable_find(prior_hlinks, gnum, data_when_new);
+			if (node->data == data_when_new) {
+				if (!(node->data = new_array0(char, 5)))
+					out_of_memory("match_gnums");
 				assert(gnum >= hlink_flist->ndx_start);
 				file->flags |= FLAG_HLINK_FIRST;
 				prev = -1;
 			} else if (CVAL(node->data, 0) == 0) {
 				struct file_list *flist;
-				struct file_struct *fp;
 				prev = IVAL(node->data, 1);
-				flist = flist_for_ndx(prev);
-				assert(flist != NULL);
-				fp = flist->files[prev - flist->ndx_start];
-				fp->flags &= ~FLAG_HLINK_LAST;
+				flist = flist_for_ndx(prev, NULL);
+				if (flist)
+					flist->files[prev - flist->ndx_start]->flags &= ~FLAG_HLINK_LAST;
+				else {
+					/* We skipped all prior files in this
+					 * group, so mark this as a "first". */
+					file->flags |= FLAG_HLINK_FIRST;
+					prev = -1;
+				}
 			} else
 				prev = -1;
 		} else {
@@ -144,12 +154,10 @@ static void match_gnums(int32 *ndx_list, int ndx_count)
 			if (gnum != gnum_next)
 				break;
 			F_HL_PREV(file) = prev;
-			/* The linked list must use raw ndx values. */
-#ifdef ICONV_OPTION
-			if (ic_ndx)
+			/* The linked list uses over-the-wire ndx values. */
+			if (unsort_ndx)
 				prev = F_NDX(file);
 			else
-#endif
 				prev = ndx_list[from] + hlink_flist->ndx_start;
 		}
 		if (prev < 0 && !inc_recurse) {
@@ -163,11 +171,9 @@ static void match_gnums(int32 *ndx_list, int ndx_count)
 		file->flags |= FLAG_HLINK_LAST;
 		F_HL_PREV(file) = prev;
 		if (inc_recurse && CVAL(node->data, 0) == 0) {
-#ifdef ICONV_OPTION
-			if (ic_ndx)
+			if (unsort_ndx)
 				prev = F_NDX(file);
 			else
-#endif
 				prev = ndx_list[from] + hlink_flist->ndx_start;
 			SIVAL(node->data, 1, prev);
 		}
@@ -180,29 +186,31 @@ static void match_gnums(int32 *ndx_list, int ndx_count)
  * to first when we're done. */
 void match_hard_links(struct file_list *flist)
 {
-	int i, ndx_count = 0;
-	int32 *ndx_list;
+	if (!list_only && flist->used) {
+		int i, ndx_count = 0;
+		int32 *ndx_list;
 
-	if (!(ndx_list = new_array(int32, flist->used)))
-		out_of_memory("match_hard_links");
+		if (!(ndx_list = new_array(int32, flist->used)))
+			out_of_memory("match_hard_links");
 
-	for (i = 0; i < flist->used; i++) {
-		if (F_IS_HLINKED(flist->sorted[i]))
-			ndx_list[ndx_count++] = i;
+		for (i = 0; i < flist->used; i++) {
+			if (F_IS_HLINKED(flist->sorted[i]))
+				ndx_list[ndx_count++] = i;
+		}
+
+		hlink_flist = flist;
+
+		if (ndx_count)
+			match_gnums(ndx_list, ndx_count);
+
+		free(ndx_list);
 	}
-
-	hlink_flist = flist;
-
-	if (ndx_count)
-		match_gnums(ndx_list, ndx_count);
-
-	free(ndx_list);
 	if (protocol_version < 30)
 		idev_destroy();
 }
 
 static int maybe_hard_link(struct file_struct *file, int ndx,
-			   const char *fname, int statret, stat_x *sxp,
+			   char *fname, int statret, stat_x *sxp,
 			   const char *oldname, STRUCT_STAT *old_stp,
 			   const char *realname, int itemizing, enum logcode code)
 {
@@ -214,55 +222,70 @@ static int maybe_hard_link(struct file_struct *file, int ndx,
 					ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
 					0, "");
 			}
-			if (verbose > 1 && maybe_ATTRS_REPORT)
+			if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 				rprintf(FCLIENT, "%s is uptodate\n", fname);
 			file->flags |= FLAG_HLINK_DONE;
 			return 0;
 		}
-		if (make_backups > 0) {
-			if (!make_backup(fname))
-				return -1;
-		} else if (robust_unlink(fname)) {
-			rsyserr(FERROR, errno, "unlink %s failed",
-				full_fname(fname));
-			return -1;
-		}
 	}
 
-	if (hard_link_one(file, fname, oldname, 0)) {
+	if (atomic_create(file, fname, NULL, oldname, MAKEDEV(0, 0), sxp, statret == 0 ? DEL_FOR_FILE : 0)) {
 		if (itemizing) {
 			itemize(fname, file, ndx, statret, sxp,
 				ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
 				realname);
 		}
-		if (code != FNONE && verbose)
+		if (code != FNONE && INFO_GTE(NAME, 1))
 			rprintf(code, "%s => %s\n", fname, realname);
 		return 0;
 	}
+
 	return -1;
 }
 
 /* Figure out if a prior entry is still there or if we just have a
- * cached name for it.  Never called with a FLAG_HLINK_FIRST entry. */
-static char *check_prior(int prev_ndx, int gnum, struct file_list **flist_p)
+ * cached name for it. */
+static char *check_prior(struct file_struct *file, int gnum,
+			 int *prev_ndx_p, struct file_list **flist_p)
 {
-	struct file_list *flist = flist_for_ndx(prev_ndx);
+	struct file_struct *fp;
 	struct ht_int32_node *node;
+	int prev_ndx = F_HL_PREV(file);
 
-	if (flist) {
-		*flist_p = flist;
-		return NULL;
+	while (1) {
+		struct file_list *flist;
+		if (prev_ndx < 0
+		 || (flist = flist_for_ndx(prev_ndx, NULL)) == NULL)
+			break;
+		fp = flist->files[prev_ndx - flist->ndx_start];
+		if (!(fp->flags & FLAG_SKIP_HLINK)) {
+			*prev_ndx_p = prev_ndx;
+			*flist_p = flist;
+			return NULL;
+		}
+		F_HL_PREV(file) = prev_ndx = F_HL_PREV(fp);
 	}
 
-	node = hashtable_find(prior_hlinks, gnum, 0);
-	assert(node != NULL && node->data);
-	assert(CVAL(node->data, 0) != 0);
-	return node->data;
+	if (inc_recurse
+	 && (node = hashtable_find(prior_hlinks, gnum, NULL)) != NULL) {
+		assert(node->data != NULL);
+		if (CVAL(node->data, 0) != 0) {
+			*prev_ndx_p = -1;
+			*flist_p = NULL;
+			return node->data;
+		}
+		/* The prior file must have been skipped. */
+		F_HL_PREV(file) = -1;
+	}
+
+	*prev_ndx_p = -1;
+	*flist_p = NULL;
+	return NULL;
 }
 
 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
-int hard_link_check(struct file_struct *file, int ndx, const char *fname,
+int hard_link_check(struct file_struct *file, int ndx, char *fname,
 		    int statret, stat_x *sxp, int itemizing,
 		    enum logcode code)
 {
@@ -271,34 +294,53 @@ int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 	char *realname, *prev_name;
 	struct file_list *flist;
 	int gnum = inc_recurse ? F_HL_GNUM(file) : -1;
-	int prev_ndx = F_HL_PREV(file);
+	int prev_ndx;
 
-	prev_name = realname = check_prior(prev_ndx, gnum, &flist);
+	prev_name = realname = check_prior(file, gnum, &prev_ndx, &flist);
 
 	if (!prev_name) {
-		struct file_struct *prev_file = flist->files[prev_ndx - flist->ndx_start];
+		struct file_struct *prev_file;
 
-		/* Is the previous link is not complete yet? */
+		if (!flist) {
+			/* The previous file was skipped, so this one is
+			 * treated as if it were the first in its group. */
+			if (DEBUG_GTE(HLINK, 2)) {
+				rprintf(FINFO, "hlink for %d (%s,%d): virtual first\n",
+					ndx, f_name(file, NULL), gnum);
+			}
+			return 0;
+		}
+
+		prev_file = flist->files[prev_ndx - flist->ndx_start];
+
+		/* Is the previous link not complete yet? */
 		if (!(prev_file->flags & FLAG_HLINK_DONE)) {
 			/* Is the previous link being transferred? */
 			if (prev_file->flags & FLAG_FILE_SENT) {
-				/* Add ourselves to the list of files that will be
-				 * updated when the transfer completes, and mark
-				 * ourself as waiting for the transfer. */
+				/* Add ourselves to the list of files that will
+				 * be updated when the transfer completes, and
+				 * mark ourself as waiting for the transfer. */
 				F_HL_PREV(file) = F_HL_PREV(prev_file);
 				F_HL_PREV(prev_file) = ndx;
 				file->flags |= FLAG_FILE_SENT;
 				cur_flist->in_progress++;
+				if (DEBUG_GTE(HLINK, 2)) {
+					rprintf(FINFO, "hlink for %d (%s,%d): waiting for %d\n",
+						ndx, f_name(file, NULL), gnum, F_HL_PREV(file));
+				}
 				return 1;
 			}
+			if (DEBUG_GTE(HLINK, 2)) {
+				rprintf(FINFO, "hlink for %d (%s,%d): looking for a leader\n",
+					ndx, f_name(file, NULL), gnum);
+			}
 			return 0;
 		}
 
 		/* There is a finished file to link with! */
 		if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
 			/* The previous previous is FIRST when prev is not. */
-			prev_ndx = F_HL_PREV(prev_file);
-			prev_name = realname = check_prior(prev_ndx, gnum, &flist);
+			prev_name = realname = check_prior(prev_file, gnum, &prev_ndx, &flist);
 			/* Update our previous pointer to point to the FIRST. */
 			F_HL_PREV(file) = prev_ndx;
 		}
@@ -306,9 +348,14 @@ int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 		if (!prev_name) {
 			int alt_dest;
 
+			assert(flist != NULL);
 			prev_file = flist->files[prev_ndx - flist->ndx_start];
 			/* F_HL_PREV() is alt_dest value when DONE && FIRST. */
 			alt_dest = F_HL_PREV(prev_file);
+			if (DEBUG_GTE(HLINK, 2)) {
+				rprintf(FINFO, "hlink for %d (%s,%d): found flist match (alt %d)\n",
+					ndx, f_name(file, NULL), gnum, alt_dest);
+			}
 
 			if (alt_dest >= 0 && dry_run) {
 				pathjoin(namebuf, MAXPATHLEN, basis_dir[alt_dest],
@@ -322,10 +369,19 @@ int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 		}
 	}
 
+	if (DEBUG_GTE(HLINK, 2)) {
+		rprintf(FINFO, "hlink for %d (%s,%d): leader is %d (%s)\n",
+			ndx, f_name(file, NULL), gnum, prev_ndx, prev_name);
+	}
+
 	if (link_stat(prev_name, &prev_st, 0) < 0) {
-		rsyserr(FERROR, errno, "stat %s failed",
-			full_fname(prev_name));
-		return -1;
+		if (!dry_run || errno != ENOENT) {
+			rsyserr(FERROR_XFER, errno, "stat %s failed", full_fname(prev_name));
+			return -1;
+		}
+		/* A new hard-link will get a new dev & inode, so approximate
+		 * those values in dry-run mode by zeroing them. */
+		memset(&prev_st, 0, sizeof prev_st);
 	}
 
 	if (statret < 0 && basis_dir[0] != NULL) {
@@ -333,22 +389,21 @@ int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 		char cmpbuf[MAXPATHLEN];
 		stat_x alt_sx;
 		int j = 0;
-#ifdef SUPPORT_ACLS
-		alt_sx.acc_acl = alt_sx.def_acl = NULL;
-#endif
+		init_stat_x(&alt_sx);
 		do {
 			pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
 			if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
 				continue;
-			if (link_dest) {
+			if (alt_dest_type == LINK_DEST) {
 				if (prev_st.st_dev != alt_sx.st.st_dev
 				 || prev_st.st_ino != alt_sx.st.st_ino)
 					continue;
 				statret = 1;
-				if (verbose < 2 || !stdout_format_has_i) {
+				if (stdout_format_has_i == 0
+				 || (!INFO_GTE(NAME, 2) && stdout_format_has_i < 2)) {
 					itemizing = 0;
 					code = FNONE;
-					if (verbose > 1 && maybe_ATTRS_REPORT)
+					if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 						rprintf(FCLIENT, "%s is uptodate\n", fname);
 				}
 				break;
@@ -363,19 +418,29 @@ int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 			sxp->st = alt_sx.st;
 #ifdef SUPPORT_ACLS
 			if (preserve_acls && !S_ISLNK(file->mode)) {
-				if (!ACL_READY(*sxp))
+				free_acl(sxp);
+				if (!ACL_READY(alt_sx))
 					get_acl(cmpbuf, sxp);
 				else {
 					sxp->acc_acl = alt_sx.acc_acl;
 					sxp->def_acl = alt_sx.def_acl;
+					alt_sx.acc_acl = alt_sx.def_acl = NULL;
 				}
 			}
 #endif
-		}
-#ifdef SUPPORT_ACLS
-		else if (preserve_acls)
-			free_acl(&alt_sx);
+#ifdef SUPPORT_XATTRS
+			if (preserve_xattrs) {
+				free_xattr(sxp);
+				if (!XATTR_READY(alt_sx))
+					get_xattr(cmpbuf, sxp);
+				else {
+					sxp->xattr = alt_sx.xattr;
+					alt_sx.xattr = NULL;
+				}
+			}
 #endif
+		} else
+			free_stat_x(&alt_sx);
 	}
 
 	if (maybe_hard_link(file, ndx, fname, statret, sxp, prev_name, &prev_st,
@@ -394,11 +459,11 @@ int hard_link_one(struct file_struct *file, const char *fname,
 	if (do_link(oldname, fname) < 0) {
 		enum logcode code;
 		if (terse) {
-			if (!verbose)
-				return -1;
+			if (!INFO_GTE(NAME, 1))
+				return 0;
 			code = FINFO;
 		} else
-			code = FERROR;
+			code = FERROR_XFER;
 		rsyserr(code, errno, "link %s => %s failed",
 			full_fname(fname), oldname);
 		return 0;
@@ -415,14 +480,14 @@ void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
 {
 	stat_x prev_sx;
 	STRUCT_STAT st;
-	char alt_name[MAXPATHLEN], *prev_name;
+	char prev_name[MAXPATHLEN], alt_name[MAXPATHLEN];
 	const char *our_name;
 	struct file_list *flist;
 	int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
 
 	if (stp == NULL && prev_ndx >= 0) {
-		if (link_stat(fname, &st, 0) < 0) {
-			rsyserr(FERROR, errno, "stat %s failed",
+		if (link_stat(fname, &st, 0) < 0 && !dry_run) {
+			rsyserr(FERROR_XFER, errno, "stat %s failed",
 				full_fname(fname));
 			return;
 		}
@@ -439,27 +504,20 @@ void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
 	} else
 		our_name = fname;
 
-#ifdef SUPPORT_ACLS
-	prev_sx.acc_acl = prev_sx.def_acl = NULL;
-#endif
+	init_stat_x(&prev_sx);
 
 	while ((ndx = prev_ndx) >= 0) {
 		int val;
-		flist = flist_for_ndx(ndx);
-		assert(flist != NULL);
+		flist = flist_for_ndx(ndx, "finish_hard_link");
 		file = flist->files[ndx - flist->ndx_start];
 		file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
 		prev_ndx = F_HL_PREV(file);
 		F_HL_PREV(file) = fin_ndx;
-		prev_name = f_name(file, NULL);
-		prev_statret = link_stat(prev_name, &prev_sx.st, 0);
+		prev_statret = link_stat(f_name(file, prev_name), &prev_sx.st, 0);
 		val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
 				      our_name, stp, fname, itemizing, code);
 		flist->in_progress--;
-#ifdef SUPPORT_ACLS
-		if (preserve_acls)
-			free_acl(&prev_sx);
-#endif
+		free_stat_x(&prev_sx);
 		if (val < 0)
 			continue;
 		if (remove_source_files == 1 && do_xfers)
@@ -468,12 +526,44 @@ void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
 
 	if (inc_recurse) {
 		int gnum = F_HL_GNUM(file);
-		struct ht_int32_node *node = hashtable_find(prior_hlinks, gnum, 0);
-		assert(node != NULL && node->data != NULL);
-		assert(CVAL(node->data, 0) == 0);
+		struct ht_int32_node *node = hashtable_find(prior_hlinks, gnum, NULL);
+		if (node == NULL) {
+			rprintf(FERROR, "Unable to find a hlink node for %d (%s)\n", gnum, f_name(file, prev_name));
+			exit_cleanup(RERR_MESSAGEIO);
+		}
+		if (node->data == NULL) {
+			rprintf(FERROR, "Hlink node data for %d is NULL (%s)\n", gnum, f_name(file, prev_name));
+			exit_cleanup(RERR_MESSAGEIO);
+		}
+		if (CVAL(node->data, 0) != 0) {
+			rprintf(FERROR, "Hlink node data for %d already has path=%s (%s)\n",
+				gnum, (char*)node->data, f_name(file, prev_name));
+			exit_cleanup(RERR_MESSAGEIO);
+		}
 		free(node->data);
 		if (!(node->data = strdup(our_name)))
 			out_of_memory("finish_hard_link");
 	}
 }
+
+int skip_hard_link(struct file_struct *file, struct file_list **flist_p)
+{
+	struct file_list *flist;
+	int prev_ndx;
+
+	file->flags |= FLAG_SKIP_HLINK;
+	if (!(file->flags & FLAG_HLINK_LAST))
+		return -1;
+
+	check_prior(file, F_HL_GNUM(file), &prev_ndx, &flist);
+	if (prev_ndx >= 0) {
+		file = flist->files[prev_ndx - flist->ndx_start];
+		if (file->flags & (FLAG_HLINK_DONE|FLAG_FILE_SENT))
+			return -1;
+		file->flags |= FLAG_HLINK_LAST;
+		*flist_p = flist;
+	}
+
+	return prev_ndx;
+}
 #endif

ifuncs.h

@@ -1,6 +1,6 @@
 /* Inline functions for rsync.
  *
- * Copyright (C) 2007 Wayne Davison
+ * Copyright (C) 2007-2019 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
@@ -35,6 +35,14 @@ realloc_xbuf(xbuf *xb, size_t sz)
 	xb->size = sz;
 }
 
+static inline void
+free_xbuf(xbuf *xb)
+{
+	if (xb->buf)
+		free(xb->buf);
+	memset(xb, 0, sizeof (xbuf));
+}
+
 static inline int
 to_wire_mode(mode_t mode)
 {
@@ -57,44 +65,42 @@ from_wire_mode(int mode)
 	return mode;
 }
 
-static inline int
-isDigit(const char *ptr)
+static inline char *
+d_name(struct dirent *di)
 {
-	return isdigit(*(unsigned char *)ptr);
+#ifdef HAVE_BROKEN_READDIR
+	return (di->d_name - 2);
+#else
+	return di->d_name;
+#endif
 }
 
-static inline int
-isPrint(const char *ptr)
+static inline void
+init_stat_x(stat_x *sx_p)
 {
-	return isprint(*(unsigned char *)ptr);
+#ifdef SUPPORT_ACLS
+	sx_p->acc_acl = sx_p->def_acl = NULL;
+#endif
+#ifdef SUPPORT_XATTRS
+	sx_p->xattr = NULL;
+#endif
 }
 
-static inline int
-isSpace(const char *ptr)
+static inline void
+free_stat_x(stat_x *sx_p)
 {
-	return isspace(*(unsigned char *)ptr);
-}
-
-static inline int
-isLower(const char *ptr)
-{
-	return islower(*(unsigned char *)ptr);
-}
-
-static inline int
-isUpper(const char *ptr)
-{
-	return isupper(*(unsigned char *)ptr);
-}
-
-static inline int
-toLower(const char *ptr)
-{
-	return tolower(*(unsigned char *)ptr);
-}
-
-static inline int
-toUpper(const char *ptr)
-{
-	return toupper(*(unsigned char *)ptr);
+#ifdef SUPPORT_ACLS
+    {
+	extern int preserve_acls;
+	if (preserve_acls)
+		free_acl(sx_p);
+    }
+#endif
+#ifdef SUPPORT_XATTRS
+    {
+	extern int preserve_xattrs;
+	if (preserve_xattrs)
+		free_xattr(sx_p);
+    }
+#endif
 }

inums.h

@@ -0,0 +1,57 @@
+/* Inline functions for rsync.
+ *
+ * Copyright (C) 2008-2019 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
+ * the Free Software Foundation; either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, visit the http://fsf.org website.
+ */
+
+static inline char *
+big_num(int64 num)
+{
+	return do_big_num(num, 0, NULL);
+}
+
+static inline char *
+comma_num(int64 num)
+{
+	extern int human_readable;
+	return do_big_num(num, human_readable != 0, NULL);
+}
+
+static inline char *
+human_num(int64 num)
+{
+	extern int human_readable;
+	return do_big_num(num, human_readable, NULL);
+}
+
+static inline char *
+big_dnum(double dnum, int decimal_digits)
+{
+	return do_big_dnum(dnum, 0, decimal_digits);
+}
+
+static inline char *
+comma_dnum(double dnum, int decimal_digits)
+{
+	extern int human_readable;
+	return do_big_dnum(dnum, human_readable != 0, decimal_digits);
+}
+
+static inline char *
+human_dnum(double dnum, int decimal_digits)
+{
+	extern int human_readable;
+	return do_big_dnum(dnum, human_readable, decimal_digits);
+}

io.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2007 Wayne Davison
+ * Copyright (C) 2007-2019 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

itypes.h

@@ -0,0 +1,65 @@
+/* Inline functions for rsync.
+ *
+ * Copyright (C) 2007-2020 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
+ * the Free Software Foundation; either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, visit the http://fsf.org website.
+ */
+
+static inline int
+isDigit(const char *ptr)
+{
+	return isdigit(*(unsigned char *)ptr);
+}
+
+static inline int
+isHexDigit(const char *ptr)
+{
+	return isxdigit(*(unsigned char *)ptr);
+}
+
+static inline int
+isPrint(const char *ptr)
+{
+	return isprint(*(unsigned char *)ptr);
+}
+
+static inline int
+isSpace(const char *ptr)
+{
+	return isspace(*(unsigned char *)ptr);
+}
+
+static inline int
+isLower(const char *ptr)
+{
+	return islower(*(unsigned char *)ptr);
+}
+
+static inline int
+isUpper(const char *ptr)
+{
+	return isupper(*(unsigned char *)ptr);
+}
+
+static inline int
+toLower(const char *ptr)
+{
+	return tolower(*(unsigned char *)ptr);
+}
+
+static inline int
+toUpper(const char *ptr)
+{
+	return toupper(*(unsigned char *)ptr);
+}

latest-year.h

@@ -0,0 +1 @@
+#define LATEST_YEAR "2020"

lib/.cvsignore

@@ -1 +0,0 @@
-dummy

lib/addrinfo.h

@@ -0,0 +1,180 @@
+/*
+PostgreSQL Database Management System
+(formerly known as Postgres, then as Postgres95)
+
+Portions Copyright (c) 1996-2005, The PostgreSQL Global Development Group
+
+Portions Copyright (c) 1994, The Regents of the University of California
+
+Permission to use, copy, modify, and distribute this software and its
+documentation for any purpose, without fee, and without a written agreement
+is hereby granted, provided that the above copyright notice and this paragraph
+and the following two paragraphs appear in all copies.
+
+IN NO EVENT SHALL THE UNIVERSITY OF CALIFORNIA BE LIABLE TO ANY PARTY FOR
+DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING
+LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION,
+EVEN IF THE UNIVERSITY OF CALIFORNIA HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGE.
+
+THE UNIVERSITY OF CALIFORNIA SPECIFICALLY DISCLAIMS ANY WARRANTIES,
+INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
+AND FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
+ON AN "AS IS" BASIS, AND THE UNIVERSITY OF CALIFORNIA HAS NO OBLIGATIONS
+TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
+
+*/
+
+/*-------------------------------------------------------------------------
+ *
+ * getaddrinfo.h
+ *	  Support getaddrinfo() on platforms that don't have it.
+ *
+ * Note: we use our own routines on platforms that don't HAVE_STRUCT_ADDRINFO,
+ * whether or not the library routine getaddrinfo() can be found.  This
+ * policy is needed because on some platforms a manually installed libbind.a
+ * may provide getaddrinfo(), yet the system headers may not provide the
+ * struct definitions needed to call it.  To avoid conflict with the libbind
+ * definition in such cases, we rename our routines to pg_xxx() via macros.
+ *
+ * This code will also work on platforms where struct addrinfo is defined
+ * in the system headers but no getaddrinfo() can be located.
+ *
+ * Copyright (c) 2003-2007, PostgreSQL Global Development Group
+ *
+ *-------------------------------------------------------------------------
+ */
+#ifndef ADDRINFO_H
+#define ADDRINFO_H
+
+
+/* Various macros that ought to be in <netdb.h>, but might not be */
+
+#ifndef EAI_FAIL
+#define EAI_BADFLAGS	(-1)
+#define EAI_NONAME		(-2)
+#define EAI_AGAIN		(-3)
+#define EAI_FAIL		(-4)
+#define EAI_FAMILY		(-6)
+#define EAI_SOCKTYPE	(-7)
+#define EAI_SERVICE		(-8)
+#define EAI_MEMORY		(-10)
+#define EAI_SYSTEM		(-11)
+#endif   /* !EAI_FAIL */
+
+#ifndef AI_PASSIVE
+#define AI_PASSIVE		0x0001
+#endif
+
+#ifndef AI_NUMERICHOST
+/*
+ * some platforms don't support AI_NUMERICHOST; define as zero if using
+ * the system version of getaddrinfo...
+ */
+#if defined(HAVE_STRUCT_ADDRINFO) && defined(HAVE_GETADDRINFO)
+#define AI_NUMERICHOST	0
+#else
+#define AI_NUMERICHOST	0x0004
+#endif
+#endif
+
+#ifndef AI_CANONNAME
+#if defined(HAVE_STRUCT_ADDRINFO) && defined(HAVE_GETADDRINFO)
+#define AI_CANONNAME 0
+#else
+#define AI_CANONNAME 0x0008
+#endif
+#endif
+
+#ifndef AI_NUMERICSERV
+#if defined(HAVE_STRUCT_ADDRINFO) && defined(HAVE_GETADDRINFO)
+#define AI_NUMERICSERV 0
+#else
+#define AI_NUMERICSERV 0x0010
+#endif
+#endif
+
+#ifndef NI_NUMERICHOST
+#define NI_NUMERICHOST	1
+#endif
+
+#ifndef NI_NUMERICSERV
+#define NI_NUMERICSERV	2
+#endif
+
+#ifndef NI_NOFQDN
+#define NI_NOFQDN	4
+#endif
+
+#ifndef NI_NAMEREQD
+#define NI_NAMEREQD 	8
+#endif
+
+#ifndef NI_DGRAM
+#define NI_DGRAM	16
+#endif
+
+
+#ifndef NI_MAXHOST
+#define NI_MAXHOST	1025
+#endif
+
+#ifndef NI_MAXSERV
+#define NI_MAXSERV	32
+#endif
+
+#ifndef HAVE_STRUCT_ADDRINFO
+struct addrinfo
+{
+	int			ai_flags;
+	int			ai_family;
+	int			ai_socktype;
+	int			ai_protocol;
+	size_t		ai_addrlen;
+	struct sockaddr *ai_addr;
+	char	   *ai_canonname;
+	struct addrinfo *ai_next;
+};
+#endif   /* !HAVE_STRUCT_ADDRINFO */
+
+#ifndef HAVE_STRUCT_SOCKADDR_STORAGE
+struct sockaddr_storage {
+	unsigned short ss_family;
+	unsigned long ss_align;
+	char ss_padding[128 - sizeof (unsigned long)];
+};
+#endif	/* !HAVE_STRUCT_SOCKADDR_STORAGE */
+
+#ifndef HAVE_GETADDRINFO
+
+/* Rename private copies per comments above */
+#ifdef getaddrinfo
+#undef getaddrinfo
+#endif
+#define getaddrinfo pg_getaddrinfo
+
+#ifdef freeaddrinfo
+#undef freeaddrinfo
+#endif
+#define freeaddrinfo pg_freeaddrinfo
+
+#ifdef gai_strerror
+#undef gai_strerror
+#endif
+#define gai_strerror pg_gai_strerror
+
+#ifdef getnameinfo
+#undef getnameinfo
+#endif
+#define getnameinfo pg_getnameinfo
+
+extern int getaddrinfo(const char *node, const char *service,
+			const struct addrinfo * hints, struct addrinfo ** res);
+extern void freeaddrinfo(struct addrinfo * res);
+extern const char *gai_strerror(int errcode);
+extern int getnameinfo(const struct sockaddr * sa, socklen_t salen,
+			char *node, size_t nodelen,
+			char *service, size_t servicelen, int flags);
+#endif   /* !HAVE_GETADDRINFO */
+
+#endif   /* ADDRINFO_H */

lib/compat.c

@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1998 Andrew Tridgell
  * Copyright (C) 2002 Martin Pool
- * Copyright (C) 2004, 2005, 2006 Wayne Davison
+ * Copyright (C) 2004-2020 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
@@ -20,6 +20,9 @@
  */
 
 #include "rsync.h"
+#include "itypes.h"
+
+static char number_separator;
 
 #ifndef HAVE_STRDUP
  char *strdup(char *s)
@@ -76,7 +79,7 @@
 
 #ifndef HAVE_STRPBRK
 /**
- * Find the first ocurrence in @p s of any character in @p accept.
+ * Find the first occurrence in @p s of any character in @p accept.
  *
  * Derived from glibc
  **/
@@ -151,3 +154,122 @@ int sys_gettimeofday(struct timeval *tv)
 	return gettimeofday(tv);
 #endif
 }
+
+#define HUMANIFY(mult) \
+	do { \
+		if (num >= mult || num <= -mult) { \
+			double dnum = (double)num / mult; \
+			char units; \
+			if (num < 0) \
+				dnum = -dnum; \
+			if (dnum < mult) \
+				units = 'K'; \
+			else if ((dnum /= mult) < mult) \
+				units = 'M'; \
+			else if ((dnum /= mult) < mult) \
+				units = 'G'; \
+			else { \
+				dnum /= mult; \
+				units = 'T'; \
+			} \
+			if (num < 0) \
+				dnum = -dnum; \
+			snprintf(bufs[n], sizeof bufs[0], "%.2f%c", dnum, units); \
+			return bufs[n]; \
+		} \
+	} while (0)
+
+/* Return the int64 number as a string.  If the human_flag arg is non-zero,
+ * we may output the number in K, M, G, or T units.  If we don't add a unit
+ * suffix, we will append the fract string, if it is non-NULL.  We can
+ * return up to 4 buffers at a time. */
+char *do_big_num(int64 num, int human_flag, const char *fract)
+{
+	static char bufs[4][128]; /* more than enough room */
+	static unsigned int n;
+	char *s;
+	int len, negated;
+
+	if (human_flag && !number_separator) {
+		char buf[32];
+		snprintf(buf, sizeof buf, "%f", 3.14);
+		if (strchr(buf, '.') != NULL)
+			number_separator = ',';
+		else
+			number_separator = '.';
+	}
+
+	n = (n + 1) % (sizeof bufs / sizeof bufs[0]);
+
+	if (human_flag > 1) {
+		if (human_flag == 2)
+			HUMANIFY(1000);
+		else
+			HUMANIFY(1024);
+	}
+
+	s = bufs[n] + sizeof bufs[0] - 1;
+	if (fract) {
+		len = strlen(fract);
+		s -= len;
+		strlcpy(s, fract, len + 1);
+	} else
+		*s = '\0';
+
+	len = 0;
+
+	if (!num)
+		*--s = '0';
+	if (num < 0) {
+		/* A maximum-size negated number can't fit as a positive,
+		 * so do one digit in negated form to start us off. */
+		*--s = (char)(-(num % 10)) + '0';
+		num = -(num / 10);
+		len++;
+		negated = 1;
+	} else
+		negated = 0;
+
+	while (num) {
+		if (human_flag) {
+			if (len == 3) {
+				*--s = number_separator;
+				len = 1;
+			} else
+				len++;
+		}
+		*--s = (char)(num % 10) + '0';
+		num /= 10;
+	}
+
+	if (negated)
+		*--s = '-';
+
+	return s;
+}
+
+/* Return the double number as a string.  If the human_flag option is > 1,
+ * we may output the number in K, M, G, or T units.  The buffer we use for
+ * our result is either a single static buffer defined here, or a buffer
+ * we get from do_big_num(). */
+char *do_big_dnum(double dnum, int human_flag, int decimal_digits)
+{
+	static char tmp_buf[128];
+#if SIZEOF_INT64 >= 8
+	char *fract;
+
+	snprintf(tmp_buf, sizeof tmp_buf, "%.*f", decimal_digits, dnum);
+
+	if (!human_flag || (dnum < 1000.0 && dnum > -1000.0))
+		return tmp_buf;
+
+	for (fract = tmp_buf+1; isDigit(fract); fract++) {}
+
+	return do_big_num((int64)dnum, human_flag, fract);
+#else
+	/* A big number might lose digits converting to a too-short int64,
+	 * so let's just return the raw double conversion. */
+	snprintf(tmp_buf, sizeof tmp_buf, "%.*f", decimal_digits, dnum);
+	return tmp_buf;
+#endif
+}

lib/getaddrinfo.c

@@ -0,0 +1,504 @@
+/*
+PostgreSQL Database Management System
+(formerly known as Postgres, then as Postgres95)
+
+Portions Copyright (c) 1996-2005, The PostgreSQL Global Development Group
+
+Portions Copyright (c) 1994, The Regents of the University of California
+
+Permission to use, copy, modify, and distribute this software and its
+documentation for any purpose, without fee, and without a written agreement
+is hereby granted, provided that the above copyright notice and this paragraph
+and the following two paragraphs appear in all copies.
+
+IN NO EVENT SHALL THE UNIVERSITY OF CALIFORNIA BE LIABLE TO ANY PARTY FOR
+DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING
+LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION,
+EVEN IF THE UNIVERSITY OF CALIFORNIA HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGE.
+
+THE UNIVERSITY OF CALIFORNIA SPECIFICALLY DISCLAIMS ANY WARRANTIES,
+INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
+AND FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
+ON AN "AS IS" BASIS, AND THE UNIVERSITY OF CALIFORNIA HAS NO OBLIGATIONS
+TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
+
+*/
+
+/*-------------------------------------------------------------------------
+ *
+ * getaddrinfo.c
+ *	  Support getaddrinfo() on platforms that don't have it.
+ *
+ * We also supply getnameinfo() here, assuming that the platform will have
+ * it if and only if it has getaddrinfo().	If this proves false on some
+ * platform, we'll need to split this file and provide a separate configure
+ * test for getnameinfo().
+ *
+ * Copyright (c) 2003-2007, PostgreSQL Global Development Group
+ *
+ * Copyright (C) 2007 Jeremy Allison.
+ * Modified to return multiple IPv4 addresses for Samba.
+ *
+ *-------------------------------------------------------------------------
+ */
+
+#include "rsync.h"
+
+#ifndef SMB_MALLOC
+#define SMB_MALLOC(s) malloc(s)
+#endif
+
+#ifndef SMB_STRDUP
+#define SMB_STRDUP(s) strdup(s)
+#endif
+
+#ifndef HOST_NAME_MAX
+#define HOST_NAME_MAX 255
+#endif
+
+static int check_hostent_err(struct hostent *hp)
+{
+#ifndef INET6
+	extern int h_errno;
+#endif
+	if (!hp) {
+		switch (h_errno) {
+			case HOST_NOT_FOUND:
+			case NO_DATA:
+				return EAI_NONAME;
+			case TRY_AGAIN:
+				return EAI_AGAIN;
+			case NO_RECOVERY:
+			default:
+				return EAI_FAIL;
+		}
+	}
+	if (!hp->h_name || hp->h_addrtype != AF_INET) {
+		return EAI_FAIL;
+	}
+	return 0;
+}
+
+static char *canon_name_from_hostent(struct hostent *hp,
+				int *perr)
+{
+	char *ret = NULL;
+
+	*perr = check_hostent_err(hp);
+	if (*perr) {
+		return NULL;
+	}
+	ret = SMB_STRDUP(hp->h_name);
+	if (!ret) {
+		*perr = EAI_MEMORY;
+	}
+	return ret;
+}
+
+static char *get_my_canon_name(int *perr)
+{
+	char name[HOST_NAME_MAX+1];
+
+	if (gethostname(name, HOST_NAME_MAX) == -1) {
+		*perr = EAI_FAIL;
+		return NULL;
+	}
+	/* Ensure null termination. */
+	name[HOST_NAME_MAX] = '\0';
+	return canon_name_from_hostent(gethostbyname(name), perr);
+}
+
+static char *get_canon_name_from_addr(struct in_addr ip,
+				int *perr)
+{
+	return canon_name_from_hostent(
+			gethostbyaddr((void *)&ip, sizeof ip, AF_INET),
+			perr);
+}
+
+static struct addrinfo *alloc_entry(const struct addrinfo *hints,
+				struct in_addr ip,
+				unsigned short port)
+{
+	struct sockaddr_in *psin = NULL;
+	struct addrinfo *ai = SMB_MALLOC(sizeof(*ai));
+
+	if (!ai) {
+		return NULL;
+	}
+	memset(ai, '\0', sizeof(*ai));
+
+	psin = SMB_MALLOC(sizeof(*psin));
+	if (!psin) {
+		free(ai);
+		return NULL;
+	}
+
+	memset(psin, '\0', sizeof(*psin));
+
+	psin->sin_family = AF_INET;
+	psin->sin_port = htons(port);
+	psin->sin_addr = ip;
+
+	ai->ai_flags = 0;
+	ai->ai_family = AF_INET;
+	ai->ai_socktype = hints->ai_socktype;
+	ai->ai_protocol = hints->ai_protocol;
+	ai->ai_addrlen = sizeof(*psin);
+	ai->ai_addr = (struct sockaddr *) psin;
+	ai->ai_canonname = NULL;
+	ai->ai_next = NULL;
+
+	return ai;
+}
+
+/*
+ * get address info for a single ipv4 address.
+ *
+ *	Bugs:	- servname can only be a number, not text.
+ */
+
+static int getaddr_info_single_addr(const char *service,
+				uint32 addr,
+				const struct addrinfo *hints,
+				struct addrinfo **res)
+{
+
+	struct addrinfo *ai = NULL;
+	struct in_addr ip;
+	unsigned short port = 0;
+
+	if (service) {
+		port = (unsigned short)atoi(service);
+	}
+	ip.s_addr = htonl(addr);
+
+	ai = alloc_entry(hints, ip, port);
+	if (!ai) {
+		return EAI_MEMORY;
+	}
+
+	/* If we're asked for the canonical name,
+	 * make sure it returns correctly. */
+	if (!(hints->ai_flags & AI_NUMERICSERV) &&
+			hints->ai_flags & AI_CANONNAME) {
+		int err;
+		if (addr == INADDR_LOOPBACK || addr == INADDR_ANY) {
+			ai->ai_canonname = get_my_canon_name(&err);
+		} else {
+			ai->ai_canonname =
+			get_canon_name_from_addr(ip,&err);
+		}
+		if (ai->ai_canonname == NULL) {
+			freeaddrinfo(ai);
+			return err;
+		}
+	}
+
+	*res = ai;
+	return 0;
+}
+
+/*
+ * get address info for multiple ipv4 addresses.
+ *
+ *	Bugs:	- servname can only be a number, not text.
+ */
+
+static int getaddr_info_name(const char *node,
+				const char *service,
+				const struct addrinfo *hints,
+				struct addrinfo **res)
+{
+	struct addrinfo *listp = NULL, *prevp = NULL;
+	char **pptr = NULL;
+	int err;
+	struct hostent *hp = NULL;
+	unsigned short port = 0;
+
+	if (service) {
+		port = (unsigned short)atoi(service);
+	}
+
+	hp = gethostbyname(node);
+	err = check_hostent_err(hp);
+	if (err) {
+		return err;
+	}
+
+	for(pptr = hp->h_addr_list; *pptr; pptr++) {
+		struct in_addr ip = *(struct in_addr *)*pptr;
+		struct addrinfo *ai = alloc_entry(hints, ip, port);
+
+		if (!ai) {
+			freeaddrinfo(listp);
+			return EAI_MEMORY;
+		}
+
+		if (!listp) {
+			listp = ai;
+			prevp = ai;
+			ai->ai_canonname = SMB_STRDUP(hp->h_name);
+			if (!ai->ai_canonname) {
+				freeaddrinfo(listp);
+				return EAI_MEMORY;
+			}
+		} else {
+			prevp->ai_next = ai;
+			prevp = ai;
+		}
+	}
+	*res = listp;
+	return 0;
+}
+
+/*
+ * get address info for ipv4 sockets.
+ *
+ *	Bugs:	- servname can only be a number, not text.
+ */
+
+int getaddrinfo(const char *node,
+		const char *service,
+		const struct addrinfo * hintp,
+		struct addrinfo ** res)
+{
+	struct addrinfo hints;
+
+	/* Setup the hints struct. */
+	if (hintp == NULL) {
+		memset(&hints, 0, sizeof(hints));
+		hints.ai_family = AF_INET;
+		hints.ai_socktype = SOCK_STREAM;
+	} else {
+		memcpy(&hints, hintp, sizeof(hints));
+	}
+
+	if (hints.ai_family != AF_INET && hints.ai_family != AF_UNSPEC) {
+		return EAI_FAMILY;
+	}
+
+	if (hints.ai_socktype == 0) {
+		hints.ai_socktype = SOCK_STREAM;
+	}
+
+	if (!node && !service) {
+		return EAI_NONAME;
+	}
+
+	if (node) {
+		if (node[0] == '\0') {
+			return getaddr_info_single_addr(service,
+					INADDR_ANY,
+					&hints,
+					res);
+		} else if (hints.ai_flags & AI_NUMERICHOST) {
+			struct in_addr ip;
+			if (inet_pton(AF_INET, node, &ip) <= 0)
+				return EAI_FAIL;
+			return getaddr_info_single_addr(service,
+					ntohl(ip.s_addr),
+					&hints,
+					res);
+		} else {
+			return getaddr_info_name(node,
+						service,
+						&hints,
+						res);
+		}
+	} else if (hints.ai_flags & AI_PASSIVE) {
+		return getaddr_info_single_addr(service,
+					INADDR_ANY,
+					&hints,
+					res);
+	}
+	return getaddr_info_single_addr(service,
+					INADDR_LOOPBACK,
+					&hints,
+					res);
+}
+
+
+void freeaddrinfo(struct addrinfo *res)
+{
+	struct addrinfo *next = NULL;
+
+	for (;res; res = next) {
+		next = res->ai_next;
+		if (res->ai_canonname) {
+			free(res->ai_canonname);
+		}
+		if (res->ai_addr) {
+			free(res->ai_addr);
+		}
+		free(res);
+	}
+}
+
+
+const char *gai_strerror(int errcode)
+{
+#ifdef HAVE_HSTRERROR
+	int			hcode;
+
+	switch (errcode)
+	{
+		case EAI_NONAME:
+			hcode = HOST_NOT_FOUND;
+			break;
+		case EAI_AGAIN:
+			hcode = TRY_AGAIN;
+			break;
+		case EAI_FAIL:
+		default:
+			hcode = NO_RECOVERY;
+			break;
+	}
+
+	return hstrerror(hcode);
+#else							/* !HAVE_HSTRERROR */
+
+	switch (errcode)
+	{
+		case EAI_NONAME:
+			return "Unknown host";
+		case EAI_AGAIN:
+			return "Host name lookup failure";
+#ifdef EAI_BADFLAGS
+		case EAI_BADFLAGS:
+			return "Invalid argument";
+#endif
+#ifdef EAI_FAMILY
+		case EAI_FAMILY:
+			return "Address family not supported";
+#endif
+#ifdef EAI_MEMORY
+		case EAI_MEMORY:
+			return "Not enough memory";
+#endif
+#ifdef EAI_NODATA
+		case EAI_NODATA:
+			return "No host data of that type was found";
+#endif
+#ifdef EAI_SERVICE
+		case EAI_SERVICE:
+			return "Class type not found";
+#endif
+#ifdef EAI_SOCKTYPE
+		case EAI_SOCKTYPE:
+			return "Socket type not supported";
+#endif
+		default:
+			return "Unknown server error";
+	}
+#endif   /* HAVE_HSTRERROR */
+}
+
+static int gethostnameinfo(const struct sockaddr *sa,
+			char *node,
+			size_t nodelen,
+			int flags)
+{
+	int ret = -1;
+	char *p = NULL;
+
+	if (!(flags & NI_NUMERICHOST)) {
+		struct hostent *hp = gethostbyaddr(
+				(void *)&((struct sockaddr_in *)sa)->sin_addr,
+				sizeof (struct in_addr),
+				sa->sa_family);
+		ret = check_hostent_err(hp);
+		if (ret == 0) {
+			/* Name looked up successfully. */
+			ret = snprintf(node, nodelen, "%s", hp->h_name);
+			if (ret < 0 || (size_t)ret >= nodelen) {
+				return EAI_MEMORY;
+			}
+			if (flags & NI_NOFQDN) {
+				p = strchr(node,'.');
+				if (p) {
+					*p = '\0';
+				}
+			}
+			return 0;
+		}
+
+		if (flags & NI_NAMEREQD) {
+			/* If we require a name and didn't get one,
+			 * automatically fail. */
+			return ret;
+		}
+		/* Otherwise just fall into the numeric host code... */
+	}
+	p = inet_ntoa(((struct sockaddr_in *)sa)->sin_addr);
+	ret = snprintf(node, nodelen, "%s", p);
+	if (ret < 0 || (size_t)ret >= nodelen) {
+		return EAI_MEMORY;
+	}
+	return 0;
+}
+
+static int getservicenameinfo(const struct sockaddr *sa,
+			char *service,
+			size_t servicelen,
+			int flags)
+{
+	int ret = -1;
+	int port = ntohs(((struct sockaddr_in *)sa)->sin_port);
+
+	if (!(flags & NI_NUMERICSERV)) {
+		struct servent *se = getservbyport(
+				port,
+				(flags & NI_DGRAM) ? "udp" : "tcp");
+		if (se && se->s_name) {
+			/* Service name looked up successfully. */
+			ret = snprintf(service, servicelen, "%s", se->s_name);
+			if (ret < 0 || (size_t)ret >= servicelen) {
+				return EAI_MEMORY;
+			}
+			return 0;
+		}
+		/* Otherwise just fall into the numeric service code... */
+	}
+	ret = snprintf(service, servicelen, "%d", port);
+	if (ret < 0 || (size_t)ret >= servicelen) {
+		return EAI_MEMORY;
+	}
+	return 0;
+}
+
+/*
+ * Convert an ipv4 address to a hostname.
+ *
+ * Bugs:	- No IPv6 support.
+ */
+int getnameinfo(const struct sockaddr *sa, socklen_t salen,
+			char *node, size_t nodelen,
+			char *service, size_t servicelen, int flags)
+{
+
+	/* Invalid arguments. */
+	if (sa == NULL || (node == NULL && service == NULL)) {
+		return EAI_FAIL;
+	}
+
+	if (sa->sa_family != AF_INET) {
+		return EAI_FAIL;
+	}
+
+	if (salen < (socklen_t)sizeof (struct sockaddr_in)) {
+		return EAI_FAIL;
+	}
+
+	if (node) {
+		int ret = gethostnameinfo(sa, node, nodelen, flags);
+		if (ret)
+			return ret;
+	}
+
+	if (service) {
+		return getservicenameinfo(sa, service, servicelen, flags);
+	}
+	return 0;
+}

lib/getpass.c

@@ -0,0 +1,72 @@
+/*
+ * An implementation of getpass for systems that lack one.
+ *
+ * Copyright (C) 2013 Roman Donchenko
+ *
+ * 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
+ * the Free Software Foundation; either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, visit the http://fsf.org website.
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <termios.h>
+
+#include "rsync.h"
+
+char *getpass(const char *prompt)
+{
+	static char password[256];
+
+	BOOL tty_changed = False, read_success;
+	struct termios tty_old, tty_new;
+	FILE *in = stdin, *out = stderr;
+	FILE *tty = fopen("/dev/tty", "w+");
+
+	if (tty)
+		in = out = tty;
+
+	if (tcgetattr(fileno(in), &tty_old) == 0) {
+		tty_new = tty_old;
+		tty_new.c_lflag &= ~(ECHO | ISIG);
+
+		if (tcsetattr(fileno(in), TCSAFLUSH, &tty_new) == 0)
+			tty_changed = True;
+	}
+
+	if (!tty_changed)
+		fputs("(WARNING: will be visible) ", out);
+	fputs(prompt, out);
+	fflush(out);
+
+	read_success = fgets(password, sizeof password, in) != NULL;
+
+	/* Print the newline that hasn't been echoed. */
+	fputc('\n', out);
+
+	if (tty_changed)
+		tcsetattr(fileno(in), TCSAFLUSH, &tty_old);
+
+	if (tty)
+		fclose(tty);
+
+	if (read_success) {
+		/* Remove the trailing newline. */
+		size_t password_len = strlen(password);
+		if (password_len && password[password_len - 1] == '\n')
+			password[password_len - 1] = '\0';
+
+		return password;
+	}
+
+	return NULL;
+}

lib/md5-asm-x86_64.s

@@ -0,0 +1,693 @@
+/*
+ * x86-64 optimized assembler MD5 implementation
+ *
+ * Author: Marc Bevand, 2004
+ *
+ * This code was placed in the public domain by the author. The original
+ * publication can be found at:
+ *
+ * https://www.zorinaq.com/papers/md5-amd64.html
+ */
+/*
+ * No modifications were made aside from changing the function and file names.
+ * The MD5_CTX structure as expected here (from OpenSSL) is binary compatible
+ * with the md_context used by rsync, for the fields accessed.
+ *
+ * Benchmarks (in MB/s)            C     ASM
+ * - Intel Atom D2700            302     334
+ * - Intel i7-7700hq             351     376
+ * - AMD ThreadRipper 2950x      728     784
+ *
+ * The original code was also incorporated into OpenSSL. It has since been
+ * modified there. Those changes have not been made here due to licensing
+ * incompatibilities. Benchmarks of those changes on the above CPUs did not
+ * show any significant difference in performance, though.
+ */
+
+.text
+.align 16
+
+.globl md5_process_asm
+.type md5_process_asm,@function
+md5_process_asm:
+	push	%rbp
+	push	%rbx
+	push	%r12
+	push	%r13			# not really useful (r13 is unused)
+	push	%r14
+	push	%r15
+
+	# rdi = arg #1 (ctx, MD5_CTX pointer)
+	# rsi = arg #2 (ptr, data pointer)
+	# rdx = arg #3 (nbr, number of 16-word blocks to process)
+	mov	%rdi,		%rbp	# rbp = ctx
+	shl	$6,		%rdx	# rdx = nbr in bytes
+	lea	(%rsi,%rdx),	%rdi	# rdi = end
+	mov	0*4(%rbp),	%eax	# eax = ctx->A
+	mov	1*4(%rbp),	%ebx	# ebx = ctx->B
+	mov	2*4(%rbp),	%ecx	# ecx = ctx->C
+	mov	3*4(%rbp),	%edx	# edx = ctx->D
+	# end is 'rdi'
+	# ptr is 'rsi'
+	# A is 'eax'
+	# B is 'ebx'
+	# C is 'ecx'
+	# D is 'edx'
+
+	cmp	%rdi,		%rsi		# cmp end with ptr
+	je	1f				# jmp if ptr == end
+
+	# BEGIN of loop over 16-word blocks
+2:	# save old values of A, B, C, D
+	mov	%eax,		%r8d
+	mov	%ebx,		%r9d
+	mov	%ecx,		%r14d
+	mov	%edx,		%r15d
+ mov	0*4(%rsi),	%r10d		/* (NEXT STEP) X[0] */
+ mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	xor	%ecx,		%r11d		/* y ^ ... */
+	lea	-680876936(%eax,%r10d),%eax		/* Const + dst + ... */
+	and	%ebx,		%r11d		/* x & ... */
+	xor	%edx,		%r11d		/* z ^ ... */
+	mov	1*4(%rsi),%r10d		/* (NEXT STEP) X[1] */
+	add	%r11d,		%eax		/* dst += ... */
+	rol	$7,		%eax		/* dst <<< s */
+	mov	%ecx,		%r11d		/* (NEXT STEP) z' = %ecx */
+	add	%ebx,		%eax		/* dst += x */
+	xor	%ebx,		%r11d		/* y ^ ... */
+	lea	-389564586(%edx,%r10d),%edx		/* Const + dst + ... */
+	and	%eax,		%r11d		/* x & ... */
+	xor	%ecx,		%r11d		/* z ^ ... */
+	mov	2*4(%rsi),%r10d		/* (NEXT STEP) X[2] */
+	add	%r11d,		%edx		/* dst += ... */
+	rol	$12,		%edx		/* dst <<< s */
+	mov	%ebx,		%r11d		/* (NEXT STEP) z' = %ebx */
+	add	%eax,		%edx		/* dst += x */
+	xor	%eax,		%r11d		/* y ^ ... */
+	lea	606105819(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	and	%edx,		%r11d		/* x & ... */
+	xor	%ebx,		%r11d		/* z ^ ... */
+	mov	3*4(%rsi),%r10d		/* (NEXT STEP) X[3] */
+	add	%r11d,		%ecx		/* dst += ... */
+	rol	$17,		%ecx		/* dst <<< s */
+	mov	%eax,		%r11d		/* (NEXT STEP) z' = %eax */
+	add	%edx,		%ecx		/* dst += x */
+	xor	%edx,		%r11d		/* y ^ ... */
+	lea	-1044525330(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	and	%ecx,		%r11d		/* x & ... */
+	xor	%eax,		%r11d		/* z ^ ... */
+	mov	4*4(%rsi),%r10d		/* (NEXT STEP) X[4] */
+	add	%r11d,		%ebx		/* dst += ... */
+	rol	$22,		%ebx		/* dst <<< s */
+	mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	add	%ecx,		%ebx		/* dst += x */
+	xor	%ecx,		%r11d		/* y ^ ... */
+	lea	-176418897(%eax,%r10d),%eax		/* Const + dst + ... */
+	and	%ebx,		%r11d		/* x & ... */
+	xor	%edx,		%r11d		/* z ^ ... */
+	mov	5*4(%rsi),%r10d		/* (NEXT STEP) X[5] */
+	add	%r11d,		%eax		/* dst += ... */
+	rol	$7,		%eax		/* dst <<< s */
+	mov	%ecx,		%r11d		/* (NEXT STEP) z' = %ecx */
+	add	%ebx,		%eax		/* dst += x */
+	xor	%ebx,		%r11d		/* y ^ ... */
+	lea	1200080426(%edx,%r10d),%edx		/* Const + dst + ... */
+	and	%eax,		%r11d		/* x & ... */
+	xor	%ecx,		%r11d		/* z ^ ... */
+	mov	6*4(%rsi),%r10d		/* (NEXT STEP) X[6] */
+	add	%r11d,		%edx		/* dst += ... */
+	rol	$12,		%edx		/* dst <<< s */
+	mov	%ebx,		%r11d		/* (NEXT STEP) z' = %ebx */
+	add	%eax,		%edx		/* dst += x */
+	xor	%eax,		%r11d		/* y ^ ... */
+	lea	-1473231341(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	and	%edx,		%r11d		/* x & ... */
+	xor	%ebx,		%r11d		/* z ^ ... */
+	mov	7*4(%rsi),%r10d		/* (NEXT STEP) X[7] */
+	add	%r11d,		%ecx		/* dst += ... */
+	rol	$17,		%ecx		/* dst <<< s */
+	mov	%eax,		%r11d		/* (NEXT STEP) z' = %eax */
+	add	%edx,		%ecx		/* dst += x */
+	xor	%edx,		%r11d		/* y ^ ... */
+	lea	-45705983(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	and	%ecx,		%r11d		/* x & ... */
+	xor	%eax,		%r11d		/* z ^ ... */
+	mov	8*4(%rsi),%r10d		/* (NEXT STEP) X[8] */
+	add	%r11d,		%ebx		/* dst += ... */
+	rol	$22,		%ebx		/* dst <<< s */
+	mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	add	%ecx,		%ebx		/* dst += x */
+	xor	%ecx,		%r11d		/* y ^ ... */
+	lea	1770035416(%eax,%r10d),%eax		/* Const + dst + ... */
+	and	%ebx,		%r11d		/* x & ... */
+	xor	%edx,		%r11d		/* z ^ ... */
+	mov	9*4(%rsi),%r10d		/* (NEXT STEP) X[9] */
+	add	%r11d,		%eax		/* dst += ... */
+	rol	$7,		%eax		/* dst <<< s */
+	mov	%ecx,		%r11d		/* (NEXT STEP) z' = %ecx */
+	add	%ebx,		%eax		/* dst += x */
+	xor	%ebx,		%r11d		/* y ^ ... */
+	lea	-1958414417(%edx,%r10d),%edx		/* Const + dst + ... */
+	and	%eax,		%r11d		/* x & ... */
+	xor	%ecx,		%r11d		/* z ^ ... */
+	mov	10*4(%rsi),%r10d		/* (NEXT STEP) X[10] */
+	add	%r11d,		%edx		/* dst += ... */
+	rol	$12,		%edx		/* dst <<< s */
+	mov	%ebx,		%r11d		/* (NEXT STEP) z' = %ebx */
+	add	%eax,		%edx		/* dst += x */
+	xor	%eax,		%r11d		/* y ^ ... */
+	lea	-42063(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	and	%edx,		%r11d		/* x & ... */
+	xor	%ebx,		%r11d		/* z ^ ... */
+	mov	11*4(%rsi),%r10d		/* (NEXT STEP) X[11] */
+	add	%r11d,		%ecx		/* dst += ... */
+	rol	$17,		%ecx		/* dst <<< s */
+	mov	%eax,		%r11d		/* (NEXT STEP) z' = %eax */
+	add	%edx,		%ecx		/* dst += x */
+	xor	%edx,		%r11d		/* y ^ ... */
+	lea	-1990404162(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	and	%ecx,		%r11d		/* x & ... */
+	xor	%eax,		%r11d		/* z ^ ... */
+	mov	12*4(%rsi),%r10d		/* (NEXT STEP) X[12] */
+	add	%r11d,		%ebx		/* dst += ... */
+	rol	$22,		%ebx		/* dst <<< s */
+	mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	add	%ecx,		%ebx		/* dst += x */
+	xor	%ecx,		%r11d		/* y ^ ... */
+	lea	1804603682(%eax,%r10d),%eax		/* Const + dst + ... */
+	and	%ebx,		%r11d		/* x & ... */
+	xor	%edx,		%r11d		/* z ^ ... */
+	mov	13*4(%rsi),%r10d		/* (NEXT STEP) X[13] */
+	add	%r11d,		%eax		/* dst += ... */
+	rol	$7,		%eax		/* dst <<< s */
+	mov	%ecx,		%r11d		/* (NEXT STEP) z' = %ecx */
+	add	%ebx,		%eax		/* dst += x */
+	xor	%ebx,		%r11d		/* y ^ ... */
+	lea	-40341101(%edx,%r10d),%edx		/* Const + dst + ... */
+	and	%eax,		%r11d		/* x & ... */
+	xor	%ecx,		%r11d		/* z ^ ... */
+	mov	14*4(%rsi),%r10d		/* (NEXT STEP) X[14] */
+	add	%r11d,		%edx		/* dst += ... */
+	rol	$12,		%edx		/* dst <<< s */
+	mov	%ebx,		%r11d		/* (NEXT STEP) z' = %ebx */
+	add	%eax,		%edx		/* dst += x */
+	xor	%eax,		%r11d		/* y ^ ... */
+	lea	-1502002290(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	and	%edx,		%r11d		/* x & ... */
+	xor	%ebx,		%r11d		/* z ^ ... */
+	mov	15*4(%rsi),%r10d		/* (NEXT STEP) X[15] */
+	add	%r11d,		%ecx		/* dst += ... */
+	rol	$17,		%ecx		/* dst <<< s */
+	mov	%eax,		%r11d		/* (NEXT STEP) z' = %eax */
+	add	%edx,		%ecx		/* dst += x */
+	xor	%edx,		%r11d		/* y ^ ... */
+	lea	1236535329(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	and	%ecx,		%r11d		/* x & ... */
+	xor	%eax,		%r11d		/* z ^ ... */
+	mov	0*4(%rsi),%r10d		/* (NEXT STEP) X[0] */
+	add	%r11d,		%ebx		/* dst += ... */
+	rol	$22,		%ebx		/* dst <<< s */
+	mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	add	%ecx,		%ebx		/* dst += x */
+ mov	1*4(%rsi),	%r10d		/* (NEXT STEP) X[1] */
+ mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+ mov	%edx,		%r12d		/* (NEXT STEP) z' = %edx */
+	not	%r11d				/* not z */
+	lea	-165796510(%eax,%r10d),%eax		/* Const + dst + ... */
+	and	%ebx,		%r12d		/* x & z */
+	and	%ecx,		%r11d		/* y & (not z) */
+	mov	6*4(%rsi),%r10d		/* (NEXT STEP) X[6] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%ecx,		%r11d		/* (NEXT STEP) z' = %ecx */
+	add	%r12d,		%eax		/* dst += ... */
+	mov	%ecx,		%r12d		/* (NEXT STEP) z' = %ecx */
+	rol	$5,		%eax		/* dst <<< s */
+	add	%ebx,		%eax		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-1069501632(%edx,%r10d),%edx		/* Const + dst + ... */
+	and	%eax,		%r12d		/* x & z */
+	and	%ebx,		%r11d		/* y & (not z) */
+	mov	11*4(%rsi),%r10d		/* (NEXT STEP) X[11] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%ebx,		%r11d		/* (NEXT STEP) z' = %ebx */
+	add	%r12d,		%edx		/* dst += ... */
+	mov	%ebx,		%r12d		/* (NEXT STEP) z' = %ebx */
+	rol	$9,		%edx		/* dst <<< s */
+	add	%eax,		%edx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	643717713(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	and	%edx,		%r12d		/* x & z */
+	and	%eax,		%r11d		/* y & (not z) */
+	mov	0*4(%rsi),%r10d		/* (NEXT STEP) X[0] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%eax,		%r11d		/* (NEXT STEP) z' = %eax */
+	add	%r12d,		%ecx		/* dst += ... */
+	mov	%eax,		%r12d		/* (NEXT STEP) z' = %eax */
+	rol	$14,		%ecx		/* dst <<< s */
+	add	%edx,		%ecx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-373897302(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	and	%ecx,		%r12d		/* x & z */
+	and	%edx,		%r11d		/* y & (not z) */
+	mov	5*4(%rsi),%r10d		/* (NEXT STEP) X[5] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	add	%r12d,		%ebx		/* dst += ... */
+	mov	%edx,		%r12d		/* (NEXT STEP) z' = %edx */
+	rol	$20,		%ebx		/* dst <<< s */
+	add	%ecx,		%ebx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-701558691(%eax,%r10d),%eax		/* Const + dst + ... */
+	and	%ebx,		%r12d		/* x & z */
+	and	%ecx,		%r11d		/* y & (not z) */
+	mov	10*4(%rsi),%r10d		/* (NEXT STEP) X[10] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%ecx,		%r11d		/* (NEXT STEP) z' = %ecx */
+	add	%r12d,		%eax		/* dst += ... */
+	mov	%ecx,		%r12d		/* (NEXT STEP) z' = %ecx */
+	rol	$5,		%eax		/* dst <<< s */
+	add	%ebx,		%eax		/* dst += x */
+	not	%r11d				/* not z */
+	lea	38016083(%edx,%r10d),%edx		/* Const + dst + ... */
+	and	%eax,		%r12d		/* x & z */
+	and	%ebx,		%r11d		/* y & (not z) */
+	mov	15*4(%rsi),%r10d		/* (NEXT STEP) X[15] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%ebx,		%r11d		/* (NEXT STEP) z' = %ebx */
+	add	%r12d,		%edx		/* dst += ... */
+	mov	%ebx,		%r12d		/* (NEXT STEP) z' = %ebx */
+	rol	$9,		%edx		/* dst <<< s */
+	add	%eax,		%edx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-660478335(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	and	%edx,		%r12d		/* x & z */
+	and	%eax,		%r11d		/* y & (not z) */
+	mov	4*4(%rsi),%r10d		/* (NEXT STEP) X[4] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%eax,		%r11d		/* (NEXT STEP) z' = %eax */
+	add	%r12d,		%ecx		/* dst += ... */
+	mov	%eax,		%r12d		/* (NEXT STEP) z' = %eax */
+	rol	$14,		%ecx		/* dst <<< s */
+	add	%edx,		%ecx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-405537848(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	and	%ecx,		%r12d		/* x & z */
+	and	%edx,		%r11d		/* y & (not z) */
+	mov	9*4(%rsi),%r10d		/* (NEXT STEP) X[9] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	add	%r12d,		%ebx		/* dst += ... */
+	mov	%edx,		%r12d		/* (NEXT STEP) z' = %edx */
+	rol	$20,		%ebx		/* dst <<< s */
+	add	%ecx,		%ebx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	568446438(%eax,%r10d),%eax		/* Const + dst + ... */
+	and	%ebx,		%r12d		/* x & z */
+	and	%ecx,		%r11d		/* y & (not z) */
+	mov	14*4(%rsi),%r10d		/* (NEXT STEP) X[14] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%ecx,		%r11d		/* (NEXT STEP) z' = %ecx */
+	add	%r12d,		%eax		/* dst += ... */
+	mov	%ecx,		%r12d		/* (NEXT STEP) z' = %ecx */
+	rol	$5,		%eax		/* dst <<< s */
+	add	%ebx,		%eax		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-1019803690(%edx,%r10d),%edx		/* Const + dst + ... */
+	and	%eax,		%r12d		/* x & z */
+	and	%ebx,		%r11d		/* y & (not z) */
+	mov	3*4(%rsi),%r10d		/* (NEXT STEP) X[3] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%ebx,		%r11d		/* (NEXT STEP) z' = %ebx */
+	add	%r12d,		%edx		/* dst += ... */
+	mov	%ebx,		%r12d		/* (NEXT STEP) z' = %ebx */
+	rol	$9,		%edx		/* dst <<< s */
+	add	%eax,		%edx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-187363961(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	and	%edx,		%r12d		/* x & z */
+	and	%eax,		%r11d		/* y & (not z) */
+	mov	8*4(%rsi),%r10d		/* (NEXT STEP) X[8] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%eax,		%r11d		/* (NEXT STEP) z' = %eax */
+	add	%r12d,		%ecx		/* dst += ... */
+	mov	%eax,		%r12d		/* (NEXT STEP) z' = %eax */
+	rol	$14,		%ecx		/* dst <<< s */
+	add	%edx,		%ecx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	1163531501(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	and	%ecx,		%r12d		/* x & z */
+	and	%edx,		%r11d		/* y & (not z) */
+	mov	13*4(%rsi),%r10d		/* (NEXT STEP) X[13] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	add	%r12d,		%ebx		/* dst += ... */
+	mov	%edx,		%r12d		/* (NEXT STEP) z' = %edx */
+	rol	$20,		%ebx		/* dst <<< s */
+	add	%ecx,		%ebx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-1444681467(%eax,%r10d),%eax		/* Const + dst + ... */
+	and	%ebx,		%r12d		/* x & z */
+	and	%ecx,		%r11d		/* y & (not z) */
+	mov	2*4(%rsi),%r10d		/* (NEXT STEP) X[2] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%ecx,		%r11d		/* (NEXT STEP) z' = %ecx */
+	add	%r12d,		%eax		/* dst += ... */
+	mov	%ecx,		%r12d		/* (NEXT STEP) z' = %ecx */
+	rol	$5,		%eax		/* dst <<< s */
+	add	%ebx,		%eax		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-51403784(%edx,%r10d),%edx		/* Const + dst + ... */
+	and	%eax,		%r12d		/* x & z */
+	and	%ebx,		%r11d		/* y & (not z) */
+	mov	7*4(%rsi),%r10d		/* (NEXT STEP) X[7] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%ebx,		%r11d		/* (NEXT STEP) z' = %ebx */
+	add	%r12d,		%edx		/* dst += ... */
+	mov	%ebx,		%r12d		/* (NEXT STEP) z' = %ebx */
+	rol	$9,		%edx		/* dst <<< s */
+	add	%eax,		%edx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	1735328473(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	and	%edx,		%r12d		/* x & z */
+	and	%eax,		%r11d		/* y & (not z) */
+	mov	12*4(%rsi),%r10d		/* (NEXT STEP) X[12] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%eax,		%r11d		/* (NEXT STEP) z' = %eax */
+	add	%r12d,		%ecx		/* dst += ... */
+	mov	%eax,		%r12d		/* (NEXT STEP) z' = %eax */
+	rol	$14,		%ecx		/* dst <<< s */
+	add	%edx,		%ecx		/* dst += x */
+	not	%r11d				/* not z */
+	lea	-1926607734(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	and	%ecx,		%r12d		/* x & z */
+	and	%edx,		%r11d		/* y & (not z) */
+	mov	0*4(%rsi),%r10d		/* (NEXT STEP) X[0] */
+	or	%r11d,		%r12d		/* (y & (not z)) | (x & z) */
+	mov	%edx,		%r11d		/* (NEXT STEP) z' = %edx */
+	add	%r12d,		%ebx		/* dst += ... */
+	mov	%edx,		%r12d		/* (NEXT STEP) z' = %edx */
+	rol	$20,		%ebx		/* dst <<< s */
+	add	%ecx,		%ebx		/* dst += x */
+ mov	5*4(%rsi),	%r10d		/* (NEXT STEP) X[5] */
+ mov	%ecx,		%r11d		/* (NEXT STEP) y' = %ecx */
+	lea	-378558(%eax,%r10d),%eax		/* Const + dst + ... */
+	mov	8*4(%rsi),%r10d		/* (NEXT STEP) X[8] */
+	xor	%edx,		%r11d		/* z ^ ... */
+	xor	%ebx,		%r11d		/* x ^ ... */
+	add	%r11d,		%eax		/* dst += ... */
+	rol	$4,		%eax		/* dst <<< s */
+	mov	%ebx,		%r11d		/* (NEXT STEP) y' = %ebx */
+	add	%ebx,		%eax		/* dst += x */
+	lea	-2022574463(%edx,%r10d),%edx		/* Const + dst + ... */
+	mov	11*4(%rsi),%r10d		/* (NEXT STEP) X[11] */
+	xor	%ecx,		%r11d		/* z ^ ... */
+	xor	%eax,		%r11d		/* x ^ ... */
+	add	%r11d,		%edx		/* dst += ... */
+	rol	$11,		%edx		/* dst <<< s */
+	mov	%eax,		%r11d		/* (NEXT STEP) y' = %eax */
+	add	%eax,		%edx		/* dst += x */
+	lea	1839030562(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	mov	14*4(%rsi),%r10d		/* (NEXT STEP) X[14] */
+	xor	%ebx,		%r11d		/* z ^ ... */
+	xor	%edx,		%r11d		/* x ^ ... */
+	add	%r11d,		%ecx		/* dst += ... */
+	rol	$16,		%ecx		/* dst <<< s */
+	mov	%edx,		%r11d		/* (NEXT STEP) y' = %edx */
+	add	%edx,		%ecx		/* dst += x */
+	lea	-35309556(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	mov	1*4(%rsi),%r10d		/* (NEXT STEP) X[1] */
+	xor	%eax,		%r11d		/* z ^ ... */
+	xor	%ecx,		%r11d		/* x ^ ... */
+	add	%r11d,		%ebx		/* dst += ... */
+	rol	$23,		%ebx		/* dst <<< s */
+	mov	%ecx,		%r11d		/* (NEXT STEP) y' = %ecx */
+	add	%ecx,		%ebx		/* dst += x */
+	lea	-1530992060(%eax,%r10d),%eax		/* Const + dst + ... */
+	mov	4*4(%rsi),%r10d		/* (NEXT STEP) X[4] */
+	xor	%edx,		%r11d		/* z ^ ... */
+	xor	%ebx,		%r11d		/* x ^ ... */
+	add	%r11d,		%eax		/* dst += ... */
+	rol	$4,		%eax		/* dst <<< s */
+	mov	%ebx,		%r11d		/* (NEXT STEP) y' = %ebx */
+	add	%ebx,		%eax		/* dst += x */
+	lea	1272893353(%edx,%r10d),%edx		/* Const + dst + ... */
+	mov	7*4(%rsi),%r10d		/* (NEXT STEP) X[7] */
+	xor	%ecx,		%r11d		/* z ^ ... */
+	xor	%eax,		%r11d		/* x ^ ... */
+	add	%r11d,		%edx		/* dst += ... */
+	rol	$11,		%edx		/* dst <<< s */
+	mov	%eax,		%r11d		/* (NEXT STEP) y' = %eax */
+	add	%eax,		%edx		/* dst += x */
+	lea	-155497632(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	mov	10*4(%rsi),%r10d		/* (NEXT STEP) X[10] */
+	xor	%ebx,		%r11d		/* z ^ ... */
+	xor	%edx,		%r11d		/* x ^ ... */
+	add	%r11d,		%ecx		/* dst += ... */
+	rol	$16,		%ecx		/* dst <<< s */
+	mov	%edx,		%r11d		/* (NEXT STEP) y' = %edx */
+	add	%edx,		%ecx		/* dst += x */
+	lea	-1094730640(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	mov	13*4(%rsi),%r10d		/* (NEXT STEP) X[13] */
+	xor	%eax,		%r11d		/* z ^ ... */
+	xor	%ecx,		%r11d		/* x ^ ... */
+	add	%r11d,		%ebx		/* dst += ... */
+	rol	$23,		%ebx		/* dst <<< s */
+	mov	%ecx,		%r11d		/* (NEXT STEP) y' = %ecx */
+	add	%ecx,		%ebx		/* dst += x */
+	lea	681279174(%eax,%r10d),%eax		/* Const + dst + ... */
+	mov	0*4(%rsi),%r10d		/* (NEXT STEP) X[0] */
+	xor	%edx,		%r11d		/* z ^ ... */
+	xor	%ebx,		%r11d		/* x ^ ... */
+	add	%r11d,		%eax		/* dst += ... */
+	rol	$4,		%eax		/* dst <<< s */
+	mov	%ebx,		%r11d		/* (NEXT STEP) y' = %ebx */
+	add	%ebx,		%eax		/* dst += x */
+	lea	-358537222(%edx,%r10d),%edx		/* Const + dst + ... */
+	mov	3*4(%rsi),%r10d		/* (NEXT STEP) X[3] */
+	xor	%ecx,		%r11d		/* z ^ ... */
+	xor	%eax,		%r11d		/* x ^ ... */
+	add	%r11d,		%edx		/* dst += ... */
+	rol	$11,		%edx		/* dst <<< s */
+	mov	%eax,		%r11d		/* (NEXT STEP) y' = %eax */
+	add	%eax,		%edx		/* dst += x */
+	lea	-722521979(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	mov	6*4(%rsi),%r10d		/* (NEXT STEP) X[6] */
+	xor	%ebx,		%r11d		/* z ^ ... */
+	xor	%edx,		%r11d		/* x ^ ... */
+	add	%r11d,		%ecx		/* dst += ... */
+	rol	$16,		%ecx		/* dst <<< s */
+	mov	%edx,		%r11d		/* (NEXT STEP) y' = %edx */
+	add	%edx,		%ecx		/* dst += x */
+	lea	76029189(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	mov	9*4(%rsi),%r10d		/* (NEXT STEP) X[9] */
+	xor	%eax,		%r11d		/* z ^ ... */
+	xor	%ecx,		%r11d		/* x ^ ... */
+	add	%r11d,		%ebx		/* dst += ... */
+	rol	$23,		%ebx		/* dst <<< s */
+	mov	%ecx,		%r11d		/* (NEXT STEP) y' = %ecx */
+	add	%ecx,		%ebx		/* dst += x */
+	lea	-640364487(%eax,%r10d),%eax		/* Const + dst + ... */
+	mov	12*4(%rsi),%r10d		/* (NEXT STEP) X[12] */
+	xor	%edx,		%r11d		/* z ^ ... */
+	xor	%ebx,		%r11d		/* x ^ ... */
+	add	%r11d,		%eax		/* dst += ... */
+	rol	$4,		%eax		/* dst <<< s */
+	mov	%ebx,		%r11d		/* (NEXT STEP) y' = %ebx */
+	add	%ebx,		%eax		/* dst += x */
+	lea	-421815835(%edx,%r10d),%edx		/* Const + dst + ... */
+	mov	15*4(%rsi),%r10d		/* (NEXT STEP) X[15] */
+	xor	%ecx,		%r11d		/* z ^ ... */
+	xor	%eax,		%r11d		/* x ^ ... */
+	add	%r11d,		%edx		/* dst += ... */
+	rol	$11,		%edx		/* dst <<< s */
+	mov	%eax,		%r11d		/* (NEXT STEP) y' = %eax */
+	add	%eax,		%edx		/* dst += x */
+	lea	530742520(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	mov	2*4(%rsi),%r10d		/* (NEXT STEP) X[2] */
+	xor	%ebx,		%r11d		/* z ^ ... */
+	xor	%edx,		%r11d		/* x ^ ... */
+	add	%r11d,		%ecx		/* dst += ... */
+	rol	$16,		%ecx		/* dst <<< s */
+	mov	%edx,		%r11d		/* (NEXT STEP) y' = %edx */
+	add	%edx,		%ecx		/* dst += x */
+	lea	-995338651(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	mov	0*4(%rsi),%r10d		/* (NEXT STEP) X[0] */
+	xor	%eax,		%r11d		/* z ^ ... */
+	xor	%ecx,		%r11d		/* x ^ ... */
+	add	%r11d,		%ebx		/* dst += ... */
+	rol	$23,		%ebx		/* dst <<< s */
+	mov	%ecx,		%r11d		/* (NEXT STEP) y' = %ecx */
+	add	%ecx,		%ebx		/* dst += x */
+ mov	0*4(%rsi),	%r10d		/* (NEXT STEP) X[0] */
+ mov	$0xffffffff,	%r11d
+ xor	%edx,		%r11d		/* (NEXT STEP) not z' = not %edx*/
+	lea	-198630844(%eax,%r10d),%eax		/* Const + dst + ... */
+	or	%ebx,		%r11d		/* x | ... */
+	xor	%ecx,		%r11d		/* y ^ ... */
+	add	%r11d,		%eax		/* dst += ... */
+	mov	7*4(%rsi),%r10d		/* (NEXT STEP) X[7] */
+	mov	$0xffffffff,	%r11d
+	rol	$6,		%eax		/* dst <<< s */
+	xor	%ecx,		%r11d		/* (NEXT STEP) not z' = not %ecx */
+	add	%ebx,		%eax		/* dst += x */
+	lea	1126891415(%edx,%r10d),%edx		/* Const + dst + ... */
+	or	%eax,		%r11d		/* x | ... */
+	xor	%ebx,		%r11d		/* y ^ ... */
+	add	%r11d,		%edx		/* dst += ... */
+	mov	14*4(%rsi),%r10d		/* (NEXT STEP) X[14] */
+	mov	$0xffffffff,	%r11d
+	rol	$10,		%edx		/* dst <<< s */
+	xor	%ebx,		%r11d		/* (NEXT STEP) not z' = not %ebx */
+	add	%eax,		%edx		/* dst += x */
+	lea	-1416354905(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	or	%edx,		%r11d		/* x | ... */
+	xor	%eax,		%r11d		/* y ^ ... */
+	add	%r11d,		%ecx		/* dst += ... */
+	mov	5*4(%rsi),%r10d		/* (NEXT STEP) X[5] */
+	mov	$0xffffffff,	%r11d
+	rol	$15,		%ecx		/* dst <<< s */
+	xor	%eax,		%r11d		/* (NEXT STEP) not z' = not %eax */
+	add	%edx,		%ecx		/* dst += x */
+	lea	-57434055(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	or	%ecx,		%r11d		/* x | ... */
+	xor	%edx,		%r11d		/* y ^ ... */
+	add	%r11d,		%ebx		/* dst += ... */
+	mov	12*4(%rsi),%r10d		/* (NEXT STEP) X[12] */
+	mov	$0xffffffff,	%r11d
+	rol	$21,		%ebx		/* dst <<< s */
+	xor	%edx,		%r11d		/* (NEXT STEP) not z' = not %edx */
+	add	%ecx,		%ebx		/* dst += x */
+	lea	1700485571(%eax,%r10d),%eax		/* Const + dst + ... */
+	or	%ebx,		%r11d		/* x | ... */
+	xor	%ecx,		%r11d		/* y ^ ... */
+	add	%r11d,		%eax		/* dst += ... */
+	mov	3*4(%rsi),%r10d		/* (NEXT STEP) X[3] */
+	mov	$0xffffffff,	%r11d
+	rol	$6,		%eax		/* dst <<< s */
+	xor	%ecx,		%r11d		/* (NEXT STEP) not z' = not %ecx */
+	add	%ebx,		%eax		/* dst += x */
+	lea	-1894986606(%edx,%r10d),%edx		/* Const + dst + ... */
+	or	%eax,		%r11d		/* x | ... */
+	xor	%ebx,		%r11d		/* y ^ ... */
+	add	%r11d,		%edx		/* dst += ... */
+	mov	10*4(%rsi),%r10d		/* (NEXT STEP) X[10] */
+	mov	$0xffffffff,	%r11d
+	rol	$10,		%edx		/* dst <<< s */
+	xor	%ebx,		%r11d		/* (NEXT STEP) not z' = not %ebx */
+	add	%eax,		%edx		/* dst += x */
+	lea	-1051523(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	or	%edx,		%r11d		/* x | ... */
+	xor	%eax,		%r11d		/* y ^ ... */
+	add	%r11d,		%ecx		/* dst += ... */
+	mov	1*4(%rsi),%r10d		/* (NEXT STEP) X[1] */
+	mov	$0xffffffff,	%r11d
+	rol	$15,		%ecx		/* dst <<< s */
+	xor	%eax,		%r11d		/* (NEXT STEP) not z' = not %eax */
+	add	%edx,		%ecx		/* dst += x */
+	lea	-2054922799(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	or	%ecx,		%r11d		/* x | ... */
+	xor	%edx,		%r11d		/* y ^ ... */
+	add	%r11d,		%ebx		/* dst += ... */
+	mov	8*4(%rsi),%r10d		/* (NEXT STEP) X[8] */
+	mov	$0xffffffff,	%r11d
+	rol	$21,		%ebx		/* dst <<< s */
+	xor	%edx,		%r11d		/* (NEXT STEP) not z' = not %edx */
+	add	%ecx,		%ebx		/* dst += x */
+	lea	1873313359(%eax,%r10d),%eax		/* Const + dst + ... */
+	or	%ebx,		%r11d		/* x | ... */
+	xor	%ecx,		%r11d		/* y ^ ... */
+	add	%r11d,		%eax		/* dst += ... */
+	mov	15*4(%rsi),%r10d		/* (NEXT STEP) X[15] */
+	mov	$0xffffffff,	%r11d
+	rol	$6,		%eax		/* dst <<< s */
+	xor	%ecx,		%r11d		/* (NEXT STEP) not z' = not %ecx */
+	add	%ebx,		%eax		/* dst += x */
+	lea	-30611744(%edx,%r10d),%edx		/* Const + dst + ... */
+	or	%eax,		%r11d		/* x | ... */
+	xor	%ebx,		%r11d		/* y ^ ... */
+	add	%r11d,		%edx		/* dst += ... */
+	mov	6*4(%rsi),%r10d		/* (NEXT STEP) X[6] */
+	mov	$0xffffffff,	%r11d
+	rol	$10,		%edx		/* dst <<< s */
+	xor	%ebx,		%r11d		/* (NEXT STEP) not z' = not %ebx */
+	add	%eax,		%edx		/* dst += x */
+	lea	-1560198380(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	or	%edx,		%r11d		/* x | ... */
+	xor	%eax,		%r11d		/* y ^ ... */
+	add	%r11d,		%ecx		/* dst += ... */
+	mov	13*4(%rsi),%r10d		/* (NEXT STEP) X[13] */
+	mov	$0xffffffff,	%r11d
+	rol	$15,		%ecx		/* dst <<< s */
+	xor	%eax,		%r11d		/* (NEXT STEP) not z' = not %eax */
+	add	%edx,		%ecx		/* dst += x */
+	lea	1309151649(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	or	%ecx,		%r11d		/* x | ... */
+	xor	%edx,		%r11d		/* y ^ ... */
+	add	%r11d,		%ebx		/* dst += ... */
+	mov	4*4(%rsi),%r10d		/* (NEXT STEP) X[4] */
+	mov	$0xffffffff,	%r11d
+	rol	$21,		%ebx		/* dst <<< s */
+	xor	%edx,		%r11d		/* (NEXT STEP) not z' = not %edx */
+	add	%ecx,		%ebx		/* dst += x */
+	lea	-145523070(%eax,%r10d),%eax		/* Const + dst + ... */
+	or	%ebx,		%r11d		/* x | ... */
+	xor	%ecx,		%r11d		/* y ^ ... */
+	add	%r11d,		%eax		/* dst += ... */
+	mov	11*4(%rsi),%r10d		/* (NEXT STEP) X[11] */
+	mov	$0xffffffff,	%r11d
+	rol	$6,		%eax		/* dst <<< s */
+	xor	%ecx,		%r11d		/* (NEXT STEP) not z' = not %ecx */
+	add	%ebx,		%eax		/* dst += x */
+	lea	-1120210379(%edx,%r10d),%edx		/* Const + dst + ... */
+	or	%eax,		%r11d		/* x | ... */
+	xor	%ebx,		%r11d		/* y ^ ... */
+	add	%r11d,		%edx		/* dst += ... */
+	mov	2*4(%rsi),%r10d		/* (NEXT STEP) X[2] */
+	mov	$0xffffffff,	%r11d
+	rol	$10,		%edx		/* dst <<< s */
+	xor	%ebx,		%r11d		/* (NEXT STEP) not z' = not %ebx */
+	add	%eax,		%edx		/* dst += x */
+	lea	718787259(%ecx,%r10d),%ecx		/* Const + dst + ... */
+	or	%edx,		%r11d		/* x | ... */
+	xor	%eax,		%r11d		/* y ^ ... */
+	add	%r11d,		%ecx		/* dst += ... */
+	mov	9*4(%rsi),%r10d		/* (NEXT STEP) X[9] */
+	mov	$0xffffffff,	%r11d
+	rol	$15,		%ecx		/* dst <<< s */
+	xor	%eax,		%r11d		/* (NEXT STEP) not z' = not %eax */
+	add	%edx,		%ecx		/* dst += x */
+	lea	-343485551(%ebx,%r10d),%ebx		/* Const + dst + ... */
+	or	%ecx,		%r11d		/* x | ... */
+	xor	%edx,		%r11d		/* y ^ ... */
+	add	%r11d,		%ebx		/* dst += ... */
+	mov	0*4(%rsi),%r10d		/* (NEXT STEP) X[0] */
+	mov	$0xffffffff,	%r11d
+	rol	$21,		%ebx		/* dst <<< s */
+	xor	%edx,		%r11d		/* (NEXT STEP) not z' = not %edx */
+	add	%ecx,		%ebx		/* dst += x */
+	# add old values of A, B, C, D
+	add	%r8d,	%eax
+	add	%r9d,	%ebx
+	add	%r14d,	%ecx
+	add	%r15d,	%edx
+
+	# loop control
+	add	$64,		%rsi		# ptr += 64
+	cmp	%rdi,		%rsi		# cmp end with ptr
+	jb	2b				# jmp if ptr < end
+	# END of loop over 16-word blocks
+1:
+	mov	%eax,		0*4(%rbp)	# ctx->A = A
+	mov	%ebx,		1*4(%rbp)	# ctx->B = B
+	mov	%ecx,		2*4(%rbp)	# ctx->C = C
+	mov	%edx,		3*4(%rbp)	# ctx->D = D
+
+	pop	%r15
+	pop	%r14
+	pop	%r13				# not really useful (r13 is unused)
+	pop	%r12
+	pop	%rbx
+	pop	%rbp
+	ret
+.L_md5_process_asm_end:
+.size md5_process_asm,.L_md5_process_asm_end-md5_process_asm

lib/md5.c

@@ -2,6 +2,7 @@
  * RFC 1321 compliant MD5 implementation
  *
  * Copyright (C) 2001-2003 Christophe Devine
+ * Copyright (C) 2007-2020 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
@@ -19,6 +20,7 @@
 
 #include "rsync.h"
 
+#ifndef USE_OPENSSL
 void md5_begin(md_context *ctx)
 {
 	ctx->A = 0x67452301;
@@ -38,22 +40,22 @@ static void md5_process(md_context *ctx, const uchar data[CSUM_CHUNK])
 	C = ctx->C;
 	D = ctx->D;
 
-	X[0] = IVAL(data, 0);
-	X[1] = IVAL(data, 4);
-	X[2] = IVAL(data, 8);
-	X[3] = IVAL(data, 12);
-	X[4] = IVAL(data, 16);
-	X[5] = IVAL(data, 20);
-	X[6] = IVAL(data, 24);
-	X[7] = IVAL(data, 28);
-	X[8] = IVAL(data, 32);
-	X[9] = IVAL(data, 36);
-	X[10] = IVAL(data, 40);
-	X[11] = IVAL(data, 44);
-	X[12] = IVAL(data, 48);
-	X[13] = IVAL(data, 52);
-	X[14] = IVAL(data, 56);
-	X[15] = IVAL(data, 60);
+	X[0] = IVALu(data, 0);
+	X[1] = IVALu(data, 4);
+	X[2] = IVALu(data, 8);
+	X[3] = IVALu(data, 12);
+	X[4] = IVALu(data, 16);
+	X[5] = IVALu(data, 20);
+	X[6] = IVALu(data, 24);
+	X[7] = IVALu(data, 28);
+	X[8] = IVALu(data, 32);
+	X[9] = IVALu(data, 36);
+	X[10] = IVALu(data, 40);
+	X[11] = IVALu(data, 44);
+	X[12] = IVALu(data, 48);
+	X[13] = IVALu(data, 52);
+	X[14] = IVALu(data, 56);
+	X[15] = IVALu(data, 60);
 
 #define S(x,n) ((x << n) | ((x & 0xFFFFFFFF) >> (32 - n)))
 
@@ -146,6 +148,10 @@ static void md5_process(md_context *ctx, const uchar data[CSUM_CHUNK])
 	ctx->D += D;
 }
 
+#if defined(HAVE_SIMD) && (CSUM_CHUNK == 64)
+extern void md5_process_asm(md_context *ctx, const void *data, size_t num);
+#endif
+
 void md5_update(md_context *ctx, const uchar *input, uint32 length)
 {
 	uint32 left, fill;
@@ -170,11 +176,20 @@ void md5_update(md_context *ctx, const uchar *input, uint32 length)
 		left = 0;
 	}
 
+#if defined(HAVE_SIMD) && (CSUM_CHUNK == 64)
+	if (length >= CSUM_CHUNK) {
+		uint32 chunks = length / CSUM_CHUNK;
+		md5_process_asm(ctx, input, chunks);
+		length -= chunks * CSUM_CHUNK;
+		input += chunks * CSUM_CHUNK;
+	}
+#else
 	while (length >= CSUM_CHUNK) {
 		md5_process(ctx, input);
 		length -= CSUM_CHUNK;
 		input  += CSUM_CHUNK;
 	}
+#endif
 
 	if (length)
 		memcpy(ctx->buffer + left, input, length);
@@ -192,8 +207,8 @@ void md5_result(md_context *ctx, uchar digest[MD5_DIGEST_LEN])
 	     | (ctx->totalN2 <<  3);
 	low  = (ctx->totalN <<  3);
 
-	SIVAL(msglen, 0, low);
-	SIVAL(msglen, 4, high);
+	SIVALu(msglen, 0, low);
+	SIVALu(msglen, 4, high);
 
 	last = ctx->totalN & 0x3F;
 	padn = last < 56 ? 56 - last : 120 - last;
@@ -201,11 +216,14 @@ void md5_result(md_context *ctx, uchar digest[MD5_DIGEST_LEN])
 	md5_update(ctx, md5_padding, padn);
 	md5_update(ctx, msglen, 8);
 
-	SIVAL(digest, 0, ctx->A);
-	SIVAL(digest, 4, ctx->B);
-	SIVAL(digest, 8, ctx->C);
-	SIVAL(digest, 12, ctx->D);
+	SIVALu(digest, 0, ctx->A);
+	SIVALu(digest, 4, ctx->B);
+	SIVALu(digest, 8, ctx->C);
+	SIVALu(digest, 12, ctx->D);
 }
+#endif
+
+#ifdef TEST_MD5
 
 void get_md5(uchar *out, const uchar *input, int n)
 {
@@ -215,8 +233,6 @@ void get_md5(uchar *out, const uchar *input, int n)
 	md5_result(&ctx, out);
 }
 
-#ifdef TEST_MD5
-
 #include <stdlib.h>
 #include <stdio.h>
 

lib/mdfour.c

@@ -4,7 +4,7 @@
  * An implementation of MD4 designed for use in the SMB authentication protocol.
  *
  * Copyright (C) 1997-1998 Andrew Tridgell
- * Copyright (C) 2005-2007 Wayne Davison
+ * Copyright (C) 2005-2020 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
@@ -193,6 +193,8 @@ void mdfour_result(md_context *md, uchar digest[MD4_DIGEST_LEN])
 	copy4(digest+12, m->D);
 }
 
+#ifdef TEST_MDFOUR
+
 void mdfour(uchar digest[MD4_DIGEST_LEN], uchar *in, int length)
 {
 	md_context md;
@@ -201,7 +203,6 @@ void mdfour(uchar digest[MD4_DIGEST_LEN], uchar *in, int length)
 	mdfour_result(&md, digest);
 }
 
-#ifdef TEST_MDFOUR
 int protocol_version = 28;
 
 static void file_checksum1(char *fname)

lib/mdigest.h

@@ -1,11 +1,24 @@
 /* The include file for both the MD4 and MD5 routines. */
 
+#ifdef USE_OPENSSL
+#include "openssl/md4.h"
+#include "openssl/md5.h"
+#endif
+
 #define MD4_DIGEST_LEN 16
 #define MD5_DIGEST_LEN 16
 #define MAX_DIGEST_LEN MD5_DIGEST_LEN
 
 #define CSUM_CHUNK 64
 
+#define CSUM_NONE 0
+#define CSUM_MD4_ARCHAIC 1
+#define CSUM_MD4_BUSTED 2
+#define CSUM_MD4_OLD 3
+#define CSUM_MD4 4
+#define CSUM_MD5 5
+#define CSUM_XXH64 6
+
 typedef struct {
 	uint32 A, B, C, D;
 	uint32 totalN;          /* bit count, lower 32 bits */
@@ -17,10 +30,13 @@ 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]);
 
-void get_mdfour(uchar digest[MD4_DIGEST_LEN], const uchar *in, int length);
+#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)
 
 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]);
-
-void get_md5(uchar digest[MD5_DIGEST_LEN], const uchar *input, int n);
+#endif

lib/permstring.c

@@ -4,7 +4,7 @@
  * Copyright (C) 1996 Andrew Tridgell
  * Copyright (C) 1996 Paul Mackerras
  * Copyright (C) 2001 Martin Pool <mbp@samba.org>
- * Copyright (C) 2003, 2006 Wayne Davison
+ * Copyright (C) 2003-2019 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

lib/pool_alloc.3

@@ -95,25 +95,39 @@ for
 .I quantum
 will produce a quantum that should meet maximal alignment
 on most platforms.
-If
-.B POOL_QALIGN
+Unless
+.B POOL_NO_QALIGN
 is set in the
 .IR flags ,
 allocations will be aligned to addresses that are a
 multiple of
 .IR quantum .
+A
+.B NULL
+may be specified for the
+.I bomb
+function pointer if it is not needed.  (See the
+.B pool_alloc()
+function for how it is used.)
 If
 .B POOL_CLEAR
 is set in the
 .IR flags ,
 all allocations from the pool will be initialized to zeros.
-You may specify a
-.B NULL
-for the
-.I bomb
-function pointer if you don't wish to use it.  (See the
-.B pool_alloc()
-function for how it is used.)
+If either
+.B POOL_PREPEND
+or
+.B POOL_INTERN
+is specified in the
+.IR flags ,
+each extent's data structure will be allocated at the start of the
+.IR size -length
+buffer (rather than as a separate, non-pool allocation), with the
+former extending the
+.I size
+to hold the structure, and the latter subtracting the structure's
+length from the indicated
+.IR size .
 .P
 .B pool_destroy()
 destroys an allocation
@@ -131,8 +145,8 @@ is
 .BR 0 ,
 .I quantum
 bytes will be allocated.
-If the pool has been created with
-.BR POOL_QALIGN ,
+If the pool has been created without
+.BR POOL_NO_QALIGN ,
 every chunk of memory that is returned will be suitably aligned.
 You can use this with the default
 .I quantum
@@ -169,7 +183,7 @@ an extent), its memory will be completely freed back to the system.
 If
 .I addr
 is
-.BR 0 ,
+.BR NULL ,
 no memory will be freed, but subsequent allocations will come
 from a new extent.
 .P

lib/pool_alloc.c

@@ -2,6 +2,8 @@
 
 #define POOL_DEF_EXTENT	(32 * 1024)
 
+#define POOL_QALIGN_P2		(1<<16)		/* power-of-2 qalign	*/
+
 struct alloc_pool
 {
 	size_t			size;		/* extent size		*/
@@ -13,7 +15,7 @@ struct alloc_pool
 
 	/* statistical data */
 	unsigned long		e_created;	/* extents created	*/
-	unsigned long		e_freed;	/* extents detroyed	*/
+	unsigned long		e_freed;	/* extents destroyed	*/
 	int64			n_allocated;	/* calls to alloc	*/
 	int64			n_freed;	/* calls to free	*/
 	int64			b_allocated;	/* cum. bytes allocated	*/
@@ -22,16 +24,18 @@ struct alloc_pool
 
 struct pool_extent
 {
+	struct pool_extent	*next;
 	void			*start;		/* starting address	*/
 	size_t			free;		/* free bytecount	*/
-	size_t			bound;		/* bytes bound by padding,
-						 * overhead and freed	*/
-	struct pool_extent	*next;
+	size_t			bound;		/* trapped free bytes	*/
 };
 
 struct align_test {
-	void *foo;
-	int64 bar;
+	uchar foo;
+	union {
+	    int64 i;
+	    void *p;
+	} bar;
 };
 
 #define MINALIGN	offsetof(struct align_test, bar)
@@ -43,20 +47,42 @@ struct align_test {
 alloc_pool_t
 pool_create(size_t size, size_t quantum, void (*bomb)(const char *), int flags)
 {
-	struct alloc_pool	*pool;
+	struct alloc_pool *pool;
 
-	if (!(pool = new(struct alloc_pool)))
-		return pool;
-	memset(pool, 0, sizeof (struct alloc_pool));
-
-	pool->size = size	/* round extent size to min alignment reqs */
-	    ? (size + MINALIGN - 1) & ~(MINALIGN - 1)
-	    : POOL_DEF_EXTENT;
-	if (flags & POOL_INTERN) {
-		pool->size -= sizeof (struct pool_extent);
-		flags |= POOL_APPEND;
+	if ((MINALIGN & (MINALIGN - 1)) != 0) {
+		if (bomb)
+			(*bomb)("Compiler error: MINALIGN is not a power of 2\n");
+		return NULL;
 	}
-	pool->quantum = quantum ? quantum : MINALIGN;
+
+	if (!(pool = new0(struct alloc_pool)))
+		return NULL;
+
+	if (!size)
+		size = POOL_DEF_EXTENT;
+	if (!quantum)
+		quantum = MINALIGN;
+
+	if (flags & POOL_INTERN) {
+		if (size <= sizeof (struct pool_extent))
+			size = quantum;
+		else
+			size -= sizeof (struct pool_extent);
+		flags |= POOL_PREPEND;
+	}
+
+	if (quantum <= 1)
+		flags = (flags | POOL_NO_QALIGN) & ~POOL_QALIGN_P2;
+	else if (!(flags & POOL_NO_QALIGN)) {
+		if (size % quantum)
+			size += quantum - size % quantum;
+		/* If quantum is a power of 2, we'll avoid using modulus. */
+		if (!(quantum & (quantum - 1)))
+			flags |= POOL_QALIGN_P2;
+	}
+
+	pool->size = size;
+	pool->quantum = quantum;
 	pool->bomb = bomb;
 	pool->flags = flags;
 
@@ -67,17 +93,21 @@ void
 pool_destroy(alloc_pool_t p)
 {
 	struct alloc_pool *pool = (struct alloc_pool *) p;
-	struct pool_extent	*cur, *next;
+	struct pool_extent *cur, *next;
 
 	if (!pool)
 		return;
 
 	for (cur = pool->extents; cur; cur = next) {
 		next = cur->next;
-		free(cur->start);
-		if (!(pool->flags & POOL_APPEND))
+		if (pool->flags & POOL_PREPEND)
+			free(PTR_ADD(cur->start, -sizeof (struct pool_extent)));
+		else {
+			free(cur->start);
 			free(cur);
+		}
 	}
+
 	free(pool);
 }
 
@@ -90,45 +120,40 @@ pool_alloc(alloc_pool_t p, size_t len, const char *bomb_msg)
 
 	if (!len)
 		len = pool->quantum;
-	else if (pool->quantum > 1 && len % pool->quantum)
-		len += pool->quantum - len % pool->quantum;
+	else if (pool->flags & POOL_QALIGN_P2) {
+		if (len & (pool->quantum - 1))
+			len += pool->quantum - (len & (pool->quantum - 1));
+	} else if (!(pool->flags & POOL_NO_QALIGN)) {
+		if (len % pool->quantum)
+			len += pool->quantum - len % pool->quantum;
+	}
 
 	if (len > pool->size)
 		goto bomb_out;
 
 	if (!pool->extents || len > pool->extents->free) {
-		void	*start;
-		size_t	free;
-		size_t	bound;
-		size_t	skew;
-		size_t	asize;
+		void *start;
+		size_t asize;
 		struct pool_extent *ext;
 
-		free = pool->size;
-		bound = 0;
-
 		asize = pool->size;
-		if (pool->flags & POOL_APPEND)
+		if (pool->flags & POOL_PREPEND)
 			asize += sizeof (struct pool_extent);
 
 		if (!(start = new_array(char, asize)))
 			goto bomb_out;
 
 		if (pool->flags & POOL_CLEAR)
-			memset(start, 0, free);
+			memset(start, 0, asize);
 
-		if (pool->flags & POOL_APPEND)
-			ext = PTR_ADD(start, free);
-		else if (!(ext = new(struct pool_extent)))
+		if (pool->flags & POOL_PREPEND) {
+			ext = start;
+			start = PTR_ADD(start, sizeof (struct pool_extent));
+		} else if (!(ext = new(struct pool_extent)))
 			goto bomb_out;
-		if (pool->flags & POOL_QALIGN && pool->quantum > 1
-		    && (skew = (size_t)PTR_ADD(start, free) % pool->quantum)) {
-			bound  += skew;
-			free -= skew;
-		}
 		ext->start = start;
-		ext->free = free;
-		ext->bound = bound;
+		ext->free = pool->size;
+		ext->bound = 0;
 		ext->next = pool->extents;
 		pool->extents = ext;
 
@@ -160,10 +185,24 @@ pool_free(alloc_pool_t p, size_t len, void *addr)
 	if (!pool)
 		return;
 
+	if (!addr) {
+		/* A NULL addr starts a fresh extent for new allocations. */
+		if ((cur = pool->extents) != NULL && cur->free != pool->size) {
+			cur->bound += cur->free;
+			cur->free = 0;
+		}
+		return;
+	}
+
 	if (!len)
 		len = pool->quantum;
-	else if (pool->quantum > 1 && len % pool->quantum)
-		len += pool->quantum - len % pool->quantum;
+	else if (pool->flags & POOL_QALIGN_P2) {
+		if (len & (pool->quantum - 1))
+			len += pool->quantum - (len & (pool->quantum - 1));
+	} else if (!(pool->flags & POOL_NO_QALIGN)) {
+		if (len % pool->quantum)
+			len += pool->quantum - len % pool->quantum;
+	}
 
 	pool->n_freed++;
 	pool->b_freed += len;
@@ -179,19 +218,12 @@ pool_free(alloc_pool_t p, size_t len, void *addr)
 	if (!prev) {
 		/* The "live" extent is kept ready for more allocations. */
 		if (cur->free + cur->bound + len >= pool->size) {
-			size_t skew;
-
 			if (pool->flags & POOL_CLEAR) {
 				memset(PTR_ADD(cur->start, cur->free), 0,
 				       pool->size - cur->free);
 			}
 			cur->free = pool->size;
 			cur->bound = 0;
-			if (pool->flags & POOL_QALIGN && pool->quantum > 1
-			    && (skew = (size_t)PTR_ADD(cur->start, cur->free) % pool->quantum)) {
-				cur->bound += skew;
-				cur->free -= skew;
-			}
 		} else if (addr == PTR_ADD(cur->start, cur->free)) {
 			if (pool->flags & POOL_CLEAR)
 				memset(addr, 0, len);
@@ -203,9 +235,12 @@ pool_free(alloc_pool_t p, size_t len, void *addr)
 
 		if (cur->free + cur->bound >= pool->size) {
 			prev->next = cur->next;
-			free(cur->start);
-			if (!(pool->flags & POOL_APPEND))
+			if (pool->flags & POOL_PREPEND)
+				free(PTR_ADD(cur->start, -sizeof (struct pool_extent)));
+			else {
+				free(cur->start);
 				free(cur);
+			}
 			pool->e_freed++;
 		} else if (prev != pool->extents) {
 			/* Move the extent to be the first non-live extent. */
@@ -242,18 +277,11 @@ pool_free_old(alloc_pool_t p, void *addr)
 			prev->next = NULL;
 			next = cur;
 		} else {
-			size_t skew;
-
 			/* The most recent live extent can just be reset. */
 			if (pool->flags & POOL_CLEAR)
 				memset(addr, 0, pool->size - cur->free);
 			cur->free = pool->size;
 			cur->bound = 0;
-			if (pool->flags & POOL_QALIGN && pool->quantum > 1
-			    && (skew = (size_t)PTR_ADD(cur->start, cur->free) % pool->quantum)) {
-				cur->bound += skew;
-				cur->free -= skew;
-			}
 			next = cur->next;
 			cur->next = NULL;
 		}
@@ -264,9 +292,12 @@ pool_free_old(alloc_pool_t p, void *addr)
 
 	while ((cur = next) != NULL) {
 		next = cur->next;
-		free(cur->start);
-		if (!(pool->flags & POOL_APPEND))
+		if (pool->flags & POOL_PREPEND)
+			free(PTR_ADD(cur->start, -sizeof (struct pool_extent)));
+		else {
+			free(cur->start);
 			free(cur);
+		}
 		pool->e_freed++;
 	}
 }
@@ -295,24 +326,30 @@ pool_boundary(alloc_pool_t p, size_t len)
 }
 
 #define FDPRINT(label, value) \
-	snprintf(buf, sizeof buf, label, value), \
-	write(fd, buf, strlen(buf))
+	do { \
+		int len = snprintf(buf, sizeof buf, label, value); \
+		if (write(fd, buf, len) != len) \
+			ret = -1; \
+	} while (0)
 
 #define FDEXTSTAT(ext) \
-	snprintf(buf, sizeof buf, "  %12ld  %5ld\n", \
-		(long) ext->free, \
-		(long) ext->bound), \
-	write(fd, buf, strlen(buf))
+	do { \
+		int len = snprintf(buf, sizeof buf, "  %12ld  %5ld\n", \
+				   (long)ext->free, (long)ext->bound); \
+		if (write(fd, buf, len) != len) \
+			ret = -1; \
+	} while (0)
 
-void
+int
 pool_stats(alloc_pool_t p, int fd, int summarize)
 {
 	struct alloc_pool *pool = (struct alloc_pool *) p;
-	struct pool_extent	*cur;
+	struct pool_extent *cur;
 	char buf[BUFSIZ];
+	int ret = 0;
 
 	if (!pool)
-		return;
+		return ret;
 
 	FDPRINT("  Extent size:       %12ld\n",	(long)	pool->size);
 	FDPRINT("  Alloc quantum:     %12ld\n",	(long)	pool->quantum);
@@ -324,13 +361,16 @@ pool_stats(alloc_pool_t p, int fd, int summarize)
 	FDPRINT("  Bytes freed:       %12.0f\n", (double) pool->b_freed);
 
 	if (summarize)
-		return;
+		return ret;
 
 	if (!pool->extents)
-		return;
+		return ret;
 
-	write(fd, "\n", 1);
+	if (write(fd, "\n", 1) != 1)
+		ret = -1;
 
 	for (cur = pool->extents; cur; cur = cur->next)
 		FDEXTSTAT(cur);
+
+	return ret;
 }

lib/pool_alloc.h

@@ -1,9 +1,9 @@
 #include <stddef.h>
 
 #define POOL_CLEAR	(1<<0)		/* zero fill allocations	*/
-#define POOL_QALIGN	(1<<1)		/* align data to quanta		*/
+#define POOL_NO_QALIGN	(1<<1)		/* don't align data to quanta	*/
 #define POOL_INTERN	(1<<2)		/* Allocate extent structures	*/
-#define POOL_APPEND	(1<<3)		/*   or appended to extent data	*/
+#define POOL_PREPEND	(1<<3)		/*   or prepend to extent data	*/
 
 typedef void *alloc_pool_t;
 

lib/sysacls.c

@@ -2,6 +2,7 @@
  * Unix SMB/CIFS implementation.
  * Based on the Samba ACL support code.
  * Copyright (C) Jeremy Allison 2000.
+ * Copyright (C) 2007-2020 Wayne Davison
  *
  * The permission functions have been changed to get/set all bits via
  * one call.  Some functions that rsync doesn't need were also removed.
@@ -449,7 +450,7 @@ SMB_ACL_T sys_acl_get_file(const char *path_p, SMB_ACL_TYPE_T type)
 	 *
 	 * Note: we assume that the acl() system call returned a
 	 * well formed ACL which is sorted so that all of the
-	 * access ACL entries preceed any default ACL entries
+	 * access ACL entries precede any default ACL entries
 	 */
 	for (naccess = 0; naccess < count; naccess++) {
 		if (acl_d->acl[naccess].a_type & ACL_DEFAULT)
@@ -550,7 +551,7 @@ SMB_ACL_T sys_acl_init(int count)
 	 * acl[] array, this actually allocates an ACL with room
 	 * for (count+1) entries
 	 */
-	if ((a = (SMB_ACL_T)SMB_MALLOC(sizeof(struct SMB_ACL_T) + count * sizeof(struct acl))) == NULL) {
+	if ((a = (SMB_ACL_T)SMB_MALLOC(sizeof a[0] + count * sizeof (struct acl))) == NULL) {
 		errno = ENOMEM;
 		return NULL;
 	}
@@ -872,6 +873,10 @@ int sys_acl_get_tag_type(SMB_ACL_ENTRY_T entry_d, SMB_ACL_TAG_T *type_p)
 
 #define	INITIAL_ACL_SIZE	16
 
+#ifndef NACLENTRIES
+#define NACLENTRIES 0
+#endif
+
 SMB_ACL_T sys_acl_get_file(const char *path_p, SMB_ACL_TYPE_T type)
 {
 	SMB_ACL_T	acl_d;
@@ -908,7 +913,7 @@ SMB_ACL_T sys_acl_get_file(const char *path_p, SMB_ACL_TYPE_T type)
 
 		sys_acl_free_acl(acl_d);
 
-		if ((count = acl(path_p, ACL_CNT, 0, NULL)) < 0) {
+		if ((count = acl(path_p, ACL_CNT, NACLENTRIES, NULL)) < 0) {
 			return NULL;
 		}
 
@@ -927,7 +932,7 @@ SMB_ACL_T sys_acl_get_file(const char *path_p, SMB_ACL_TYPE_T type)
 	 *
 	 * Note: we assume that the acl() system call returned a
 	 * well formed ACL which is sorted so that all of the
-	 * access ACL entries preceed any default ACL entries
+	 * access ACL entries precede any default ACL entries
 	 */
 	for (naccess = 0; naccess < count; naccess++) {
 		if (acl_d->acl[naccess].a_type & ACL_DEFAULT)
@@ -1006,7 +1011,7 @@ SMB_ACL_T sys_acl_init(int count)
 	 * acl[] array, this actually allocates an ACL with room
 	 * for (count+1) entries
 	 */
-	if ((a = SMB_MALLOC(sizeof(struct SMB_ACL_T) + count * sizeof(struct acl))) == NULL) {
+	if ((a = (SMB_ACL_T)SMB_MALLOC(sizeof a[0] + count * sizeof(struct acl))) == NULL) {
 		errno = ENOMEM;
 		return NULL;
 	}
@@ -1090,7 +1095,7 @@ struct hpux_acl_types {
  * structures.
  * Inputs:
  *
- * acl_count      - Count of ACLs in the array of ACL strucutres.
+ * acl_count      - Count of ACLs in the array of ACL structures.
  * aclp           - Array of ACL structures.
  * acl_type_count - Pointer to acl_types structure. Should already be
  *                  allocated.
@@ -1251,7 +1256,7 @@ static int hpux_acl_sort(int acl_count, int calclass, struct acl *aclp)
 {
 #if !defined(HAVE_HPUX_ACLSORT)
 	/*
-	 * The aclsort() system call is availabe on the latest HPUX General
+	 * The aclsort() system call is available on the latest HPUX General
 	 * Patch Bundles. So for HPUX, we developed our version of acl_sort 
 	 * function. Because, we don't want to update to a new 
 	 * HPUX GR bundle just for aclsort() call.
@@ -1306,7 +1311,7 @@ or DEF_USER_OBJ or DEF_GROUP_OBJ or DEF_OTHER_OBJ\n"));
 	 * Sorting crieteria - First sort by ACL type. If there are multiple entries of
 	 * same ACL type, sort by ACL id.
 	 *
-	 * I am using the trival kind of sorting method here because, performance isn't 
+	 * I am using the trivial kind of sorting method here because, performance isn't 
 	 * really effected by the ACLs feature. More over there aren't going to be more
 	 * than 17 entries on HPUX. 
 	 */
@@ -1637,14 +1642,14 @@ SMB_ACL_T sys_acl_init(int count)
 		return NULL;
 	}
 
-	if ((a = SMB_MALLOC(sizeof(struct SMB_ACL_T) + sizeof(struct acl))) == NULL) {
+	if ((a = (SMB_ACL_T)SMB_MALLOC(sizeof a[0] + sizeof (struct acl))) == NULL) {
 		errno = ENOMEM;
 		return NULL;
 	}
 
 	a->next = -1;
 	a->freeaclp = False;
-	a->aclp = (struct acl *)(&a->aclp + sizeof(struct acl *));
+	a->aclp = (struct acl *)((char *)a + sizeof a[0]);
 	a->aclp->acl_cnt = 0;
 
 	return a;
@@ -1735,6 +1740,13 @@ int sys_acl_get_entry( SMB_ACL_T theacl, int entry_id, SMB_ACL_ENTRY_T *entry_p)
 	struct new_acl_entry *entry;
 	int keep_going;
 
+	if (entry_id == SMB_ACL_FIRST_ENTRY)
+		theacl->count = 0;
+	else if (entry_id != SMB_ACL_NEXT_ENTRY) {
+		errno = EINVAL;
+		return -1;
+	}
+
 	DEBUG(10,("This is the count: %d\n",theacl->count));
 
 	/* Check if count was previously set to -1. *
@@ -1804,11 +1816,14 @@ SMB_ACL_T sys_acl_get_file( const char *path_p, SMB_ACL_TYPE_T type)
 	struct acl_entry_link *acl_entry_link_head;
 	int i;
 	int rc = 0;
-	uid_t user_id;
 
 	/* AIX has no DEFAULT */
 	if  ( type == SMB_ACL_TYPE_DEFAULT ) {
+#ifdef ENOTSUP
 		errno = ENOTSUP;
+#else
+		errno = ENOSYS;
+#endif
 		return NULL;
 	}
 
@@ -2025,7 +2040,6 @@ SMB_ACL_T sys_acl_get_fd(int fd)
 	struct acl_entry_link *acl_entry_link_head;
 	int i;
 	int rc = 0;
-	uid_t user_id;
 
 	/* Get the acl using fstatacl */
    
@@ -2255,6 +2269,11 @@ SMB_ACL_T sys_acl_init( int count)
 {
 	struct acl_entry_link *theacl = NULL;
  
+	if (count < 0) {
+		errno = EINVAL;
+		return NULL;
+	}
+
 	DEBUG(10,("Entering sys_acl_init\n"));
 
 	theacl = SMB_MALLOC_P(struct acl_entry_link);
@@ -2373,7 +2392,6 @@ int sys_acl_set_file( const char *name, SMB_ACL_TYPE_T acltype, SMB_ACL_T theacl
 	struct acl_entry *acl_entry = NULL;
 	struct ace_id *ace_id = NULL;
 	uint id_type;
-	uint ace_access;
 	uint user_id;
 	uint acl_length;
 	uint rc;
@@ -2551,7 +2569,7 @@ int sys_acl_set_fd( int fd, SMB_ACL_T theacl)
 }
 #endif
 
-int sys_acl_delete_def_file(const char *name)
+int sys_acl_delete_def_file(UNUSED(const char *name))
 {
 	/* AIX has no default ACL */
 	return 0;
@@ -2694,6 +2712,8 @@ int sys_acl_set_info(SMB_ACL_ENTRY_T entry, SMB_ACL_TAG_T tag_type, uint32 bits,
 		rc = mbr_uid_to_uuid(u_g_id, uu);
 	else
 		rc = mbr_gid_to_uuid(u_g_id, uu);
+	if (rc != 0)
+		return rc;
 
 	if (acl_set_tag_type(entry, tag_type) != 0
 	 || acl_set_qualifier(entry, &uu) != 0
@@ -2765,6 +2785,11 @@ int no_acl_syscall_error(int err)
 		return 1;
 	}
 #endif
+	if (err == EINVAL) {
+		/* If the type of SMB_ACL_TYPE_ACCESS or SMB_ACL_TYPE_DEFAULT
+		 * isn't valid, then the ACLs must be non-POSIX. */
+		return 1;
+	}
 	return 0;
 }
 

lib/sysacls.h

@@ -3,6 +3,7 @@
  * Version 2.2.x
  * Portable SMB ACL interface
  * Copyright (C) Jeremy Allison 2000
+ * Copyright (C) 2007-2019 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

lib/sysxattrs.c

@@ -2,6 +2,7 @@
  * Extended attribute support for rsync.
  *
  * Copyright (C) 2004 Red Hat, Inc.
+ * Copyright (C) 2003-2019 Wayne Davison
  * Written by Jay Fenlason.
  *
  * This program is free software; you can redistribute it and/or modify
@@ -23,6 +24,10 @@
 
 #ifdef SUPPORT_XATTRS
 
+#ifdef HAVE_OSX_XATTRS
+#define GETXATTR_FETCH_LIMIT (64*1024*1024)
+#endif
+
 #if defined HAVE_LINUX_XATTRS
 
 ssize_t sys_lgetxattr(const char *path, const char *name, void *value, size_t size)
@@ -54,7 +59,24 @@ ssize_t sys_llistxattr(const char *path, char *list, size_t size)
 
 ssize_t sys_lgetxattr(const char *path, const char *name, void *value, size_t size)
 {
-	return getxattr(path, name, value, size, 0, XATTR_NOFOLLOW);
+	ssize_t len = getxattr(path, name, value, size, 0, XATTR_NOFOLLOW);
+
+	/* If we're retrieving data, handle resource forks > 64MB specially */
+	if (value != NULL && len == GETXATTR_FETCH_LIMIT && (size_t)len < size) {
+		/* getxattr will only return 64MB of data at a time, need to call again with a new offset */
+		u_int32_t offset = len;
+		size_t data_retrieved = len;
+		while (data_retrieved < size) {
+			len = getxattr(path, name, value + offset, size - data_retrieved, offset, XATTR_NOFOLLOW);
+			if (len <= 0)
+				break;
+			data_retrieved += len;
+			offset += (u_int32_t)len;
+		}
+		len = data_retrieved;
+	}
+
+	return len;
 }
 
 ssize_t sys_fgetxattr(int filedes, const char *name, void *value, size_t size)
@@ -125,6 +147,150 @@ ssize_t sys_llistxattr(const char *path, char *list, size_t size)
 	return len;
 }
 
+#elif HAVE_SOLARIS_XATTRS
+
+static ssize_t read_xattr(int attrfd, void *buf, size_t buflen)
+{
+	STRUCT_STAT sb;
+	ssize_t ret;
+
+	if (fstat(attrfd, &sb) < 0)
+		ret = -1;
+	else if (sb.st_size > SSIZE_MAX) {
+		errno = ERANGE;
+		ret = -1;
+	} else if (buflen == 0)
+		ret = sb.st_size;
+	else if (sb.st_size > buflen) {
+		errno = ERANGE;
+		ret = -1;
+	} else {
+		size_t bufpos;
+		for (bufpos = 0; bufpos < sb.st_size; ) {
+			ssize_t cnt = read(attrfd, buf + bufpos, sb.st_size - bufpos);
+			if (cnt <= 0) {
+				if (cnt < 0 && errno == EINTR)
+					continue;
+				bufpos = -1;
+				break;
+			}
+			bufpos += cnt;
+		}
+		ret = bufpos;
+	}
+
+	close(attrfd);
+
+	return ret;
+}
+
+ssize_t sys_lgetxattr(const char *path, const char *name, void *value, size_t size)
+{
+	int attrfd;
+
+	if ((attrfd = attropen(path, name, O_RDONLY)) < 0) {
+		errno = ENOATTR;
+		return -1;
+	}
+
+	return read_xattr(attrfd, value, size);
+}
+
+ssize_t sys_fgetxattr(int filedes, const char *name, void *value, size_t size)
+{
+	int attrfd;
+
+	if ((attrfd = openat(filedes, name, O_RDONLY|O_XATTR, 0)) < 0) {
+		errno = ENOATTR;
+		return -1;
+	}
+
+	return read_xattr(attrfd, value, size);
+}
+
+int sys_lsetxattr(const char *path, const char *name, const void *value, size_t size)
+{
+	int attrfd;
+	size_t bufpos;
+	mode_t mode = S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP;
+
+	if ((attrfd = attropen(path, name, O_CREAT|O_TRUNC|O_WRONLY, mode)) < 0)
+		return -1;
+
+	for (bufpos = 0; bufpos < size; ) {
+		ssize_t cnt = write(attrfd, value+bufpos, size);
+		if (cnt <= 0) {
+			if (cnt < 0 && errno == EINTR)
+				continue;
+			bufpos = -1;
+			break;
+		}
+		bufpos += cnt;
+	}
+
+	close(attrfd);
+
+	return bufpos > 0 ? 0 : -1;
+}
+
+int sys_lremovexattr(const char *path, const char *name)
+{
+	int attrdirfd;
+	int ret;
+
+	if ((attrdirfd = attropen(path, ".", O_RDONLY)) < 0)
+		return -1;
+
+	ret = unlinkat(attrdirfd, name, 0);
+
+	close(attrdirfd);
+
+	return ret;
+}
+
+ssize_t sys_llistxattr(const char *path, char *list, size_t size)
+{
+	int attrdirfd;
+	DIR *dirp;
+	struct dirent *dp;
+	ssize_t ret = 0;
+
+	if ((attrdirfd = attropen(path, ".", O_RDONLY)) < 0) {
+		errno = ENOTSUP;
+		return -1;
+	}
+
+	if ((dirp = fdopendir(attrdirfd)) == NULL) {
+		close(attrdirfd);
+		return -1;
+	}
+
+	while ((dp = readdir(dirp))) {
+		int len = strlen(dp->d_name);
+
+		if (dp->d_name[0] == '.' && (len == 1 || (len == 2 && dp->d_name[1] == '.')))
+			continue;
+		if (len == 11 && dp->d_name[0] == 'S' && strncmp(dp->d_name, "SUNWattr_r", 10) == 0
+		 && (dp->d_name[10] == 'o' || dp->d_name[10] == 'w'))
+			continue;
+
+		if ((ret += len+1) > size) {
+			if (size == 0)
+				continue;
+			ret = -1;
+			errno = ERANGE;
+			break;
+		}
+		memcpy(list, dp->d_name, len+1);
+		list += len+1;
+	}
+
+	closedir(dirp);
+	close(attrdirfd);
+
+	return ret;
+}
+
 #else
 
 #error You need to create xattr compatibility functions.

lib/sysxattrs.h

@@ -1,9 +1,9 @@
 #ifdef SUPPORT_XATTRS
 
-#if defined HAVE_ATTR_XATTR_H
-#include <attr/xattr.h>
-#elif defined HAVE_SYS_XATTR_H
+#if defined HAVE_SYS_XATTR_H
 #include <sys/xattr.h>
+#elif defined HAVE_ATTR_XATTR_H
+#include <attr/xattr.h>
 #elif defined HAVE_SYS_EXTATTR_H
 #include <sys/extattr.h>
 #endif

log.c

@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1998-2001 Andrew Tridgell <tridge@samba.org>
  * Copyright (C) 2000-2001 Martin Pool <mbp@samba.org>
- * Copyright (C) 2003-2007 Wayne Davison
+ * Copyright (C) 2003-2020 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
@@ -20,9 +20,9 @@
  */
 
 #include "rsync.h"
-#include "ifuncs.h"
+#include "itypes.h"
+#include "inums.h"
 
-extern int verbose;
 extern int dry_run;
 extern int am_daemon;
 extern int am_server;
@@ -31,16 +31,20 @@ extern int am_generator;
 extern int local_server;
 extern int quiet;
 extern int module_id;
-extern int msg_fd_out;
 extern int allow_8bit_chars;
 extern int protocol_version;
+extern int always_checksum;
 extern int preserve_times;
-extern int uid_ndx;
-extern int gid_ndx;
+extern int msgs2stderr;
+extern int xfersum_type;
+extern int checksum_type;
 extern int stdout_format_has_i;
 extern int stdout_format_has_o_or_i;
 extern int logfile_format_has_i;
 extern int logfile_format_has_o_or_i;
+extern int receiver_symlink_times;
+extern int64 total_data_written;
+extern int64 total_data_read;
 extern mode_t orig_umask;
 extern char *auth_user;
 extern char *stdout_format;
@@ -50,22 +54,29 @@ extern char *logfile_name;
 extern iconv_t ic_chck;
 #endif
 #ifdef ICONV_OPTION
-extern iconv_t ic_send, ic_recv;
+extern iconv_t ic_recv;
 #endif
-extern char curr_dir[];
-extern char *module_dir;
+extern char curr_dir[MAXPATHLEN];
+extern char *full_module_path;
 extern unsigned int module_dirlen;
+extern char sender_file_sum[MAX_DIGEST_LEN];
+extern const char undetermined_hostname[];
 
 static int log_initialised;
 static int logfile_was_closed;
 static FILE *logfile_fp;
 struct stats stats;
 
-int log_got_error = 0;
+int got_xfer_error = 0;
+int output_needs_newline = 0;
+int send_msgs_to_gen = 0;
+
+static int64 initial_data_written;
+static int64 initial_data_read;
 
 struct {
-        int code;
-        char const *name;
+	int code;
+	char const *name;
 } const rerr_names[] = {
 	{ RERR_SYNTAX     , "syntax or usage error" },
 	{ RERR_PROTOCOL   , "protocol incompatibility" },
@@ -83,14 +94,15 @@ struct {
 	{ RERR_SIGNAL     , "received SIGINT, SIGTERM, or SIGHUP" },
 	{ RERR_WAITCHILD  , "waitpid() failed" },
 	{ RERR_MALLOC     , "error allocating core memory buffers" },
-	{ RERR_PARTIAL    , "some files could not be transferred" },
+	{ RERR_PARTIAL    , "some files/attrs were not transferred (see previous errors)" },
 	{ RERR_VANISHED   , "some files vanished before they could be transferred" },
+	{ RERR_DEL_LIMIT  , "the --max-delete limit stopped deletions" },
 	{ RERR_TIMEOUT    , "timeout in data send/receive" },
+	{ RERR_CONTIMEOUT , "timeout waiting for daemon connection" },
 	{ RERR_CMD_FAILED , "remote shell failed" },
 	{ RERR_CMD_KILLED , "remote shell killed" },
 	{ RERR_CMD_RUN    , "remote command could not be run" },
 	{ RERR_CMD_NOTFOUND,"remote command not found" },
-	{ RERR_DEL_LIMIT  , "the --max-delete limit stopped deletions" },
 	{ 0, NULL }
 };
 
@@ -112,8 +124,7 @@ static void logit(int priority, const char *buf)
 	if (logfile_was_closed)
 		logfile_reopen();
 	if (logfile_fp) {
-		fprintf(logfile_fp, "%s [%d] %s",
-			timestring(time(NULL)), (int)getpid(), buf);
+		fprintf(logfile_fp, "%s [%d] %s", timestring(time(NULL)), (int)getpid(), buf);
 		fflush(logfile_fp);
 	} else {
 		syslog(priority, "%s", buf);
@@ -122,21 +133,16 @@ static void logit(int priority, const char *buf)
 
 static void syslog_init()
 {
-	static int been_here = 0;
 	int options = LOG_PID;
 
-	if (been_here)
-		return;
-	been_here = 1;
-
 #ifdef LOG_NDELAY
 	options |= LOG_NDELAY;
 #endif
 
 #ifdef LOG_DAEMON
-	openlog("rsyncd", options, lp_syslog_facility(module_id));
+	openlog(lp_syslog_tag(module_id), options, lp_syslog_facility(module_id));
 #else
-	openlog("rsyncd", options);
+	openlog(lp_syslog_tag(module_id), options);
 #endif
 
 #ifndef LOG_NDELAY
@@ -156,14 +162,16 @@ static void logfile_open(void)
 		rsyserr(FERROR, fopen_errno,
 			"failed to open log-file %s", logfile_name);
 		rprintf(FINFO, "Ignoring \"log file\" setting.\n");
+		logfile_name = "";
 	}
 }
 
 void log_init(int restart)
 {
 	if (log_initialised) {
-		if (!restart)
+		if (!restart) /* Note: a restart only happens with am_daemon */
 			return;
+		assert(logfile_name); /* all am_daemon procs got at least an empty string */
 		if (strcmp(logfile_name, lp_log_file(module_id)) != 0) {
 			if (logfile_fp) {
 				fclose(logfile_fp);
@@ -173,7 +181,8 @@ void log_init(int restart)
 			logfile_name = NULL;
 		} else if (*logfile_name)
 			return; /* unchanged, non-empty "log file" names */
-		else if (lp_syslog_facility(-1) != lp_syslog_facility(module_id))
+		else if (lp_syslog_facility(-1) != lp_syslog_facility(module_id)
+		      || strcmp(lp_syslog_tag(-1), lp_syslog_tag(module_id)) != 0)
 			closelog();
 		else
 			return; /* unchanged syslog settings */
@@ -195,6 +204,7 @@ void log_init(int restart)
 		syslog_init();
 }
 
+/* Note that this close & reopen idiom intentionally ignores syslog logging. */
 void logfile_close(void)
 {
 	if (logfile_fp) {
@@ -212,35 +222,36 @@ void logfile_reopen(void)
 	}
 }
 
-static void filtered_fwrite(FILE *f, const char *buf, int len, int use_isprint)
+static void filtered_fwrite(FILE *f, const char *in_buf, int in_len, int use_isprint, char end_char)
 {
-	const char *s, *end = buf + len;
-	for (s = buf; s < end; s++) {
-		if ((s < end - 4
-		  && *s == '\\' && s[1] == '#'
-		  && isDigit(s + 2)
-		  && isDigit(s + 3)
-		  && isDigit(s + 4))
-		 || (*s != '\t'
-		  && ((use_isprint && !isPrint(s))
-		   || *(uchar*)s < ' '))) {
-			if (s != buf && fwrite(buf, s - buf, 1, f) != 1)
+	char outbuf[1024], *ob = outbuf;
+	const char *end = in_buf + in_len;
+	while (in_buf < end) {
+		if (ob - outbuf >= (int)sizeof outbuf - 10) {
+			if (fwrite(outbuf, ob - outbuf, 1, f) != 1)
 				exit_cleanup(RERR_MESSAGEIO);
-			fprintf(f, "\\#%03o", *(uchar*)s);
-			buf = s + 1;
+			ob = outbuf;
 		}
+		if ((in_buf < end - 4 && *in_buf == '\\' && in_buf[1] == '#'
+		  && isDigit(in_buf + 2) && isDigit(in_buf + 3) && isDigit(in_buf + 4))
+		 || (*in_buf != '\t' && ((use_isprint && !isPrint(in_buf)) || *(uchar*)in_buf < ' ')))
+			ob += snprintf(ob, 6, "\\#%03o", *(uchar*)in_buf++);
+		else
+			*ob++ = *in_buf++;
 	}
-	if (buf != end && fwrite(buf, end - buf, 1, f) != 1)
+	if (end_char) /* The "- 10" above means that there is always room for one more char here. */
+		*ob++ = end_char;
+	if (ob != outbuf && fwrite(outbuf, ob - outbuf, 1, f) != 1)
 		exit_cleanup(RERR_MESSAGEIO);
 }
 
 /* this is the underlying (unformatted) rsync debugging function. Call
- * it with FINFO, FERROR or FLOG.  Note: recursion can happen with
- * certain fatal conditions. */
+ * it with FINFO, FERROR_*, FWARNING, FLOG, or FCLIENT.  Note: recursion
+ * can happen with certain fatal conditions. */
 void rwrite(enum logcode code, const char *buf, int len, int is_utf8)
 {
-	int trailing_CR_or_NL;
-	FILE *f = NULL;
+	char trailing_CR_or_NL;
+	FILE *f = msgs2stderr ? stderr : stdout;
 #ifdef ICONV_OPTION
 	iconv_t ic = is_utf8 && ic_recv != (iconv_t)-1 ? ic_recv : ic_chck;
 #else
@@ -252,22 +263,34 @@ void rwrite(enum logcode code, const char *buf, int len, int is_utf8)
 	if (len < 0)
 		exit_cleanup(RERR_MESSAGEIO);
 
-	if (am_server && msg_fd_out >= 0) {
+	if (msgs2stderr) {
+		/* A normal daemon can get msgs2stderr set if the socket is busted, so we
+		 * change the message destination into an FLOG message in order to try to
+		 * get some info about an abnormal-exit into the log file. An rsh daemon
+		 * can have this set via user request, so we'll leave the code alone so
+		 * that the msg gets logged and then sent to stderr after that. */
+		if (am_daemon > 0 && code != FCLIENT)
+			code = FLOG;
+	} else if (send_msgs_to_gen) {
 		assert(!is_utf8);
-		/* Pass the message to our sibling. */
+		/* Pass the message to our sibling in native charset. */
 		send_msg((enum msgcode)code, buf, len, 0);
 		return;
 	}
 
-	if (code == FSOCKERR) /* This gets simplified for a non-sibling. */
+	if (code == FERROR_SOCKET) /* This gets simplified for a non-sibling. */
 		code = FERROR;
+	else if (code == FERROR_UTF8) {
+		is_utf8 = 1;
+		code = FERROR;
+	}
 
 	if (code == FCLIENT)
 		code = FINFO;
 	else if (am_daemon || logfile_name) {
 		static int in_block;
 		char msg[2048];
-		int priority = code == FERROR ? LOG_WARNING : LOG_INFO;
+		int priority = code == FINFO || code == FLOG ? LOG_INFO :  LOG_WARNING;
 
 		if (in_block)
 			return;
@@ -283,34 +306,59 @@ void rwrite(enum logcode code, const char *buf, int len, int is_utf8)
 	} else if (code == FLOG)
 		return;
 
-	if (quiet && code != FERROR)
-		return;
-
-	if (am_server) {
-		/* Pass the message to the non-server side. */
-		if (send_msg((enum msgcode)code, buf, len, !is_utf8))
-			return;
-		if (am_daemon) {
-			/* TODO: can we send the error to the user somehow? */
-			return;
-		}
-	}
-
 	switch (code) {
+	case FERROR_XFER:
+		got_xfer_error = 1;
+		/* FALL THROUGH */
 	case FERROR:
-		log_got_error = 1;
+	case FWARNING:
 		f = stderr;
 		break;
 	case FINFO:
-		f = am_server ? stderr : stdout;
+		if (quiet)
+			return;
 		break;
+	/*case FLOG:*/
+	/*case FCLIENT:*/
+	/*case FERROR_UTF8:*/
+	/*case FERROR_SOCKET:*/
 	default:
+		fprintf(stderr, "Bad logcode in rwrite(): %d [%s]\n", (int)code, who_am_i());
 		exit_cleanup(RERR_MESSAGEIO);
 	}
 
+	if (am_server && !msgs2stderr) {
+		enum msgcode msg = (enum msgcode)code;
+		if (protocol_version < 30) {
+			if (msg == MSG_ERROR)
+				msg = MSG_ERROR_XFER;
+			else if (msg == MSG_WARNING)
+				msg = MSG_INFO;
+		}
+		/* Pass the message to the non-server side. */
+		if (send_msg(msg, buf, len, !is_utf8))
+			return;
+		if (am_daemon > 0) {
+			/* TODO: can we send the error to the user somehow? */
+			return;
+		}
+		f = stderr;
+	}
+
+	if (output_needs_newline) {
+		fputc('\n', f);
+		output_needs_newline = 0;
+	}
+
 	trailing_CR_or_NL = len && (buf[len-1] == '\n' || buf[len-1] == '\r')
 			  ? buf[--len] : 0;
 
+	if (len && buf[0] == '\r') {
+		fputc('\r', f);
+		buf++;
+		len--;
+	}
+
 #ifdef ICONV_CONST
 	if (ic != (iconv_t)-1) {
 		xbuf outbuf, inbuf;
@@ -318,32 +366,39 @@ void rwrite(enum logcode code, const char *buf, int len, int is_utf8)
 		int ierrno;
 
 		INIT_CONST_XBUF(outbuf, convbuf);
-		INIT_XBUF(inbuf, (char*)buf, len, -1);
+		INIT_XBUF(inbuf, (char*)buf, len, (size_t)-1);
 
 		while (inbuf.len) {
-			iconvbufs(ic, &inbuf, &outbuf, 0);
+			iconvbufs(ic, &inbuf, &outbuf, inbuf.pos ? 0 : ICB_INIT);
 			ierrno = errno;
 			if (outbuf.len) {
-				filtered_fwrite(f, convbuf, outbuf.len, 0);
+				filtered_fwrite(f, convbuf, outbuf.len, 0, 0);
 				outbuf.len = 0;
 			}
-			if (!ierrno || ierrno == E2BIG)
-				continue;
-			fprintf(f, "\\#%03o", CVAL(inbuf.buf, inbuf.pos++));
-			inbuf.len--;
+			/* Log one byte of illegal/incomplete sequence and continue with
+			 * the next character. Check that the buffer is non-empty for the
+			 * sake of robustness. */
+			if ((ierrno == EILSEQ || ierrno == EINVAL) && inbuf.len) {
+				fprintf(f, "\\#%03o", CVAL(inbuf.buf, inbuf.pos++));
+				inbuf.len--;
+			}
+		}
+
+		if (trailing_CR_or_NL) {
+			fputc(trailing_CR_or_NL, f);
+			fflush(f);
 		}
 	} else
 #endif
-		filtered_fwrite(f, buf, len, !allow_8bit_chars);
-
-	if (trailing_CR_or_NL) {
-		fputc(trailing_CR_or_NL, f);
-		fflush(f);
+	{
+		filtered_fwrite(f, buf, len, !allow_8bit_chars, trailing_CR_or_NL);
+		if (trailing_CR_or_NL)
+			fflush(f);
 	}
 }
 
-/* This is the rsync debugging function. Call it with FINFO, FERROR or
- * FLOG. */
+/* This is the rsync debugging function. Call it with FINFO, FERROR_*,
+ * FWARNING, FLOG, or FCLIENT. */
 void rprintf(enum logcode code, const char *format, ...)
 {
 	va_list ap;
@@ -397,8 +452,7 @@ void rsyserr(enum logcode code, int errcode, const char *format, ...)
 	char buf[BIGPATHBUFLEN];
 	size_t len;
 
-	strlcpy(buf, RSYNC_NAME ": ", sizeof buf);
-	len = (sizeof RSYNC_NAME ": ") - 1;
+	len = snprintf(buf, sizeof buf, RSYNC_NAME ": [%s] ", who_am_i());
 
 	va_start(ap, format);
 	len += vsnprintf(buf + len, sizeof buf - len, format, ap);
@@ -416,23 +470,28 @@ void rsyserr(enum logcode code, int errcode, const char *format, ...)
 
 void rflush(enum logcode code)
 {
-	FILE *f = NULL;
+	FILE *f;
 
 	if (am_daemon || code == FLOG)
 		return;
 
-	if (code == FERROR || am_server)
-		f = stderr;
-	else
+	if (!am_server && (code == FINFO || code == FCLIENT))
 		f = stdout;
+	else
+		f = stderr;
 
 	fflush(f);
 }
 
+void remember_initial_stats(void)
+{
+	initial_data_read = total_data_read;
+	initial_data_written = total_data_written;
+}
+
 /* A generic logging routine for send/recv, with parameter substitiution. */
 static void log_formatted(enum logcode code, const char *format, const char *op,
-			  struct file_struct *file, const char *fname,
-			  struct stats *initial_stats, int iflags,
+			  struct file_struct *file, const char *fname, int iflags,
 			  const char *hlink)
 {
 	char buf[MAXPATHLEN+1024], buf2[MAXPATHLEN], fmt[32];
@@ -455,30 +514,45 @@ static void log_formatted(enum logcode code, const char *format, const char *op,
 	buf[total] = '\0';
 
 	for (p = buf; (p = strchr(p, '%')) != NULL; ) {
+		int humanize = 0;
 		s = p++;
 		c = fmt + 1;
+		while (*p == '\'') {
+			humanize++;
+			p++;
+		}
 		if (*p == '-')
 			*c++ = *p++;
 		while (isDigit(p) && c - fmt < (int)(sizeof fmt) - 8)
 			*c++ = *p++;
+		while (*p == '\'') {
+			humanize++;
+			p++;
+		}
 		if (!*p)
 			break;
 		*c = '\0';
 		n = NULL;
 
+		/* Note for %h and %a: it doesn't matter what fd we pass to
+		 * client_{name,addr} because rsync_module will already have
+		 * forced the answer to be cached (assuming, of course, for %h
+		 * that lp_reverse_lookup(module_id) is true). */
 		switch (*p) {
 		case 'h':
-			if (am_daemon)
-				n = client_name(0);
+			if (am_daemon) {
+				n = lp_reverse_lookup(module_id)
+				  ? client_name(0) : undetermined_hostname;
+			}
 			break;
 		case 'a':
 			if (am_daemon)
 				n = client_addr(0);
 			break;
 		case 'l':
-			strlcat(fmt, ".0f", sizeof fmt);
+			strlcat(fmt, "s", sizeof fmt);
 			snprintf(buf2, sizeof buf2, fmt,
-				 (double)F_LENGTH(file));
+				 do_big_num(F_LENGTH(file), humanize, NULL));
 			n = buf2;
 			break;
 		case 'U':
@@ -498,14 +572,13 @@ static void log_formatted(enum logcode code, const char *format, const char *op,
 			}
 			break;
 		case 'p':
-			strlcat(fmt, "ld", sizeof fmt);
-			snprintf(buf2, sizeof buf2, fmt,
-				 (long)getpid());
+			strlcat(fmt, "d", sizeof fmt);
+			snprintf(buf2, sizeof buf2, fmt, (int)getpid());
 			n = buf2;
 			break;
 		case 'M':
 			n = c = timestring(file->modtime);
-			while ((c = strchr(p, ' ')) != NULL)
+			while ((c = strchr(c, ' ')) != NULL)
 				*c = '-';
 			break;
 		case 'B':
@@ -581,57 +654,70 @@ static void log_formatted(enum logcode code, const char *format, const char *op,
 			n = timestring(time(NULL));
 			break;
 		case 'P':
-			n = module_dir;
+			n = full_module_path;
 			break;
 		case 'u':
 			n = auth_user;
 			break;
 		case 'b':
-			if (am_sender) {
-				b = stats.total_written -
-					initial_stats->total_written;
-			} else {
-				b = stats.total_read -
-					initial_stats->total_read;
-			}
-			strlcat(fmt, ".0f", sizeof fmt);
-			snprintf(buf2, sizeof buf2, fmt, (double)b);
+		case 'c':
+			if (!(iflags & ITEM_TRANSFER))
+				b = 0;
+			else if ((!!am_sender) ^ (*p == 'c'))
+				b = total_data_written - initial_data_written;
+			else
+				b = total_data_read - initial_data_read;
+			strlcat(fmt, "s", sizeof fmt);
+			snprintf(buf2, sizeof buf2, fmt,
+				 do_big_num(b, humanize, NULL));
 			n = buf2;
 			break;
-		case 'c':
-			if (!am_sender) {
-				b = stats.total_written -
-					initial_stats->total_written;
-			} else {
-				b = stats.total_read -
-					initial_stats->total_read;
+		case 'C':
+			n = NULL;
+			if (S_ISREG(file->mode)) {
+				if (always_checksum)
+					n = sum_as_hex(checksum_type, F_SUM(file), 1);
+				else if (iflags & ITEM_TRANSFER)
+					n = sum_as_hex(xfersum_type, sender_file_sum, 0);
+			}
+			if (!n) {
+				int sum_len = csum_len_for_type(always_checksum ? checksum_type : xfersum_type,
+								always_checksum);
+				memset(buf2, ' ', sum_len*2);
+				buf2[sum_len*2] = '\0';
+				n = buf2;
 			}
-			strlcat(fmt, ".0f", sizeof fmt);
-			snprintf(buf2, sizeof buf2, fmt, (double)b);
-			n = buf2;
 			break;
 		case 'i':
 			if (iflags & ITEM_DELETED) {
-				n = "*deleting";
+				n = "*deleting  ";
 				break;
 			}
 			n  = c = buf2 + MAXPATHLEN - 32;
 			c[0] = iflags & ITEM_LOCAL_CHANGE
-			      ? iflags & ITEM_XNAME_FOLLOWS ? 'h' : 'c'
+			     ? iflags & ITEM_XNAME_FOLLOWS ? 'h' : 'c'
 			     : !(iflags & ITEM_TRANSFER) ? '.'
 			     : !local_server && *op == 's' ? '<' : '>';
-			c[1] = S_ISDIR(file->mode) ? 'd'
-			     : IS_SPECIAL(file->mode) ? 'S'
-			     : IS_DEVICE(file->mode) ? 'D'
-			     : S_ISLNK(file->mode) ? 'L' : 'f';
-			c[2] = !(iflags & ITEM_REPORT_CHECKSUM) ? '.' : 'c';
-			c[3] = !(iflags & ITEM_REPORT_SIZE) ? '.' : 's';
-			c[4] = !(iflags & ITEM_REPORT_TIME) ? '.'
-			     : !preserve_times || S_ISLNK(file->mode) ? 'T' : 't';
+			if (S_ISLNK(file->mode)) {
+				c[1] = 'L';
+				c[3] = '.';
+				c[4] = !(iflags & ITEM_REPORT_TIME) ? '.'
+				     : !preserve_times || !receiver_symlink_times
+				    || (iflags & ITEM_REPORT_TIMEFAIL) ? 'T' : 't';
+			} else {
+				c[1] = S_ISDIR(file->mode) ? 'd'
+				     : IS_SPECIAL(file->mode) ? 'S'
+				     : IS_DEVICE(file->mode) ? 'D' : 'f';
+				c[3] = !(iflags & ITEM_REPORT_SIZE) ? '.' : 's';
+				c[4] = !(iflags & ITEM_REPORT_TIME) ? '.'
+				     : !preserve_times ? 'T' : 't';
+			}
+			c[2] = !(iflags & ITEM_REPORT_CHANGE) ? '.' : 'c';
 			c[5] = !(iflags & ITEM_REPORT_PERMS) ? '.' : 'p';
 			c[6] = !(iflags & ITEM_REPORT_OWNER) ? '.' : 'o';
 			c[7] = !(iflags & ITEM_REPORT_GROUP) ? '.' : 'g';
-			c[8] = !(iflags & ITEM_REPORT_ATIME) ? '.' : 'u';
+			c[8] = !(iflags & ITEM_REPORT_ATIME) ? '.'
+			     : S_ISLNK(file->mode) ? 'U' : 'u';
 			c[9] = !(iflags & ITEM_REPORT_ACL) ? '.' : 'a';
 			c[10] = !(iflags & ITEM_REPORT_XATTR) ? '.' : 'x';
 			c[11] = '\0';
@@ -701,10 +787,12 @@ int log_format_has(const char *format, char esc)
 		return 0;
 
 	for (p = format; (p = strchr(p, '%')) != NULL; ) {
-		if (*++p == '-')
+		for (p++; *p == '\''; p++) {} /*SHARED ITERATOR*/
+		if (*p == '-')
 			p++;
 		while (isDigit(p))
 			p++;
+		while (*p == '\'') p++;
 		if (!*p)
 			break;
 		if (*p == esc)
@@ -716,36 +804,30 @@ int log_format_has(const char *format, char esc)
 /* Log the transfer of a file.  If the code is FCLIENT, the output just goes
  * to stdout.  If it is FLOG, it just goes to the log file.  Otherwise we
  * output to both. */
-void log_item(enum logcode code, struct file_struct *file,
-	      struct stats *initial_stats, int iflags, const char *hlink)
+void log_item(enum logcode code, struct file_struct *file, int iflags, const char *hlink)
 {
 	const char *s_or_r = am_sender ? "send" : "recv";
 
-	if (code != FLOG && stdout_format && !am_server) {
-		log_formatted(FCLIENT, stdout_format, s_or_r,
-			      file, NULL, initial_stats, iflags, hlink);
-	}
-	if (code != FCLIENT && logfile_format && *logfile_format) {
-		log_formatted(FLOG, logfile_format, s_or_r,
-			      file, NULL, initial_stats, iflags, hlink);
-	}
+	if (code != FLOG && stdout_format && !am_server)
+		log_formatted(FCLIENT, stdout_format, s_or_r, file, NULL, iflags, hlink);
+	if (code != FCLIENT && logfile_format && *logfile_format)
+		log_formatted(FLOG, logfile_format, s_or_r, file, NULL, iflags, hlink);
 }
 
-void maybe_log_item(struct file_struct *file, int iflags, int itemizing,
-		    const char *buf)
+void maybe_log_item(struct file_struct *file, int iflags, int itemizing, const char *buf)
 {
 	int significant_flags = iflags & SIGNIFICANT_ITEM_FLAGS;
 	int see_item = itemizing && (significant_flags || *buf
-		|| stdout_format_has_i > 1 || (verbose > 1 && stdout_format_has_i));
+		|| stdout_format_has_i > 1 || (INFO_GTE(NAME, 2) && stdout_format_has_i));
 	int local_change = iflags & ITEM_LOCAL_CHANGE && significant_flags;
 	if (am_server) {
 		if (logfile_name && !dry_run && see_item
 		 && (significant_flags || logfile_format_has_i))
-			log_item(FLOG, file, &stats, iflags, buf);
+			log_item(FLOG, file, iflags, buf);
 	} else if (see_item || local_change || *buf
 	    || (S_ISDIR(file->mode) && significant_flags)) {
 		enum logcode code = significant_flags || logfile_format_has_i ? FINFO : FCLIENT;
-		log_item(code, file, &stats, iflags, buf);
+		log_item(code, file, iflags, buf);
 	}
 }
 
@@ -754,29 +836,28 @@ void log_delete(const char *fname, int mode)
 	static struct {
 		union file_extras ex[4]; /* just in case... */
 		struct file_struct file;
-	} x;
+	} x; /* Zero-initialized due to static declaration. */
 	int len = strlen(fname);
 	const char *fmt;
 
 	x.file.mode = mode;
 
-	if (!verbose && !stdout_format)
-		;
-	else if (am_server && protocol_version >= 29 && len < MAXPATHLEN) {
+	if (am_server && protocol_version >= 29 && len < MAXPATHLEN) {
 		if (S_ISDIR(mode))
 			len++; /* directories include trailing null */
 		send_msg(MSG_DELETED, fname, len, am_generator);
-	} else {
+	} else if (!INFO_GTE(DEL, 1) && !stdout_format)
+		;
+	else {
 		fmt = stdout_format_has_o_or_i ? stdout_format : "deleting %n";
-		log_formatted(FCLIENT, fmt, "del.", &x.file, fname, &stats,
-			      ITEM_DELETED, NULL);
+		log_formatted(FCLIENT, fmt, "del.", &x.file, fname, ITEM_DELETED, NULL);
 	}
 
 	if (!logfile_name || dry_run || !logfile_format)
 		return;
 
 	fmt = logfile_format_has_o_or_i ? logfile_format : "deleting %n";
-	log_formatted(FLOG, fmt, "del.", &x.file, fname, &stats, ITEM_DELETED, NULL);
+	log_formatted(FLOG, fmt, "del.", &x.file, fname, ITEM_DELETED, NULL);
 }
 
 /*
@@ -787,12 +868,15 @@ void log_delete(const char *fname, int mode)
  */
 void log_exit(int code, const char *file, int line)
 {
-	if (code == 0) {
-		rprintf(FLOG,"sent %.0f bytes  received %.0f bytes  total size %.0f\n",
-			(double)stats.total_written,
-			(double)stats.total_read,
-			(double)stats.total_size);
-	} else if (am_server != 2) {
+	/* The receiving side's stats are split between 2 procs until the
+	 * end of the run, so only the sender can output non-final info. */
+	if (code == 0 || am_sender) {
+		rprintf(FLOG,"sent %s bytes  received %s bytes  total size %s\n",
+			big_num(stats.total_written),
+			big_num(stats.total_read),
+			big_num(stats.total_size));
+	}
+	if (code != 0 && am_server != 2) {
 		const char *name;
 
 		name = rerr_name(code);
@@ -801,7 +885,7 @@ void log_exit(int code, const char *file, int line)
 
 		/* VANISHED is not an error, only a warning */
 		if (code == RERR_VANISHED) {
-			rprintf(FINFO, "rsync warning: %s (code %d) at %s(%d) [%s=%s]\n",
+			rprintf(FWARNING, "rsync warning: %s (code %d) at %s(%d) [%s=%s]\n",
 				name, code, file, line, who_am_i(), RSYNC_VERSION);
 		} else {
 			rprintf(FERROR, "rsync error: %s (code %d) at %s(%d) [%s=%s]\n",

m4/have_type.m4

@@ -0,0 +1,22 @@
+dnl AC_HAVE_TYPE(TYPE,INCLUDES)
+AC_DEFUN([AC_HAVE_TYPE], [
+AC_REQUIRE([AC_HEADER_STDC])
+cv=`echo "$1" | sed 'y%./+- %__p__%'`
+AC_MSG_CHECKING(for $1)
+AC_CACHE_VAL([ac_cv_type_$cv],
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[
+AC_INCLUDES_DEFAULT
+$2]],
+[[$1 foo;]])],
+[eval "ac_cv_type_$cv=yes"],
+[eval "ac_cv_type_$cv=no"]))dnl
+ac_foo=`eval echo \\$ac_cv_type_$cv`
+AC_MSG_RESULT($ac_foo)
+if test "$ac_foo" = yes; then
+  ac_tr_hdr=HAVE_`echo $1 | sed 'y%abcdefghijklmnopqrstuvwxyz./- %ABCDEFGHIJKLMNOPQRSTUVWXYZ____%'`
+if false; then
+	AC_CHECK_TYPES($1)
+fi
+  AC_DEFINE_UNQUOTED($ac_tr_hdr, 1, [Define if you have type `$1'])
+fi
+])

m4/header_major_fixed.m4

@@ -0,0 +1,27 @@
+AC_DEFUN([AC_HEADER_MAJOR_FIXED],
+[AC_CACHE_CHECK(whether sys/types.h defines makedev,
+		ac_cv_header_sys_types_h_makedev,
+[AC_LINK_IFELSE([AC_LANG_PROGRAM([[@%:@include <sys/types.h>]],
+				 [[return makedev(0, 0);]])],
+		[if grep sys/sysmacros.h conftest.err >/dev/null; then
+		   ac_cv_header_sys_types_h_makedev=no
+		 else
+		   ac_cv_header_sys_types_h_makedev=yes
+		 fi],
+		[ac_cv_header_sys_types_h_makedev=no])
+])
+
+if test $ac_cv_header_sys_types_h_makedev = no; then
+AC_CHECK_HEADER(sys/mkdev.h,
+		[AC_DEFINE(MAJOR_IN_MKDEV, 1,
+			   [Define to 1 if `major', `minor', and `makedev' are
+			    declared in <mkdev.h>.])])
+
+  if test $ac_cv_header_sys_mkdev_h = no; then
+    AC_CHECK_HEADER(sys/sysmacros.h,
+		    [AC_DEFINE(MAJOR_IN_SYSMACROS, 1,
+			       [Define to 1 if `major', `minor', and `makedev'
+				are declared in <sysmacros.h>.])])
+  fi
+fi
+])

m4/socklen_t.m4

@@ -1,27 +1,3 @@
-dnl AC_VALIDATE_CACHE_SYSTEM_TYPE[(cmd)]
-dnl if the cache file is inconsistent with the current host,
-dnl target and build system types, execute CMD or print a default
-dnl error message.
-AC_DEFUN(AC_VALIDATE_CACHE_SYSTEM_TYPE, [
-    AC_REQUIRE([AC_CANONICAL_SYSTEM])
-    AC_MSG_CHECKING([config.cache system type])
-    if { test x"${ac_cv_host_system_type+set}" = x"set" &&
-         test x"$ac_cv_host_system_type" != x"$host"; } ||
-       { test x"${ac_cv_build_system_type+set}" = x"set" &&
-         test x"$ac_cv_build_system_type" != x"$build"; } ||
-       { test x"${ac_cv_target_system_type+set}" = x"set" &&
-         test x"$ac_cv_target_system_type" != x"$target"; }; then
-	AC_MSG_RESULT([different])
-	ifelse($#, 1, [$1],
-		[AC_MSG_ERROR(["you must remove config.cache and restart configure"])])
-    else
-	AC_MSG_RESULT([same])
-    fi
-    ac_cv_host_system_type="$host"
-    ac_cv_build_system_type="$build"
-    ac_cv_target_system_type="$target"
-])
-
 dnl Check for socklen_t: historically on BSD it is an int, and in
 dnl POSIX 1g it is a type of its own, but some platforms use different
 dnl types for the argument to getsockopt, getpeername, etc.  So we
@@ -42,15 +18,15 @@ AC_DEFUN([TYPE_SOCKLEN_T],
          rsync_cv_socklen_t_equiv=
          for arg2 in "struct sockaddr" void; do
             for t in int size_t unsigned long "unsigned long"; do
-               AC_TRY_COMPILE([
+               AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[
 #include <sys/types.h>
 #include <sys/socket.h>
 
                   int getpeername (int, $arg2 *, $t *);
-               ],[
+               ]],[[
                   $t len;
                   getpeername(0,0,&len);
-               ],[
+               ]])],[
                   rsync_cv_socklen_t_equiv="$t"
                   break
                ])
@@ -67,5 +43,3 @@ AC_DEFUN([TYPE_SOCKLEN_T],
       [#include <sys/types.h>
 #include <sys/socket.h>])
 ])
-
-

m4/validate_cache_system_type.m4

@@ -0,0 +1,23 @@
+dnl AC_VALIDATE_CACHE_SYSTEM_TYPE[(cmd)]
+dnl if the cache file is inconsistent with the current host,
+dnl target and build system types, execute CMD or print a default
+dnl error message.
+AC_DEFUN([AC_VALIDATE_CACHE_SYSTEM_TYPE], [
+    AC_REQUIRE([AC_CANONICAL_SYSTEM])
+    AC_MSG_CHECKING([config.cache system type])
+    if { test x"${ac_cv_host_system_type+set}" = x"set" &&
+         test x"$ac_cv_host_system_type" != x"$host"; } ||
+       { test x"${ac_cv_build_system_type+set}" = x"set" &&
+         test x"$ac_cv_build_system_type" != x"$build"; } ||
+       { test x"${ac_cv_target_system_type+set}" = x"set" &&
+         test x"$ac_cv_target_system_type" != x"$target"; }; then
+	AC_MSG_RESULT([different])
+	ifelse($#, 1, [$1],
+		[AC_MSG_ERROR(["you must remove config.cache and restart configure"])])
+    else
+	AC_MSG_RESULT([same])
+    fi
+    ac_cv_host_system_type="$host"
+    ac_cv_build_system_type="$build"
+    ac_cv_target_system_type="$target"
+])

match.c

@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1996 Andrew Tridgell
  * Copyright (C) 1996 Paul Mackerras
- * Copyright (C) 2003-2007 Wayne Davison
+ * Copyright (C) 2003-2020 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
@@ -20,13 +20,14 @@
  */
 
 #include "rsync.h"
+#include "inums.h"
 
-extern int verbose;
-extern int do_progress;
 extern int checksum_seed;
 extern int append_mode;
+extern int xfersum_type;
 
 int updating_basis_file;
+char sender_file_sum[MAX_DIGEST_LEN];
 
 static int false_alarms;
 static int hash_hits;
@@ -39,29 +40,50 @@ static int total_matches;
 
 extern struct stats stats;
 
-#define TABLESIZE (1<<16)
+#define TRADITIONAL_TABLESIZE (1<<16)
 
+static uint32 tablesize;
 static int32 *hash_table;
 
 #define SUM2HASH2(s1,s2) (((s1) + (s2)) & 0xFFFF)
 #define SUM2HASH(sum) SUM2HASH2((sum)&0xFFFF,(sum)>>16)
 
+#define BIG_SUM2HASH(sum) ((sum)%tablesize)
+
 static void build_hash_table(struct sum_struct *s)
 {
+	static uint32 alloc_size;
 	int32 i;
 
-	if (!hash_table) {
-		hash_table = new_array(int32, TABLESIZE);
+	/* Dynamically calculate the hash table size so that the hash load
+	 * for big files is about 80%.  A number greater than the traditional
+	 * size must be odd or s2 will not be able to span the entire set. */
+	tablesize = (uint32)(s->count/8) * 10 + 11;
+	if (tablesize < TRADITIONAL_TABLESIZE)
+		tablesize = TRADITIONAL_TABLESIZE;
+	if (tablesize > alloc_size || tablesize < alloc_size - 16*1024) {
+		if (hash_table)
+			free(hash_table);
+		hash_table = new_array(int32, tablesize);
 		if (!hash_table)
 			out_of_memory("build_hash_table");
+		alloc_size = tablesize;
 	}
 
-	memset(hash_table, 0xFF, TABLESIZE * sizeof hash_table[0]);
+	memset(hash_table, 0xFF, tablesize * sizeof hash_table[0]);
 
-	for (i = 0; i < s->count; i++) {
-		uint32 t = SUM2HASH(s->sums[i].sum1);
-		s->sums[i].chain = hash_table[t];
-		hash_table[t] = i;
+	if (tablesize == TRADITIONAL_TABLESIZE) {
+		for (i = 0; i < s->count; i++) {
+			uint32 t = SUM2HASH(s->sums[i].sum1);
+			s->sums[i].chain = hash_table[t];
+			hash_table[t] = i;
+		}
+	} else {
+		for (i = 0; i < s->count; i++) {
+			uint32 t = BIG_SUM2HASH(s->sums[i].sum1);
+			s->sums[i].chain = hash_table[t];
+			hash_table[t] = i;
+		}
 	}
 }
 
@@ -69,8 +91,7 @@ static void build_hash_table(struct sum_struct *s)
 static OFF_T last_match;
 
 
-/**
- * Transmit a literal and/or match token.
+/* Transmit a literal and/or match token.
  *
  * This delightfully-named function is called either when we find a
  * match and need to transmit all the unmatched data leading up to it,
@@ -78,19 +99,18 @@ static OFF_T last_match;
  * transmit it.  As a result of this second case, it is called even if
  * we have not matched at all!
  *
- * @param i If >0, the number of a matched token.  If 0, indicates we
- * have only literal data.
- **/
-static void matched(int f, struct sum_struct *s, struct map_struct *buf,
-		    OFF_T offset, int32 i)
+ * If i >= 0, the number of a matched token.  If < 0, indicates we have
+ * only literal data.  A -1 will send a 0-token-int too, and a -2 sends
+ * only literal data, w/o any token-int. */
+static void matched(int f, struct sum_struct *s, struct map_struct *buf, OFF_T offset, int32 i)
 {
 	int32 n = (int32)(offset - last_match); /* max value: block_size (int32) */
 	int32 j;
 
-	if (verbose > 2 && i >= 0) {
+	if (DEBUG_GTE(DELTASUM, 2) && i >= 0) {
 		rprintf(FINFO,
-			"match at %.0f last_match=%.0f j=%d len=%ld n=%ld\n",
-			(double)offset, (double)last_match, i,
+			"match at %s last_match=%s j=%d len=%ld n=%ld\n",
+			big_num(offset), big_num(last_match), i,
 			(long)s->sums[i].len, (long)n);
 	}
 
@@ -112,7 +132,7 @@ static void matched(int f, struct sum_struct *s, struct map_struct *buf,
 	else
 		last_match = offset;
 
-	if (buf && do_progress)
+	if (buf && INFO_GTE(PROGRESS, 1))
 		show_progress(last_match, buf->file_size);
 }
 
@@ -120,8 +140,8 @@ static void matched(int f, struct sum_struct *s, struct map_struct *buf,
 static void hash_search(int f,struct sum_struct *s,
 			struct map_struct *buf, OFF_T len)
 {
-	OFF_T offset, end;
-	int32 k, want_i, backup;
+	OFF_T offset, aligned_offset, end;
+	int32 k, want_i, aligned_i, backup;
 	char sum2[SUM_LENGTH];
 	uint32 s1, s2, sum;
 	int more;
@@ -131,9 +151,9 @@ static void hash_search(int f,struct sum_struct *s,
 	 * coding of the output to work more efficiently. */
 	want_i = 0;
 
-	if (verbose > 2) {
-		rprintf(FINFO, "hash search b=%ld len=%.0f\n",
-			(long)s->blength, (double)len);
+	if (DEBUG_GTE(DELTASUM, 2)) {
+		rprintf(FINFO, "hash search b=%ld len=%s\n",
+			(long)s->blength, big_num(len));
 	}
 
 	k = (int32)MIN(len, (OFF_T)s->blength);
@@ -143,36 +163,55 @@ static void hash_search(int f,struct sum_struct *s,
 	sum = get_checksum1((char *)map, k);
 	s1 = sum & 0xFFFF;
 	s2 = sum >> 16;
-	if (verbose > 3)
+	if (DEBUG_GTE(DELTASUM, 3))
 		rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k);
 
-	offset = 0;
+	offset = aligned_offset = aligned_i = 0;
 
 	end = len + 1 - s->sums[s->count-1].len;
 
-	if (verbose > 3) {
-		rprintf(FINFO, "hash search s->blength=%ld len=%.0f count=%.0f\n",
-			(long)s->blength, (double)len, (double)s->count);
+	if (DEBUG_GTE(DELTASUM, 3)) {
+		rprintf(FINFO, "hash search s->blength=%ld len=%s count=%s\n",
+			(long)s->blength, big_num(len), big_num(s->count));
 	}
 
 	do {
 		int done_csum2 = 0;
-		int32 i;
+		uint32 hash_entry;
+		int32 i, *prev;
 
-		if (verbose > 4) {
-			rprintf(FINFO, "offset=%.0f sum=%04x%04x\n",
-				(double)offset, s2 & 0xFFFF, s1 & 0xFFFF);
+		if (DEBUG_GTE(DELTASUM, 4)) {
+			rprintf(FINFO, "offset=%s sum=%04x%04x\n",
+				big_num(offset), s2 & 0xFFFF, s1 & 0xFFFF);
 		}
 
-		i = hash_table[SUM2HASH2(s1,s2)];
-		if (i < 0)
-			goto null_hash;
+		if (tablesize == TRADITIONAL_TABLESIZE) {
+			hash_entry = SUM2HASH2(s1,s2);
+			if ((i = hash_table[hash_entry]) < 0)
+				goto null_hash;
+			sum = (s1 & 0xffff) | (s2 << 16);
+		} else {
+			sum = (s1 & 0xffff) | (s2 << 16);
+			hash_entry = BIG_SUM2HASH(sum);
+			if ((i = hash_table[hash_entry]) < 0)
+				goto null_hash;
+		}
+		prev = &hash_table[hash_entry];
 
-		sum = (s1 & 0xffff) | (s2 << 16);
 		hash_hits++;
 		do {
 			int32 l;
 
+			/* When updating in-place, the chunk's offset must be
+			 * either >= our offset or identical data at that offset.
+			 * Remove any bypassed entries that we can never use. */
+			if (updating_basis_file && s->sums[i].offset < offset
+			 && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) {
+				*prev = s->sums[i].chain;
+				continue;
+			}
+			prev = &s->sums[i].chain;
+
 			if (sum != s->sums[i].sum1)
 				continue;
 
@@ -181,16 +220,10 @@ static void hash_search(int f,struct sum_struct *s,
 			if (l != s->sums[i].len)
 				continue;
 
-			/* in-place: ensure chunk's offset is either >= our
-			 * offset or that the data didn't move. */
-			if (updating_basis_file && s->sums[i].offset < offset
-			    && !(s->sums[i].flags & SUMFLG_SAME_OFFSET))
-				continue;
-
-			if (verbose > 3) {
+			if (DEBUG_GTE(DELTASUM, 3)) {
 				rprintf(FINFO,
-					"potential match at %.0f i=%ld sum=%08x\n",
-					(double)offset, (long)i, sum);
+					"potential match at %s i=%ld sum=%08x\n",
+					big_num(offset), (long)i, sum);
 			}
 
 			if (!done_csum2) {
@@ -206,39 +239,62 @@ static void hash_search(int f,struct sum_struct *s,
 
 			/* When updating in-place, the best possible match is
 			 * one with an identical offset, so we prefer that over
-			 * the following want_i optimization. */
+			 * the adjacent want_i optimization. */
 			if (updating_basis_file) {
-				int32 i2;
-				for (i2 = i; i2 >= 0; i2 = s->sums[i2].chain) {
-					if (s->sums[i2].offset != offset)
-						continue;
-					if (i2 != i) {
-						if (sum != s->sums[i2].sum1)
-							break;
-						if (memcmp(sum2, s->sums[i2].sum2,
-							   s->s2length) != 0)
-							break;
-						i = i2;
+				/* All the generator's chunks start at blength boundaries. */
+				while (aligned_offset < offset) {
+					aligned_offset += s->blength;
+					aligned_i++;
+				}
+				if ((offset == aligned_offset
+				  || (sum == 0 && l == s->blength && aligned_offset + l <= len))
+				 && aligned_i < s->count) {
+					if (i != aligned_i) {
+						if (sum != s->sums[aligned_i].sum1
+						 || l != s->sums[aligned_i].len
+						 || memcmp(sum2, s->sums[aligned_i].sum2, s->s2length) != 0)
+							goto check_want_i;
+						i = aligned_i;
 					}
-					/* This chunk was at the same offset on
-					 * both the sender and the receiver. */
+					if (offset != aligned_offset) {
+						/* We've matched some zeros in a spot that is also zeros
+						 * further along in the basis file, if we find zeros ahead
+						 * in the sender's file, we'll output enough literal data
+						 * to re-align with the basis file, and get back to seeking
+						 * instead of writing. */
+						backup = (int32)(aligned_offset - last_match);
+						if (backup < 0)
+							backup = 0;
+						map = (schar *)map_ptr(buf, aligned_offset - backup, l + backup)
+						    + backup;
+						sum = get_checksum1((char *)map, l);
+						if (sum != s->sums[i].sum1)
+							goto check_want_i;
+						get_checksum2((char *)map, l, sum2);
+						if (memcmp(sum2, s->sums[i].sum2, s->s2length) != 0)
+							goto check_want_i;
+						/* OK, we have a re-alignment match.  Bump the offset
+						 * forward to the new match point. */
+						offset = aligned_offset;
+					}
+					/* This identical chunk is in the same spot in the old and new file. */
 					s->sums[i].flags |= SUMFLG_SAME_OFFSET;
-					goto set_want_i;
+					want_i = i;
 				}
 			}
 
+		  check_want_i:
 			/* we've found a match, but now check to see
 			 * if want_i can hint at a better match. */
 			if (i != want_i && want_i < s->count
-			    && (!updating_basis_file || s->sums[want_i].offset >= offset
-			     || s->sums[want_i].flags & SUMFLG_SAME_OFFSET)
-			    && sum == s->sums[want_i].sum1
-			    && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) {
+			 && (!updating_basis_file || s->sums[want_i].offset >= offset
+			  || s->sums[want_i].flags & SUMFLG_SAME_OFFSET)
+			 && sum == s->sums[want_i].sum1
+			 && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) {
 				/* we've found an adjacent match - the RLL coder
 				 * will be happy */
 				i = want_i;
 			}
-		    set_want_i:
 			want_i = i + 1;
 
 			matched(f,s,buf,offset,i);
@@ -260,8 +316,7 @@ static void hash_search(int f,struct sum_struct *s,
 
 		/* Trim off the first byte from the checksum */
 		more = offset + k < len;
-		map = (schar *)map_ptr(buf, offset - backup, k + more + backup)
-		    + backup;
+		map = (schar *)map_ptr(buf, offset - backup, k + more + backup) + backup;
 		s1 -= map[0] + CHAR_OFFSET;
 		s2 -= k * (map[0]+CHAR_OFFSET);
 
@@ -303,7 +358,6 @@ static void hash_search(int f,struct sum_struct *s,
  **/
 void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len)
 {
-	char file_sum[MAX_DIGEST_LEN];
 	int sum_len;
 
 	last_match = 0;
@@ -312,13 +366,13 @@ void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len)
 	matches = 0;
 	data_transfer = 0;
 
-	sum_init(checksum_seed);
+	sum_init(xfersum_type, checksum_seed);
 
 	if (append_mode > 0) {
 		if (append_mode == 2) {
 			OFF_T j = 0;
 			for (j = CHUNK_SIZE; j < s->flength; j += CHUNK_SIZE) {
-				if (buf && do_progress)
+				if (buf && INFO_GTE(PROGRESS, 1))
 					show_progress(last_match, buf->file_size);
 				sum_update(map_ptr(buf, last_match, CHUNK_SIZE),
 					   CHUNK_SIZE);
@@ -326,7 +380,7 @@ void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len)
 			}
 			if (last_match < s->flength) {
 				int32 n = (int32)(s->flength - last_match);
-				if (buf && do_progress)
+				if (buf && INFO_GTE(PROGRESS, 1))
 					show_progress(last_match, buf->file_size);
 				sum_update(map_ptr(buf, last_match, n), n);
 			}
@@ -338,12 +392,12 @@ void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len)
 	if (len > 0 && s->count > 0) {
 		build_hash_table(s);
 
-		if (verbose > 2)
+		if (DEBUG_GTE(DELTASUM, 2))
 			rprintf(FINFO,"built hash table\n");
 
 		hash_search(f, s, buf, len);
 
-		if (verbose > 2)
+		if (DEBUG_GTE(DELTASUM, 2))
 			rprintf(FINFO,"done hash search\n");
 	} else {
 		OFF_T j;
@@ -353,18 +407,27 @@ void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len)
 		matched(f, s, buf, len, -1);
 	}
 
-	sum_len = sum_end(file_sum);
-	/* If we had a read error, send a bad checksum. */
-	if (buf && buf->status != 0)
-		file_sum[0]++;
+	sum_len = sum_end(sender_file_sum);
 
-	if (verbose > 2)
+	/* If we had a read error, send a bad checksum.  We use all bits
+	 * off as long as the checksum doesn't happen to be that, in
+	 * which case we turn the last 0 bit into a 1. */
+	if (buf && buf->status != 0) {
+		int i;
+		for (i = 0; i < sum_len && sender_file_sum[i] == 0; i++) {}
+		memset(sender_file_sum, 0, sum_len);
+		if (i == sum_len)
+			sender_file_sum[i-1]++;
+	}
+
+	if (DEBUG_GTE(DELTASUM, 2))
 		rprintf(FINFO,"sending file_sum\n");
-	write_buf(f, file_sum, sum_len);
+	write_buf(f, sender_file_sum, sum_len);
 
-	if (verbose > 2)
+	if (DEBUG_GTE(DELTASUM, 2)) {
 		rprintf(FINFO, "false_alarms=%d hash_hits=%d matches=%d\n",
 			false_alarms, hash_hits, matches);
+	}
 
 	total_hash_hits += hash_hits;
 	total_false_alarms += false_alarms;
@@ -374,11 +437,11 @@ void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len)
 
 void match_report(void)
 {
-	if (verbose <= 1)
+	if (!DEBUG_GTE(DELTASUM, 1))
 		return;
 
 	rprintf(FINFO,
-		"total: matches=%d  hash_hits=%d  false_alarms=%d data=%.0f\n",
+		"total: matches=%d  hash_hits=%d  false_alarms=%d data=%s\n",
 		total_matches, total_hash_hits, total_false_alarms,
-		(double)stats.literal_data);
+		big_num(stats.literal_data));
 }

maybe-make-man

@@ -0,0 +1,40 @@
+#!/bin/sh
+
+if [ x"$2" = x ]; then
+    echo "Usage: $0 SRC_DIR NAME.NUM.md" 1>&2
+    exit 1
+fi
+
+srcdir="$1"
+inname="$2"
+flagfile="$srcdir/.md2man-works"
+
+if [ ! -d "$srcdir" ]; then
+    echo "The specified SRC_DIR is not a directory: $srcdir" 1>&2
+    exit 1
+fi
+
+if [ ! -f "$flagfile" ]; then
+    # We test our smallest manpage just to see if the python setup works.
+    if "$srcdir/md2man" --test "$srcdir/rsync-ssl.1.md" >/dev/null 2>&1; then
+	touch $flagfile
+    else
+	outname=`echo "$inname" | sed 's/\.md$//'`
+	if [ -f "$outname" ]; then
+	    exit 0
+	elif [ -f "$srcdir/$outname" ]; then
+	    echo "Copying $srcdir/$outname"
+	    cp -p "$srcdir/$outname" .
+	    exit 0
+	else
+	    echo "ERROR: $outname cannot be created."
+	    if [ -f "$HOME/build_farm/build_test.fns" ]; then
+		exit 0 # No exit errorno to avoid a build failure in the samba build farm
+	    else
+		exit 1
+	    fi
+	fi
+    fi
+fi
+
+"$srcdir/md2man" "$srcdir/$inname"

md2man

@@ -0,0 +1,381 @@
+#!/usr/bin/env python3
+
+# This script takes a manpage written in markdown and turns it into an html web
+# page and a nroff man page.  The input file must have the name of the program
+# and the section in this format: NAME.NUM.md.  The output files are written
+# into the current directory named NAME.NUM.html and NAME.NUM.  The input
+# format has one extra extension: if a numbered list starts at 0, it is turned
+# into a description list. The dl's dt tag is taken from the contents of the
+# first tag inside the li, which is usually a p, code, or strong tag.  The
+# cmarkgfm or commonmark lib is used to transforms the input file into html.
+# The html.parser is used as a state machine that both tweaks the html and
+# outputs the nroff data based on the html tags.
+#
+# Copyright (C) 2020 Wayne Davison
+#
+# This program is freely redistributable.
+
+import sys, os, re, argparse, subprocess, time
+from html.parser import HTMLParser
+
+CONSUMES_TXT = set('h1 h2 p li pre'.split())
+
+HTML_START = """\
+<html><head>
+<title>%s</title>
+<link href="https://fonts.googleapis.com/css2?family=Roboto&family=Roboto+Mono&display=swap" rel="stylesheet">
+<style>
+body {
+  max-width: 50em;
+  margin: auto;
+}
+body, b, strong, u {
+  font-family: 'Roboto', sans-serif;
+}
+code {
+  font-family: 'Roboto Mono', monospace;
+  font-weight: bold;
+  white-space: pre;
+}
+pre code {
+  display: block;
+  font-weight: normal;
+}
+blockquote pre code {
+  background: #f1f1f1;
+}
+dd p:first-of-type {
+  margin-block-start: 0em;
+}
+</style>
+</head><body>
+"""
+
+HTML_END = """\
+<div style="float: right"><p><i>%s</i></p></div>
+</body></html>
+"""
+
+MAN_START = r"""
+.TH "%s" "%s" "%s" "%s" "User Commands"
+""".lstrip()
+
+MAN_END = """\
+"""
+
+NORM_FONT = ('\1', r"\fP")
+BOLD_FONT = ('\2', r"\fB")
+UNDR_FONT = ('\3', r"\fI")
+NBR_DASH = ('\4', r"\-")
+NBR_SPACE = ('\xa0', r"\ ")
+
+md_parser = None
+
+def main():
+    fi = re.match(r'^(?P<fn>(?P<srcdir>.+/)?(?P<name>(?P<prog>[^/]+)\.(?P<sect>\d+))\.md)$', args.mdfile)
+    if not fi:
+        die('Failed to parse NAME.NUM.md out of input file:', args.mdfile)
+    fi = argparse.Namespace(**fi.groupdict())
+
+    if not fi.srcdir:
+        fi.srcdir = './'
+
+    fi.title = fi.prog + '(' + fi.sect + ') man page'
+    fi.mtime = 0
+
+    if os.path.lexists(fi.srcdir + '.git'):
+        fi.mtime = int(subprocess.check_output('git log -1 --format=%at'.split()))
+
+    env_subs = { 'prefix': os.environ.get('RSYNC_OVERRIDE_PREFIX', None) }
+
+    if args.test:
+        env_subs['VERSION'] = '1.0.0'
+        env_subs['libdir'] = '/usr'
+    else:
+        for fn in 'NEWS.md Makefile'.split():
+            try:
+                st = os.lstat(fi.srcdir + fn)
+            except:
+                die('Failed to find', fi.srcdir + fn)
+            if not fi.mtime:
+                fi.mtime = st.st_mtime
+
+        with open(fi.srcdir + 'Makefile', 'r', encoding='utf-8') as fh:
+            for line in fh:
+                m = re.match(r'^(\w+)=(.+)', line)
+                if not m:
+                    continue
+                var, val = (m.group(1), m.group(2))
+                if var == 'prefix' and env_subs[var] is not None:
+                    continue
+                while re.search(r'\$\{', val):
+                    val = re.sub(r'\$\{(\w+)\}', lambda m: env_subs[m.group(1)], val)
+                env_subs[var] = val
+                if var == 'VERSION':
+                    break
+
+    with open(fi.fn, 'r', encoding='utf-8') as fh:
+        txt = fh.read()
+
+    txt = re.sub(r'@VERSION@', env_subs['VERSION'], txt)
+    txt = re.sub(r'@LIBDIR@', env_subs['libdir'], txt)
+
+    fi.html_in = md_parser(txt)
+    txt = None
+
+    fi.date = time.strftime('%d %b %Y', time.localtime(fi.mtime))
+    fi.man_headings = (fi.prog, fi.sect, fi.date, fi.prog + ' ' + env_subs['VERSION'])
+
+    HtmlToManPage(fi)
+
+    if args.test:
+        print("The test was successful.")
+        return
+
+    for fn, txt in ((fi.name + '.html', fi.html_out), (fi.name, fi.man_out)):
+        print("Wrote:", fn)
+        with open(fn, 'w', encoding='utf-8') as fh:
+            fh.write(txt)
+
+
+def html_via_cmarkgfm(txt):
+    return cmarkgfm.markdown_to_html(txt)
+
+
+def html_via_commonmark(txt):
+    return commonmark.HtmlRenderer().render(commonmark.Parser().parse(txt))
+
+
+class HtmlToManPage(HTMLParser):
+    def __init__(self, fi):
+        HTMLParser.__init__(self, convert_charrefs=True)
+
+        st = self.state = argparse.Namespace(
+                list_state = [ ],
+                p_macro = ".P\n",
+                at_first_tag_in_li = False,
+                at_first_tag_in_dd = False,
+                dt_from = None,
+                in_pre = False,
+                in_code = False,
+                html_out = [ HTML_START % fi.title ],
+                man_out = [ MAN_START % fi.man_headings ],
+                txt = '',
+                )
+
+        self.feed(fi.html_in)
+        fi.html_in = None
+
+        st.html_out.append(HTML_END % fi.date)
+        st.man_out.append(MAN_END)
+
+        fi.html_out = ''.join(st.html_out)
+        st.html_out = None
+
+        fi.man_out = ''.join(st.man_out)
+        st.man_out = None
+
+
+    def handle_starttag(self, tag, attrs_list):
+        st = self.state
+        if args.debug:
+            self.output_debug('START', (tag, attrs_list))
+        if st.at_first_tag_in_li:
+            if st.list_state[-1] == 'dl':
+                st.dt_from = tag
+                if tag == 'p':
+                    tag = 'dt'
+                else:
+                    st.html_out.append('<dt>')
+            elif tag == 'p':
+                st.at_first_tag_in_dd = True # Kluge to suppress a .P at the start of an li.
+            st.at_first_tag_in_li = False
+        if tag == 'p':
+            if not st.at_first_tag_in_dd:
+                st.man_out.append(st.p_macro)
+        elif tag == 'li':
+            st.at_first_tag_in_li = True
+            lstate = st.list_state[-1]
+            if lstate == 'dl':
+                return
+            if lstate == 'o':
+                st.man_out.append(".IP o\n")
+            else:
+                st.man_out.append(".IP " + str(lstate) + ".\n")
+                st.list_state[-1] += 1
+        elif tag == 'blockquote':
+            st.man_out.append(".RS 4\n")
+        elif tag == 'pre':
+            st.in_pre = True
+            st.man_out.append(st.p_macro + ".nf\n")
+        elif tag == 'code' and not st.in_pre:
+            st.in_code = True
+            st.txt += BOLD_FONT[0]
+        elif tag == 'strong' or tag == 'b':
+            st.txt += BOLD_FONT[0]
+        elif tag == 'em' or  tag == 'i':
+            tag = 'u' # Change it into underline to be more like the man page
+            st.txt += UNDR_FONT[0]
+        elif tag == 'ol':
+            start = 1
+            for var, val in attrs_list:
+                if var == 'start':
+                    start = int(val) # We only support integers.
+                    break
+            if st.list_state:
+                st.man_out.append(".RS\n")
+            if start == 0:
+                tag = 'dl'
+                attrs_list = [ ]
+                st.list_state.append('dl')
+            else:
+                st.list_state.append(start)
+            st.man_out.append(st.p_macro)
+            st.p_macro = ".IP\n"
+        elif tag == 'ul':
+            st.man_out.append(st.p_macro)
+            if st.list_state:
+                st.man_out.append(".RS\n")
+                st.p_macro = ".IP\n"
+            st.list_state.append('o')
+        st.html_out.append('<' + tag + ''.join(' ' + var + '="' + htmlify(val) + '"' for var, val in attrs_list) + '>')
+        st.at_first_tag_in_dd = False
+
+
+    def handle_endtag(self, tag):
+        st = self.state
+        if args.debug:
+            self.output_debug('END', (tag,))
+        if tag in CONSUMES_TXT or st.dt_from == tag:
+            txt = st.txt.strip()
+            st.txt = ''
+        else:
+            txt = None
+        add_to_txt = None
+        if tag == 'h1':
+            st.man_out.append(st.p_macro + '.SH "' + manify(txt) + '"\n')
+        elif tag == 'h2':
+            st.man_out.append(st.p_macro + '.SS "' + manify(txt) + '"\n')
+        elif tag == 'p':
+            if st.dt_from == 'p':
+                tag = 'dt'
+                st.man_out.append('.IP "' + manify(txt) + '"\n')
+                st.dt_from = None
+            elif txt != '':
+                st.man_out.append(manify(txt) + "\n")
+        elif tag == 'li':
+            if st.list_state[-1] == 'dl':
+                if st.at_first_tag_in_li:
+                    die("Invalid 0. -> td translation")
+                tag = 'dd'
+            if txt != '':
+                st.man_out.append(manify(txt) + "\n")
+            st.at_first_tag_in_li = False
+        elif tag == 'blockquote':
+            st.man_out.append(".RE\n")
+        elif tag == 'pre':
+            st.in_pre = False
+            st.man_out.append(manify(txt) + "\n.fi\n")
+        elif (tag == 'code' and not st.in_pre):
+            st.in_code = False
+            add_to_txt = NORM_FONT[0]
+        elif tag == 'strong' or tag == 'b':
+            add_to_txt = NORM_FONT[0]
+        elif tag == 'em' or  tag == 'i':
+            tag = 'u' # Change it into underline to be more like the man page
+            add_to_txt = NORM_FONT[0]
+        elif tag == 'ol' or tag == 'ul':
+            if st.list_state.pop() == 'dl':
+                tag = 'dl'
+            if st.list_state:
+                st.man_out.append(".RE\n")
+            else:
+                st.p_macro = ".P\n"
+            st.at_first_tag_in_dd = False
+        st.html_out.append('</' + tag + '>')
+        if add_to_txt:
+            if txt is None:
+                st.txt += add_to_txt
+            else:
+                txt += add_to_txt
+        if st.dt_from == tag:
+            st.man_out.append('.IP "' + manify(txt) + '"\n')
+            st.html_out.append('</dt><dd>')
+            st.at_first_tag_in_dd = True
+            st.dt_from = None
+        elif tag == 'dt':
+            st.html_out.append('<dd>')
+            st.at_first_tag_in_dd = True
+
+
+    def handle_data(self, txt):
+        st = self.state
+        if args.debug:
+            self.output_debug('DATA', (txt,))
+        if st.in_pre:
+            html = htmlify(txt)
+        else:
+            txt = re.sub(r'\s--(\s)', NBR_SPACE[0] + r'--\1', txt).replace('--', NBR_DASH[0]*2)
+            txt = re.sub(r'(^|\W)-', r'\1' + NBR_DASH[0], txt)
+            html = htmlify(txt)
+            if st.in_code:
+                txt = re.sub(r'\s', NBR_SPACE[0], txt)
+                html = html.replace(NBR_DASH[0], '-').replace(NBR_SPACE[0], ' ') # <code> is non-breaking in CSS
+        st.html_out.append(html.replace(NBR_SPACE[0], '&nbsp;').replace(NBR_DASH[0], '-&#8288;'))
+        st.txt += txt
+
+
+    def output_debug(self, event, extra):
+        import pprint
+        st = self.state
+        if args.debug < 2:
+            st = argparse.Namespace(**vars(st))
+            if len(st.html_out) > 2:
+                st.html_out = ['...'] + st.html_out[-2:]
+            if len(st.man_out) > 2:
+                st.man_out = ['...'] + st.man_out[-2:]
+        print(event, extra)
+        pprint.PrettyPrinter(indent=2).pprint(vars(st))
+
+
+def manify(txt):
+    return re.sub(r"^(['.])", r'\&\1', txt.replace('\\', '\\\\')
+            .replace(NBR_SPACE[0], NBR_SPACE[1])
+            .replace(NBR_DASH[0], NBR_DASH[1])
+            .replace(NORM_FONT[0], NORM_FONT[1])
+            .replace(BOLD_FONT[0], BOLD_FONT[1])
+            .replace(UNDR_FONT[0], UNDR_FONT[1]), flags=re.M)
+
+
+def htmlify(txt):
+    return txt.replace('&', '&amp;').replace('<', '&lt;').replace('>', '&gt;').replace('"', '&quot;')
+
+
+def warn(*msg):
+    print(*msg, file=sys.stderr)
+
+
+def die(*msg):
+    warn(*msg)
+    sys.exit(1)
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description='Transform a NAME.NUM.md markdown file into a NAME.NUM.html web page & a NAME.NUM man page.', add_help=False)
+    parser.add_argument('--test', action='store_true', help='Test if we can parse the input w/o updating any files.')
+    parser.add_argument('--debug', '-D', action='count', default=0, help='Output copious info on the html parsing. Repeat for even more.')
+    parser.add_argument("--help", "-h", action="help", help="Output this help message and exit.")
+    parser.add_argument('mdfile', help="The NAME.NUM.md file to parse.")
+    args = parser.parse_args()
+
+    try:
+        import cmarkgfm
+        md_parser = html_via_cmarkgfm
+    except:
+        try:
+            import commonmark
+            md_parser = html_via_commonmark
+        except:
+            die("Failed to find cmarkgfm or commonmark for python3.")
+
+    main()

mkproto.awk

@@ -1,74 +1,39 @@
-# generate prototypes for Samba C code
-# tridge, June 1996
+#!/usr/bin/awk -f
 
 BEGIN {
-  inheader=0;
-  print "/* This file is automatically generated with \"make proto\". DO NOT EDIT */"
-  print ""
+    while ((getline i < "proto.h") > 0) old_protos = old_protos ? old_protos "\n" i : i
+    protos = "/* This file is automatically generated with \"make proto\". DO NOT EDIT */\n"
 }
 
-{
-  if (inheader) {
-    if (match($0,"[)][ \t]*$")) {
-      inheader = 0;
-      printf "%s;\n",$0;
-    } else {
-      printf "%s\n",$0;
-    }
-    next;
-  }
+inheader {
+    protos = protos "\n" ((inheader = /\)[ \t]*$/ ? 0 : 1) ? $0 : $0 ";")
+    next
 }
 
-/^FN_LOCAL_BOOL/ {
-  split($0,a,"[,()]")
-  printf "BOOL %s(int );\n", a[2]
+/^FN_(LOCAL|GLOBAL)_[^(]+\([^,()]+/ {
+    local = /^FN_LOCAL/
+    gsub(/^FN_(LOC|GLOB)AL_|,.*$/, "")
+    sub(/^BOOL\(/, "BOOL ")
+    sub(/^CHAR\(/, "char ")
+    sub(/^INTEGER\(/, "int ")
+    sub(/^STRING\(/, "char *")
+    protos = protos "\n" $0 (local ? "(int module_id);" : "(void);")
+    next
 }
 
-/^FN_LOCAL_STRING/ {
-  split($0,a,"[,()]")
-  printf "char *%s(int );\n", a[2]
+/^static|^extern|;/||!/^[A-Za-z][A-Za-z0-9_]* / { next }
+
+/\(.*\)[ \t]*$/ {
+    protos = protos "\n" $0 ";"
+    next
 }
 
-/^FN_LOCAL_INT/ {
-  split($0,a,"[,()]")
-  printf "int %s(int );\n", a[2]
+/\(/ {
+    inheader = 1
+    protos = protos "\n" $0
 }
 
-/^FN_LOCAL_CHAR/ {
-  split($0,a,"[,()]")
-  printf "char %s(int );\n", a[2]
-}
-
-/^FN_GLOBAL_BOOL/ {
-  split($0,a,"[,()]")
-  printf "BOOL %s(void);\n", a[2]
-}
-
-/^FN_GLOBAL_STRING/ {
-  split($0,a,"[,()]")
-  printf "char *%s(void);\n", a[2]
-}
-
-/^FN_GLOBAL_INT/ {
-  split($0,a,"[,()]")
-  printf "int %s(void);\n", a[2]
-}
-
-/^static|^extern/ || /[;]/ {
-  next;
-}
-
-!/^[A-Za-z][A-Za-z0-9_]* / {
-  next;
-}
-
-/[(].*[)][ \t]*$/ {
-    printf "%s;\n",$0;
-    next;
-}
-
-/[(]/ {
-  inheader=1;
-  printf "%s\n",$0;
-  next;
+END {
+    if (old_protos != protos) print protos > "proto.h"
+    printf "" > "proto.h-tstamp"
 }

mkrounding.c

@@ -1,66 +0,0 @@
-/*
- * A pre-compilation helper program to aid in the creation of rounding.h.
- *
- * Copyright (C) 2007 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
- * the Free Software Foundation; either version 3 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License along
- * with this program; if not, visit the http://fsf.org website.
- */
-
-#include "rsync.h"
-
-struct test1 {
-    union file_extras extras1[1];
-    struct {
-#	include "mkrounding.h"
-    } file;
-};
-
-struct test2 {
-    union file_extras extras2[2];
-    struct {
-#	include "mkrounding.h"
-    } file;
-};
-
-struct test4 {
-    union file_extras extras4[4];
-    struct {
-#	include "mkrounding.h"
-    } file;
-};
-
-#define SIZE_TEST(n) (sizeof (struct test ## n) == EXTRA_LEN * n + sizeof (struct file_struct))
-
- int main(UNUSED(int argc), UNUSED(char *argv[]))
-{
-    int cnt;
-    if (SIZE_TEST(1))
-	cnt = 0;
-    else if (SIZE_TEST(2))
-	cnt = 1;
-    else if (SIZE_TEST(4))
-	cnt = 3;
-    else {
-	fprintf(stderr, "Unable to determine required file_extras rounding!\n");
-	cnt = 3;
-    }
-    if (cnt)
-	fprintf(stderr, "Rounding file_extras in multiples of %d", cnt + 1);
-    else
-	fprintf(stderr, "No rounding needed for file_extras");
-    fprintf(stderr, " (EXTRA_LEN=%d, FILE_STRUCT_LEN=%d)\n",
-	    (int)EXTRA_LEN, (int)FILE_STRUCT_LEN);
-    printf("#define EXTRA_ROUNDING %d\n", cnt);
-    return 0;
-}

packaging/branch-from-patch

@@ -0,0 +1,174 @@
+#!/usr/bin/env -S python3 -B
+
+# This script turns one or more diff files in the patches dir (which is
+# expected to be a checkout of the rsync-patches git repo) into a branch
+# in the main rsync git checkout. This allows the applied patch to be
+# merged with the latest rsync changes and tested.  To update the diff
+# with the resulting changes, see the patch-update script.
+
+import os, sys, re, argparse, glob
+
+sys.path = ['packaging'] + sys.path
+
+from pkglib import *
+
+def main():
+    global created, info, local_branch
+
+    cur_branch, args.base_branch = check_git_state(args.base_branch, not args.skip_check, args.patches_dir)
+
+    local_branch = get_patch_branches(args.base_branch)
+
+    if args.delete_local_branches:
+        for name in sorted(local_branch):
+            branch = f"patch/{args.base_branch}/{name}"
+            cmd_chk(['git', 'branch', '-D', branch])
+        local_branch = set()
+
+    if args.add_missing:
+        for fn in sorted(glob.glob(f"{args.patches_dir}/*.diff")):
+            name = re.sub(r'\.diff$', '', re.sub(r'.+/', '', fn))
+            if name not in local_branch and fn not in args.patch_files:
+                args.patch_files.append(fn)
+
+    if not args.patch_files:
+        return
+
+    for fn in args.patch_files:
+        if not fn.endswith('.diff'):
+            die(f"Filename is not a .diff file: {fn}")
+        if not os.path.isfile(fn):
+            die(f"File not found: {fn}")
+
+    scanned = set()
+    info = { }
+
+    patch_list = [ ]
+    for fn in args.patch_files:
+        m = re.match(r'^(?P<dir>.*?)(?P<name>[^/]+)\.diff$', fn)
+        patch = argparse.Namespace(**m.groupdict())
+        if patch.name in scanned:
+            continue
+        patch.fn = fn
+
+        lines = [ ]
+        commit_hash = None
+        with open(patch.fn, 'r', encoding='utf-8') as fh:
+            for line in fh:
+                m = re.match(r'^based-on: (\S+)', line)
+                if m:
+                    commit_hash = m[1]
+                    break
+                if (re.match(r'^index .*\.\..* \d', line)
+                 or re.match(r'^diff --git ', line)
+                 or re.match(r'^--- (old|a)/', line)):
+                    break
+                lines.append(re.sub(r'\s*\Z', "\n", line, 1))
+        info_txt = ''.join(lines).strip() + "\n"
+        lines = None
+
+        parent = args.base_branch
+        patches = re.findall(r'patch -p1 <%s/(\S+)\.diff' % args.patches_dir, info_txt)
+        if patches:
+            last = patches.pop()
+            if last != patch.name:
+                warn(f"No identity patch line in {patch.fn}")
+                patches.append(last)
+            if patches:
+                parent = patches.pop()
+                if parent not in scanned:
+                    diff_fn = patch.dir + parent + '.diff'
+                    if not os.path.isfile(diff_fn):
+                        die(f"Failed to find parent of {patch.fn}: {parent}")
+                    # Add parent to args.patch_files so that we will look for the
+                    # parent's parent.  Any duplicates will be ignored.
+                    args.patch_files.append(diff_fn)
+        else:
+            warn(f"No patch lines found in {patch.fn}")
+
+        info[patch.name] = [ parent, info_txt, commit_hash ]
+
+        patch_list.append(patch)
+
+    created = set()
+    for patch in patch_list:
+        create_branch(patch)
+
+    cmd_chk(['git', 'checkout', args.base_branch])
+
+
+def create_branch(patch):
+    if patch.name in created:
+        return
+    created.add(patch.name)
+
+    parent, info_txt, commit_hash = info[patch.name]
+    parent = argparse.Namespace(dir=patch.dir, name=parent, fn=patch.dir + parent + '.diff')
+
+    if parent.name == args.base_branch:
+        parent_branch = commit_hash if commit_hash else args.base_branch
+    else:
+        create_branch(parent)
+        parent_branch = '/'.join(['patch', args.base_branch, parent.name])
+
+    branch = '/'.join(['patch', args.base_branch, patch.name])
+    print("\n" + '=' * 64)
+    print(f"Processing {branch} ({parent_branch})")
+
+    if patch.name in local_branch:
+        cmd_chk(['git', 'branch', '-D', branch])
+
+    cmd_chk(['git', 'checkout', '-b', branch, parent_branch])
+
+    info_fn = 'PATCH.' + patch.name
+    with open(info_fn, 'w', encoding='utf-8') as fh:
+        fh.write(info_txt)
+    cmd_chk(['git', 'add', info_fn])
+
+    with open(patch.fn, 'r', encoding='utf-8') as fh:
+        patch_txt = fh.read()
+
+    cmd_run('patch -p1'.split(), input=patch_txt)
+
+    for fn in glob.glob('*.orig') + glob.glob('*/*.orig'):
+        os.unlink(fn)
+
+    pos = 0
+    new_file_re = re.compile(r'\nnew file mode (?P<mode>\d+)\s+--- /dev/null\s+\+\+\+ b/(?P<fn>.+)')
+    while True:
+        m = new_file_re.search(patch_txt, pos)
+        if not m:
+            break
+        os.chmod(m['fn'], int(m['mode'], 8))
+        cmd_chk(['git', 'add', m['fn']])
+        pos = m.end()
+
+    while True:
+        cmd_chk('git status'.split())
+        ans = input('Press Enter to commit, Ctrl-C to abort, or type a wild-name to add a new file: ')
+        if ans == '':
+            break
+        cmd_chk("git add " + ans, shell=True)
+
+    while True:
+        s = cmd_run(['git', 'commit', '-a', '-m', f"Creating branch from {patch.name}.diff."])
+        if not s.returncode:
+            break
+        s = cmd_run(['/bin/zsh'])
+        if s.returncode:
+            die('Aborting due to shell error code')
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description="Create a git patch branch from an rsync patch file.", add_help=False)
+    parser.add_argument('--branch', '-b', dest='base_branch', metavar='BASE_BRANCH', default='master', help="The branch the patch is based on. Default: master.")
+    parser.add_argument('--add-missing', '-a', action='store_true', help="Add a branch for every patches/*.diff that doesn't have a branch.")
+    parser.add_argument('--skip-check', action='store_true', help="Skip the check that ensures starting with a clean branch.")
+    parser.add_argument('--delete', dest='delete_local_branches', action='store_true', help="Delete all the local patch/BASE/* branches, not just the ones that are being recreated.")
+    parser.add_argument('--patches-dir', '-p', metavar='DIR', default='patches', help="Override the location of the rsync-patches dir. Default: patches.")
+    parser.add_argument('patch_files', metavar='patches/DIFF_FILE', nargs='*', help="Specify what patch diff files to process. Default: all of them.")
+    parser.add_argument("--help", "-h", action="help", help="Output this help message and exit.")
+    args = parser.parse_args()
+    main()
+
+# vim: sw=4 et

packaging/cull_options

@@ -1,16 +1,36 @@
-#!/usr/bin/perl
+#!/usr/bin/env perl
 # This script outputs some perl code that parses all possible options
 # that the code in options.c might send to the server.  This perl code
 # is included in the rrsync script.
 use strict;
 
-our(%short_no_arg, %short_with_num, %long_opt);
+our %short_no_arg;
+our %short_with_num = ( '@' => 1 );
+our %long_opt = ( # These include some extra long-args that BackupPC uses:
+    'block-size' => 1,
+    'daemon' => -1,
+    'debug' => 1,
+    'fake-super' => 0,
+    'fuzzy' => 0,
+    'group' => 0,
+    'hard-links' => 0,
+    'ignore-times' => 0,
+    'info' => 1,
+    'links' => 0,
+    'log-file' => 3,
+    'one-file-system' => 0,
+    'owner' => 0,
+    'perms' => 0,
+    'recursive' => 0,
+    'times' => 0,
+    'write-devices' => -1,
+);
 our $last_long_opt;
 
 open(IN, '../options.c') or die "Unable to open ../options.c: $!\n";
 
 while (<IN>) {
-    if (/\Qargstr[x++]\E = '(.)'/) {
+    if (/\Qargstr[x++]\E = '([^.ie])'/) {
 	$short_no_arg{$1} = 1;
 	undef $last_long_opt;
     } elsif (/\Qasprintf(\E[^,]+, "-([a-zA-Z0-9])\%l?[ud]"/) {
@@ -18,15 +38,15 @@ while (<IN>) {
 	undef $last_long_opt;
     } elsif (/\Qargs[ac++]\E = "--([^"=]+)"/) {
 	$last_long_opt = $1;
-	$long_opt{$1} = 0;
+	$long_opt{$1} = 0 unless exists $long_opt{$1};
     } elsif (defined($last_long_opt)
-	&& /\Qargs[ac++]\E = ([^["\s]+);/ && $1 ne 'dest_option') {
+	&& /\Qargs[ac++]\E = ([^["\s]+);/) {
 	$long_opt{$last_long_opt} = 2;
 	undef $last_long_opt;
-    } elsif (/dest_option = "--([^"]+)"/) {
+    } elsif (/return "--([^"]+-dest)";/) {
 	$long_opt{$1} = 2;
 	undef $last_long_opt;
-    } elsif (/\Qasprintf(\E[^,]+, "--([^"=]+)=/ || /\Qargs[ac++]\E = "--([^"=]+)=/) {
+    } elsif (/\Qasprintf(\E[^,]+, "--([^"=]+)=/ || /\Qargs[ac++]\E = "--([^"=]+)=/ || /fmt = .*: "--([^"=]+)=/) {
 	$long_opt{$1} = 1;
 	undef $last_long_opt;
     }
@@ -57,7 +77,8 @@ foreach my $opt (sort keys %long_opt) {
     my $val = $long_opt{$opt};
     $val = 1 if $opt =~ /^(max-|min-)/;
     $val = 3 if $opt eq 'files-from';
-    $val = '$ro ? -1 : ' . $val if $opt =~ /^remove-/;
+    $val = q"$only eq 'r' ? -1 : " . $val if $opt =~ /^(remove-|log-file)/;
+    $val = q"$only eq 'w' ? -1 : " . $val if $opt eq 'sender';
     print "  '$opt' => $val,\n";
 }
 

packaging/lsb/rsync.spec

@@ -1,88 +1,87 @@
-Summary: A program for synchronizing files over a network.
+Summary: A fast, versatile, remote (and local) file-copying tool
 Name: rsync
-Version: 3.0.0pre2
+Version: 3.2.0
+%define fullversion %{version}
 Release: 1
+%define srcdir src
 Group: Applications/Internet
-Source:	ftp://rsync.samba.org/pub/rsync/rsync-%{version}.tar.gz
+License: GPL
+Source0: http://rsync.samba.org/ftp/rsync/%{srcdir}/rsync-%{fullversion}.tar.gz
+#Source1: http://rsync.samba.org/ftp/rsync/%{srcdir}/rsync-patches-%{fullversion}.tar.gz
 URL: http://rsync.samba.org/
 
 Prefix: %{_prefix}
 BuildRoot: /var/tmp/%{name}-root
-License: GPL
+
+%package ssl-daemon
+Summary: An stunnel config file to support ssl rsync daemon connections.
+Group: Applications/Internet
+Requires: rsync, stunnel >= 4
 
 %description
-Rsync uses a reliable algorithm to bring remote and host files into
-sync very quickly. Rsync is fast because it just sends the differences
-in the files over the network instead of sending the complete
-files. Rsync is often used as a very powerful mirroring process or
-just as a more capable replacement for the rcp command. A technical
-report which describes the rsync algorithm is included in this
-package.
+Rsync is a fast and extraordinarily versatile file copying tool.  It can
+copy locally, to/from another host over any remote shell, or to/from a
+remote rsync daemon.  It offers a large number of options that control
+every aspect of its behavior and permit very flexible specification of the
+set of files to be copied.  It is famous for its delta-transfer algorithm,
+which reduces the amount of data sent over the network by sending only the
+differences between the source files and the existing files in the
+destination.  Rsync is widely used for backups and mirroring and as an
+improved copy command for everyday use.
+
+%description ssl-daemon
+Provides a config file for stunnel that will (if you start your stunnel
+service) cause stunnel to listen for ssl rsync-daemon connections and run
+"rsync --daemon" to handle them.
 
 %prep
-%setup -q
+# Choose one -- setup source only, or setup source + rsync-patches:
+%setup -q -n rsync-%{fullversion}
+#%setup -q -b1 -n rsync-%{fullversion}
+
+# If you you used "%setup -q -b1 ...", choose the patches you wish to apply:
+#patch -p1 <patches/acls.diff
+#patch -p1 <patches/xattrs.diff
+#patch -p1 <patches/remote-option.diff
+#patch -p1 <patches/db.diff
+
+# Avoid extra perl dependencies for scripts going into doc dir.
+chmod -x support/*
 
 %build
+#./prepare-source
 %configure
 
 make
 
 %install
 rm -rf $RPM_BUILD_ROOT
+make install install-ssl-daemon DESTDIR=$RPM_BUILD_ROOT
 
-%makeinstall
+mkdir -p $RPM_BUILD_ROOT/etc/xinetd.d $RPM_BUILD_ROOT/etc/rsync-ssl/certs
+install -m 644 packaging/lsb/rsync.xinetd $RPM_BUILD_ROOT/etc/xinetd.d/rsync
 
 %clean
 rm -rf $RPM_BUILD_ROOT
 
 %files
 %defattr(-,root,root)
-%doc COPYING README tech_report.tex
+%doc COPYING NEWS.md OLDNEWS.md README.md support/ tech_report.tex
+%config(noreplace) /etc/xinetd.d/rsync
 %{_prefix}/bin/rsync
+%{_prefix}/bin/rsync-ssl
 %{_mandir}/man1/rsync.1*
+%{_mandir}/man1/rsync-ssl.1*
 %{_mandir}/man5/rsyncd.conf.5*
 
+%files ssl-daemon
+%config(noreplace) /etc/stunnel/rsyncd.conf
+%dir /etc/rsync-ssl/certs
+
 %changelog
-* Thu Jan 30 2003 Horst von Brand <vonbrand@inf.utfsm.cl>
-  Fixed "Sept" date in %changelog here
-  Use %{_mandir} to point to manpages
-  Support for compressed manpages (* at end catches them in %files)
-  Add doc/README-SGML and doc/rsync.sgml to %doc
+* Fri Jun 19 2020 Wayne Davison <wayne@opencoder.net>
+Released 3.2.0.
 
-* Mon Sep 11 2000 John H Terpstra <jht@turbolinux.com>
-  Changed target paths to be Linux Standards Base compliant
-
-* Mon Jan 25 1999 Stefan Hornburg <racke@linuxia.de>
-  quoted RPM_OPT_FLAGS for the sake of robustness
-
-* Mon May 18 1998 Andrew Tridgell <tridge@samba.anu.edu.au>
-  reworked for auto-building when I release rsync (tridge@samba.anu.edu.au)
-
-* Sat May 16 1998 John H Terpstra <jht@aquasoft.com.au>
-  Upgraded to Rsync 2.0.6
-    -new feature anonymous rsync
-
-* Mon Apr  6 1998 Douglas N. Arnold <dna@math.psu.edu>
-
-Upgrade to rsync version 1.7.2.
-
-* Sun Mar  1 1998 Douglas N. Arnold <dna@math.psu.edu>
-
-Built 1.6.9-1 based on the 1.6.3-2 spec file of John A. Martin.
-Changes from 1.6.3-2 packaging: added latex and dvips commands
-to create tech_report.ps.
-
-* Mon Aug 25 1997 John A. Martin <jam@jamux.com>
-
-Built 1.6.3-2 after finding no rsync-1.6.3-1.src.rpm although there
-was an ftp://ftp.redhat.com/pub/contrib/alpha/rsync-1.6.3-1.alpha.rpm
-showing no packager nor signature but giving 
-"Source RPM: rsync-1.6.3-1.src.rpm".
-
-Changes from 1.6.2-1 packaging: added '$RPM_OPT_FLAGS' to make, strip
-to '%build', removed '%prefix'.
-
-* Thu Apr 10 1997 Michael De La Rue <miked@ed.ac.uk>
-
-rsync-1.6.2-1 packaged.  (This entry by jam to credit Michael for the
-previous package(s).)
+* Fri Mar 21 2008 Wayne Davison <wayne@opencoder.net>
+Added installation of /etc/xinetd.d/rsync file and some commented-out
+lines that demonstrate how to use the rsync-patches tar file.

packaging/lsb/rsync.xinetd

@@ -0,0 +1,13 @@
+# default: off
+# description: The rsync server is a good addition to an ftp server, as it
+#	allows crc checksumming etc.
+service rsync
+{
+	disable	= yes
+	socket_type     = stream
+	wait            = no
+	user            = root
+	server          = /usr/bin/rsync
+	server_args     = --daemon
+	log_on_failure  += USERID
+}

packaging/md2html

@@ -0,0 +1,99 @@
+#!/usr/bin/env python3
+
+# Copyright (C) 2020 Wayne Davison
+#
+# This program is freely redistributable.
+
+import os, re, argparse
+
+HTML_START = """\
+<html><head>
+<title>%s</title>
+<link href="https://fonts.googleapis.com/css2?family=Roboto&family=Roboto+Mono&display=swap" rel="stylesheet">
+<style>
+body {
+  max-width: 50em;
+  margin: auto;
+}
+body, b, strong, u {
+  font-family: 'Roboto', sans-serif;
+}
+code {
+  font-family: 'Roboto Mono', monospace;
+  font-weight: bold;
+}
+pre code {
+  display: block;
+  font-weight: normal;
+}
+blockquote pre code {
+  background: #f1f1f1;
+}
+dd p:first-of-type {
+  margin-block-start: 0em;
+}
+</style>
+</head><body>
+"""
+
+HTML_END = """\
+</body></html>
+"""
+
+md_parser = None
+
+def main():
+    for mdfn in args.mdfiles:
+        if not mdfn.endswith('.md'):
+            print('Ignoring non-md input file:', mdfn)
+            continue
+        title = re.sub(r'.*/', '', mdfn).replace('.md', '')
+        htfn = mdfn.replace('.md', '.html')
+
+        print("Parsing", mdfn, '->', htfn)
+
+        with open(mdfn, 'r', encoding='utf-8') as fh:
+            txt = fh.read()
+
+        txt = re.sub(r'\s--\s', '\xa0-- ', txt)
+
+        html = md_parser(txt)
+
+        html = re.sub(r'(<code>)([\s\S]*?)(</code>)', lambda m: m[1] + re.sub(r'\s', '\xa0', m[2]) + m[3], html)
+        html = html.replace('--', '&#8209;&#8209;').replace("\xa0-", '&nbsp;&#8209;').replace("\xa0", '&nbsp;')
+        html = re.sub(r'(\W)-', r'\1&#8209;', html)
+
+        if os.path.lexists(htfn):
+            os.unlink(htfn)
+
+        with open(htfn, 'w', encoding='utf-8') as fh:
+            fh.write(HTML_START % title)
+            fh.write(html)
+            fh.write(HTML_END)
+
+
+def html_via_cmarkgfm(txt):
+    return cmarkgfm.markdown_to_html(txt)
+
+
+def html_via_commonmark(txt):
+    return commonmark.HtmlRenderer().render(commonmark.Parser().parse(txt))
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description='Output html for md pages.', add_help=False)
+    parser.add_argument("--help", "-h", action="help", help="Output this help message and exit.")
+    parser.add_argument("mdfiles", nargs='+', help="The .md files to turn into .html files.")
+    args = parser.parse_args()
+
+    try:
+        import cmarkgfm
+        md_parser = html_via_cmarkgfm
+    except:
+        try:
+            import commonmark
+            md_parser = html_via_commonmark
+        except:
+            die("Failed to find cmarkgfm or commonmark for python3.")
+
+    main()

packaging/nightly-rsync

@@ -1,130 +1,106 @@
-#!/usr/bin/perl
-use strict;
+#!/usr/bin/env -S python3 -B
 
 # This script expects the directory ~/samba-rsync-ftp to exist and to be a
 # copy of the /home/ftp/pub/rsync dir on samba.org.  It also requires a
-# pristine CVS checkout of rsync (don't use your normal rsync build dir
-# unless you're 100% sure that there are not unchecked-in changes).
+# git checkout of rsync (feel free to use your normal rsync build dir as
+# long as it doesn't have any uncommitted changes).
 #
-# If this is run with -ctu, it will make an updated "nightly" tar file in
+# If this is run with -tu, it will make an updated "nightly" tar file in
 # the nightly dir.  It will also remove any old tar files, regenerate the
 # HTML man pages in the nightly dir, and then rsync the changes to the
 # samba.org server.
 
-use Getopt::Long;
-use Date::Format;
+import os, sys, re, argparse, glob
+from datetime import datetime, timezone
+from getpass import getpass
 
-# Choose any dir where a pristine rsync has been checked out of CVS.
-our $unpacked = $ENV{HOME} . '/release/nightly';
-# Where the local copy of /home/ftp/pub/rsync/nightly should be updated.
-our $nightly = $ENV{HOME} . '/samba-rsync-ftp/nightly';
-our $nightly_symlink = "$nightly/rsync-HEAD.tar.gz";
+sys.path = ['packaging'] + sys.path
 
-our($cvs_update, $make_tar, $upload, $help_opt);
-&Getopt::Long::Configure('bundling');
-&usage if !&GetOptions(
-    'cvs-update|c' => \$cvs_update,
-    'make-tar|t' => \$make_tar,
-    'upload|u' => \$upload,
-    'help|h' => \$help_opt,
-) || $help_opt;
+from pkglib import *
 
-our $name = time2str('rsync-HEAD-%Y%m%d-%H%M%Z', time, 'GMT');
-our $ztoday = time2str('%d %b %Y', time);
-our $today = $ztoday;
+# Where the local copy of /home/ftp/pub/rsync/dev/nightly should be updated.
+dest = os.environ['HOME'] + '/samba-rsync-ftp/dev/nightly'
+samba_host = os.environ['SAMBA_HOST']
+nightly_symlink = f"{dest}/rsync-HEAD.tar.gz"
 
-chdir($unpacked) or die $!;
+def main():
+    now = datetime.now(timezone.utc)
+    name = now.strftime('rsync-HEAD-%Y%m%d-%H%MGMT')
+    ztoday = now.strftime('%d %b %Y')
+    today = ztoday.lstrip('0')
+    gen_target = 'gensend' if args.upload else 'gen'
 
-if ($cvs_update) {
-    print "Updating from cvs...\n";
-    system 'cvs -q up' and die $!;
-}
+    if not os.path.isdir(dest):
+        die("$dest does not exist")
+    if not os.path.isdir('.git'):
+        die("There is no .git dir in the current directory.")
+    if not os.path.exists('rsyncd.conf.5.md'):
+        die("There is no rsync checkout in the current directory.")
 
-if ($make_tar) {
-    print "Generating list of active CVS files...\n";
-    my($dir, @files);
-    open(CVS, '-|', 'cvs status 2>&1') or die $!;
-    while (<CVS>) {
-	if (/^cvs status: Examining (.*)/) {
-	    if ($1 eq '.') {
-		$dir = '';
-	    } else {
-		push(@files, $1);
-		$dir = $1 . '/';
-	    }
-	} elsif (/^File: (.*?)\s+Status: (.*)/ && $1 ne '.cvsignore') {
-	    push(@files, $dir . $1);
-	    if ($2 ne 'Up-to-date') {
-		print "*** Not up-to-date: $dir$1\n";
-	    }
-	}
-    }
-    close CVS;
+    mandate_gensend_hook()
 
-    print "Creating $unpacked/$name.tar.gz\n";
-    chdir('..') or die $!;
-    rename($unpacked, $name) or die $!;
-    open(TAR, '|-', "fakeroot tar --files-from=- --no-recursion --mode=g-w -czf $nightly/$name.tar.gz $name") or die $!;
-    foreach (@files) {
-	print TAR "$name/$_\n";
-    }
-    close TAR;
-    rename($name, $unpacked) or die $!;
-    unlink($nightly_symlink);
-    symlink("$name.tar.gz", $nightly_symlink);
-}
+    if args.make_tar:
+        check_git_state('master')
+        cmd_chk(['touch', 'NEWS.md'])
+        cmd_chk(['make', gen_target])
+        cmd_chk(['rsync', '-a', *glob.glob('*.[1-9].html'), dest])
 
-chdir($nightly) or die $!;
+        gen_files = get_gen_files()
 
-foreach my $fn (qw( rsync.yo rsyncd.conf.yo )) {
-    my $html_fn = $fn;
-    $html_fn =~ s/\.yo/.html/;
+        confversion = get_configure_version()
 
-    open(IN, '<', "$unpacked/$fn") or die $!;
-    undef $/; $_ = <IN>; $/ = "\n";
-    close IN;
+        # All version values are strings!
+        last_version, last_protocol_version = get_OLDNEWS_version_info()
+        protocol_version, subprotocol_version = get_protocol_versions()
 
-    s/^(manpage\([^)]+\)\(\d+\)\()[^)]+(\).*)/$1$today$2/m;
-    #s/^(This man ?page is current for version) \S+ (of rsync)/$1 $version $2/m;
+        if 'dev' in confversion or 'pre' in confversion:
+            if last_protocol_version != protocol_version:
+                if subprotocol_version == '0':
+                    die("SUBPROTOCOL_VERSION must not be 0 for a non-final release with a changed PROTOCOL_VERSION.")
+            elif subprotocol_version != '0':
+                die("SUBPROTOCOL_VERSION must be 0 when the PROTOCOL_VERSION hasn't changed from the last release.")
+        elif subprotocol_version != '0':
+            die("SUBPROTOCOL_VERSION must be 0 for a final release.")
 
-    open(OUT, '>', $fn) or die $!;
-    print OUT $_;
-    close OUT;
+        name_slash = name + '/'
+        tar_name = f"{name}.tar.gz"
 
-    system "yodl2html -o $html_fn $fn";
+        print('Creating', tar_name)
 
-    unlink($fn);
-}
+        cmd_chk(['rsync', '-a', *gen_files, name_slash])
+        cmd_chk(f"git archive --format=tar --prefix={name}/ HEAD | tar xf -")
+        cmd_chk(['support/git-set-file-times', '--quiet', '--prefix', name_slash])
+        cmd_chk(['fakeroot', 'tar', 'czf', os.path.join(dest, tar_name), name])
+        cmd_chk(['rm', '-rf', name])
 
-my $cnt = 0;
-open(PIPE, '-|', 'ls -1t rsync-HEAD-*') or die $!;
-while (<PIPE>) {
-    chomp;
-    next if $cnt++ < 10;
-    unlink($_);
-}
-close PIPE;
+        if os.path.lexists(nightly_symlink):
+            os.unlink(nightly_symlink)
+        os.symlink(tar_name, nightly_symlink)
 
-system 'ls -ltr';
+    os.chdir(dest)
 
-if ($upload) {
-    my $opt = '';
-    if (defined $ENV{RSYNC_PARTIAL_DIR}) {
-	$opt = " -f 'R $ENV{RSYNC_PARTIAL_DIR}'";
-    }
-    system "rsync$opt -aviHP --delete-after . samba.org:/home/ftp/pub/rsync/nightly";
-}
+    tar_files = list(reversed(sorted(glob.glob('rsync-HEAD-*'))))
+    if len(tar_files) > 10:
+        for fn in tar_files[10:]:
+            print('Removing', fn)
+            os.unlink(fn)
 
-exit;
+    cmd_run('ls -ltr'.split())
 
-sub usage
-{
-    die <<EOT;
-Usage: nightly-rsync [OPTIONS]
+    if args.upload:
+        cmd = 'rsync -aivHP --delete-after'.split()
+        partial_dir = os.environ.get('RSYNC_PARTIAL_DIR', None)
+        if partial_dir:
+            cmd.append('-fR ' + partial_dir)
+        cmd_chk([*cmd, '.', f"{samba_host}:/home/ftp/pub/rsync/dev/nightly"])
 
- -c, --cvs-update  update $unpacked via CVS.
- -t, --make-tar    create a new tar file in $nightly
- -u, --upload      upload the revised nightly dir to samba.org
- -h, --help        display this help
-EOT
-}
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description='A helper script for "nightly" tar files.', add_help=False)
+    parser.add_argument('--make-tar', '-t', action='store_true', help=f"Create a new tar file in {dest}.")
+    parser.add_argument('--upload', '-u', action='store_true', help="Upload the revised nightly dir to {samba_host}.")
+    parser.add_argument("--help", "-h", action="help", help="Output this help message and exit.")
+    args = parser.parse_args()
+    main()
+
+# vim: sw=4 et

packaging/patch-update

@@ -0,0 +1,210 @@
+#!/usr/bin/env -S python3 -B
+
+# This script is used to turn one or more of the "patch/BASE/*" branches
+# into one or more diffs in the "patches" directory.  Pass the option
+# --gen if you want generated files in the diffs.  Pass the name of
+# one or more diffs if you want to just update a subset of all the
+# diffs.
+
+import os, sys, re, argparse, time, shutil
+
+sys.path = ['packaging'] + sys.path
+
+from pkglib import *
+
+MAKE_GEN_CMDS = [
+        'make -f prepare-source.mak conf'.split(),
+        './config.status'.split(),
+        'make gen'.split(),
+        ]
+TMP_DIR = "patches.gen"
+
+os.environ['GIT_MERGE_AUTOEDIT'] = 'no'
+
+def main():
+    global master_commit, parent_patch, description, completed, last_touch
+
+    if not os.path.isdir(args.patches_dir):
+        die(f'No "{args.patches_dir}" directory was found.')
+    if not os.path.isdir('.git'):
+        die('No ".git" directory present in the current dir.')
+
+    starting_branch, args.base_branch = check_git_state(args.base_branch, not args.skip_check, args.patches_dir)
+
+    master_commit = latest_git_hash(args.base_branch)
+
+    if args.gen:
+        if os.path.lexists(TMP_DIR):
+            die(f'"{TMP_DIR}" must not exist in the current directory.')
+        gen_files = get_gen_files()
+        os.mkdir(TMP_DIR, 0o700)
+        for cmd in MAKE_GEN_CMDS:
+            cmd_chk(cmd)
+        cmd_chk(['rsync', '-a', *gen_files, f'{TMP_DIR}/master/'])
+
+    last_touch = time.time()
+
+    # Start by finding all patches so that we can load all possible parents.
+    patches = sorted(list(get_patch_branches(args.base_branch)))
+
+    parent_patch = { }
+    description = { }
+
+    for patch in patches:
+        branch = f"patch/{args.base_branch}/{patch}"
+        desc = ''
+        proc = cmd_pipe(['git', 'diff', '-U1000', f"{args.base_branch}...{branch}", '--', f"PATCH.{patch}"])
+        in_diff = False
+        for line in proc.stdout:
+            if in_diff:
+                if not re.match(r'^[ +]', line):
+                    continue
+                line = line[1:]
+                m = re.search(r'patch -p1 <patches/(\S+)\.diff', line)
+                if m and m[1] != patch:
+                    parpat = parent_patch[patch] = m[1]
+                    if not parpat in patches:
+                        die(f"Parent of {patch} is not a local branch: {parpat}")
+                desc += line
+            elif re.match(r'^@@ ', line):
+                in_diff = True
+        description[patch] = desc
+        proc.communicate()
+
+    if args.patch_files: # Limit the list of patches to actually process
+        valid_patches = patches
+        patches = [ ]
+        for fn in args.patch_files:
+            name = re.sub(r'\.diff$', '', re.sub(r'.+/', '', fn))
+            if name not in valid_patches:
+                die(f"Local branch not available for patch: {name}")
+            patches.append(name)
+
+    completed = set()
+
+    for patch in patches:
+        if patch in completed:
+            continue
+        if not update_patch(patch):
+            break
+
+    if args.gen:
+        shutil.rmtree(TMP_DIR)
+
+    while last_touch >= time.time():
+        time.sleep(1)
+    cmd_chk(['git', 'checkout', starting_branch])
+
+
+def update_patch(patch):
+    global last_touch
+
+    completed.add(patch) # Mark it as completed early to short-circuit any (bogus) dependency loops.
+
+    parent = parent_patch.get(patch, None)
+    if parent:
+        if parent not in completed:
+            if not update_patch(parent):
+                return 0
+        based_on = parent = f"patch/{args.base_branch}/{parent}"
+    else:
+        parent = args.base_branch
+        based_on = master_commit
+
+    print(f"======== {patch} ========")
+
+    while args.gen and last_touch >= time.time():
+        time.sleep(1)
+    s = cmd_run(f"git checkout patch/{args.base_branch}/{patch}".split())
+    if s.returncode != 0:
+        return 0
+
+    s = cmd_run(['git', 'merge', based_on])
+    ok = s.returncode == 0
+    if not ok or args.shell:
+        m = re.search(r'([^/]+)$', parent)
+        parent_dir = m[1]
+        if not ok:
+            print(f'"git merge {based_on}" incomplete -- please fix.')
+        os.environ['PS1'] = f"[{parent_dir}] {patch}: "
+        while True:
+            s = cmd_run([os.environ.get('SHELL', '/bin/sh')])
+            if s.returncode != 0:
+                ans = input("Abort? [n/y] ")
+                if re.match(r'^y', ans, flags=re.I):
+                    return 0
+                continue
+            cur_branch, is_clean, status_txt = check_git_status(0)
+            if is_clean:
+                break
+            print(status_txt, end='')
+
+    with open(f"{args.patches_dir}/{patch}.diff", 'w', encoding='utf-8') as fh:
+        fh.write(description[patch])
+        fh.write(f"\nbased-on: {based_on}\n")
+
+        if args.gen:
+            gen_files = get_gen_files()
+            for cmd in MAKE_GEN_CMDS:
+                cmd_chk(cmd)
+            cmd_chk(['rsync', '-a', *gen_files, f"{TMP_DIR}/{patch}/"])
+        else:
+            gen_files = [ ]
+        last_touch = time.time()
+
+        proc = cmd_pipe(['git', 'diff', based_on])
+        skipping = False
+        for line in proc.stdout:
+            if skipping:
+                if not re.match(r'^diff --git a/', line):
+                    continue
+                skipping = False
+            elif re.match(r'^diff --git a/PATCH', line):
+                skipping = True
+                continue
+            if not re.match(r'^index ', line):
+                fh.write(line)
+        proc.communicate()
+
+        if args.gen:
+            e_tmp_dir = re.escape(TMP_DIR)
+            diff_re  = re.compile(r'^(diff -Nurp) %s/[^/]+/(.*?) %s/[^/]+/(.*)' % (e_tmp_dir, e_tmp_dir))
+            minus_re = re.compile(r'^\-\-\- %s/[^/]+/([^\t]+)\t.*' % e_tmp_dir)
+            plus_re  = re.compile(r'^\+\+\+ %s/[^/]+/([^\t]+)\t.*' % e_tmp_dir)
+
+            if parent == args.base_branch:
+                parent_dir = 'master'
+            else:
+                m = re.search(r'([^/]+)$', parent)
+                parent_dir = m[1]
+
+            proc = cmd_pipe(['diff', '-Nurp', f"{TMP_DIR}/{parent_dir}", f"{TMP_DIR}/{patch}"])
+            for line in proc.stdout:
+                line = diff_re.sub(r'\1 a/\2 b/\3', line)
+                line = minus_re.sub(r'--- a/\1', line)
+                line =  plus_re.sub(r'+++ b/\1', line)
+                fh.write(line)
+            proc.communicate()
+            for fn in gen_files:
+                os.unlink(fn)
+
+    return 1
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description="Turn a git branch back into a diff files in the patches dir.", add_help=False)
+    parser.add_argument('--branch', '-b', dest='base_branch', metavar='BASE_BRANCH', default='master', help="The branch the patch is based on. Default: master.")
+    parser.add_argument('--skip-check', action='store_true', help="Skip the check that ensures starting with a clean branch.")
+    parser.add_argument('--shell', '-s', action='store_true', help="Launch a shell for every patch/BASE/* branch updated, not just when a conflict occurs.")
+    parser.add_argument('--gen', metavar='DIR', nargs='?', const='', help='Include generated files. Optional DIR value overrides the default of using the "patches" dir.')
+    parser.add_argument('--patches-dir', '-p', metavar='DIR', default='patches', help="Override the location of the rsync-patches dir. Default: patches.")
+    parser.add_argument('patch_files', metavar='patches/DIFF_FILE', nargs='*', help="Specify what patch diff files to process. Default: all of them.")
+    parser.add_argument("--help", "-h", action="help", help="Output this help message and exit.")
+    args = parser.parse_args()
+    if args.gen == '':
+        args.gen = args.patches_dir
+    elif args.gen is not None:
+        args.patches_dir = args.gen
+    main()
+
+# vim: sw=4 et

packaging/pkglib.py

@@ -0,0 +1,261 @@
+import os, sys, re, subprocess
+
+# This python3 library provides a few helpful routines that are
+# used by the latest packaging scripts.
+
+default_encoding = 'utf-8'
+
+# Output the msg args to stderr.  Accepts all the args that print() accepts.
+def warn(*msg):
+    print(*msg, file=sys.stderr)
+
+
+# Output the msg args to stderr and die with a non-zero return-code.
+# Accepts all the args that print() accepts.
+def die(*msg):
+    warn(*msg)
+    sys.exit(1)
+
+
+# Set this to an encoding name or set it to None to avoid the default encoding idiom.
+def set_default_encoding(enc):
+    default_encoding = enc
+
+
+# Set shell=True if the cmd is a string; sets a default encoding unless raw=True was specified.
+def _tweak_opts(cmd, opts, **maybe_set):
+    # This sets any maybe_set value that isn't already set AND creates a copy of opts for us.
+    opts = {**maybe_set, **opts}
+
+    if type(cmd) == str:
+        opts = {'shell': True, **opts}
+
+    want_raw = opts.pop('raw', False)
+    if default_encoding and not want_raw:
+        opts = {'encoding': default_encoding, **opts}
+
+    capture = opts.pop('capture', None)
+    if capture:
+        if capture == 'stdout':
+            opts = {'stdout': subprocess.PIPE, **opts}
+        elif capture == 'stderr':
+            opts = {'stderr': subprocess.PIPE, **opts}
+        elif capture == 'output':
+            opts = {'stdout': subprocess.PIPE, 'stderr': subprocess.PIPE, **opts}
+        elif capture == 'combined':
+            opts = {'stdout': subprocess.PIPE, 'stderr': subprocess.STDOUT, **opts}
+
+    discard = opts.pop('discard', None)
+    if discard:
+        # We DO want to override any already set stdout|stderr values (unlike above).
+        if discard == 'stdout' or discard == 'output':
+            opts['stdout'] = subprocess.DEVNULL
+        if discard == 'stderr' or discard == 'output':
+            opts['stderr'] = subprocess.DEVNULL
+
+    return opts
+
+
+# This does a normal subprocess.run() with some auto-args added to make life easier.
+def cmd_run(cmd, **opts):
+    return subprocess.run(cmd, **_tweak_opts(cmd, opts))
+
+
+# Like cmd_run() with a default check=True specified.
+def cmd_chk(cmd, **opts):
+    return subprocess.run(cmd, **_tweak_opts(cmd, opts, check=True))
+
+
+# Capture stdout in a string and return the (output, return_code) tuple.
+# Use capture='combined' opt to get both stdout and stderr together.
+def cmd_txt_status(cmd, **opts):
+    input = opts.pop('input', None)
+    if input is not None:
+        opts['stdin'] = subprocess.PIPE
+    proc = subprocess.Popen(cmd, **_tweak_opts(cmd, opts, capture='stdout'))
+    out = proc.communicate(input=input)[0]
+    return (out, proc.returncode)
+
+
+# Like cmd_txt_status() but just return the output.
+def cmd_txt(cmd, **opts):
+    return cmd_txt_status(cmd, **opts)[0]
+
+
+# Capture stdout in a string and return the output if the command has a 0 return code.
+# Otherwise it throws an exception that indicates the return code and the output.
+def cmd_txt_chk(cmd, **opts):
+    out, rc = cmd_txt_status(cmd, **opts)
+    if rc != 0:
+        cmd_err = f'Command "{cmd}" returned non-zero exit status "{rc}" and output:\n{out}'
+        raise Exception(cmd_err)
+    return out
+
+
+# Starts a piped-output command of stdout (by default) and leaves it up to you to read
+# the output and call communicate() on the returned object.
+def cmd_pipe(cmd, **opts):
+    return subprocess.Popen(cmd, **_tweak_opts(cmd, opts, capture='stdout'))
+
+
+# Runs a "git status" command and dies if the checkout is not clean (the
+# arg fatal_unless_clean can be used to make that non-fatal.  Returns a
+# tuple of the current branch, the is_clean flag, and the status text.
+def check_git_status(fatal_unless_clean=True, subdir='.'):
+    status_txt = cmd_txt_chk(f"cd '{subdir}' && git status")
+    is_clean = re.search(r'\nnothing to commit.+working (directory|tree) clean', status_txt) != None
+
+    if not is_clean and fatal_unless_clean:
+        if subdir == '.':
+            subdir = ''
+        else:
+            subdir = f" *{subdir}*"
+        die(f"The{subdir} checkout is not clean:\n" + status_txt)
+
+    m = re.match(r'^(?:# )?On branch (.+)\n', status_txt)
+    cur_branch = m[1] if m else None
+
+    return (cur_branch, is_clean, status_txt)
+
+
+# Calls check_git_status() on the current git checkout and (optionally) a subdir path's
+# checkout. Use fatal_unless_clean to indicate if an unclean checkout is fatal or not.
+# The master_branch arg indicates what branch we want both checkouts to be using, and
+# if the branch is wrong the user is given the option of either switching to the right
+# branch or aborting.
+def check_git_state(master_branch, fatal_unless_clean=True, check_extra_dir=None):
+    cur_branch = check_git_status(fatal_unless_clean)[0]
+    branch = re.sub(r'^patch/([^/]+)/[^/]+$', r'\1', cur_branch) # change patch/BRANCH/PATCH_NAME into BRANCH
+    if branch != master_branch:
+        print(f"The checkout is not on the {master_branch} branch.")
+        if master_branch != 'master':
+            sys.exit(1)
+        ans = input(f"Do you want me to continue with --branch={branch}? [n] ")
+        if not ans or not re.match(r'^y', ans, flags=re.I):
+            sys.exit(1)
+        master_branch = branch
+
+    if check_extra_dir and os.path.isdir(os.path.join(check_extra_dir, '.git')):
+        branch = check_git_status(fatal_unless_clean, check_extra_dir)[0]
+        if branch != master_branch:
+            print(f"The *{check_extra_dir}* checkout is on branch {branch}, not branch {master_branch}.")
+            ans = input(f"Do you want to change it to branch {master_branch}? [n] ")
+            if not ans or not re.match(r'^y', ans, flags=re.I):
+                sys.exit(1)
+            subdir.check_call(f"cd {check_extra_dir} && git checkout '{master_branch}'", shell=True)
+
+    return (cur_branch, master_branch)
+
+
+# Return the git hash of the most recent commit.
+def latest_git_hash(branch):
+    out = cmd_txt_chk(['git', 'log', '-1', '--no-color', branch])
+    m = re.search(r'^commit (\S+)', out, flags=re.M)
+    if not m:
+        die(f"Unable to determine commit hash for master branch: {branch}")
+    return m[1]
+
+
+# Return a set of all branch names that have the format "patch/BASE_BRANCH/NAME"
+# for the given base_branch string.  Just the NAME portion is put into the set.
+def get_patch_branches(base_branch):
+    branches = set()
+    proc = cmd_pipe('git branch -l'.split())
+    for line in proc.stdout:
+        m = re.search(r' patch/([^/]+)/(.+)', line)
+        if m and m[1] == base_branch:
+            branches.add(m[2])
+    proc.communicate()
+    return branches
+
+
+def mandate_gensend_hook():
+    hook = '.git/hooks/pre-push'
+    if not os.path.exists(hook):
+        print('Creating hook file:', hook)
+        cmd_chk(['./rsync', '-a', 'packaging/pre-push', hook])
+    else:
+        out, rc = cmd_txt_status(['fgrep', 'make gensend', hook], discard='output')
+        if rc:
+            die('Please add a "make gensend" into your', hook, 'script.')
+
+
+# Snag the GENFILES values out of the Makefile.in file and return them as a list.
+def get_gen_files():
+    cont_re = re.compile(r'\\\n')
+
+    extras = [ ]
+
+    with open('Makefile.in', 'r', encoding='utf-8') as fh:
+        for line in fh:
+            if not extras:
+                chk = re.sub(r'^GENFILES=', '', line)
+                if line == chk:
+                    continue
+                line = chk
+            m = re.search(r'\\$', line)
+            line = re.sub(r'^\s+|\s*\\\n?$|\s+$', '', line)
+            extras += line.split()
+            if not m:
+                break
+
+    return extras
+
+
+def get_configure_version():
+    with open('configure.ac', 'r', encoding='utf-8') as fh:
+        for line in fh:
+            m = re.match(r'^AC_INIT\(\[rsync\],\s*\[(\d.+?)\]', line)
+            if m:
+                return m[1]
+    die("Unable to find AC_INIT with version in configure.ac")
+
+
+def get_OLDNEWS_version_info():
+    rel_re = re.compile(r'^\| \d{2} \w{3} \d{4}\s+\|\s+(?P<ver>\d+\.\d+\.\d+)\s+\|\s+(?P<pdate>\d{2} \w{3} \d{4}\s+)?\|\s+(?P<pver>\d+)\s+\|')
+    last_version = last_protocol_version = None
+    pdate = { }
+
+    with open('OLDNEWS.md', 'r', encoding='utf-8') as fh:
+        for line in fh:
+            if not last_version:
+                m = re.search(r'(\d+\.\d+\.\d+)', line)
+                if m:
+                    last_version = m[1]
+            m = rel_re.match(line)
+            if m:
+                if m['pdate']:
+                    pdate[m['ver']] = m['pdate']
+                if m['ver'] == last_version:
+                    last_protocol_version = m['pver']
+                    break
+
+    if not last_protocol_version:
+        die(f"Unable to determine protocol_version for {last_version}.")
+
+    return last_version, last_protocol_version
+
+
+def get_protocol_versions():
+    protocol_version = subprotocol_version = None
+
+    with open('rsync.h', 'r', encoding='utf-8') as fh:
+        for line in fh:
+            m = re.match(r'^#define\s+PROTOCOL_VERSION\s+(\d+)', line)
+            if m:
+                protocol_version = m[1]
+                continue
+            m = re.match(r'^#define\s+SUBPROTOCOL_VERSION\s+(\d+)', line)
+            if m:
+                subprotocol_version = m[1]
+                break
+
+    if not protocol_version:
+        die("Unable to determine the current PROTOCOL_VERSION.")
+
+    if not subprotocol_version:
+        die("Unable to determine the current SUBPROTOCOL_VERSION.")
+
+    return protocol_version, subprotocol_version
+
+# vim: sw=4 et

packaging/pre-push

@@ -0,0 +1,3 @@
+#!/bin/sh
+cat >/dev/null # Just discard stdin data
+make gensend

packaging/release-rsync

@@ -1,328 +1,394 @@
-#!/usr/bin/perl
-use strict;
+#!/usr/bin/env -S python3 -B
 
-# This script expects the directory ~/samba-rsync-ftp to exist and to
-# be a copy of the /home/ftp/pub/rsync dir on samba.org.  If it is run
-# in test mode, it instead expects a dir named ~/tmp/samba-rsync-ftp
-# (e.g. copy ~/samba-rsync-ftp into ~/tmp and you can do a trial-run of
-# a release without affecting the files in the ~/samba-rsync-ftp dir).
-#
-# Run this as "release-rsync live" to affect ~/samba-rsync-ftp instead
-# of ~/tmp/samba-rsync-ftp.
+# This script expects the directory ~/samba-rsync-ftp to exist and to be a
+# copy of the /home/ftp/pub/rsync dir on samba.org.  When the script is done,
+# the git repository in the current directory will be updated, and the local
+# ~/samba-rsync-ftp dir will be ready to be rsynced to samba.org.
 
-use Date::Format;
+import os, sys, re, argparse, glob, shutil, signal
+from datetime import datetime
+from getpass import getpass
 
-my $dest = $ENV{HOME} . '/samba-rsync-ftp';
-my $releasedir = $ENV{HOME} . '/release';
-my $cvsroot = $ENV{CVSROOT} = 'samba.org:/data/cvs';
+sys.path = ['packaging'] + sys.path
 
-my $ztoday = time2str('%d %b %Y', time);
-my $today = $ztoday;
-$today =~ s/^0//;
+from pkglib import *
 
-my $break = <<EOT;
-==========================================================================
-EOT
-my $note = <<EOT;
-== Note: type "-a u,n" if you want to auto-accept the U,N suggestions.  ==
-EOT
+dest = os.environ['HOME'] + '/samba-rsync-ftp'
+ORIGINAL_PATH = os.environ['PATH']
 
-my $live = shift;
-my $skipping = '';
+def main():
+    now = datetime.now()
+    cl_today = now.strftime('* %a %b %d %Y')
+    year = now.strftime('%Y')
+    ztoday = now.strftime('%d %b %Y')
+    today = ztoday.lstrip('0')
 
-print $break;
-if ($live) {
-    print <<EOT;
+    mandate_gensend_hook()
+
+    curdir = os.getcwd()
+
+    signal.signal(signal.SIGINT, signal_handler)
+
+    gen_files = get_gen_files()
+
+    dash_line = '=' * 74
+
+    print(f"""\
+{dash_line}
 == This will release a new version of rsync onto an unsuspecting world. ==
-EOT
-} else {
-    print <<EOT;
-==     **** TESTMODE ****    (Add "live" arg to avoid this.)            ==
-EOT
-    $dest =~ s#([^/]+$)#tmp/$1#;
-    $skipping = ' ** SKIPPING **';
-}
-die "$dest does not exist\n" unless -d $dest;
+{dash_line}
+""")
 
-print $break, "\nChecking out the latest rsync into $releasedir ...\n";
+    if not os.path.isdir(dest):
+        die(dest, "dest does not exist")
+    if not os.path.isdir('.git'):
+        die("There is no .git dir in the current directory.")
+    if os.path.lexists('a'):
+        die('"a" must not exist in the current directory.')
+    if os.path.lexists('b'):
+        die('"b" must not exist in the current directory.')
+    if os.path.lexists('patches.gen'):
+        die('"patches.gen" must not exist in the current directory.')
 
-mkdir($releasedir, 0755) or die $! unless -d $releasedir;
-chdir($releasedir) or die $!;
+    check_git_state(args.master_branch, True, 'patches')
 
-system 'rm -rf rsync';
+    confversion = get_configure_version()
 
-my(%dirs, @files);
-open(CVS, '-|', 'cvs checkout -P rsync') or die $!;
-while (<CVS>) {
-    print $_;
-    next if /\.(cvs)?ignore$/;
-    if (m#^[UP] rsync/(.*)#) {
-	my $fn = $1;
-	my($dir) = $fn =~ m#^(.+)/#;
-	push(@files, $dir) if defined($dir) && !$dirs{$1}++;
-	push(@files, $fn);
-    }
-}
+    # All version values are strings!
+    lastversion, last_protocol_version = get_OLDNEWS_version_info()
+    protocol_version, subprotocol_version = get_protocol_versions()
 
-chdir('rsync') or die $!;
+    version = confversion
+    m = re.search(r'pre(\d+)', version)
+    if m:
+        version = re.sub(r'pre\d+', 'pre' + str(int(m[1]) + 1), version)
+    else:
+        version = version.replace('dev', 'pre1')
 
-my($version, $lastversion);
-open(IN, '<', 'configure.in') or die $!;
-while (<IN>) {
-    if (/^RSYNC_VERSION=(.*)/) {
-	$version = $lastversion = $1;
-	last;
-    }
-}
-close IN;
+    ans = input(f"Please enter the version number of this release: [{version}] ")
+    if ans == '.':
+        version = re.sub(r'pre\d+', '', version)
+    elif ans != '':
+        version = ans
+    if not re.match(r'^[\d.]+(pre\d+)?$', version):
+        die(f'Invalid version: "{version}"')
 
-if ($lastversion =~ /cvs$/) {
-    open(IN, '<', 'OLDNEWS') or die $!;
-    $_ = <IN>;
-    close IN;
-    ($lastversion) = /(\d+\.\d+\.\d+)/;
-}
+    v_ver = 'v' + version
+    rsync_ver = 'rsync-' + version
 
-$version =~ s/cvs/pre1/ || $version =~ s/pre(\d+)/ 'pre' . ($1 + 1) /e;
+    if os.path.lexists(rsync_ver):
+        die(f'"{rsync_ver}" must not exist in the current directory.')
 
-print $break, "\nPlease enter the version number of this release: [$version] ";
-chomp($_ = <STDIN>);
-if ($_ eq '.') {
-    $version =~ s/pre\d+//;
-} elsif ($_ ne '') {
-    $version = $_;
-}
-$version =~ s/[-.]*pre[-.]*/pre/;
+    out = cmd_txt_chk(['git', 'tag', '-l', v_ver])
+    if out != '':
+        print(f"Tag {v_ver} already exists.")
+        ans = input("\nDelete tag or quit? [Q/del] ")
+        if not re.match(r'^del', ans, flags=re.I):
+            die("Aborted")
+        cmd_chk(['git', 'tag', '-d', v_ver])
 
-$lastversion =~ s/(\d+)pre\d+$/ $1 - 1 /e unless $version =~ /pre/;
+    version = re.sub(r'[-.]*pre[-.]*', 'pre', version)
+    if 'pre' in version and not confversion.endswith('dev'):
+        lastversion = confversion
 
-my $cvstag = "release-$version";
-$cvstag =~ s/[.]/-/g;
-$cvstag =~ s/pre/-pre/;
+    ans = input(f"Enter the previous version to produce a patch against: [{lastversion}] ")
+    if ans != '':
+        lastversion = ans
+    lastversion = re.sub(r'[-.]*pre[-.]*', 'pre', lastversion)
 
-print "Enter the previous version to produce a patch against: [$lastversion] ";
-chomp($_ = <STDIN>);
-$lastversion = $_ if $_ ne '';
-$lastversion =~ s/[-.]*pre[-.]*/pre/;
+    rsync_lastver = 'rsync-' + lastversion
+    if os.path.lexists(rsync_lastver):
+        die(f'"{rsync_lastver}" must not exist in the current directory.')
 
-my $release = 1;
-print "Please enter the RPM release number of this release: [$release] ";
-chomp($_ = <STDIN>);
-$release = $_ if $_ ne '';
+    m = re.search(r'(pre\d+)', version)
+    pre = m[1] if m else ''
 
-my $diffdir;
-my $skipping2;
-if ($lastversion =~ /pre/) {
-    if ($version !~ /pre/) {
-	die "You should not diff a release version against a pre-release version.\n";
-    }
-    $diffdir = "$dest/old-previews";
-    $skipping2 = ' ** SKIPPING **';
-} elsif ($version =~ /pre/) {
-    $diffdir = $dest;
-    $skipping2 = ' ** SKIPPING **';
-} else {
-    $diffdir = "$dest/old-versions";
-    $skipping2 = '';
-}
+    release = '0.1' if pre else '1'
+    ans = input(f"Please enter the RPM release number of this release: [{release}] ")
+    if ans != '':
+        release = ans
+    if pre:
+        release += '.' + pre
 
-print "\n", $break, <<EOT;
-\$version is "$version"
-\$lastversion is "$lastversion"
-\$cvstag is "$cvstag"
-\$dest is "$dest"
-\$releasedir is "$releasedir"
-\$diffdir is "$diffdir"
-\$release is "$release"
+    finalversion = re.sub(r'pre\d+', '', version)
+    if protocol_version == last_protocol_version:
+        proto_changed = 'unchanged'
+        proto_change_date = ' ' * 11
+    else:
+        proto_changed = 'changed'
+        if finalversion in pdate:
+            proto_change_date = pdate[finalversion]
+        else:
+            while True:
+                ans = input("On what date did the protocol change to {protocol_version} get checked in? (dd Mmm yyyy) ")
+                if re.match(r'^\d\d \w\w\w \d\d\d\d$', ans):
+                    break
+            proto_change_date = ans
+
+    if 'pre' in lastversion:
+        if not pre:
+            die("You should not diff a release version against a pre-release version.")
+        srcdir = srcdiffdir = lastsrcdir = 'src-previews'
+        skipping = ' ** SKIPPING **'
+    elif pre:
+        srcdir = srcdiffdir = 'src-previews'
+        lastsrcdir = 'src'
+        skipping = ' ** SKIPPING **'
+    else:
+        srcdir = lastsrcdir = 'src'
+        srcdiffdir = 'src-diffs'
+        skipping = ''
+
+    print(f"""
+{dash_line}
+version is "{version}"
+lastversion is "{lastversion}"
+dest is "{dest}"
+curdir is "{curdir}"
+srcdir is "{srcdir}"
+srcdiffdir is "{srcdiffdir}"
+lastsrcdir is "{lastsrcdir}"
+release is "{release}"
 
 About to:
-    - make sure that SUBPROTOCOL_VERSION is 0$skipping2
-    - tweak the version in configure.in, configure, and the spec files
-    - make sure that configure, config.h.in, and proto.h are updated
-    - tweak NEWS and OLDNEWS to update the release date$skipping2
-    - tweak the date in the *.yo files and re-generate the man pages
-    - make sure that the patches dir has been updated
-    - page through the "cvs diff" output
+    - tweak SUBPROTOCOL_VERSION in rsync.h, if needed
+    - tweak the version in configure.ac and the spec files
+    - tweak NEWS.md and OLDNEWS.md to ensure header values are correct
+    - generate configure.sh, config.h.in, and proto.h
+    - page through the differences
+""")
+    ans = input("<Press Enter to continue> ")
 
-EOT
-print "<Press Enter to continue> ";
-$_ = <STDIN>;
-my $f_opt = /f/ ? ' -f' : '';
+    specvars = {
+        'Version:': finalversion,
+        'Release:': release,
+        '%define fullversion': f'%{{version}}{pre}',
+        'Released': version + '.',
+        '%define srcdir': srcdir,
+        }
 
-print $break;
-system "./prepare-source && touch proto.h";
+    tweak_files = 'configure.ac rsync.h NEWS.md OLDNEWS.md'.split()
+    tweak_files += glob.glob('packaging/*.spec')
+    tweak_files += glob.glob('packaging/*/*.spec')
 
-my @tweak_files = ( glob('packaging/*.spec'), glob('packaging/*/*.spec'),
-		    glob('*.yo'), qw( configure.in configure ) );
-if ($version !~ /pre/) {
-    push(@tweak_files, qw( rsync.h NEWS OLDNEWS ));
-}
-foreach my $fn (@tweak_files) {
-    open(IN, '<', $fn) or die $!;
-    undef $/; $_ = <IN>; $/ = "\n";
-    close IN;
-    if ($fn =~ /configure/) {
-	s/^RSYNC_VERSION=.*/RSYNC_VERSION=$version/m;
-    } elsif ($fn =~ /\.spec/) {
-	s/^(Version:) .*/$1 $version/m;
-	s/^(Release:) .*/$1 $release/m;
-    } elsif ($fn =~ /\.yo/) {
-	s/^(manpage\([^)]+\)\(\d+\)\()[^)]+(\).*)/$1$today$2/m;
-	s/^(This man ?page is current for version) \S+ (of rsync)/$1 $version $2/m;
-    } elsif ($fn eq 'NEWS') {
-	s/^(NEWS for rsync \Q$version\E) \(UNRELEASED\)\s*\n/$1 ($today)\n/mi
-	    or die "Couldn't update NEWS file with release date!\n";
-    } elsif ($fn eq 'rsync.h') {
-	s/(#define\s+SUBPROTOCOL_VERSION)\s+\d+/$1 0/;
-    } elsif ($fn eq 'OLDNEWS') {
-	s/^\t\S\S\s\S\S\S\s\d\d\d\d(\t\Q$version\E)/\t$ztoday$1/m
-	    or die "Couldn't update OLDNEWS file with release date!\n";
-    } else {
-	die "Unrecognized file in \@tweak_files: $fn\n";
-    }
-    open(OUT, '>', $fn) or die $!;
-    print OUT $_;
-    close OUT;
-}
+    for fn in tweak_files:
+        with open(fn, 'r', encoding='utf-8') as fh:
+            old_txt = txt = fh.read()
+        if 'configure' in fn:
+            x_re = re.compile(r'^(AC_INIT\(\[rsync\],\s*\[)\d.+?(\])', re.M)
+            txt = replace_or_die(x_re, r'\g<1>%s\2' % version, txt, f"Unable to update AC_INIT with version in {fn}")
+        elif '.spec' in fn:
+            for var, val in specvars.items():
+                x_re = re.compile(r'^%s .*' % re.escape(var), re.M)
+                txt = replace_or_die(x_re, var + ' ' + val, txt, f"Unable to update {var} in {fn}")
+            x_re = re.compile(r'^\* \w\w\w \w\w\w \d\d \d\d\d\d (.*)', re.M)
+            txt = replace_or_die(x_re, r'%s \1' % cl_today, txt, f"Unable to update ChangeLog header in {fn}")
+        elif fn == 'rsync.h':
+            x_re = re.compile('(#define\s+SUBPROTOCOL_VERSION)\s+(\d+)')
+            repl = lambda m: m[1] + ' ' + '0' if not pre or proto_changed != 'changed' else 1 if m[2] == '0' else m[2]
+            txt = replace_or_die(x_re, repl, txt, f"Unable to find SUBPROTOCOL_VERSION define in {fn}")
+        elif fn == 'NEWS.md':
+            x_re = re.compile(
+                    r'^(# NEWS for rsync %s )(\(UNRELEASED\))\s*(\n\nProtocol: )(\d+) (\([^)]+\))\n' % re.escape(finalversion),
+                    re.I)
+            repl = lambda m: m[1] + (m[2] if pre else f"({today})") + m[3] + f"{protocol_version} ({proto_changed})\n"
+            msg = (f"The first 3 lines of {fn} are not in the right format.  They must be:\n"
+                    + f"# NEWS for rsync {finalversion} (UNRELEASED)\n\n"
+                    + f"Protocol: {protocol_version} ({proto_changed})")
+            txt = replace_or_die(x_re, repl, txt, msg)
+        elif fn == 'OLDNEWS.md':
+            efv = re.escape(finalversion)
+            x_re = re.compile(r'^(\| )(\S{2} \S{3} \d{4})(\s+\|\s+%s\s+\| ).{11}(\s+\| )\S{2}(\s+\|+)$' % efv, re.M)
+            repl = lambda m: m[1] + (m[2] if pre else ztoday) + m[3] + proto_change_date + m[4] + protocol_version + m[5]
+            txt = replace_or_die(x_re, repl, txt, f'Unable to find "| ?? ??? {year} | {finalversion} | ... |" line in {fn}')
+        else:
+            die(f"Unrecognized file in tweak_files: {fn}")
 
-system "yodl2man -o rsync.1 rsync.yo; ./tweak_manpage rsync.1";
-system "yodl2man -o rsyncd.conf.5 rsyncd.conf.yo; ./tweak_manpage rsyncd.conf.5";
+        if txt != old_txt:
+            print(f"Updating {fn}")
+            with open(fn, 'w', encoding='utf-8') as fh:
+                fh.write(txt)
 
-mkdir('patches/tmp') or die $!;
-system "rsync -a --exclude=patches/ --exclude-from=.cvsignore . patches/tmp/cvsdir/";
+    cmd_chk(['packaging/year-tweak'])
 
-print "\n", $break, $note, $break;
-system "patches/verify-patches -n -an$f_opt";
+    print(dash_line)
+    cmd_run("git diff --color | less -p '^diff .*'")
 
-print $break;
-system "cvs -q diff | egrep -v '^(===============|RCS file: |retrieving revision |Index: )' | less -p '^diff .*'";
+    srctar_name = f"{rsync_ver}.tar.gz"
+    pattar_name = f"rsync-patches-{version}.tar.gz"
+    diff_name = f"{rsync_lastver}-{version}.diffs.gz"
+    srctar_file = f"{dest}/{srcdir}/{srctar_name}"
+    pattar_file = f"{dest}/{srcdir}/{pattar_name}"
+    diff_file = f"{dest}/{srcdiffdir}/{diff_name}"
+    news_file = f"{dest}/{srcdir}/{rsync_ver}-NEWS.md"
+    lasttar_file = f"{dest}/{lastsrcdir}/{rsync_lastver}.tar.gz"
 
-print $break, <<EOT;
+    print(f"""\
+{dash_line}
 
 About to:
-    - "cvs commit" all changes$skipping
-    - "cvs tag" this release as $cvstag$skipping
-    - change the diffs in the patches dir to include generated files
+    - git commit all changes
+    - generate the manpages
+    - merge the {args.master_branch} branch into the patch/{args.master_branch}/* branches
+    - update the files in the "patches" dir and OPTIONALLY
+      (if you type 'y') to launch a shell for each patch
+""")
+    ans = input("<Press Enter OR 'y' to continue> ")
 
-EOT
-print "<Press Enter to continue> ";
-$_ = <STDIN>;
+    s = cmd_run(['git', 'commit', '-a', '-m', f'Preparing for release of {version}'])
+    if s.returncode:
+        die('Aborting')
 
-if ($live) {
-    system "cvs commit -m 'Preparing for release of $version'";
-    system "cvs tag -F $cvstag .";
-}
+    cmd_chk('make reconfigure ; make gen')
+    cmd_chk(['rsync', '-a', *gen_files, 'SaVeDiR/'])
 
-if (!/skip/i) {
-    print "\n", $break, $note, $break;
-    system "patches/verify-patches -pun -an";
-}
+    print(f'Creating any missing patch branches.')
+    s = cmd_run(f'packaging/branch-from-patch --branch={args.master_branch} --add-missing')
+    if s.returncode:
+        die('Aborting')
 
-my $tar_name = "rsync-$version.tar.gz";
-my $diff_name = "rsync-$lastversion-$version.diffs.gz";
-my $tar_file = "$dest/$tar_name";
-my $diff_file = "$dest/$diff_name";
+    print('Updating files in "patches" dir ...')
+    s = cmd_run(f'packaging/patch-update --branch={args.master_branch}')
+    if s.returncode:
+        die('Aborting')
 
-print $break, <<EOT;
+    if re.match(r'^y', ans, re.I):
+        print(f'\nVisiting all "patch/{args.master_branch}/*" branches ...')
+        cmd_run(f"packaging/patch-update --branch={args.master_branch} --skip-check --shell")
 
-About to do the following in the samba-rsync-ftp dir:
-    - move the old tar/diff files into the appropriate old-* dirs
-    - hard-link the moved tar/diff files on samba.org$skipping
-    - create release tar, "$tar_name"
-    - create release diffs, "$diff_name"
-    - update README, *NEWS, TODO, and cvs.log
-    - update rsync*.html man pages
-    - gpg-sign the release files$skipping
+    cmd_run("rm -f *.[o15] *.html")
+    cmd_chk('rsync -a SaVeDiR/ .'.split())
 
-EOT
-print "<Press Enter to continue> ";
-$_ = <STDIN>;
+    if os.path.isdir('patches/.git'):
+        s = cmd_run(f"cd patches && git commit -a -m 'The patches for {version}.'")
+        if s.returncode:
+            die('Aborting')
 
-chdir($releasedir) or die $!;
+    print(f"""\
+{dash_line}
 
-print $break;
-system "rm -rf rsync-$version";
-rename('rsync', "rsync-$version") or die $!;
+About to:
+    - create signed tag for this release: {v_ver}
+    - create release diffs, "{diff_name}"
+    - create release tar, "{srctar_name}"
+    - generate {rsync_ver}/patches/* files
+    - create patches tar, "{pattar_name}"
+    - update top-level README.md, *NEWS.md, TODO, and ChangeLog
+    - update top-level rsync*.html manpages
+    - gpg-sign the release files
+    - update hard-linked top-level release files{skipping}
+""")
+    ans = input("<Press Enter to continue> ")
 
-# When creating a pre-release after a normal release, there's nothing to move.
-if ($diffdir ne $dest) {
-    chdir($dest) or die $!;
+    # TODO: is there a better way to ensure that our passphrase is in the agent?
+    cmd_run("touch TeMp; gpg --sign TeMp; rm TeMp*")
 
-    print "Shuffling old files ...\n";
+    out = cmd_txt(f"git tag -s -m 'Version {version}.' {v_ver}", capture='combined')
+    print(out, end='')
+    if 'bad passphrase' in out or 'failed' in out:
+        die('Aborting')
 
-    # We need to run this regardless of $lastversion's "pre"ness.
-    my @moved_files;
-    foreach my $fn (glob('rsync*pre*.tar.gz*'), glob('rsync*pre*-NEWS')) {
-	link($fn, "old-previews/$fn") or die $!;
-	push(@moved_files, $fn);
-    }
+    if os.path.isdir('patches/.git'):
+        out = cmd_txt(f"cd patches && git tag -s -m 'Version {version}.' {v_ver}", capture='combined')
+        print(out, end='')
+        if 'bad passphrase' in out or 'failed' in out:
+            die('Aborting')
 
-    if ($version !~ /pre/) {
-	foreach my $fn (glob('rsync*.tar.gz*'), glob('rsync*-NEWS')) {
-	    next if $fn =~ /^rsync.*pre/;
-	    link($fn, "old-versions/$fn") or die $!;
-	    push(@moved_files, $fn);
-	}
+    os.environ['PATH'] = ORIGINAL_PATH
 
-	foreach my $fn (glob('rsync*pre*.diffs.gz*')) {
-	    unlink($fn);
-	}
+    # Extract the generated files from the old tar.
+    tweaked_gen_files = [ f"{rsync_lastver}/{x}" for x in gen_files ]
+    cmd_run(['tar', 'xzf', lasttar_file, *tweaked_gen_files])
+    os.rename(rsync_lastver, 'a')
 
-	foreach my $fn (glob('rsync*.diffs.gz*')) {
-	    link($fn, "old-patches/$fn") or die $!;
-	    push(@moved_files, $fn);
-	}
-    }
+    print(f"Creating {diff_file} ...")
+    cmd_chk(['rsync', '-a', *gen_files, 'b/'])
 
-    # Optimize our future upload (in the absence of --detect-renamed) by
-    # using rsync to hard-link the above files on samba.org.
-    if ($live) {
-	system "rsync -avHOC --include='rsync*.gz*' --include='old-*/' --exclude='*' . samba.org:/home/ftp/pub/rsync";
-    }
-    foreach (@moved_files) {
-	unlink($_);
-    }
+    sed_script = r's:^((---|\+\+\+) [ab]/[^\t]+)\t.*:\1:' # CAUTION: must not contain any single quotes!
+    cmd_chk(f"(git diff v{lastversion} {v_ver} -- ':!.github'; diff -upN a b | sed -r '{sed_script}') | gzip -9 >{diff_file}")
+    shutil.rmtree('a')
+    os.rename('b', rsync_ver)
 
-    chdir($releasedir) or die $!;
-}
+    print(f"Creating {srctar_file} ...")
+    cmd_chk(f"git archive --format=tar --prefix={rsync_ver}/ {v_ver} | tar xf -")
+    cmd_chk(f"support/git-set-file-times --quiet --prefix={rsync_ver}/")
+    cmd_chk(['fakeroot', 'tar', 'czf', srctar_file, '--exclude=.github', rsync_ver])
+    shutil.rmtree(rsync_ver)
 
-print "Creating $tar_file ...\n";
-system "fakeroot tar czf $tar_file rsync-$version";
-open(TAR, '|-', "fakeroot tar --files-from=- --no-recursion --mode=g+w -czf $tar_file rsync-$version") or die $!;
-foreach (@files) {
-    print TAR "rsync-$version/$_\n";
-}
-close TAR;
+    print(f'Updating files in "{rsync_ver}/patches" dir ...')
+    os.mkdir(rsync_ver, 0o755)
+    os.mkdir(f"{rsync_ver}/patches", 0o755)
+    cmd_chk(f"packaging/patch-update --skip-check --branch={args.master_branch} --gen={rsync_ver}/patches".split())
 
-print "Creating $diff_file ...\n";
-system "rm -rf rsync-$version rsync-$lastversion";
-system "tar xzf $tar_file; tar xzf $diffdir/rsync-$lastversion.tar.gz";
-## TWEAK THE VERSIONS AS DESIRED HERE ##
-#mkdir("rsync-$lastversion/support", 0755) or die $!;
-#rename("rsync-$lastversion/rsyncstats", "rsync-$lastversion/support/rsyncstats");
-#unlink("rsync-$lastversion/.ignore");
-## END ##
-system "diff -urN --exclude=patches rsync-$lastversion rsync-$version| gzip -9 >$diff_file";
+    cmd_run("rm -f *.[o15] *.html")
+    cmd_chk('rsync -a SaVeDiR/ .'.split())
+    shutil.rmtree('SaVeDiR')
+    cmd_chk('make gen'.split())
 
-print "Updating the other files in $dest ...\n";
-system "rsync -a rsync-$version/{README,NEWS,OLDNEWS,TODO} $dest";
-unlink("$dest/rsync-$version-NEWS");
-link("$dest/NEWS", "$dest/rsync-$version-NEWS");
-system "rsync -a $cvsroot/CVSROOT/rsync.updates $dest/cvs.log";
+    print(f"Creating {pattar_file} ...")
+    cmd_chk(['fakeroot', 'tar', 'chzf', pattar_file, rsync_ver + '/patches'])
+    shutil.rmtree(rsync_ver)
 
-system "yodl2html -o $dest/rsync.html rsync-$version/rsync.yo";
-system "yodl2html -o $dest/rsyncd.conf.html rsync-$version/rsyncd.conf.yo";
+    print(f"Updating the other files in {dest} ...")
+    md_files = 'README.md NEWS.md OLDNEWS.md'.split()
+    html_files = [ fn for fn in gen_files if fn.endswith('.html') ]
+    cmd_chk(['rsync', '-a', *md_files, *html_files, dest])
+    cmd_chk(["packaging/md2html"] + [ dest +'/'+ fn for fn in md_files ])
 
-system "rm -rf rsync-*";
+    for topfn, verfn in (('NEWS.md', news_file), ('NEWS.html', news_file.replace('.md', '.html'))):
+        topfn = dest + '/' + topfn
+        if os.path.lexists(verfn):
+            os.unlink(verfn)
+        os.link(topfn, verfn)
 
-if ($live) {
-    chdir($dest) or die $!;
-    system "gpg -ba $tar_name; gpg -ba $diff_name";
-    print $break, <<EOT;
+    cmd_chk(f"git log --name-status | gzip -9 >{dest}/ChangeLog.gz")
 
-All done.  Remember to announce the release on *BOTH*
-rsync-announce\@lists.samba.org and rsync\@lists.samba.org!
-EOT
-} else {
-    print $break, "All done.\n";
-}
+    for fn in (srctar_file, pattar_file, diff_file):
+        asc_fn = fn + '.asc'
+        if os.path.lexists(asc_fn):
+            os.unlink(asc_fn)
+        res = cmd_run(['gpg', '--batch', '-ba', fn])
+        if res.returncode != 0 and res.returncode != 2:
+            die("gpg signing failed")
+
+    if not pre:
+        for find in f'{dest}/rsync-*.gz {dest}/rsync-*.asc {dest}/rsync-*-NEWS.md {dest}/src-previews/rsync-*diffs.gz*'.split():
+            for fn in glob.glob(find):
+                os.unlink(fn)
+        top_link = [
+                srctar_file, f"{srctar_file}.asc",
+                pattar_file, f"{pattar_file}.asc",
+                diff_file, f"{diff_file}.asc",
+                news_file,
+                ]
+        for fn in top_link:
+            os.link(fn, re.sub(r'/src(-\w+)?/', '/', fn))
+
+    print(f"""\
+{dash_line}
+
+Local changes are done.  When you're satisfied, push the git repository
+and rsync the release files.  Remember to announce the release on *BOTH*
+rsync-announce@lists.samba.org and rsync@lists.samba.org (and the web)!
+""")
+
+
+def replace_or_die(regex, repl, txt, die_msg):
+    m = regex.search(txt)
+    if not m:
+        die(die_msg)
+    return regex.sub(repl, txt, 1)
+
+
+def signal_handler(sig, frame):
+    die("\nAborting due to SIGINT.")
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description="Prepare a new release of rsync in the git repo & ftp dir.", add_help=False)
+    parser.add_argument('--branch', '-b', dest='master_branch', default='master', help="The branch to release. Default: master.")
+    parser.add_argument("--help", "-h", action="help", help="Output this help message and exit.")
+    args = parser.parse_args()
+    main()
+
+# vim: sw=4 et

packaging/solaris/build_pkg.sh

@@ -41,7 +41,7 @@ mkdir -p $FAKE_ROOT/man/man5
 cp ../../../rsync $FAKE_ROOT/bin/rsync
 cp ../../../rsync.1 $FAKE_ROOT/man/man1/rsync.1
 cp ../../../rsyncd.conf.5 $FAKE_ROOT/man/man5/rsyncd.conf.5
-cp ../../../README $FAKE_ROOT/doc/rsync/README
+cp ../../../README.md $FAKE_ROOT/doc/rsync/README.md
 cp ../../../COPYING $FAKE_ROOT/doc/rsync/COPYING
 cp ../../../tech_report.pdf $FAKE_ROOT/doc/rsync/tech_report.pdf
 cp ../../../COPYING $FAKE_ROOT/COPYING
@@ -68,7 +68,7 @@ d none bin 0755 bin bin
 f none bin/rsync 0755 bin bin
 d none doc 0755 bin bin
 d none doc/$NAME 0755 bin bin
-f none doc/$NAME/README 0644 bin bin
+f none doc/$NAME/README.md 0644 bin bin
 f none doc/$NAME/COPYING 0644 bin bin
 f none doc/$NAME/tech_report.pdf 0644 bin bin
 d none man 0755 bin bin

packaging/systemd/rsync.service

@@ -0,0 +1,30 @@
+[Unit]
+Description=fast remote file copy program daemon
+ConditionPathExists=/etc/rsyncd.conf
+After=network.target
+
+[Service]
+ExecStart=/usr/bin/rsync --daemon --no-detach
+RestartSec=1
+
+# Citing README.md:
+#
+#   [...] Using ssh is recommended for its security features.
+#
+#   Alternatively, rsync can run in `daemon' mode, listening on a socket.
+#   This is generally used for public file distribution, [...]
+#
+# So let's assume some extra security is more than welcome here. We do full
+# system protection (which makes it read-only) and hide users' homes and
+# devices. To override these defaults, it's best to do so in the drop-in
+# directory, often done via `systemctl edit rsync.service`. The file needs
+# just the bare minimum of the right [heading] and override values.
+# See systemd.unit(5) and search for "drop-in" for full details.
+
+ProtectSystem=full
+ProtectHome=on
+PrivateDevices=on
+NoNewPrivileges=on
+
+[Install]
+WantedBy=multi-user.target

packaging/systemd/rsync.socket

@@ -0,0 +1,10 @@
+[Unit]
+Description=socket for fast remote file copy program daemon
+Conflicts=rsync.service
+
+[Socket]
+ListenStream=873
+Accept=true
+
+[Install]
+WantedBy=sockets.target

packaging/systemd/rsync@.service

@@ -0,0 +1,28 @@
+[Unit]
+Description=fast remote file copy program daemon
+After=network.target
+
+[Service]
+ExecStart=-/usr/bin/rsync --daemon
+StandardInput=socket
+StandardOutput=inherit
+StandardError=journal
+
+# Citing README.md:
+#
+#   [...] Using ssh is recommended for its security features.
+#
+#   Alternatively, rsync can run in `daemon' mode, listening on a socket.
+#   This is generally used for public file distribution, [...]
+#
+# So let's assume some extra security is more than welcome here. We do full
+# system protection (which makes it read-only) and hide users' homes and
+# devices. To override these defaults, it's best to do so in the drop-in
+# directory, often done via `systemctl edit rsync@.service`. The file needs
+# just the bare minimum of the right [heading] and override values.
+# See systemd.unit(5) and search for "drop-in" for full details.
+
+ProtectSystem=full
+ProtectHome=on
+PrivateDevices=on
+NoNewPrivileges=on

packaging/var-checker

@@ -0,0 +1,87 @@
+#!/usr/bin/env -S python3 -B
+
+# This script checks the *.c files for extraneous "extern" variables,
+# for vars that are defined but not used, and for inconsistent array
+# sizes.  Run it from inside the main rsync directory.
+
+import re, argparse, glob
+
+VARS_RE = re.compile(r'^(?!(?:extern|enum)\s)([a-zA-Z]\S*\s+.*);', re.M)
+EXTERNS_RE = re.compile(r'^extern\s+(.*);', re.M)
+
+sizes = { }
+
+def main():
+    add_syscall_c = set('t_stub.c t_unsafe.c tls.c trimslash.c'.split())
+    add_util_c = set('t_stub.c t_unsafe.c'.split())
+
+    syscall_c = slurp_file('syscall.c', True)
+    util_c = slurp_file('util.c', True)
+
+    for fn in sorted(glob.glob('*.c')):
+        txt = slurp_file(fn)
+
+        var_list = parse_vars(fn, VARS_RE.findall(txt))
+        extern_list = parse_vars(fn, EXTERNS_RE.findall(txt))
+        if not var_list and not extern_list:
+            continue
+
+        if fn in add_syscall_c:
+            txt += syscall_c
+        if fn in add_util_c:
+            txt += util_c
+
+        txt = re.sub(r'INFO_GTE', 'info_levels ', txt)
+        txt = re.sub(r'DEBUG_GTE', 'debug_levels ', txt)
+        txt = re.sub(r'SIGACTION\(', 'sigact (', txt)
+
+        find = '|'.join([ re.escape(x) for x in var_list + extern_list ])
+        var_re = re.compile(r'(?<!\sstruct )\b(%s)(?!\w)' % find)
+
+        found = { x: 0 for x in var_list + extern_list }
+        for var in var_re.findall(txt):
+            found[var] += 1
+
+        for var in sorted(var_list + extern_list):
+            if found[var] == 1:
+                vtype = 'var' if var in var_list else 'extern'
+                print(fn, f'has extraneous {vtype}: "{var}"')
+
+
+def slurp_file(fn, drop_externs=False):
+    with open(fn, 'r', encoding='utf-8') as fh:
+        txt = fh.read()
+    if drop_externs:
+        txt = EXTERNS_RE.sub('', txt)
+    return txt
+
+
+def parse_vars(fn, lines):
+    ret = [ ]
+    for line in lines:
+        line = re.sub(r'\s*\{.*\}', '', line)
+        line = re.sub(r'\s*\(.*\)', '', line)
+        for item in re.split(r'\s*,\s*', line):
+            item = re.sub(r'\s*=.*', '', item)
+            m = re.search(r'(?P<var>\w+)(?P<sz>\[.*?\])?$', item)
+            if not m:
+                print(f"Bogus match? ({item})")
+                continue
+            if m['sz']:
+                if m['var'] in sizes:
+                    if sizes[m['var']] != m['sz']:
+                        var = m['var']
+                        print(fn, f'has inconsistent size for "{var}":', m['sz'], 'vs', sizes[var])
+                else:
+                    sizes[m['var']] = m['sz']
+            ret.append(m['var'])
+    return ret
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description='Check the *.c files for extraneous extern vars.', add_help=False)
+    parser.add_argument("--help", "-h", action="help", help="Output this help message and exit.")
+    args = parser.parse_args()
+    main()
+
+# vim: sw=4 et

packaging/year-tweak

@@ -0,0 +1,94 @@
+#!/usr/bin/env python3
+
+# This uses the output from "support/git-set-file-times --list" to discern
+# the last-modified year of each *.c & *.h file and updates the copyright
+# year if it isn't set right.
+
+import sys, os, re, argparse, subprocess
+from datetime import datetime
+
+MAINTAINER_NAME = 'Wayne Davison'
+MAINTAINER_SUF = ' ' + MAINTAINER_NAME + "\n"
+
+def main():
+    latest_year = '2000'
+
+    proc = subprocess.Popen('support/git-set-file-times --list'.split(), stdout=subprocess.PIPE, encoding='utf-8')
+    for line in proc.stdout:
+        m = re.match(r'^\S\s+(?P<year>\d\d\d\d)\S+\s+\S+\s+(?P<fn>.+)', line)
+        if not m:
+            print("Failed to parse line from git-set-file-times:", line)
+            sys.exit(1)
+        m = argparse.Namespace(**m.groupdict())
+        if m.year > latest_year:
+            latest_year = m.year
+        if m.fn.startswith('zlib/') or m.fn.startswith('popt/'):
+            continue
+        if re.search(r'\.(c|h|sh|test)$', m.fn):
+            maybe_edit_copyright_year(m.fn, m.year)
+    proc.communicate()
+
+    fn = 'latest-year.h'
+    with open(fn, 'r', encoding='utf-8') as fh:
+        old_txt = fh.read()
+
+    txt = f'#define LATEST_YEAR "{latest_year}"\n'
+    if txt != old_txt:
+        print(f"Updating {fn} with year {latest_year}")
+        with open(fn, 'w', encoding='utf-8') as fh:
+            fh.write(txt)
+
+
+def maybe_edit_copyright_year(fn, year):
+    opening_lines = [ ]
+    copyright_line = None
+
+    with open(fn, 'r', encoding='utf-8') as fh:
+        for lineno, line in enumerate(fh):
+            opening_lines.append(line)
+            if lineno > 3 and not re.search(r'\S', line):
+                break
+            m = re.match(r'^(?P<pre>.*Copyright\s+\S+\s+)(?P<year>\d\d\d\d(?:-\d\d\d\d)?(,\s+\d\d\d\d)*)(?P<suf>.+)', line)
+            if not m:
+                continue
+            copyright_line = argparse.Namespace(**m.groupdict())
+            copyright_line.lineno = len(opening_lines)
+            copyright_line.is_maintainer_line = MAINTAINER_NAME in copyright_line.suf
+            copyright_line.txt = line
+            if copyright_line.is_maintainer_line:
+                break
+
+        if not copyright_line:
+            return
+
+        if copyright_line.is_maintainer_line:
+            cyears = copyright_line.year.split('-')
+            if year == cyears[0]:
+                cyears = [ year ]
+            else:
+                cyears = [ cyears[0], year ]
+            txt = copyright_line.pre + '-'.join(cyears) + MAINTAINER_SUF
+            if txt == copyright_line.txt:
+                return
+            opening_lines[copyright_line.lineno - 1] = txt
+        else:
+            if fn.startswith('lib/') or fn.startswith('testsuite/'):
+                return
+            txt = copyright_line.pre + year + MAINTAINER_SUF
+            opening_lines[copyright_line.lineno - 1] += txt
+
+        remaining_txt = fh.read()
+
+    print(f"Updating {fn} with year {year}")
+
+    with open(fn, 'w', encoding='utf-8') as fh:
+        fh.write(''.join(opening_lines))
+        fh.write(remaining_txt)
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(description="Grab the year of last mod for our c & h files and make sure the Copyright comment is up-to-date.")
+    args = parser.parse_args()
+    main()
+
+# vim: sw=4 et

params.c

@@ -1,5 +1,5 @@
 /* This modules is based on the params.c module from Samba, written by Karl Auer
-   and much modifed by Christopher Hertel. */
+   and much modified by Christopher Hertel. */
 
 /*
  * This program is free software; you can redistribute it and/or modify
@@ -59,7 +59,7 @@
  *  beginning with either a semicolon (';') or a pound sign ('#').
  *
  *  All whitespace in section names and parameter names is compressed
- *  to single spaces.  Leading and trailing whitespace is stipped from
+ *  to single spaces.  Leading and trailing whitespace is stripped from
  *  both names and values.
  *
  *  Only the first equals sign in a parameter line is significant.
@@ -75,6 +75,7 @@
 
 #include "rsync.h"
 #include "ifuncs.h"
+#include "itypes.h"
 
 /* -------------------------------------------------------------------------- **
  * Constants...
@@ -93,6 +94,8 @@
 
 static char *bufr  = NULL;
 static int   bSize = 0;
+static BOOL  (*the_sfunc)(char *);
+static BOOL  (*the_pfunc)(char *, char *);
 
 /* -------------------------------------------------------------------------- **
  * Functions...
@@ -150,7 +153,7 @@ static int EatComment( FILE *InFile )
 
 static int Continuation( char *line, int pos )
   /* ------------------------------------------------------------------------ **
-   * Scan backards within a string to discover if the last non-whitespace
+   * Scan backwards within a string to discover if the last non-whitespace
    * character is a line-continuation character ('\\').
    *
    *  Input:  line  - A pointer to a buffer containing the string to be
@@ -211,7 +214,7 @@ static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) )
       bufr   = realloc_array( bufr, char, bSize );
       if( NULL == bufr )
         {
-        rprintf(FERROR, "%s Memory re-allocation failure.", func);
+        rprintf(FLOG, "%s Memory re-allocation failure.", func);
         return( False );
         }
       }
@@ -223,7 +226,7 @@ static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) )
         bufr[end] = '\0';
         if( 0 == end )                  /* Don't allow an empty name.       */
           {
-          rprintf(FERROR, "%s Empty section name in configuration file.\n", func );
+          rprintf(FLOG, "%s Empty section name in config file.\n", func );
           return( False );
           }
         if( !sfunc( bufr ) )            /* Got a valid name.  Deal with it. */
@@ -236,7 +239,7 @@ static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) )
         if( i < 0 )
           {
           bufr[end] = '\0';
-          rprintf(FERROR, "%s Badly formed line in configuration file: %s\n",
+          rprintf(FLOG, "%s Badly formed line in config file: %s\n",
                    func, bufr );
           return( False );
           }
@@ -261,7 +264,7 @@ static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) )
     }
 
   /* We arrive here if we've met the EOF before the closing bracket. */
-  rprintf(FERROR, "%s Unexpected EOF in the configuration file: %s\n", func, bufr );
+  rprintf(FLOG, "%s Unexpected EOF in the config file: %s\n", func, bufr );
   return( False );
   } /* Section */
 
@@ -305,7 +308,7 @@ static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
       bufr   = realloc_array( bufr, char, bSize );
       if( NULL == bufr )
         {
-        rprintf(FERROR, "%s Memory re-allocation failure.", func) ;
+        rprintf(FLOG, "%s Memory re-allocation failure.", func) ;
         return( False );
         }
       }
@@ -315,13 +318,12 @@ static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
       case '=':                 /* Equal sign marks end of param name. */
         if( 0 == end )              /* Don't allow an empty name.      */
           {
-          rprintf(FERROR, "%s Invalid parameter name in config. file.\n", func );
+          rprintf(FLOG, "%s Invalid parameter name in config file.\n", func );
           return( False );
           }
         bufr[end++] = '\0';         /* Mark end of string & advance.   */
-        i       = end;              /* New string starts here.         */
-        vstart  = end;              /* New string is parameter value.  */
-        bufr[i] = '\0';             /* New string is nul, for now.     */
+        i = vstart = end;           /* New string starts here.         */
+        c = EatWhitespace(InFile);
         break;
 
       case '\n':                /* Find continuation char, else error. */
@@ -329,7 +331,7 @@ static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
         if( i < 0 )
           {
           bufr[end] = '\0';
-          rprintf(FERROR, "%s Ignoring badly formed line in configuration file: %s\n",
+          rprintf(FLOG, "%s Ignoring badly formed line in config file: %s\n",
                    func, bufr );
           return( True );
           }
@@ -340,9 +342,22 @@ static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
       case '\0':                /* Shouldn't have EOF within param name. */
       case EOF:
         bufr[i] = '\0';
-        rprintf(FERROR, "%s Unexpected end-of-file at: %s\n", func, bufr );
+        rprintf(FLOG, "%s Unexpected end-of-file at: %s\n", func, bufr );
         return( True );
 
+      case ' ':
+      case '\t':
+        /* A directive divides at the first space or tab. */
+        if (*bufr == '&') {
+          bufr[end++] = '\0';
+          i = vstart = end;
+          c = EatWhitespace(InFile);
+          if (c == '=')
+            c = EatWhitespace(InFile);
+          break;
+        }
+        /* FALL THROUGH */
+
       default:
         if( isspace( c ) )     /* One ' ' per whitespace region.       */
           {
@@ -360,7 +375,6 @@ static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
     }
 
   /* Now parse the value. */
-  c = EatWhitespace( InFile );  /* Again, trim leading whitespace. */
   while( (EOF !=c) && (c > 0) )
     {
 
@@ -370,7 +384,7 @@ static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
       bufr   = realloc_array( bufr, char, bSize );
       if( NULL == bufr )
         {
-        rprintf(FERROR, "%s Memory re-allocation failure.", func) ;
+        rprintf(FLOG, "%s Memory re-allocation failure.", func) ;
         return( False );
         }
       }
@@ -406,7 +420,88 @@ static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
   return( pfunc( bufr, &bufr[vstart] ) );   /* Pass name & value to pfunc().  */
   } /* Parameter */
 
-static BOOL Parse( FILE *InFile,
+static int name_cmp(const void *n1, const void *n2)
+{
+    return strcmp(*(char * const *)n1, *(char * const *)n2);
+}
+
+static int include_config(char *include, int manage_globals)
+{
+    STRUCT_STAT sb;
+    char *match = manage_globals ? "*.conf" : "*.inc";
+    int ret;
+
+    if (do_stat(include, &sb) < 0) {
+	rsyserr(FLOG, errno, "unable to stat config file \"%s\"", include);
+	return 0;
+    }
+
+    if (S_ISREG(sb.st_mode)) {
+	if (manage_globals && the_sfunc)
+	    the_sfunc("]push");
+	ret = pm_process(include, the_sfunc, the_pfunc);
+	if (manage_globals && the_sfunc)
+	    the_sfunc("]pop");
+    } else if (S_ISDIR(sb.st_mode)) {
+	char buf[MAXPATHLEN], **bpp;
+	item_list conf_list;
+	struct dirent *di;
+	size_t j;
+	DIR *d;
+
+	if (!(d = opendir(include))) {
+	    rsyserr(FLOG, errno, "unable to open config dir \"%s\"", include);
+	    return 0;
+	}
+
+	memset(&conf_list, 0, sizeof conf_list);
+
+	while ((di = readdir(d)) != NULL) {
+	    char *dname = d_name(di);
+	    if (!wildmatch(match, dname))
+		continue;
+	    bpp = EXPAND_ITEM_LIST(&conf_list, char *, 32);
+	    pathjoin(buf, sizeof buf, include, dname);
+	    *bpp = strdup(buf);
+	}
+	closedir(d);
+
+	if (!(bpp = conf_list.items))
+	    return 1;
+
+	if (conf_list.count > 1)
+	    qsort(bpp, conf_list.count, sizeof (char *), name_cmp);
+
+	for (j = 0, ret = 1; j < conf_list.count; j++) {
+	    if (manage_globals && the_sfunc)
+		the_sfunc(j == 0 ? "]push" : "]reset");
+	    if ((ret = pm_process(bpp[j], the_sfunc, the_pfunc)) != 1)
+		break;
+	}
+
+	if (manage_globals && the_sfunc)
+	    the_sfunc("]pop");
+
+	for (j = 0; j < conf_list.count; j++)
+	    free(bpp[j]);
+	free(bpp);
+    } else
+	ret = 0;
+
+    return ret;
+}
+
+static int parse_directives(char *name, char *val)
+{
+    if (strcasecmp(name, "&include") == 0)
+        return include_config(val, 1);
+    if (strcasecmp(name, "&merge") == 0)
+        return include_config(val, 0);
+    rprintf(FLOG, "Unknown directive: %s.\n", name);
+    return 0;
+}
+
+static int Parse( FILE *InFile,
                    BOOL (*sfunc)(char *),
                    BOOL (*pfunc)(char *, char *) )
   /* ------------------------------------------------------------------------ **
@@ -418,7 +513,8 @@ static BOOL Parse( FILE *InFile,
    *          pfunc   - Function to be called when a parameter is scanned.
    *                    See Parameter().
    *
-   *  Output: True if the file was successfully scanned, else False.
+   *  Output: 1 if the file was successfully scanned, 2 if the file was
+   *  scanned until a section header with no section function, else 0.
    *
    *  Notes:  The input can be viewed in terms of 'lines'.  There are four
    *          types of lines:
@@ -427,7 +523,7 @@ static BOOL Parse( FILE *InFile,
    *                         The remainder of the line is ignored.
    *            Section    - First non-whitespace character is a '['.
    *            Parameter  - The default case.
-   * 
+   *
    * ------------------------------------------------------------------------ **
    */
   {
@@ -448,29 +544,39 @@ static BOOL Parse( FILE *InFile,
         break;
 
       case '[':                         /* Section Header. */
-	      if (!sfunc) return True;
-	      if( !Section( InFile, sfunc ) )
-		      return( False );
-	      c = EatWhitespace( InFile );
-	      break;
+        if (!sfunc)
+          return 2;
+        if( !Section( InFile, sfunc ) )
+          return 0;
+        c = EatWhitespace( InFile );
+        break;
 
       case '\\':                        /* Bogus backslash. */
         c = EatWhitespace( InFile );
         break;
 
+      case '&':                         /* Handle directives */
+        the_sfunc = sfunc;
+        the_pfunc = pfunc;
+        c = Parameter( InFile, parse_directives, c );
+        if (c != 1)
+          return c;
+        c = EatWhitespace( InFile );
+        break;
+
       default:                          /* Parameter line. */
         if( !Parameter( InFile, pfunc, c ) )
-          return( False );
+          return 0;
         c = EatWhitespace( InFile );
         break;
       }
     }
-  return( True );
+  return 1;
   } /* Parse */
 
 static FILE *OpenConfFile( char *FileName )
   /* ------------------------------------------------------------------------ **
-   * Open a configuration file.
+   * Open a config file.
    *
    *  Input:  FileName  - The pathname of the config file to be opened.
    *
@@ -485,21 +591,21 @@ static FILE *OpenConfFile( char *FileName )
 
   if( NULL == FileName || 0 == *FileName )
     {
-    rprintf(FERROR,"%s No configuration filename specified.\n", func);
+    rprintf(FLOG, "%s No config filename specified.\n", func);
     return( NULL );
     }
 
   OpenedFile = fopen( FileName, "r" );
   if( NULL == OpenedFile )
     {
-    rsyserr(FERROR, errno, "unable to open configuration file \"%s\"",
+    rsyserr(FLOG, errno, "unable to open config file \"%s\"",
 	    FileName);
     }
 
   return( OpenedFile );
   } /* OpenConfFile */
 
-BOOL pm_process( char *FileName,
+int pm_process( char *FileName,
                  BOOL (*sfunc)(char *),
                  BOOL (*pfunc)(char *, char *) )
   /* ------------------------------------------------------------------------ **
@@ -511,7 +617,8 @@ BOOL pm_process( char *FileName,
    *          pfunc     - A pointer to a function that will be called when
    *                      a parameter name and value are discovered.
    *
-   *  Output: TRUE if the file was successfully parsed, else FALSE.
+   *  Output: 1 if the file was successfully parsed, 2 if parsing ended at a
+   *  section header w/o a section function, else 0.
    *
    * ------------------------------------------------------------------------ **
    */
@@ -534,7 +641,7 @@ BOOL pm_process( char *FileName,
     bufr = new_array( char, bSize );
     if( NULL == bufr )
       {
-      rprintf(FERROR,"%s memory allocation failure.\n", func);
+      rprintf(FLOG, "%s memory allocation failure.\n", func);
       fclose(InFile);
       return( False );
       }
@@ -548,11 +655,11 @@ BOOL pm_process( char *FileName,
 
   if( !result )                               /* Generic failure. */
     {
-    rprintf(FERROR,"%s Failed.  Error returned from params.c:parse().\n", func);
-    return( False );
+    rprintf(FLOG, "%s Failed.  Error returned from params.c:parse().\n", func);
+    return 0;
     }
 
-  return( True );                             /* Generic success. */
+  return result;
   } /* pm_process */
 
 /* -------------------------------------------------------------------------- */