rsync.git
26 years agoprototype update
Andrew Tridgell [Mon, 23 Mar 1998 04:47:40 +0000 (04:47 +0000)]
prototype update

26 years ago- keep a list of pids and send them a SIGUSR1 for cleanup rather than
Andrew Tridgell [Mon, 23 Mar 1998 04:44:44 +0000 (04:44 +0000)]
- keep a list of pids and send them a SIGUSR1 for cleanup rather than
using setpgrp()

- adapt the block size for really large files to reduce the checksum
  size and memory overheads

26 years agofor systems with a broken select use u_sleep() to ensure the write
Andrew Tridgell [Mon, 23 Mar 1998 04:14:21 +0000 (04:14 +0000)]
for systems with a broken select use u_sleep() to ensure the write
process doesn't chew too much cpu time.

26 years agoa large change to make rsync much more memory efficient. This is done
Andrew Tridgell [Mon, 23 Mar 1998 03:45:02 +0000 (03:45 +0000)]
a large change to make rsync much more memory efficient. This is done
in 3 ways:

1) the realloc is done on a list of pointers, not a list of structures
2) only the basename of the file is now kept in the file struct an a
   util function f_name() is now used to access the full name when
   required.
3) pointers to directory names are re-used

hopefully I haven't broken anything. This will need lots of testing.

26 years agocompatibility functions now split into their own file
Andrew Tridgell [Tue, 24 Feb 1998 11:54:10 +0000 (11:54 +0000)]
compatibility functions now split into their own file

26 years ago*** empty log message ***
Andrew Tridgell [Tue, 27 Jan 1998 17:51:37 +0000 (17:51 +0000)]
*** empty log message ***

26 years ago*** empty log message *** v1.6.9
Andrew Tridgell [Tue, 13 Jan 1998 19:25:59 +0000 (19:25 +0000)]
*** empty log message ***

26 years ago*** empty log message ***
Andrew Tridgell [Tue, 13 Jan 1998 19:19:48 +0000 (19:19 +0000)]
*** empty log message ***

26 years ago*** empty log message ***
Andrew Tridgell [Tue, 13 Jan 1998 18:35:10 +0000 (18:35 +0000)]
*** empty log message ***

26 years ago*** empty log message ***
Andrew Tridgell [Tue, 13 Jan 1998 17:16:14 +0000 (17:16 +0000)]
*** empty log message ***

26 years ago*** empty log message ***
Andrew Tridgell [Tue, 13 Jan 1998 15:57:26 +0000 (15:57 +0000)]
*** empty log message ***

26 years ago*** empty log message ***
Andrew Tridgell [Tue, 6 Jan 1998 16:53:52 +0000 (16:53 +0000)]
*** empty log message ***

26 years agocosmetic changes only
Andrew Tridgell [Mon, 5 Jan 1998 23:30:44 +0000 (23:30 +0000)]
cosmetic changes only
- minor correction to --update docs
- don't print "building file list" stuff when building a local file
list for use in the --delete code.
- remove some spaces at the end of lines in Makefile.in

26 years ago*** empty log message ***
Andrew Tridgell [Thu, 1 Jan 1998 15:53:05 +0000 (15:53 +0000)]
*** empty log message ***

26 years agoremoved the read_write() call, it's not used anywhere
Andrew Tridgell [Wed, 31 Dec 1997 16:48:11 +0000 (16:48 +0000)]
removed the read_write() call, it's not used anywhere

26 years agosimplified the logic of whether to skip a file or not. The -c
Andrew Tridgell [Wed, 31 Dec 1997 16:39:34 +0000 (16:39 +0000)]
simplified the logic of whether to skip a file or not. The -c
(always_checksum) option is probably now more useful.

26 years agobuffer overflow patches from mhpower@mit.edu (Matt Power)
Andrew Tridgell [Tue, 30 Dec 1997 17:36:58 +0000 (17:36 +0000)]
buffer overflow patches from mhpower@mit.edu (Matt Power)

26 years agofixed the --suffix option. It wasn't being propogated to the remote
Andrew Tridgell [Tue, 30 Dec 1997 10:54:47 +0000 (10:54 +0000)]
fixed the --suffix option. It wasn't being propogated to the remote
end in the server_options() function.

26 years agopreparing for release of 1.6.8 v1.6.8
rsync-bugs [Sun, 28 Dec 1997 22:28:21 +0000 (22:28 +0000)]
preparing for release of 1.6.8

26 years ago- added a debug message
Andrew Tridgell [Sun, 28 Dec 1997 22:26:51 +0000 (22:26 +0000)]
- added a debug message
- rebuilt prototypes

26 years ago- fixed spelling errors in man page
Andrew Tridgell [Sun, 28 Dec 1997 22:13:40 +0000 (22:13 +0000)]
- fixed spelling errors in man page
- fixed bug in hard link handling that could lead to spurious hard
links.
- fixed bug in the string handling in flist.c

26 years agofollowing a report of problems with Linux/alpha I've changed zlib.c to
Andrew Tridgell [Thu, 18 Dec 1997 11:18:32 +0000 (11:18 +0000)]
following a report of problems with Linux/alpha I've changed zlib.c to
use uint32 instead of "long" in several places. Apparently this fixes
things on the alpha. The strange thing is that my own tests on a
OSF/alpha box and a 64 bit IRIX box showed no problems. I wonder what
was actually going wrong? I'll email the zlib maintainers and let them
know.

26 years agoadded a new mirror site to the README
Andrew Tridgell [Thu, 18 Dec 1997 11:13:27 +0000 (11:13 +0000)]
added a new mirror site to the README

26 years agopreparing for release of 1.6.7 v1.6.7
rsync-bugs [Wed, 17 Dec 1997 11:19:01 +0000 (11:19 +0000)]
preparing for release of 1.6.7

26 years ago*** empty log message ***
Andrew Tridgell [Wed, 17 Dec 1997 11:07:17 +0000 (11:07 +0000)]
*** empty log message ***

26 years agohandle things more grecefully when one machine supports hard links and
Andrew Tridgell [Tue, 16 Dec 1997 23:09:22 +0000 (23:09 +0000)]
handle things more grecefully when one machine supports hard links and
the other doesn't or one machine supports soft links and the other
doesn't.

26 years ago- check for setlinebuf() in autoconf. Apparently HPUX doesn't have it
Andrew Tridgell [Tue, 16 Dec 1997 22:39:55 +0000 (22:39 +0000)]
- check for setlinebuf() in autoconf. Apparently HPUX doesn't have it
- use @exec_prefix@ and @prefix@ in more useful ways in Makefile.in

26 years agofixed a bug in the handling of the new --relative option. The file was
Andrew Tridgell [Tue, 16 Dec 1997 20:29:35 +0000 (20:29 +0000)]
fixed a bug in the handling of the new --relative option. The file was
being opened twice but closed once. The process eventually died with
an out of file descriptors error.

26 years agopreparing for release of 1.6.6 v1.6.6
rsync-bugs [Tue, 16 Dec 1997 18:20:10 +0000 (18:20 +0000)]
preparing for release of 1.6.6

26 years agoChecker showed that zlib was using a element of its internal state
Andrew Tridgell [Tue, 16 Dec 1997 18:18:02 +0000 (18:18 +0000)]
Checker showed that zlib was using a element of its internal state
structure without initialising it. Although it looks harmless I've
added a bzero() to make absolutely sure that the code behaves
consistently across platforms.

26 years agofixed a nasty bug in the handling of the --delete option when there
Andrew Tridgell [Tue, 16 Dec 1997 17:59:49 +0000 (17:59 +0000)]
fixed a nasty bug in the handling of the --delete option when there
are duplicate file names in the list of files to be transferred
(eg. the user specifies the same file twice).

26 years ago#if 0 the write exception code for the moment. I need to work out why
Andrew Tridgell [Tue, 16 Dec 1997 09:25:17 +0000 (09:25 +0000)]
#if 0 the write exception code for the moment. I need to work out why
it gets a successful write select on a fd followed by a EAGAIN
write yet the fd is still OK.

26 years agopreparing for release of 1.6.5 v1.6.5
rsync-bugs [Tue, 16 Dec 1997 09:14:47 +0000 (09:14 +0000)]
preparing for release of 1.6.5

26 years agoadded .cvsignore to the .cvsignore file to prevent it from being
Andrew Tridgell [Tue, 16 Dec 1997 09:14:18 +0000 (09:14 +0000)]
added .cvsignore to the .cvsignore file to prevent it from being
distributed when I run the release script

26 years ago- fixed the "write exception" error. I was resetting got_select at the
Andrew Tridgell [Tue, 16 Dec 1997 08:48:36 +0000 (08:48 +0000)]
- fixed the "write exception" error. I was resetting got_select at the
wrong point
- fixed a seg fault error in flist.c
- only print hlink debug messages when using -v

26 years agopreparing for release of 1.6.4 v1.6.4
rsync-bugs [Mon, 15 Dec 1997 23:08:30 +0000 (23:08 +0000)]
preparing for release of 1.6.4

26 years ago- fixed an off by 1 bug in the hard link support
Andrew Tridgell [Mon, 15 Dec 1997 22:54:23 +0000 (22:54 +0000)]
- fixed an off by 1 bug in the hard link support
- added a note about anonymous cvs access to the source code in the
README

26 years agoprocess directory permissions and times ater hard links becuase the
Andrew Tridgell [Mon, 15 Dec 1997 22:23:30 +0000 (22:23 +0000)]
process directory permissions and times ater hard links becuase the
hard link processing can modify the directory times

26 years agocasting cleanups (rsync now compiles cleanly under IRIX 6.4)
rsync-bugs [Mon, 15 Dec 1997 21:41:33 +0000 (21:41 +0000)]
casting cleanups (rsync now compiles cleanly under IRIX 6.4)

26 years agosome people are now using rsync as a public server, using various
Andrew Tridgell [Mon, 15 Dec 1997 21:35:37 +0000 (21:35 +0000)]
some people are now using rsync as a public server, using various
patches or wrappers. One problem with this is that rsync was not
written with this in mind and wasn't very careful about possible stack
overflows etc which could lead to security breaches. This wasn't a
problem when run in the traditional way as any user that can run rsync
can login anyway and cause much more damage that way.
This patch attempts to close possible stack overflow problems. I've
checked for all strcpy(), strcat(), sprintf() and memcpy()
overflows. I would appreciate it if someone else with a devious mind
could also go through the rsync source code and see if there are any
other stack overflows possible. Let me know if you do.

26 years agoseveral changes:
Andrew Tridgell [Mon, 15 Dec 1997 21:04:04 +0000 (21:04 +0000)]
several changes:
- by popular demand I have changed the behaviour of the --delete
option. It should now work as "expected" for even those people silly
enough not to read the man page. rsync will now only look for
candidate files/directories to delete in directories that are
explicitly transferred from the sender
- updated the README a bit
- try to fail a bit more gracefully when rsync runs out of disk
space. I don't think this issues is fully resolved yet

26 years agominor cleanups (unused variables etc)
rsync-bugs [Mon, 15 Dec 1997 19:11:24 +0000 (19:11 +0000)]
minor cleanups (unused variables etc)

26 years agoonly include sys/mod.h if S_ISLNK is not already defined. This allows
rsync-bugs [Mon, 15 Dec 1997 18:36:21 +0000 (18:36 +0000)]
only include sys/mod.h if S_ISLNK is not already defined. This allows
things to work on aix without causing broken sgi headers files to spew
lots of complaints

26 years agobugfix from Kenji Miyake <kenjim@fix.co.jp>
Andrew Tridgell [Mon, 15 Dec 1997 18:01:15 +0000 (18:01 +0000)]
bugfix from Kenji Miyake <kenjim@fix.co.jp>
this fixes a off by 1 error that could cause rsync to crash when used
with --delete

26 years agopatch from Jim Meyering <meyering@eng.ascend.com>
Andrew Tridgell [Mon, 15 Dec 1997 17:57:28 +0000 (17:57 +0000)]
patch from Jim Meyering <meyering@eng.ascend.com>
- fixed typos in man page
- use @CFLAGS@ in Makefile.in, allowing override on command line

26 years agosetup line buffering for debug messages
Andrew Tridgell [Mon, 15 Dec 1997 17:46:58 +0000 (17:46 +0000)]
setup line buffering for debug messages

26 years agobe a bit more friendly on systems that behave badly (consume lots of
Andrew Tridgell [Mon, 15 Dec 1997 17:29:51 +0000 (17:29 +0000)]
be a bit more friendly on systems that behave badly (consume lots of
ram) when you realloc() up by a small amount at a time

26 years agoinclude sys/socket.h if possible (this should make rsync compile
Andrew Tridgell [Mon, 15 Dec 1997 16:27:04 +0000 (16:27 +0000)]
include sys/socket.h if possible (this should make rsync compile
cleanly on SCO)

26 years agoadded a --relative (== -R) option. This is what Anthony Thyssen
Andrew Tridgell [Mon, 15 Dec 1997 14:43:27 +0000 (14:43 +0000)]
added a --relative (== -R) option. This is what Anthony Thyssen
suggested on the list recently. See the man page entry for details but
basically it changes the behaviour so that paths are not stripped,
thus allowing you to specify a single rsync command to sync lots of
directories/files while preserving the full path name of each file.
also fixed a bug in the handling of umasks when both the source and
destination machines are local. We need to reset the umask before the
exec to ensure that the child gets a correct umask.

26 years agoadded some more debug info to the "EOF in map_ptr" error message to
Andrew Tridgell [Mon, 15 Dec 1997 13:38:06 +0000 (13:38 +0000)]
added some more debug info to the "EOF in map_ptr" error message to
try to make it easier to track down what is going wrong.
Jarom, can you let me know what output this gives?
PS: the cvs commit messages for rsync are now sent to the rsync
mailing list. This will allow everyone to see what changes are being
made.

26 years agodamn, I had implicitly assumed signed characters when writing the
Andrew Tridgell [Mon, 15 Dec 1997 12:37:40 +0000 (12:37 +0000)]
damn, I had implicitly assumed signed characters when writing the
rolling checksum code. The result is that rsync worked much more
slowly when going between two machines where one of the machines uses
signed and the other unsigned chars. The rolling checksum rarely
matched so effectively a copy was done in many cases. The data always
came through correctly so no file corruption occurred but it's pretty
pointless using rsync if it doesn't speed things up!
I've now made the sign of the chars explicit

26 years agoI think I've (finally) fixed the problem with rsync periodically
Andrew Tridgell [Mon, 15 Dec 1997 12:33:15 +0000 (12:33 +0000)]
I think I've (finally) fixed the problem with rsync periodically
freezing when used with ssh. The problem is really the ssh use of
blocking calls. rsync has had a fix for this for a while which relies
on using non-blocking calls itself and then reading any data from the
incoming fd when trying to write, thus ensuring that the incoming fd
doesn't get full.
The problem was the the incoming fd wasn't added to the select
statement, which meant that new data arriving on the fd only got read
when the select timed out, which happens every 10 seconds. Thus things
could slow to a crawl!
The incoming fd is now in the select call, and this seems to fix the
problem.

26 years agodon't use WRAP as that maybe be defined in standard includes
Andrew Tridgell [Thu, 30 Oct 1997 08:13:32 +0000 (08:13 +0000)]
don't use WRAP as that maybe be defined in standard includes

26 years agofix version number
Andrew Tridgell [Thu, 30 Oct 1997 08:07:38 +0000 (08:07 +0000)]
fix version number

26 years agosend "." filenames so the other end knows that a directory was sent
Andrew Tridgell [Thu, 30 Oct 1997 08:07:28 +0000 (08:07 +0000)]
send "." filenames so the other end knows that a directory was sent

26 years agofix spelling error
Andrew Tridgell [Thu, 30 Oct 1997 08:06:46 +0000 (08:06 +0000)]
fix spelling error

26 years agominor updates
Andrew Tridgell [Thu, 10 Jul 1997 01:12:36 +0000 (01:12 +0000)]
minor updates

27 years agominor bugfixes
Paul Mackerras [Fri, 11 Apr 1997 06:18:02 +0000 (06:18 +0000)]
minor bugfixes

27 years agoFixed bug in receiving 32kB compressible files
Paul Mackerras [Wed, 6 Nov 1996 04:49:53 +0000 (04:49 +0000)]
Fixed bug in receiving 32kB compressible files

27 years ago*** empty log message ***
Andrew Tridgell [Wed, 2 Oct 1996 06:33:58 +0000 (06:33 +0000)]
*** empty log message ***

27 years agoadded info on european mirror of rsync
Andrew Tridgell [Wed, 2 Oct 1996 06:33:49 +0000 (06:33 +0000)]
added info on european mirror of rsync

27 years agoFix compression bug where incompressible files weren't transmitted correctly.
Paul Mackerras [Wed, 2 Oct 1996 04:20:22 +0000 (04:20 +0000)]
Fix compression bug where incompressible files weren't transmitted correctly.

27 years agomove check for remsh after config.h
Andrew Tridgell [Tue, 1 Oct 1996 12:04:47 +0000 (12:04 +0000)]
move check for remsh after config.h

27 years agochange BAD to ZBAD to avoid conflict with some systems
Andrew Tridgell [Tue, 1 Oct 1996 06:44:00 +0000 (06:44 +0000)]
change BAD to ZBAD to avoid conflict with some systems

27 years ago- check for sys/select.h
Andrew Tridgell [Tue, 1 Oct 1996 06:40:36 +0000 (06:40 +0000)]
- check for sys/select.h

- check for and use lchown if possible

- fix problem with root directory on source and destination

27 years ago- added info on the mailing list
Andrew Tridgell [Tue, 1 Oct 1996 04:31:34 +0000 (04:31 +0000)]
- added info on the mailing list

27 years agoupdate to version 1.6
Andrew Tridgell [Tue, 1 Oct 1996 04:13:03 +0000 (04:13 +0000)]
update to version 1.6

27 years ago- detect presence of remsh and use it instead of rsh
Andrew Tridgell [Tue, 1 Oct 1996 04:12:30 +0000 (04:12 +0000)]
- detect presence of remsh and use it instead of rsh

- handle directory ownership and permissions much better.

- fix bug where links caused the permissions of files to be
set incorrectly

- override the default umask in setting file permissions

- better handling -o and -D being passed to non-root users

- handle rsync to a destination of /

- fix the handling of mismatched file types at either end of the
link. For example, if the destination is a link and the source is not.

27 years agoThis should fix the bug where file transfer with compression failed with
Paul Mackerras [Mon, 15 Jul 1996 05:38:20 +0000 (05:38 +0000)]
This should fix the bug where file transfer with compression failed with
the message 'inflate returned 0 (0 bytes)'.

27 years agoadd warning if corruption detected
Andrew Tridgell [Fri, 12 Jul 1996 14:37:25 +0000 (14:37 +0000)]
add warning if corruption detected

27 years agofix serious corruption bug
Andrew Tridgell [Fri, 12 Jul 1996 14:36:03 +0000 (14:36 +0000)]
fix serious corruption bug

27 years ago*** empty log message ***
Andrew Tridgell [Fri, 5 Jul 1996 04:49:32 +0000 (04:49 +0000)]
*** empty log message ***

27 years ago*** empty log message ***
Andrew Tridgell [Thu, 4 Jul 1996 13:56:53 +0000 (13:56 +0000)]
*** empty log message ***

27 years agoadded CHAR_OFFSET
Andrew Tridgell [Thu, 4 Jul 1996 13:08:07 +0000 (13:08 +0000)]
added CHAR_OFFSET

27 years ago*** empty log message ***
Andrew Tridgell [Thu, 4 Jul 1996 10:16:56 +0000 (10:16 +0000)]
*** empty log message ***

27 years ago*** empty log message ***
Andrew Tridgell [Wed, 3 Jul 1996 08:18:11 +0000 (08:18 +0000)]
*** empty log message ***

27 years ago*** empty log message ***
Andrew Tridgell [Wed, 3 Jul 1996 08:12:32 +0000 (08:12 +0000)]
*** empty log message ***

27 years ago- fixed file time bug
Andrew Tridgell [Wed, 3 Jul 1996 06:24:34 +0000 (06:24 +0000)]
- fixed file time bug
- doc updates

27 years agofix buglet in handling multiple files
Paul Mackerras [Wed, 3 Jul 1996 05:28:55 +0000 (05:28 +0000)]
fix buglet in handling multiple files

27 years ago*** empty log message ***
Andrew Tridgell [Wed, 3 Jul 1996 05:22:09 +0000 (05:22 +0000)]
*** empty log message ***

27 years agoAdded in-stream deflate compression for file reconstruction instructions.
Paul Mackerras [Wed, 3 Jul 1996 04:05:46 +0000 (04:05 +0000)]
Added in-stream deflate compression for file reconstruction instructions.

27 years agocheck for null unmap
Andrew Tridgell [Wed, 3 Jul 1996 02:21:47 +0000 (02:21 +0000)]
check for null unmap

27 years agosave some more bytes by making the checksum smaller
Andrew Tridgell [Tue, 2 Jul 1996 06:02:36 +0000 (06:02 +0000)]
save some more bytes by making the checksum smaller

27 years agoadded seed to file sum
Andrew Tridgell [Tue, 2 Jul 1996 03:22:08 +0000 (03:22 +0000)]
added seed to file sum

27 years agoadded resend logic
Andrew Tridgell [Tue, 2 Jul 1996 03:19:58 +0000 (03:19 +0000)]
added resend logic
added checksum on all files

27 years ago*** empty log message ***
Andrew Tridgell [Mon, 1 Jul 1996 23:27:19 +0000 (23:27 +0000)]
*** empty log message ***

27 years agoadded hooks for compression in token.c
Andrew Tridgell [Mon, 1 Jul 1996 05:55:05 +0000 (05:55 +0000)]
added hooks for compression in token.c

27 years agoget byte order of checksum seed right
Andrew Tridgell [Sun, 30 Jun 1996 05:02:45 +0000 (05:02 +0000)]
get byte order of checksum seed right

27 years agonew header tests for sunos
Andrew Tridgell [Sun, 30 Jun 1996 04:42:04 +0000 (04:42 +0000)]
new header tests for sunos

27 years ago*** empty log message ***
Andrew Tridgell [Sun, 30 Jun 1996 04:30:43 +0000 (04:30 +0000)]
*** empty log message ***

27 years agoadded checksum seed
Andrew Tridgell [Sun, 30 Jun 1996 04:29:46 +0000 (04:29 +0000)]
added checksum seed

27 years ago- hard links
Andrew Tridgell [Sun, 30 Jun 1996 03:57:22 +0000 (03:57 +0000)]
- hard links
- better sparse handling
- FERROR and FINFO

27 years ago*** empty log message ***
Andrew Tridgell [Fri, 28 Jun 1996 15:24:39 +0000 (15:24 +0000)]
*** empty log message ***

27 years ago*** empty log message ***
Andrew Tridgell [Fri, 28 Jun 1996 14:43:49 +0000 (14:43 +0000)]
*** empty log message ***

27 years ago*** empty log message ***
Andrew Tridgell [Fri, 28 Jun 1996 14:27:04 +0000 (14:27 +0000)]
*** empty log message ***

27 years agofixed sighup handling
Andrew Tridgell [Fri, 28 Jun 1996 14:26:15 +0000 (14:26 +0000)]
fixed sighup handling

27 years agonew exit/cleanup code
Andrew Tridgell [Fri, 28 Jun 1996 13:55:41 +0000 (13:55 +0000)]
new exit/cleanup code

27 years ago- handle no mmap for munmap
Andrew Tridgell [Fri, 28 Jun 1996 08:01:55 +0000 (08:01 +0000)]
- handle no mmap for munmap
- handle sparse files
- add MAX_PROTOCOL_VERSION

27 years agostandard input on server may be nonblocking
Paul Mackerras [Fri, 28 Jun 1996 07:06:34 +0000 (07:06 +0000)]
standard input on server may be nonblocking

27 years agoadded csum-length option
Andrew Tridgell [Thu, 27 Jun 1996 05:16:15 +0000 (05:16 +0000)]
added csum-length option