[ssh.c sshconnect.c sshconnect.h]
make ssh(1)'s ConnectTimeout option apply to both the TCP connection and
SSH banner exchange (previously it just covered the TCP connection).
This allows callers of ssh(1) to better detect and deal with stuck servers
that accept a TCP connection but don't progress the protocol, and also
makes ConnectTimeout useful for connections via a ProxyCommand;
feedback and "looks ok" markus@
[clientloop.c monitor.c monitor_fdpass.c monitor_fdpass.h]
[monitor_wrap.c ssh.c]
make file descriptor passing code return an error rather than call fatal()
when it encounters problems, and use this to make session multiplexing
masters survive slaves failing to pass all stdio FDs; ok markus@
[clientloop.c clientloop.h ssh.c]
bz#1232: ensure that any specified LocalCommand is executed after the
tunnel device is opened. Also, make failures to open a tunnel device
fatal when ExitOnForwardFailure is active.
Reported by h.goebel AT goebel-consult.de; ok dtucker markus reyk deraadt
openbsd-compat/bsd-poll.{c,h} openbsd-compat/openbsd-compat.h]
Add an implementation of poll() built on top of select(2). Code from
OpenNTPD with changes suggested by djm. ok djm@
[atomicio.c]
if the fd passed to atomicio/atomiciov() is non blocking, then poll() to
avoid a spin if it is not yet ready for reading/writing; ok dtucker@
[ssh.c]
when waiting for the multiplex exit status, read until the master end
writes an entire int of data *and* closes the client_fd; fixes mux
regression spotted by dtucker, ok dtucker@
USE_BUILTIN_RIJNDAEL compat goop to openssl-compat.h so it can be
shared with umac.c. Allows building with OpenSSL 0.9.5 again including
umac support. With tim@ djm@, ok djm.
[ssh.c ssh.1]
Add "-K" flag for ssh to set GSSAPIAuthentication=yes and
GSSAPIDelegateCredentials=yes. This is symmetric with -k (disable GSSAPI)
and is useful for hosts with /home on Kerberised NFS; bz #1312
patch from Markus.Kuhn AT cl.cam.ac.uk; ok dtucker@ markus@
[ssh.c]
fix slave exit value when a control master goes away without passing the
full exit status by ensuring that the slave reads a full int. bz#1261
reported by frekko AT gmail.com; ok markus@ dtucker@
[scp.c]
make scp try to skip FIFOs rather than blocking when nothing is listening.
depends on the platform supporting sane O_NONBLOCK semantics for open
on FIFOs (apparently POSIX does not mandate this), which OpenBSD does.
bz #856; report by cjwatson AT debian.org; ok markus@
[ssh-add.1]
better document ssh-add's -d option (delete identies from agent), bz#1224
new text based on some provided by andrewmc-debian AT celt.dias.ie;
ok dtucker@
[ssh_config]
Add a "MACs" line after "Ciphers" with the default MAC algorithms,
to ease people who want to tweak both (eg. for performance reasons).
ok deraadt@ djm@ dtucker@
[kex.h mac.c mac.h monitor_wrap.c myproposal.h packet.c ssh.1]
[ssh_config.5 sshd.8 sshd_config.5]
Add a new MAC algorithm for data integrity, UMAC-64 (not default yet,
must specify umac-64@openssh.com). Provides about 20% end-to-end speedup
compared to hmac-md5. Represents a different approach to message
authentication to that of HMAC that may be beneficial if HMAC based on
one of its underlying hash algorithms is found to be vulnerable to a
new attack. http://www.ietf.org/rfc/rfc4418.txt
in conjunction with and OK djm@
[kex.c monitor_wrap.c packet.c mac.h kex.h mac.c]
Preserve MAC ctx between packets, saving 2xhash calls per-packet.
Yields around a 12-16% end-to-end speedup for arcfour256/hmac-md5
patch from markus@ tested dtucker@ and myself, ok markus@ and me (I'm
committing at his request)
[scp.1 ssh_config.5 sftp-server.8 ssh-agent.1 sshd_config.5 sftp.1
ssh-keygen.1 ssh-keyscan.1 ssh-add.1 sshd.8 ssh.1 ssh-keysign.8]
convert to new .Dd format;
(We will need to teach mdoc2man.awk to understand this too.)
[monitor.c]
pass received SIGINT from monitor to postauth child so it can clean
up properly. bz#1196, patch from senthilkumar_sen AT hotpop.com;
ok markus@
[sshconnect2.c]
fall back to gethostname() when the outgoing connection is not
on a socket, such as is the case when ProxyCommand is used.
Gives hostbased auth an opportunity to work; bz#616, report
and feedback stuart AT kaloram.com; ok markus@
[sftp-server.c]
bz#1286 stop reading and processing commands when input or output buffer
is nearly full, otherwise sftp-server would happily try to grow the
input/output buffers past the maximum supported by the buffer API and
promptly fatal()
based on patch from Thue Janus Kristensen; feedback & ok dtucker@
[sftp-server.c]
cast "%llu" format spec to (unsigned long long); do not assume a
u_int64_t arg is the same as 'unsigned long long'.
from Dmitry V. Levin <ldv@altlinux.org>
ok markus@ 'Yes, that looks correct' millert@
- (tim) [auth.c configure.ac defines.h session.c openbsd-compat/port-uw.c
openbsd-compat/port-uw.h openbsd-compat/xcrypt.c] Rework libiaf test/defines
to account for IRIX having libiaf but not set_id(). Patch with & ok dtucker@
[ssh-agent.c]
Remove the signal handler that checks if the agent's parent process
has gone away, instead check when the select loop returns. Record when
the next key will expire when scanning for expired keys. Set the select
timeout to whichever of these two things happens next. With djm@, with &
ok deraadt@ markus@
[sshd_config]
Disable the legacy SSH protocol 1 for new installations via
a configuration override. In the future, we will change the
server's default itself so users who need the legacy protocol
will need to turn it on explicitly
[servconf.c sshd.c]
Move C/R -> kbdint special case to after the defaults have been
loaded, which makes ChallengeResponse default to yes again. This
was broken by the Match changes and not fixed properly subsequently.
Found by okan at demirmen.com, ok djm@ "please do it" deraadt@
bug in OpenSSL 0.9.8e that prevents aes256-ctr, aes192-ctr and arcfour256
ciphers from working correctly (disconnects with "Bad packet length"
errors) as found by Ben Harris. ok djm@
[auth2.c sshd_config.5 servconf.c]
Remove ChallengeResponseAuthentication support inside a Match
block as its interaction with KbdInteractive makes it difficult to
support. Also, relocate the CR/kbdint option special-case code into
servconf. "please commit" djm@, ok markus@ for the relocation.
[ssh-agent.c]
Remove expired keys periodically so they don't remain in memory when
the agent is entirely idle, as noted by David R. Piegdon. This is the
simple fix, a more efficient one will be done later. With markus,
deraadt, with & ok djm.
[sshd.c]
Clear alarm() before restarting sshd on SIGHUP. Without this, if there's
a SIGALRM pending (for SSH1 key regeneration) when sshd is SIGHUP'ed, the
newly exec'ed sshd will get the SIGALRM and not have a handler for it,
and the default action will terminate the listening sshd. Analysis and
patch from andrew at gaul.org.
an array for signatures when there are none since "calloc(0, n) returns
NULL on some platforms (eg Tru64), which is explicitly permitted by
POSIX. Diagnosis and patch by svallet genoscope.cns.fr.
[monitor_wrap.c servconf.c servconf.h monitor.c sshd_config.5]
Teach Match how handle config directives that are used before
authentication. This allows configurations such as permitting password
authentication from the local net only while requiring pubkey from
offsite. ok djm@, man page bits ok jmc@
[scp.c]
fix detection of whether we should show progress meter or not: scp
tested isatty(stderr) but wrote the progress meter to stdout. This patch
makes it test stdout. bz#1265 reported by junkmail AT bitsculpture.com;
of dtucker@
[sftp-client.c]
return error from do_upload() when a write fails. fixes bz#1252: zero
exit status from sftp when uploading to a full device. report from
jirkat AT atlas.cz; ok dtucker@
[ssh_config.5]
do not use a list for SYNOPSIS;
this is actually part of a larger report sent by eric s. raymond
and forwarded by brad, but i only read half of it. spotted by brad.
when closing a tty session when a background process still holds tty
fds open. Great detective work and patch by Marc Aurele La France,
slightly tweaked by me; ok dtucker@
library interfaces aren't very helpful. Fix up the DOPR_OUTCH macro
so it works properly and modify its callers so that they don't pre or
post decrement arguments that are conditionally evaluated. While there,
put SNPRINTF_CONST back as it prevents build failures in some
configurations. ok djm@ (for most of it)
value of snprintf replacement, similar to bugs in various libc
implementations. This overflow is not exploitable in OpenSSH.
While I'm fiddling with it, make it a fair bit faster by inlining the
append-char routine; ok dtucker@
[channels.c compat.c compat.h]
bz #1019: some ssh.com versions apparently can't cope with the
remote port forwarding bind_address being a hostname, so send
them an address for cases where they are not explicitly
specified (wildcard or localhost bind). reported by daveroth AT
acm.org; ok dtucker@ deraadt@
[monitor.c version.h]
correctly check for bad signatures in the monitor, otherwise the monitor
and the unpriv process can get out of sync. with dtucker@, ok djm@,
dtucker@
[kexdhc.c kexdhs.c kexgexc.c kexgexs.c]
check DH_compute_key() for -1 even if it should not happen because of
earlier calls to dh_pub_is_valid(); report krahmer at suse.de; ok djm
events fatal in Solaris process contract support and tell it to signal
only processes in the same process group when something happens.
Based on information from andrew.benham at thus.net and similar to
a patch from Chad Mynhier. ok djm@
[clientloop.c serverloop.c]
exit instead of doing a blocking tcp send if we detect a client/server
timeout, since the tcp sendqueue might be already full (of alive
requests); ok dtucker, report mpf
- ray@cvs.openbsd.org 2006/09/30 17:48:22
[sftp.c]
Clear errno before calling the strtol functions.
From Paul Stoeber <x0001 at x dot de1 dot cc>.
OK deraadt@.
section so additional platform specific CHECK_HEADER tests will work
correctly. Fixes "<net/if_tap.h> on FreeBSD" problem report by des AT des.no
Feedback and "seems like a good idea" dtucker@
[sftp.c]
Use S_IS* macros insted of masking with S_IF* flags. The latter may
have multiple bits set, which lead to surprising results. Spotted by
Paul Stoeber, more to come. ok millert@ pedro@ jaredy@ djm@
macro redefinitions, and if not, remove "-qlanglvl=ansi" from the flags.
Allows build out of the box with older VAC and XLC compilers. Found by
David Bronder and Bernhard Simon.
Damien Miller
Darren Tucker
Tim Rice