[readconf.c readconf.h ssh.1 ssh.c ssh_config.5]
s/canonicalise/canonicalize/ for consistency with existing spelling,
e.g. authorized_keys; pointed out by naddy@
[readconf.c readconf.h roaming_client.c ssh.1 ssh.c ssh_config.5]
[sshconnect.c sshconnect.h]
Implement client-side hostname canonicalisation to allow an explicit
search path of domain suffixes to use to convert unqualified host names
to fully-qualified ones for host key matching.
This is particularly useful for host certificates, which would otherwise
need to list unqualified names alongside fully-qualified ones (and this
causes a number of problems).
"looks fine" markus@
[canohost.c misc.c misc.h readconf.c sftp-server.c ssh.c]
refactor client config code a little:
add multistate option partsing to readconf.c, similar to servconf.c's
existing code.
move checking of options that accept "none" as an argument to readconf.c
add a lowercase() function and use it instead of explicit tolower() in
loops
part of a larger diff that was ok markus@
[readconf.c readconf.h ssh-keysign.c ssh.c ssh_config.5]
add a "Match" keyword to ssh_config that allows matching on hostname,
user and result of arbitrary commands. "nice work" markus@
[sftp-client.c]
rename flag arguments to be more clear and consistent.
reorder some internal function arguments to make adding additional flags
easier.
no functional change
[sftp-server.8 sftp-server.c]
Add ability to whitelist and/or blacklist sftp protocol requests by name.
Refactor dispatch loop and consolidate read-only mode checks.
Make global variables static, since sftp-server is linked into sshd(8).
ok dtucker@
[dh.c dh.h]
Increase the size of the Diffie-Hellman groups requested for a each
symmetric key size. New values from NIST Special Publication 800-57 with
the upper limit specified by RFC4419. Pointed out by Peter Backes, ok
djm@.
[channels.c]
bz#1297 - tell the client (via packet_send_debug) when their preferred
listen address has been overridden by the server's GatewayPorts;
ok dtucker@
[progressmeter.c]
store the initial file offset so the progress meter doesn't freak out
when resuming sftp transfers. bz#2137; patch from Iain Morgan; ok dtucker@
[ssh_config]
Remove gssapi config parts from ssh_config, as was already done for
sshd_config. Req by/ok ajacoutot@
ID SYNC ONLY for portable; kerberos/gssapi is still pretty popular
[channels.c]
avoid unaligned access in code that reused a buffer to send a
struct in_addr in a reply; simpler just use use buffer_put_int();
from portable; spotted by and ok dtucker@
[sshd_config]
Remove commented-out kerberos/gssapi config options from sample config,
kerberos support is currently not enabled in ssh in OpenBSD. Discussed with
various people; ok deraadt@
ID SYNC ONLY for portable; kerberos/gssapi is still pretty popular
[ssh-keygen.c sshconnect1.c sshd.c]
All the instances of arc4random_stir() are bogus, since arc4random()
does this itself, inside itself, and has for a very long time.. Actually,
this was probably reducing the entropy available.
ok djm
ID SYNC ONLY for portable; we don't trust other arc4random implementations
to do this right.
[ssh-keygen.c]
improve batch processing a bit by making use of the quite flag a bit
more often and exit with a non zero code if asked to find a hostname
in a known_hosts file and it wasn't there;
originally from reyk@, ok djm
[sshd.c]
Stir PRNG after post-accept fork. The child gets a different PRNG state
anyway via rexec and explicit privsep reseeds, but it's good to be sure.
ok markus@
[readconf.c readconf.h ssh_config.5 sshconnect.c]
Add a ssh_config ProxyUseFDPass option that supports the use of
ProxyCommands that establish a connection and then pass a connected
file descriptor back to ssh(1). This allows the ProxyCommand to exit
rather than have to shuffle data back and forth and enables ssh to use
getpeername, etc. to obtain address information just like it does with
regular directly-connected sockets. ok markus@
[sftp-client.c]
two problems found by a to-be-committed regress test: 1) msg_id was not
being initialised so was starting at a random value from the heap
(harmless, but confusing). 2) some error conditions were not being
propagated back to the caller
[sftp-client.c sftp-client.h sftp.c]
add a "-l" flag for the rename command to force it to use the silly
standard SSH_FXP_RENAME command instead of the POSIX-rename- like
posix-rename@openssh.com extension.
intended for use in regress tests, so no documentation.
CLOCK_MONOTONIC...) fails. Some older versions of RHEL have the
CLOCK_MONOTONIC define but don't actually support it. Found and tested
by Kevin Brott, ok djm.
blocking connecting socket will clear any stored errno that might
otherwise have been retrievable via getsockopt(). A hack to limit writes
to TTYs on AIX was triggering this. Since only AIX needs the hack, wrap
it in an #ifdef. Diagnosis and patch from Ivo Raisr.
[sftp-client.c sftp-client.h sftp.1 sftp.c]
sftp support for resuming partial downloads; patch mostly by Loganaden
Velvindron/AfriNIC with some tweaks by me; feedback and ok dtucker@
[ssh.c]
daemonise backgrounded (ControlPersist'ed) multiplexing master to ensure
it is fully detached from its controlling terminal. based on debugging
[auth.h kex.h kexdhs.c kexecdhs.c kexgexs.c monitor.c servconf.c]
[servconf.h session.c sshd.c sshd_config.5]
add ssh-agent(1) support to sshd(8); allows encrypted hostkeys,
or hostkeys on smartcards; most of the work by Zev Weiss; bz #1974
ok djm@
[ssh-keygen.c]
do_print_resource_record() can never be called with a NULL filename, so
don't attempt (and bungle) asking for one if it has not been specified
bz#2127 ok dtucker@
[misc.c]
in ssh_gai_strerror() don't fallback to strerror for EAI_SYSTEM when
errno == 0. Avoids confusing error message in some broken resolver
cases. bz#2122 patch from plautrba AT redhat.com; ok dtucker
[ssh-keygen.1 ssh.1 ssh_config.5 sshd.8 sshd_config.5]
do not use Sx for sections outwith the man page - ingo informs me that
stuff like html will render with broken links;
issue reported by Eric S. Raymond, via djm
[auth-rsa.c auth.h auth2-hostbased.c auth2-pubkey.c monitor.c]
for hostbased authentication, print the client host and user on
the auth success/failure line; bz#2064, ok dtucker@
[scp.c]
Handle time_t values as long long's when formatting them and when
parsing them from remote servers.
Improve error checking in parsing of 'T' lines.
ok dtucker@ deraadt@
contrib/cygwin/ssh-user-config] Modernizes and improve readability of
the Cygwin README file (which hasn't been updated for ages), drop
unsupported OSes from the ssh-host-config help text, and drop an
unneeded option from ssh-user-config. Patch from vinschen at redhat com.
[channels.c channels.h clientloop.c]
Add an "ABANDONED" channel state and use for mux sessions that are
disconnected via the ~. escape sequence. Channels in this state will
be able to close if the server responds, but do not count as active channels.
This means that if you ~. all of the mux clients when using ControlPersist
on a broken network, the backgrounded mux master will exit when the
Control Persist time expires rather than hanging around indefinitely.
bz#1917, also reported and tested by tedu@. ok djm@ markus@.
[sshd.c]
When running sshd -D, close stderr unless we have explicitly requesting
logging to stderr. From james.hunt at ubuntu.com via bz#1976, djm's patch
so, err, ok dtucker.
[sftp.c]
Make sftp's libedit interface marginally multibyte aware by building up
the quoted string by character instead of by byte. Prevents failures
when linked against a libedit built with wide character support (bz#1990).
"looks ok" djm
[mac.c]
force the MAC output to be 64-bit aligned so umac won't see unaligned
accesses on strict-alignment architectures. bz#2101, patch from
tomas.kuthan at oracle.com, ok djm@
[clientloop.h clientloop.c mux.c]
No need for the mux cleanup callback to be visible so restore it to static
and call it through the detach_user function pointer. ok djm@
forwarding test is extremely slow copying data on some machines so switch
back to copying the much smaller ls binary until we can figure out why
this is.
need a shell that can handle "[ file1 -nt file2 ]". Rather than keep
dealing with shell portability issues in regression tests, we let
configure find us a capable shell on those platforms with an old /bin/sh.
[scp.c sftp-client.c]
Replace S_IWRITE, which isn't standardized, with S_IWUSR, which is. Patch
from Nathan Osman via bz#2113. ok deraadt.
(note: corrected bug number from 2085)
[ssh-agent.c clientloop.c misc.h packet.c progressmeter.c misc.c
channels.c sandbox-systrace.c]
Use clock_gettime(CLOCK_MONOTONIC ...) for ssh timers so that things like
keepalives and rekeying will work properly over clock steps. Suggested by
markus@, "looks good" djm@.
[auth.h auth.c key.c monitor.c auth-rsa.c auth2.c auth1.c key.h]
Standardise logging of supplemental information during userauth. Keys
and ruser is now logged in the auth success/failure message alongside
the local username, remote host/port and protocol in use. Certificates
contents and CA are logged too.
Pushing all logging onto a single line simplifies log analysis as it is
no longer necessary to relate information scattered across multiple log
entries. "I like it" markus@
[regress/login-timeout.sh regress/reexec.sh regress/test-exec.sh]
Use SUDO when cat'ing pid files and running the sshd log wrapper so that
it works with a restrictive umask and the pid files are not world readable.
Changes from -portable. (id sync only)
[regress/agent-timeout.sh]
Pull back some portability changes from -portable:
- TIMEOUT is a read-only variable in some shells
- not all greps have -q so redirect to /dev/null instead.
(ID sync only)
methods. When the openssl version doesn't support ECDH then next one on
the list is DH group exchange, but that causes a bit more traffic which can
mean that the tests flip bits in the initial exchange rather than the MACed
traffic and we get different errors to what the tests look for.
[regress/Makefile regress/rekey.sh regress/integrity.sh
regress/sshd-log-wrapper.sh regress/forwarding.sh regress/test-exec.sh]
use -E option for ssh and sshd to write debuging logs to ssh{,d}.log and
save the output from any failing tests. If a test fails the debug output
from ssh and sshd for the failing tests (and only the failing tests) should
be available in failed-ssh{,d}.log.
[regress/rekey.sh regress/test-exec.sh regress/integrity.sh
regress/multiplex.sh Makefile regress/cfgmatch.sh]
Split the regress log into 3 parts: the debug output from ssh, the debug
log from sshd and the output from the client command (ssh, scp or sftp).
Somewhat functional now, will become more useful when ssh/sshd -E is added.
[test-exec.sh]
Only regenerate host keys if they don't exist or if ssh-keygen has changed
since they were. Reduces test runtime by 5-30% depending on machine
speed.
[sshd_config.5 servconf.c servconf.h packet.c serverloop.c monitor.c sshd_config
sshd.c] Add RekeyLimit to sshd with the same syntax as the client allowing
rekeying based on traffic volume or time. ok djm@, help & ok jmc@ for the man
page.
[ssh_config sshconnect2.c packet.c readconf.h readconf.c clientloop.c
ssh_config.5 packet.h]
Add an optional second argument to RekeyLimit in the client to allow
rekeying based on elapsed time in addition to amount of traffic.
with djm@ jmc@, ok djm
we don't get a warning on compilers that *don't* support it. Add
-Wno-unknown-warning-option. Move both to the start of the list for
maximum noise suppression. Tested with gcc 4.6.3, gcc 2.95.4 and clang 2.9.
openbsd-compat/getopt_long.c regress/modpipe.c] Remove getopt.c, add
portability code to getopt_long.c and switch over Makefile and the ugly
hack in modpipe.c. Fixes bz#1448.
[authfile.c cipher.c cipher.h kex.c kex.h kexecdh.c kexecdhc.c kexecdhs.c]
[key.c key.h mac.c mac.h packet.c ssh.1 ssh.c]
add the ability to query supported ciphers, MACs, key type and KEX
algorithms to ssh. Includes some refactoring of KEX and key type handling
to be table-driven; ok markus@
[session.c]
reintroduce 1.262 without the connection-killing bug:
fatal() when ChrootDirectory specified by running without root privileges;
ok markus@
Damien Miller
Darren Tucker
Tim Rice