[PROTOCOL PROTOCOL.agent PROTOCOL.certkeys auth2-jpake.c authfd.c]
[authfile.c buffer.h dns.c kex.c kex.h key.c key.h monitor.c]
[monitor_wrap.c myproposal.h packet.c packet.h pathnames.h readconf.c]
[ssh-add.1 ssh-add.c ssh-agent.1 ssh-agent.c ssh-keygen.1 ssh-keygen.c]
[ssh-keyscan.1 ssh-keyscan.c ssh-keysign.8 ssh.1 ssh.c ssh2.h]
[ssh_config.5 sshconnect.c sshconnect2.c sshd.8 sshd.c sshd_config.5]
[uuencode.c uuencode.h bufec.c kexecdh.c kexecdhc.c kexecdhs.c ssh-ecdsa.c]
Implement Elliptic Curve Cryptography modes for key exchange (ECDH) and
host/user keys (ECDSA) as specified by RFC5656. ECDH and ECDSA offer
better performance than plain DH and DSA at the same equivalent symmetric
key length, as well as much shorter keys.
Only the mandatory sections of RFC5656 are implemented, specifically the
three REQUIRED curves nistp256, nistp384 and nistp521 and only ECDH and
ECDSA. Point compression (optional in RFC5656 is NOT implemented).
Certificate host and user keys using the new ECDSA key types are supported.
Note that this code has not been tested for interoperability and may be
subject to change.
feedback and ok markus@
[auth-options.c auth1.c auth2.c bufaux.c buffer.h kex.c key.c packet.c]
[packet.h ssh-dss.c ssh-rsa.c]
Add buffer_get_cstring() and related functions that verify that the
string extracted from the buffer contains no embedded \0 characters*
This prevents random (possibly malicious) crap from being appended to
strings where it would not be noticed if the string is used with
a string(3) function.
Use the new API in a few sensitive places.
* actually, we allow a single one at the end of the string for now because
we don't know how many deployed implementations get this wrong, but don't
count on this to remain indefinitely.
[ssh-add.c ssh-agent.c ssh-keygen.c ssh-keysign.c ssh.c sshd.c]
OpenSSL_add_all_algorithms is the name of the function we have a man page
for, so use that. ok djm
[ssh.c]
close any extra file descriptors inherited from parent at start and
reopen stdin/stdout to /dev/null when forking for ControlPersist.
prevents tools that fork and run a captive ssh for communication from
failing to exit when the ssh completes while they wait for these fds to
close. The inherited fds may persist arbitrarily long if a background
mux master has been started by ControlPersist. cvs and scp were effected
by this.
"please commit" markus@
openbsd-compat/openbsd-compat.h openbsd-compat/strptime.c] Add strptime to
the compat library which helps on platforms like old IRIX. Based on work
by djm, tested by Tom Christensen.
regress/test-exec.sh] Under certain conditions when testing with sudo
tests would fail because the pidfile could not be read by a regular user.
"cat: cannot open ...../regress/pidfile: Permission denied (error 13)"
Make sure cat is run by $SUDO. no objection from me. djm@
[channels.c]
Fix a trio of bugs in the local/remote window calculation for datagram
data channels (i.e. TunnelForward):
Calculate local_consumed correctly in channel_handle_wfd() by measuring
the delta to buffer_len(c->output) from when we start to when we finish.
The proximal problem here is that the output_filter we use in portable
modified the length of the dequeued datagram (to futz with the headers
for !OpenBSD).
In channel_output_poll(), don't enqueue datagrams that won't fit in the
peer's advertised packet size (highly unlikely to ever occur) or which
won't fit in the peer's remaining window (more likely).
In channel_input_data(), account for the 4-byte string header in
datagram packets that we accept from the peer and enqueue in c->output.
report, analysis and testing 2/3 cases from wierbows AT us.ibm.com;
"looks good" markus@
[authfile.c]
commited the wrong version of the hostbased certificate diff; this
version replaces some strlc{py,at} verbosity with xasprintf() at
the request of markus@
[PROTOCOL.certkeys ssh-keygen.c]
tighten the rules for certificate encoding by requiring that options
appear in lexical order and make our ssh-keygen comply. ok markus@
[ssh.1 ssh_config.5 sshd.8]
Remove mentions of weird "addr/port" alternate address format for IPv6
addresses combinations. It hasn't worked for ages and we have supported
the more commen "[addr]:port" format for a long time. ok jmc@ markus@
[clientloop.c readconf.c readconf.h ssh.c ssh_config.5]
add a "ControlPersist" option that automatically starts a background
ssh(1) multiplex master when connecting. This connection can stay alive
indefinitely, or can be set to automatically close after a user-specified
duration of inactivity. bz#1330 - patch by dwmw2 AT infradead.org, but
further hacked on by wmertens AT cisco.com, apb AT cequrux.com,
martin-mindrot-bugzilla AT earth.li and myself; "looks ok" markus@
[sftp-client.c]
bz#1797: fix swapped args in upload_dir_internal(), breaking recursive
upload depth checks and causing verbose printing of transfers to always
be turned on; patch from imorgan AT nas.nasa.gov
[ssh-rsa.c]
more timing paranoia - compare all parts of the expected decrypted
data before returning. AFAIK not exploitable in the SSH protocol.
"groovy" deraadt@
[auth-rsa.c channels.c jpake.c key.c misc.c misc.h monitor.c]
[packet.c ssh-rsa.c]
implement a timing_safe_cmp() function to compare memory without leaking
timing information by short-circuiting like memcmp() and use it for
some of the more sensitive comparisons (though nothing high-value was
readily attackable anyway); "looks ok" markus@
[ssh.c ssh_config.5]
expand %h to the hostname in ssh_config Hostname options. While this
sounds useless, it is actually handy for working with unqualified
hostnames:
Host *.*
Hostname %h
Host *
Hostname %h.example.org
"I like it" markus@
[misc.c]
unbreak strdelim() skipping past quoted strings, e.g.
AllowUsers "blah blah" blah
was broken; report and fix in bz#1757 from bitman.zhou AT centrify.com
ok dtucker;
[ssh.c]
log the hostname and address that we connected to at LogLevel=verbose
after authentication is successful to mitigate "phishing" attacks by
servers with trusted keys that accept authentication silently and
automatically before presenting fake password/passphrase prompts;
"nice!" markus@
[ssh.c]
log the hostname and address that we connected to at LogLevel=verbose
after authentication is successful to mitigate "phishing" attacks by
servers with trusted keys that accept authentication silently and
automatically before presenting fake password/passphrase prompts;
"nice!" markus@
[channels.c session.c]
bz#1750: fix requirement for /dev/null inside ChrootDirectory for
internal-sftp accidentally introduced in r1.253 by removing the code
that opens and dup /dev/null to stderr and modifying the channels code
to read stderr but discard it instead; ok markus@
[channels.c mux.c readconf.c readconf.h ssh.h]
bz#1327: remove hardcoded limit of 100 permitopen clauses and port
forwards per direction; ok markus@ stevesk@
[ssh-keyscan.c]
replace verbose and overflow-prone Linebuf code with read_keyfile_line()
based on patch from joachim AT joachimschipper.nl; bz#1565; ok dtucker@
[auth.c]
queue auth debug messages for bad ownership or permissions on the user's
keyfiles. These messages will be sent after the user has successfully
authenticated (where our client will display them with LogLevel=debug).
[servconf.c sshd_config.5]
expose some more sshd_config options inside Match blocks:
AuthorizedKeysFile AuthorizedPrincipalsFile
HostbasedUsesNameFromPacketOnly PermitTunnel
bz#1764; feedback from imorgan AT nas.nasa.gov; ok dtucker@
[mux.c]
Correct sizing of object to be allocated by calloc(), replacing
sizeof(state) with sizeof(*state). This worked by accident since
the struct contained a single int at present, but could have broken
in the future. patch from hyc AT symas.com
[PROTOCOL.certkeys auth-options.c ssh-keygen.c]
Move the permit-* options to the non-critical "extensions" field for v01
certificates. The logic is that if another implementation fails to
implement them then the connection just loses features rather than fails
outright.
ok markus@
[PROTOCOL.mux clientloop.h mux.c readconf.c readconf.h ssh.1 ssh.c]
mux support for remote forwarding with dynamic port allocation,
use with
LPORT=`ssh -S muxsocket -R0:localhost:25 -O forward somehost`
feedback and ok djm@
[channels.c channels.h mux.c ssh.c]
Pause the mux channel while waiting for reply from aynch callbacks.
Prevents misordering of replies if new requests arrive while waiting.
Extend channel open confirm callback to allow signalling failure
conditions as well as success. Use this to 1) fix a memory leak, 2)
start using the above pause mechanism and 3) delay sending a success/
failure message on mux slave session open until we receive a reply from
the server.
motivated by and with feedback from markus@
[auth-options.c auth-options.h auth.c auth.h auth2-pubkey.c]
[key.c servconf.c servconf.h sshd.8 sshd_config.5]
add some optional indirection to matching of principal names listed
in certificates. Currently, a certificate must include the a user's name
to be accepted for authentication. This change adds the ability to
specify a list of certificate principal names that are acceptable.
When authenticating using a CA trusted through ~/.ssh/authorized_keys,
this adds a new principals="name1[,name2,...]" key option.
For CAs listed through sshd_config's TrustedCAKeys option, a new config
option "AuthorizedPrincipalsFile" specifies a per-user file containing
the list of acceptable names.
If either option is absent, the current behaviour of requiring the
username to appear in principals continues to apply.
These options are useful for role accounts, disjoint account namespaces
and "user@realm"-style naming policies in certificates.
feedback and ok markus@
[sshconnect2.c]
bz#1502: authctxt.success is declared as an int, but passed by
reference to function that accepts sig_atomic_t*. Convert it to
the latter; ok markus@ dtucker@
[session.c]
set stderr to /dev/null for subsystems rather than just closing it.
avoids hangs if a subsystem or shell initialisation writes to stderr.
bz#1750; ok markus@
[mux.c]
set "detach_close" flag when registering channel cleanup callbacks.
This causes the channel to close normally when its fds close and
hangs when terminating a mux slave using ~. bz#1758; ok markus@
- djm@cvs.openbsd.org 2010/04/23 01:47:41
[ssh-keygen.c]
bz#1740: display a more helpful error message when $HOME is
inaccessible while trying to create .ssh directory. Based on patch
from jchadima AT redhat.com; ok dtucker@
- djm@cvs.openbsd.org 2010/04/16 01:58:45
[regress/cert-hostkey.sh regress/cert-userkey.sh]
regression tests for v01 certificate format
includes interop tests for v00 certs
[PROTOCOL.certkeys auth-options.c auth-options.h auth-rsa.c]
[auth2-pubkey.c authfd.c key.c key.h myproposal.h ssh-add.c]
[ssh-agent.c ssh-dss.c ssh-keygen.1 ssh-keygen.c ssh-rsa.c]
[sshconnect.c sshconnect2.c sshd.c]
revised certificate format ssh-{dss,rsa}-cert-v01@openssh.com with the
following changes:
move the nonce field to the beginning of the certificate where it can
better protect against chosen-prefix attacks on the signature hash
Rename "constraints" field to "critical options"
Add a new non-critical "extensions" field
Add a serial number
The older format is still support for authentication and cert generation
(use "ssh-keygen -t v00 -s ca_key ..." to generate a v00 certificate)
ok markus@
[ssh-pkcs11.c]
retry lookup for private key if there's no matching key with CKA_SIGN
attribute enabled; this fixes fixes MuscleCard support (bugzilla #1736)
ok djm@
[sshconnect2.c]
show the key type that we are offering in debug(), helps distinguish
between certs and plain keys as the path to the private key is usually
the same.
[clientloop.c]
bz#1698: kill channel when pty allocation requests fail. Fixed
stuck client if the server refuses pty allocation.
ok dtucker@ "think so" markus@
- djm@cvs.openbsd.org 2010/03/26 03:13:17
[bufaux.c]
allow buffer_get_int_ret/buffer_get_int64_ret to take a NULL pointer
argument to allow skipping past values in a buffer
[ssh_config.5]
Reformat default value of PreferredAuthentications entry (current
formatting implies ", " is acceptable as a separator, which it's not.
ok djm@
[auth-rhosts.c monitor.c monitor_wrap.c session.c auth-options.c sshd.c]
Hold authentication debug messages until after successful authentication.
Fixes an info leak of environment variables specified in authorized_keys,
reported by Jacob Appelbaum. ok djm@
[regress/cert-hostkey.sh regress/cert-userkey.sh]
add an extra test to ensure that authentication with the wrong
certificate fails as it should (and it does)
[auth-rh-rsa.c auth-rsa.c auth.c auth.h auth2-hostbased.c auth2-pubkey.c]
[authfile.c authfile.h hostfile.c hostfile.h servconf.c servconf.h]
[ssh-keygen.c ssh.1 sshconnect.c sshd_config.5]
Add a TrustedUserCAKeys option to sshd_config to specify CA keys that
are trusted to authenticate users (in addition than doing it per-user
in authorized_keys).
Add a RevokedKeys option to sshd_config and a @revoked marker to
known_hosts to allow keys to me revoked and banned for user or host
authentication.
feedback and ok markus@
case from that matched in the system password database. On this
platform, passwords are stored case-insensitively, but sshd requires
exact case matching for Match blocks in sshd_config(5). Based on
a patch from vinschen AT redhat.com.
- djm@cvs.openbsd.org 2010/02/26 20:29:54
[PROTOCOL PROTOCOL.agent PROTOCOL.certkeys addrmatch.c auth-options.c]
[auth-options.h auth.h auth2-pubkey.c authfd.c dns.c dns.h hostfile.c]
[hostfile.h kex.h kexdhs.c kexgexs.c key.c key.h match.h monitor.c]
[myproposal.h servconf.c servconf.h ssh-add.c ssh-agent.c ssh-dss.c]
[ssh-keygen.1 ssh-keygen.c ssh-rsa.c ssh.1 ssh.c ssh2.h sshconnect.c]
[sshconnect2.c sshd.8 sshd.c sshd_config.5]
Add support for certificate key types for users and hosts.
OpenSSH certificate key types are not X.509 certificates, but a much
simpler format that encodes a public key, identity information and
some validity constraints and signs it with a CA key. CA keys are
regular SSH keys. This certificate style avoids the attack surface
of X.509 certificates and is very easy to deploy.
Certified host keys allow automatic acceptance of new host keys
when a CA certificate is marked as sh/known_hosts.
see VERIFYING HOST KEYS in ssh(1) for details.
Certified user keys allow authentication of users when the signing
CA key is marked as trusted in authorized_keys. See "AUTHORIZED_KEYS
FILE FORMAT" in sshd(8) for details.
Certificates are minted using ssh-keygen(1), documentation is in
the "CERTIFICATES" section of that manpage.
Documentation on the format of certificates is in the file
PROTOCOL.certkeys
feedback and ok markus@
[regress/test-exec.sh]
wait for sshd to fully stop in cleanup() function; avoids races in tests
that do multiple start_sshd/cleanup cycles; "I hate pidfiles" deraadt@
[regress/Makefile]
turn on all the malloc(3) checking options when running regression
tests. this has caught a few bugs for me in the past; ok dtucker@
[pathnames.h readconf.c readconf.h scp.1 sftp.1 ssh-add.1 ssh-add.c]
[ssh-agent.c ssh-keygen.1 ssh-keygen.c ssh.1 ssh.c ssh_config.5]
replace our obsolete smartcard code with PKCS#11.
ftp://ftp.rsasecurity.com/pub/pkcs/pkcs-11/v2-20/pkcs-11v2-20.pdf
ssh(1) and ssh-keygen(1) use dlopen(3) directly to talk to a PKCS#11
provider (shared library) while ssh-agent(1) delegates PKCS#11 to
a forked a ssh-pkcs11-helper process.
PKCS#11 is currently a compile time option.
feedback and ok djm@; inspired by patches from Alon Bar-Lev
`
[bufaux.c]
make buffer_get_string_ret() really non-fatal in all cases (it was
using buffer_get_int(), which could fatal() on buffer empty);
ok markus dtucker
[channels.c]
fake local addr:port when stdio fowarding as some servers (Tectia at
least) validate that they are well-formed;
reported by imorgan AT nas.nasa.gov
ok dtucker
[mux.c]
don't mark channel as read failed if it is already closing; suppresses
harmless error messages when connecting to SSH.COM Tectia server
report by imorgan AT nas.nasa.gov
[clientloop.c]
downgrade an error() to a debug() - this particular case can be hit in
normal operation for certain sequences of mux slave vs session closure
and is harmless
after registering the hardware engines, which causes the openssl.cnf file to
be processed. See OpenSSL's man page for OPENSSL_config(3) for details.
Patch from Solomon Peachy, ok djm@.
[mux.c]
fix bug introduced in mux rewrite:
In a mux master, when a socket to a mux slave closes before its server
session (as may occur when the slave has been signalled), gracefully
close the server session rather than deleting its channel immediately.
A server may have more messages on that channel to send (e.g. an exit
message) that will fatal() the client if they are sent to a channel that
has been prematurely deleted.
spotted by imorgan AT nas.nasa.gov
[channels.c channels.h clientloop.c clientloop.h mux.c nchan.c ssh.c]
rewrite ssh(1) multiplexing code to a more sensible protocol.
The new multiplexing code uses channels for the listener and
accepted control sockets to make the mux master non-blocking, so
no stalls when processing messages from a slave.
avoid use of fatal() in mux master protocol parsing so an errant slave
process cannot take down a running master.
implement requesting of port-forwards over multiplexed sessions. Any
port forwards requested by the slave are added to those the master has
established.
add support for stdio forwarding ("ssh -W host:port ...") in mux slaves.
document master/slave mux protocol so that other tools can use it to
control a running ssh(1). Note: there are no guarantees that this
protocol won't be incompatibly changed (though it is versioned).
feedback Salvador Fandino, dtucker@
channel changes ok markus@
[sftp.c]
Reset SIGTERM to SIG_DFL before executing ssh, so that even if sftp
inherited SIGTERM as ignored it will still be able to kill the ssh it
starts.
ok dtucker@
[auth.c]
when using ChrootDirectory, make sure we test for the existence of the
user's shell inside the chroot; bz #1679, patch from alex AT rtfs.hu;
ok dtucker
[sftp.1 sftp.c]
sftp.1: put ls -h in the right place
sftp.c: as above, plus add -p to get/put, and shorten their arg names
to keep the help usage nicely aligned
ok djm
[sftp.c]
don't append a space after inserting a completion of a directory (i.e.
a path ending in '/') for a slightly better user experience; ok dtucker@
Damien Miller
Darren Tucker
Tim Rice