OpenSSH 5.1: call for testing

[Damien Miller]

Hi,

OpenSSH 5.1 is almost ready for release, so we would appreciate testing
on as many platforms and systems as possible. This release is one of
the biggest in recent years, with two hackathons' worth of improvements
and fixes for some of our most recalcitrant bugs.

Snapshot releases for portable OpenSSH are available from
http://www.mindrot.org/openssh_snap/ 

The OpenBSD version is available in CVS HEAD:
http://www.openbsd.org/anoncvs.html 

Portable OpenSSH is also available via anonymous CVS using the
instructions at http://www.openssh.com/portable.html#cvs

Running the regression tests supplied with Portable OpenSSH does not
require installation and is a simply:

$ ./configure && make tests

Live testing on suitable non-production systems is also appreciated.
Please send reports of success or failure to
openssh-unix-dev at mindrot.org.

Below is a summary of changes. More detail may be found in the
ChangeLog in the portable OpenSSH tarballs.

Thanks to the many people who contributed to this release.


New features:

 * Introduce experimental SSH Fingerprint ASCII Visualisation to ssh(1)
   and ssh-keygen(1). Visual fingerprinnt display is controlled by a new
   ssh_config(5) option "VisualHostKey". The intent is to render
   SSH host keys in a visual form that is amenable to easy recall and
   rejection of changed host keys. This technique inspired by the
   graphical hash visualisation schemes known as "random art[*]", and
   by Dan Kaminsky's musings at 23C3 in Berlin.
   Fingerprint visualisation in is currently disabled by default, as the
   algorithm used to generate the random art is still subject to change.
   [*] "Hash Visualization: a New Technique to improve Real-World
       Security", Perrig A. and Song D., 1999, International Workshop on
       Cryptographic Techniques and E-Commerce (CrypTEC '99)
   http://sparrow.ece.cmu.edu/~adrian/projects/validation/validation.pdf

 * sshd_config(5) now supports CIDR address/masklen matching in "Match
   address" blocks, with a fallback to classic wildcard matching. For 
   example:
     Match address 192.0.2.0/24,3ffe:ffff::/32,!10.*
         PasswordAuthentication yes

 * sshd(8) now supports CIDR matching in ~/.ssh/authorized_keys
   from="..." restrictions, also with a fallback to classic wildcard
   matching.

 * Added an extended test mode (-T) to sshd(8) to request that it write
   its effective configuration to stdout and exit. Extended test mode
   also supports the specification of connection parameters (username,
   source address and hostname) to test the application of
   sshd_config(5) Match rules. 

 * sftp-server(8) now supports extension methods statvfs at openssh.com and
   fstatvfs at openssh.com that implement statvfs(2)-like operations.

 * sftp(1) now has a "df" command to the sftp client that uses the
   statvfs at openssh.com to produce a df(1)-like display of filesystem
   space and inode utilisation (requires statvfs at openssh.com support on
   the server)

 * Added a MaxSessions option to sshd_config(5) to allow control of the
   number of multiplexed sessions supported over a single TCP connection.
   This allows increasing the number of allowed sessions above the
   previous default of 10, disabling connection multiplexing 
   (MaxSessions=1) or disallowing login/shell/subsystem sessions
   entirely (MaxSessions=0).

 * Added a no-more-sessions at openssh.com global request extension that is
   sent from ssh(1) to sshd(8) when the client knows that it will never
   request another session (i.e. when session multiplexing is disabled). 
   This allows a server to disallow further session requests and
   terminate the session in cases where the client has been hijacked.

 * ssh-keygen(1) now supports the use of the -l option in combination
   with -F to search for a host in ~/.ssh/known_hosts and display its 
   fingerprint.

 * ssh-keyscan(1) now defaults to "rsa" (protocol 2) keys, instead of
   "rsa1".

 * Added an AllowAgentForwarding option to sshd_config(8) to control
   whether authentication agent forwarding is permitted. Note that this
   is a loose control, as a client may install their own unofficial
   forwarder.

 * Avoid unnecessary malloc/copy/free when receiving network data,
   resulting in a ~10% speedup

 * ssh(1) and sshd(8) will now try additional addresses when connecting
   to a port forward destination whose DNS name resolves to more than
   one address. The previous behaviour was to try the only first address
   and give up if that failed.

 * ssh(1) and sshd(8) now support signalling that channels are
   half-closed for writing, through a channel protocol extension
   notification "eow at openssh.com". This allows propagation of closed
   file descriptors, so that commands such as:
       "ssh -2 localhost od /bin/ls | true"
   do not send unnecessary data over the wire. (bz#85)

 * sshd(8): increased the default size of ssh protocol 1 ephemeral keys 
   from 768 to 1024 bits.

 * When ssh(1) has been requested to fork after authentication
   ("ssh -f") with ExitOnForwardFailure enabled, delay the fork until
   after replies for any -R forwards have been seen. Allows for robust
   detection of -R forward failure when using -f.

 * "Match group" blocks in sshd_config(5) now support negation of
   groups. E.g. "Match group staff,!guests"

 * sftp(1) and sftp-server(8) now allow chmod-like operations to set 
   set[ug]id/sticky bits.

 * The MaxAuthTries option is now permitted in sshd_config(5) match
   blocks.

 * Multiplexed ssh(1) sessions now support a subset of the ~ escapes
   that are available to a primary connection.

 * ssh(1) connection multiplexing will now fall back to creating a new
   connection in most error cases.

 * Added some basic interoperability tests against Twisted Conch.

 * Documented OpenSSH's extensions to and deviations from the published
   SSH protocols (the PROTOCOL file in the distribution)

 * Documented OpenSSH's ssh-agent protocol (PROTOCOL.agent).

Bugfixes

 * Make ssh(1) deal more gracefully with channel requests that fail.
   Previously it would optimistically assume that requests would always 
   succeed, which could cause hangs if they did not (e.g. when the
   server runs out of file descriptors).

 * ssh(1) now reports multiplexing errors via the multiplex slave's
   stderr where possible (subject to LogLevel in the mux master).

 * ssh(1) and sshd(8) now send terminate protocol banners with CR+LF for
   protocol 2 to comply with RFC 4253. Previously they were terminated
   with CR alone. Protocol 1 banners remain CR terminated.

 * Merged duplicate authentication file checks in sshd(8) and refuse to
   read authorised_keys and .shosts from non-regular files.

 * Ensure that sshd(8)'s umask disallows at least group and world write,
   even if a more permissive one has been inherited.

 * Suppress the warning message from sshd(8) when changing to a
   non-existent user home directory after chrooting.

 * Mention that scp(1) follows symlinks when performing recursive
   copies.

 * Prevent sshd(8) from erroneously applying public key restrictions
   leaned from ~/.ssh/authorized_keys to other authentication methods
   when public key authentication subsequently fails.

 * Fix protocol keepalive timeouts - in some cases, keepalive packets
   were being sent, but the connection was not being closed when the
   limit for missing replies was exceeded.

 * Fix ssh(1) sending invalid TTY modes when a TTY was forced (ssh -tt)
   but stdin was not a TTY.

 * ssh(1) will now exit with a non-zero exit status if
   ExitOnForwardFailure was set and forwardings were disabled due to a
   failed host key check.

 * Fix MaxAuthTries tests to disallow a free authentication try to
   clients that skipped the protocol 2 "none" authentication method.

 * bz#1363: Make keepalive timeouts apply while synchronously waiting
   for a packet, particularly during key renegotiation.

 * sshd(8) has been audited to eliminate fd leaks and calls to fatal()
   in conditions of file descriptor exhaustion.

Portable OpenSSH-specific bugfixes

 * Avoid a sshd(8) hang-on-exit on Solaris caused by depending on the 
   success of isatty() on a PTY master (undefined behaviour). Probably 
   affected other platforms too.

 * bz#1083: Fixed test for locked accounts on HP/UX with shadowed
   passwords disabled.

 * bz#1386: Disable poll() fallback in atomiciov for Tru64. readv
   doesn't seem to be a comparable object there, which lead to
   compilation errors.

 * bz#1447: Fall back to racy rename if link returns EXDEV.

 * bz#1467: Explicitly handle EWOULDBLOCK wherever we handle EAGAIN, on
   some platforms (HP nonstop) it is a distinct errno.

 * bz#1240: Avoid NULL dereferences in ancient sigaction replacement
   code.

 * bz#1276: Avoid linking against libgssapi, which despite its name
   doesn't seem to implement all of GSSAPI.

 * bz#1112: Use explicit noreturn attribute instead of __dead, fixing
   compilation problems on Interix.

 * bz#1241: Support password expiry on Tru64 SIA systems.

 * bz#1462: Fix an UMAC alignment problem that manifested on Itanium 
   platforms.


Damien Miller