Call for testing: OpenSSH 10.1p1
Damien Miller
djm at mindrot.org
Tue Sep 30 23:45:30 AEST 2025
Hi,
OpenSSH 10.1p1 is almost ready for release, so we would appreciate testing
on as many platforms and systems as possible.
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 git using the
instructions at http://www.openssh.com/portable.html#cvs
At https://anongit.mindrot.org/openssh.git/ or via a mirror at Github:
https://github.com/openssh/openssh-portable
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. Security bugs should be reported
directly to openssh at openssh.com.
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.
Potentially-incompatible changes
--------------------------------
* ssh(1): add a warning when the connection negotiates a non-post
quantum key agreement algorithm.
This warning has been added due to the risk of "store now, decrypt
later" attacks. More details at https://openssh.com/pq.html
This warning may be controlled via a new WarnWeakCrypto ssh_config
option, defaulting to on. This option is likely to control
additional weak crypto warnings in the future.
* ssh(1), sshd(8): major changes to handling of DSCP marking/IPQoS
Both the client and the server have changed the default DCSP
(a.k.a IPQos) values and the way these values are selected at
runtime.
Both endpoints now use Expedited Forward (EF) for interactive
traffic by default. This provides better prioritisation,
especially on wireless media (cf. RFC 8325). Non-interactive
traffic now uses the operating system default DSCP marking.
Both the interactive and non-interactive DSCP values may be
overridden via the IPQoS keyword in ssh_config(5) and
sshd_config(5).
The DSCP value selected may now change over the course of a
connection. ssh(1) and sshd(8) will automatically select between
the interactive and non-interactive IPQoS values depending on
the type of SSH channels open. E.g. if a sftp session is using
the connectionn, then the non-interactive value will be used.
This is important now that the default interactive IPQoS is EF
(Expedited Forwarding), as many networks are configured to allow
only relatively small amounts of traffic of this class and they will
aggressively deprioritise the entire connection if this is exceeded.
* ssh-add(1): when adding certificates to an agent, set the expiry
to the certificate expiry time plus a short (5 min) grace period.
This will cause the agent to automtically remove certificates shortly
after they expire. A new ssh-add -N option disables this behaviour.
* All: remove experimental support for XMSS keys. This was never
enabled by default. We expect to implement a new post-quantu
signature scheme in the near future.
* ssh(1), sshd(8): deprecate support for IPv4 type-of-service (TOS)
keywords in the IPQoS configuration directive.
Type of Service (ToS) was deprecated in the late nineties and
replaced with the Differentiated Services architecture. Diffserv
has significant advantages for operators because this mechanism
offers more granularity.
OpenSSH switched its default IPQoS from ToS to DSCP values in 2018.
IPQoS configurations with 'lowdelay', 'reliability', or
'throughput' will be ignored and instead the system default QoS
settings apply. Additionally, a debug message is logged about the
deprecation with a suggestion to use DSCP.
* ssh-agent(1), sshd(8): move agent listener sockets from /tmp to
under ~/.ssh/agent for both ssh-agent(1) and forwarded sockets
in sshd(8).
This ensures processes that have restricted filesystem access
that includes /tmp do not ambiently have the ability to use keys
in an agent.
Moving the default directory has the consequence that the OS will
no longer clean up stale agent sockets, so ssh-agent now gains
this ability.
To support $HOME on NFS, the socket path includes a truncated hash of
the hostname. ssh-agent will by default only clean up sockets from
the same hostname.
ssh-agent(1) gains some new flags: -U suppresses the automatic
cleanup of stale sockets when it starts. -u forces a cleanup
without keeping a running agent, -uu forces a cleanup that ignores
the hostname. -T makes ssh-agent put the socket back in /tmp.
Changes since OpenSSH 10.0
==========================
New features
------------
* ssh(1), sshd(8): add SIGINFO handlers to log active channel and
session information.
* sshd(8): when refusing a certificate for user authentication, log
enough information to identify the certificate in addition to the
reason why it was being denied. Makes debugging certificate
authorisation problems a bit easier.
* ssh(1), ssh-agent(1): support ed25519 keys hosted on PKCS#11
tokens.
* ssh(1): add a ssh_config(5) RefuseConnection option that, when
encountered while processing an active section in a
configuration terminates ssh(1) with an error message that
contains the argument to the option.
This may be useful for expressing reminders or warnings in config
files, for example:
Match host foo
RefuseConnection "foo is deprecated, use splork instead"
* sshd(8): make the X11 display number check relative to
X11DisplayOffset. This will allows people to use X11DisplayOffset
to configure much higher port ranges if they really want, while
not changing the default behaviour.
* unit tests: the unit test framework now includes some basic
benchmarking capabilities. Run with "make UNITTEST_BENCHMARK=yes"
on OpenBSD or "make unit-bench" on Portable OpenSSH.
Bugfixes
--------
* sshd(8): fix mistracking of MaxStartups process exits in some
situations. At worst, this could cause all MaxStartups slots to
fill and sshd to refuse new connections.
* ssh(1): fix delay on X client startup when ObscureKeystrokeTiming
is enabled. bz#3820
* sshd(8): increase the maximum size of the supported configuration
from 256KB to 4MB, which ought to be enough for anybody. Fail
early and visibly when this limit is breached. bz3808
* sftp(1): during sftp uploads, avoid a condition where a failed
write could be ignored if a subsequent write succeeded. This is
unlikely but technically possible because sftp servers are
allowed to reorder requests.
* sftp(1): avoid a fatal() when sftp tab-completes filenames that
share common utf-8 characters that don't encode to a complete
codepoint.
* sshd(8): avoid a race condition when the sshd-auth process exits
tha could cause a spurious error message to be logged.
* sshd(8): log at level INFO when PerSourcePenalties actually
blocks access to a source address range. Previously this was
logged at level VERBOSE, which hid enforcement actions under
default config settings.
* sshd(8): GssStrictAcceptor was missing from sshd -T output; fix
* sshd(8): Make the MaxStartups and PerSourceNetBlockSize options
first-match-wins as advertised. bz3859
* ssh(1): fix an incorrect return value check in the local forward
cancellation path that would cause failed cancellations not to be
logged.
* sshd(8): make "Match !final" not trigger a 2nd pass ssh_config
parsing pass (unless hostname canonicalisation or a separate
"Match final" does). bz3843
* ssh(1): better debug diagnostics when loading keys. Will now list
key fingerprint and algorithm (not just algorithm number) as well
as making it explicit which keys didn't load.
* All: fix a number of memory leaks found by LeakSanitizer,
Coverity and manual inspection.
* sshd(8): : Output the current name for PermitRootLogin's
"prohibit-password" in sshd -T instead of its deprecated alias
"without-password". bz#3788
* ssh(1): make writing known_hosts lines more atomic by writing
the entire line in one operation and using unbuffered stdio.
Usually writes to this file are serialised on the "Are you sure you
want to continue connecting?" prompt, but if host key checking is
disabled and connections were being made with high concurrency
then interleaved writes might have been possible.
Portability
-----------
* sshd(8): check the username didn't change during the PAM
transactions.
PAM modules can change the user during their execution, but
this is not supported by sshd(8). If such a case was incorrectly
configured by the system administrator, then sshd(8) could end up
using a different username to the one authorised by PAM.
* sshd(8): don't log audit messages with UNKNOWN hostname to avoid
slow DNS lookups in the audit subsystem.
* All: when making a copy of struct passwd, ensure struct fields are
non-NULL. Android libc can return NULL pw_gecos, for example.
* All: Remove status bits from OpenSSL >=3 version check.
* sshd(8), ssh(1): Use SSH_TUN_COMPAT_AF on FreeBSD. Otherwise tun
forwarding from other OSes fails as soon as the first IPv6 message
is sent by the other side (which is usually a Router Solicitation
ICMPv6 message which is sent as soon as the interface is up).
* ssh(1), ssh-agent(8): check for nlist function presence before
attenmpting to use it instead of relying on the presence of the
nlist.h header. Mac OS X, in particular has the header, but only
has the function in the 32bit libraries.
* All: fill in missing system header files.
Create replacement header files inside openbsd-compat for common
headers that are missing on a given platform. Usually these are
just empty, but in some cases they'll include the equivalent file.
This avoids having to wrap those includes in '#ifdef HAVE_FOO_H'
and reduces the diff between Portable OpenSSH and OpenBSD.
* sshd(8): handle futex_time64 properly in seccomp sandbox
Previously we only allowed __NR_futex, but some 32-bit systems
apparently support __NR_futex_time64. We had support for this
in the sandbox, but because of a macro error only __NR_futex was
allowlisted.
* Add contrib/gnome-ssh-askpass4 for GNOME 40+ using the GCR API.
* sshd(8): let ga_init() fail gracefully if getgrouplist does.
Apparently getgrouplist() can fail on OSX when passed a
non-existent group name. Other platforms seem to return a group
list consisting of the numeric gid passed to the function. bz3848
* ssh-agent(1): exit 0 from SIGTERM under systemd socket-activation,
preventing a graceful shutdown of an agent via systemd from
incorrectly marking the service as "failed".
* build: wrap some autoconf macros in AC_CACHE_CHECK.
This allows skipping/overriding the OSSH_CHECK_CFLAG_COMPILE and
OSSH_CHECK_CFLAG_LINK macros used to discover supported compiler
or linker flags. E.g.
$ ./configure ossh_cv_cflag__fzero_call_used_regs_used=no
[...]
checking if cc supports compile flag -fzero-call-used-regs=used and linking succeeds... (cached) no
Reporting Bugs:
===============
- Please read https://www.openssh.com/report.html
Security bugs should be reported directly to openssh at openssh.com
OpenSSH is brought to you by Markus Friedl, Niels Provos, Theo de
Raadt, Kevin Steves, Damien Miller, Darren Tucker, Jason McIntyre,
Tim Rice and Ben Lindstrom.
More information about the openssh-unix-dev
mailing list