the parts of the grace_alarm_handler() using the exact things allowed by the
signal-safe rules. This is a good rule of thumb: Handlers should be written
to either set a global volatile sig_atomic_t inspected from outside, and/or
directly perform only safe operations listed in our sigaction(2) manual page.
ok djm markus
OpenBSD-Commit-ID: 14168ae8368aab76e4ed79e17a667cb46f404ecd
Cygwin doesn't support FD passing and so used to disable post-auth
privilege separation entirely because privsep requires PTY allocation
to happen in the privileged monitor process with the PTY file
descriptors being passed back to the unprivileged process.
This brings back a minimal version of the previous special treatment
for Cygwin (and any other platform that sets DISABLE_FD_PASSING):
privilege separation remains enabled, but PTY allocation happens in
the post-auth user process rather than the monitor.
This either requires PTY allocation to not need privilege to begin
with (this appears to be the case on Cygwin), or the post-auth
privsep process retain privilege (other platforms that set the
DISABLE_FD_PASSING option).
Keeping privileges here is bad, but the non-Cygwin systems that set
DISABLE_FD_PASSING are so deeply legacy that this is likely to be the
least of their problems.
sshd-session attempting to use options.kerberos_authentication to
decide whether it needed to lookup the privsep user before the
configuration was loaded. This caused it to get a placeholder value
that caused it always to try to lookup the privsep user, breaking at
least one test environment.
problematic client behaviours, controlled by two new sshd_config(5) options:
PerSourcePenalties and PerSourcePenaltyExemptList.
When PerSourcePenalties are enabled, sshd(8) will monitor the exit
status of its child pre-auth session processes. Through the exit
status, it can observe situations where the session did not
authenticate as expected. These conditions include when the client
repeatedly attempted authentication unsucessfully (possibly indicating
an attack against one or more accounts, e.g. password guessing), or
when client behaviour caused sshd to crash (possibly indicating
attempts to exploit sshd).
When such a condition is observed, sshd will record a penalty of some
duration (e.g. 30 seconds) against the client's address. If this time
is above a minimum threshold specified by the PerSourcePenalties, then
connections from the client address will be refused (along with any
others in the same PerSourceNetBlockSize CIDR range).
Repeated offenses by the same client address will accrue greater
penalties, up to a configurable maximum. A PerSourcePenaltyExemptList
option allows certain address ranges to be exempt from all penalties.
We hope these options will make it significantly more difficult for
attackers to find accounts with weak/guessable passwords or exploit
bugs in sshd(8) itself.
PerSourcePenalties is off by default, but we expect to enable it
automatically in the near future.
much feedback markus@ and others, ok markus@
OpenBSD-Commit-ID: 89ded70eccb2b4926ef0366a4d58a693de366cca
binaries. This step splits sshd into a listener and a session binary. More
splits are planned.
After this changes, the listener binary will validate the configuration,
load the hostkeys, listen on port 22 and manage MaxStartups only. All
session handling will be performed by a new sshd-session binary that the
listener fork+execs.
This reduces the listener process to the minimum necessary and sets us
up for future work on the sshd-session binary.
feedback/ok markus@ deraadt@
NB. if you're updating via source, please restart sshd after installing,
otherwise you run the risk of locking yourself out.
OpenBSD-Commit-ID: 43c04a1ab96cdbdeb53d2df0125a6d42c5f19934
deraadt@openbsd.org
Damien Miller
djm@openbsd.org
jsg@openbsd.org