[ssh_config]
Make it clear that the example entries in ssh_config are only some of the
commonly-used options and refer the user to ssh_config(5) for more
details; ok djm@
monitor_wrap.c monitor_wrap.h session.c sshd.c]: Prepend all of the audit
defines and enums with SSH_ to prevent namespace collisions on some
platforms (eg AIX).
monitor.h monitor_wrap.c monitor_wrap.h session.c sshd.c] Bug #125:
(first stage) Add audit instrumentation to sshd, currently disabled by
default. with suggestions from and djm@
Bug #974: Teach sshd to write failed login records to btmp for failed auth
attempts (currently only for password, kbdint and C/R, only on Linux and
HP-UX), based on code from login.c from util-linux. With ashok_kovai at
hotmail.com, ok djm@
the process. Since we also unset KRB5CCNAME at startup, if it's set after
authentication it must have been set by the platform's native auth system.
This was already done for AIX; this enables it for the general case.
Make record_failed_login() call provide hostname rather than having the
implementations having to do lookups themselves. Only affects AIX and
UNICOS (the latter only uses the "user" parameter anyway). ok djm@
platforms syslog will revert to its default values. This may result in
messages from external libraries (eg libwrap) being sent to a different
facility.
[scp.c sftp.c]
Have scp and sftp wait for the spawned ssh to exit before they exit
themselves. This prevents ssh from being unable to restore terminal
modes (not normally a problem on OpenBSD but common with -Portable
on POSIX platforms). From peak at argo.troja.mff.cuni.cz (bz#950);
ok djm@ markus@
[auth-passwd.c sshd.c]
Warn in advance for password and account expiry; initialize loginmsg
buffer earlier and clear it after privsep fork. ok and help dtucker@
markus@
[cipher-ctr.c cipher.c]
remove fallback AES support for old OpenSSL, as OpenBSD has had it for
many years now; ok deraadt@
(Id sync only: Portable will continue to support older OpenSSLs)
[auth-bsdauth.c auth2-chall.c]
Have keyboard-interactive code call the drivers even for responses for
invalid logins. This allows the drivers themselves to decide how to
handle them and prevent leaking information where possible. Existing
behaviour for bsdauth is maintained by checking authctxt->valid in the
bsdauth driver. Note that any third-party kbdint drivers will now need
to be able to handle responses for invalid logins. ok markus@
"make survey" and "make send-survey". This will provide data on the
configure parameters, platform and platform features to the development
team, which will allow (among other things) better targetting of testing.
It's entirely voluntary and is off be default. ok djm@
from prngd is enabled at compile time but fails at run time, eg because
prngd is not running. Note that if you have prngd running when OpenSSH is
built, OpenSSL will consider itself internally seeded and rand-helper won't
be built at all unless explicitly enabled via --with-rand-helper. ok djm@
Darren Tucker