>change key_load_public to try to read a public from:
>rsa1 private or rsa1 public and ssh2 keys.
>this makes ssh-keygen -e fail for ssh1 keys more gracefully
>for example; report from itojun (netbsd pr 20550).
* The 'kex guesses' bugfix from OpenSSH 3.6 triggers a bug
in a few other SSH v2 implementations and causes connections to
stall. OpenSSH 3.6.1 disables this bugfix when interoperating
with these implementations.
* RSA blinding is now used by ssh(1), sshd(8) and ssh-agent(1).
in order to avoid potential timing attacks against the RSA keys.
Older versions of OpenSSH have been using RSA blinding in
ssh-keysign(1) only.
* ssh-agent(1) optionally requires user confirmation if a key gets
used, see '-c' in ssh-add(1).
* sshd(8) now handles PermitRootLogin correctly when UsePrivilegeSeparation
is enabled.
* sshd(8) now removes X11 cookies when a session gets closed.
* ssh-keysign(8) is disabled by default and only enabled if the
new EnableSSHKeysign option is set in the global ssh_config(5)
file.
* ssh(1) and sshd(8) now handle 'kex guesses' correctly (key exchange
guesses).
* ssh(1) no longer overwrites SIG_IGN. This matches behaviour from
rsh(1) and is used by backup tools.
* setting ProxyCommand to 'none' disables the proxy feature, see
ssh_config(5).
* scp(1) supports add -1 and -2.
* scp(1) supports bandwidth limiting.
* sftp(1) displays a progressmeter.
* sftp(1) has improved error handling for scripting.
hp = strtok(NULL, ",");
to end of while loop which parses "host.allow" from login.conf(5).
Otherwise, sshd(8) would just infinite loop unless there was a
(positive or negative) match in the first word of the "host.allow" list...
to debug (on the server) why a login failed with this information.
* If _PASSWORD_CHGNOW is defined (it's -1 in NetBSD), check that pw_change
is not set to that before testing if the password has expired.
Still prevent the login, but log a different failure message in this case.
XXX: we need to decide if we let interactive logins occur in this case,
but force a password change, a la login(1).
via timing by performing a MAC computation even if incorrect
block cipher padding has been found. This is a countermeasure
against active attacks where the attacker has to distinguish
between bad padding and a MAC verification error. (CAN-2003-0078)
itojun