> What exactly is the impact of this vulnerability?
It was first introduced on 2006-09-17 in Debian unstable.
If your key-pair was generated on a Debian or derivative system it must be
regenerated. If a DSA key was used on an affected system it must be
regenerated. see: http://www.debian.org/security/key-rollover
While keys generated with GnuPG or GNUTLS are not effected if they were used
for signing or authentication on an affected system they should be
regenerated. Make new key-pairs, sign with old keys, revoke old keys.
> 1) Does it let a attacker, who has listening ability on a local
> network, to intercept keys? (ie reduce security of SSH to that of telnet)
No. An attacker can not compromise the system just by sniffing traffic.
When a public key is available a bruteforce against how the private key was
generated is possible. When a client connects to a host it receives a copy
of the public key. Any one who can connect to an affected host or listen to
the connection, even if they can't log on, could break the keys by
bruteforce attacking the badly limited entropy pool used to generte the
keys instead of the keys themselves. An attacker may then impersonate the
Personal keys generated on, and or DSA keys used from, an affected system
are also vulnerable.
> 2) Does it allow an attacker, who does NOT have a listening ability,
> to log on to remote machines using known weak keys? (ie brute force a
> fully remote machine)
No, but they may be able to compromise the host key and impersonate the
host. Also DSA keys used from affected systemsmay be able to be
> Just what is the extent of this sad story?
> As I use ssh and keys a lot, this means that I had to spend a lot of
> time fixing all the trust network that I have. I think that I am done,