Pending further investigation, we now allege that Ethan Benson wrote: > On Wed, Dec 20, 2000 at 12:01:51PM -0500, Andy Bastien wrote: > > [snip] > > NTLMv2 improves on NTLMv1 by going from an 56 bit MD4 hash to a 128 > > bit MD5 hash. NTLMv2 is very difficult to attack with a brute-force > > your still missing the point, NT still just sends this md5 hash (which > is think is still unsalted) over the network where its used _as_ the > password. that is the hash is simply compared with the hash stored in > the password database. so if you sniff the hash you can use it to > authenticate yourself against an NT server, even if you *don't* know > the actual password. (this only requires a slightly modified > smbclient utility)
That only works if the password is less than eight characters. This because there are two session keys returned by the server. One is the nthash, and the other is half of the lmhash. If your password is no more than seven characters, the whole thing fits into that half of the lmhash and you can use it to authenticate in the future. The nthash does you very little good. Of course, you can disable lmhashes on the server and fix this problem. That's what I was referring to when I mentioned that registry key. Let's not forget that NFS can be subject to address spoofing. You can sniff telnet, POP3, HTTP, and FTP sessions to get plaintext usernames and passwords. All of these exploits are at least as bad as the lmhash thing with SMB. There are other ways to get around the security that SMB provides. You can hijack sessions, or pretend to be a DC, for example. I would never say that it's perfect, but just because it is possible to crack it doesn't mean you should throw away the security that it does provide.
