FreeTDS Development Group

[Peter Deacon <peterd AT iea-software.com>]

On Fri, 1 Aug 2008, Frediano Ziglio wrote:

> Il giorno gio, 31/07/2008 alle 18.25 -0400, Sergey Sikorskiy ha scritto:
> > Hi,

> > I've just checked with Microsoft web-site and found detailed
> > documentation for NTLM.
> > http://msdn.microsoft.com/en-us/library/cc207842.aspx

> > This documentation clearly says that "(1.7) "The NTLM authentication
> > version is not negotiated by the protocol. It must be configured on
> > both the client and the server prior to authentication. The version is
> > selected by the client, and requested during the protocol negotiation.
> > If the server does not support the version selected by the client,
> > authentication fails."

> > The rest is negotiable, including sending of LMv2 hash, which won't be
> > sent in my implementation.

> yes, but I think that client should say: I support all these and server
> reply, you can use such... If we support all versions we should be
> always able to negotiate the correct (most secure) authentication.
> I think that is NTLM2 is negotiated either NTLMv2 or NTLM2 should be
> used. At this point I think that if we have valid Target Info NTLMv2
> should be negotiated...

IMHO with these sorts of things where there is no serious machinary to 
perform secure protocol negotiation both client and server side 
configuration is required to prevent attacks which can cause less secure 
protocols being negotiated in order to take advantage of a weakness.

For example v2 supports mutual authentication, earlier protocols did not. 
If your able to MITM the connection and negotiate a protocol without 
mutual auth ... there is little point remaining in having v2 as its 
enhanced security properties can simply be bypassed.

take care,
Peter