Home page logo
/

pen-test logo Penetration Testing mailing list archives

Re: Passwords with Lan Manager (LM) under Windows
From: "Thor (Hammer of God)" <thor () hammerofgod com>
Date: Thu, 22 Sep 2005 01:30:01 -0700

Last one-- I already said why you can't pre-compile NTLMv2: The hash generated for the challenge/response exchange not only includes the password, but user/domain data as well.

If my password was "I'mNotGoingToArgueAboutThisAnymore!" then that could be pre-compiled into a rainbow table for NTLM (if you actually went that far out.) But with NTLMv2, the hash is generated from "I'mNotGoingToArgueAboutThisAnymore!" plus usernameATdomainDOTcom. You would have to precompile every possible password combination with every possible user/domain name. Ain't gonna happen.

And I'm not sure where you are getting your info regarding Microsoft "dropping NTLMv2 for backward compatibility." Backward compatibility is already there! Regardless, the new settings I have for LanMan auth levels in addition to the NTLM SSP settings for both client and server that I currently have configured in my install of Longhorn say that you are mistaken...

Not to be rude, but I've said all I have to say on the subject, and don't see any point in bickering... I'm available off-line if you want to discuss this any further.

t

----- Original Message ----- From: "Craig Wright" <cwright () bdosyd com au> To: "Thor (Hammer of God)" <thor () hammerofgod com>; <pand0ra.usa () gmail com>; <pen-test () securityfocus com>
Sent: Wednesday, September 21, 2005 11:41 PM
Subject: RE: Passwords with Lan Manager (LM) under Windows


First "You can't precompile that data into a rainbow, you know?". Please
explain why not. HMAC-MD5 message authentication and MD4 are used to
create the challenge. There are rainbow tables for both of these - yes
they take more time to search the challenges but they exist. Publicly
available MD4 rainbow tables are still in early development and mostly
only cover "alpha numeric"...but they exist.

Yes (before you state that this does not supply an attacker with the
password itself) I do understand that using a rainbow table against the
challenge hash does not give you the password itself. But it can give
you access to the system. Personally what I care about is if the system
is accessed not if the password is known by the attacker. If I have
access to the system I can do damage. Install a keylogger for example if
all I wanted was the password.

"Well, that's an issue with the client"...well we are talking a client
server environment. Both are necessary. Yes IBM created Lanman, I read
the RFC when they first submitted it, but MS implemented it, not IBM. PS
to this, MSFT are dropping NTLMv2 at the expense of causing issues with
backward compatibility... Please explain why if "NTLMv2 is tight". This
is a BIG step for MSFT and one I do really applaud.

***---***---*** Gratuitous MSFT Quote ***---***---***

"All three algorithms (MD4, MD5 and DES - my my... NTLMv2) show signs of
"extreme weakness" and have been banned", Michael Howard - Microsoft.

***---***---*** Gratuitous MSFT Quote ***---***---***

As I said earlier "Kerberos support with IPsec" And by this yes
certificates are the best. I still stick by IPsec based auth as I did
state in an earlier post "Kerberos support with IPsec" which is a valid
option as I originally stated it, the alternative being Kerberos without
IPsec. IPsec is used in this case to protect the traffic between the
systems.

Why authentication as a term for IPsec. Because AH is used to negotiate
between the client and the server. (and I also know that the domain
controllers do not use IPsec to send encrypted Kerberos info to each
other and that this needs to be taken into account in the design and
placement of AD domain systems).

Though preshared keys are supported - certificates are better. These are
used to setup the IPsec session which is used to protect the Kerberos
key exchange.

Thus the first stage - before the initial "TGS request for a ticket" is
to have encryption between the client and the AD server thus further
protecting the process. AN IPsec session is later used from the Kerberos
client to the application server (and yes I have not gone into the
stages used by the TGS and AS, have just passed over the TGT, not even
mentioned the 3 subprotocols used etc) protecting the ticket exchange.

Craig

PS - I know that this is simplified and that I could expand this further
(into a few hundred page paper) to cover all the stages and miss
nothing.

PPS I have not even gone into collision analysis against MD5 or MD4 both
of which are viable options

-----Original Message-----
From: Thor (Hammer of God) [mailto:thor () hammerofgod com]
Sent: 22 September 2005 3:46
To: Craig Wright; pand0ra.usa () gmail com; pen-test () securityfocus com
Subject: Re: Passwords with Lan Manager (LM) under Windows

Well, that's an issue with the client, not NTLMv2.  NTLMv2 is tight.  LM
sucks- that's obvious (and it was IBM, not MS that gave us that one.)
And yes, you can use precomputed tables against NTLM hashes, but not
against NTLMv2... The NTLM hash is keyed off of the password, but NTLMv2
hashes up the password with the user's domain/user data when generating
the key...
You can't precompile that data into a rainbow, you know?

Regarding the "IPsec based auth" reference (here I go again), I'd have
to say that there is no such thing... IPSec negotiation in Windows can
be based on one of three mechanisms:  A pre-shared key, Kerberos, or a
cert-- it is not an authentication protocol in itself... (the cert being
the strongest IMO).

t


----- Original Message -----
From: "Craig Wright" <cwright () bdosyd com au>
To: "Thor (Hammer of God)" <thor () hammerofgod com>;
<pand0ra.usa () gmail com>; <pen-test () securityfocus com>
Sent: Wednesday, September 21, 2005 10:05 PM
Subject: RE: Passwords with Lan Manager (LM) under Windows


Further to the last post
There are a number of issues with NTLMv2 and legacy applications such as
Windows RAS that cause lower levels of authentication

I still say that Kerberos or IPsec based auth is the best policy in
windows. LanMan, NTLMv1 or V2 are vulnerable.

Precomputed tables may have been uncommon 12 months ago - but that was
then and this is now.

Cain & Abel will use sorted Rainbow Tables for Cryptanalysis attacks

Craig

-----Original Message-----
From: Thor (Hammer of God) [mailto:thor () hammerofgod com]
Sent: 22 September 2005 12:00
To: Craig Wright; pand0ra.usa () gmail com; pen-test () securityfocus com
Subject: Re: Passwords with Lan Manager (LM) under Windows


----- Original Message -----
From: "Craig Wright" <cwright () bdosyd com au>
To: <pand0ra.usa () gmail com>; <pen-test () securityfocus com>
Sent: Wednesday, September 21, 2005 12:32 PM
Subject: RE: Passwords with Lan Manager (LM) under Windows


Even NTLMv2  will break the hashing into chunks which are able to be
individually broken down.

I'm not sure what you mean... NTLMv2 uses a single 128bit key for the
hash, challenge and response...  Or are you referring to the NTLM2
session response key (56+56+16)? If so, that is not the same thing as
NTLMv2...
Can
you elaborate please ?

t






------------------------------------------------------------------------------
Audit your website security with Acunetix Web Vulnerability Scanner: Hackers are concentrating their efforts on attacking applications on your website. Up to 75% of cyber attacks are launched on shopping carts, forms, login pages, dynamic content etc. Firewalls, SSL and locked-down servers are futile against web application hacking. Check your website for vulnerabilities to SQL injection, Cross site scripting and other web attacks before hackers do! Download Trial at:

http://www.securityfocus.com/sponsor/pen-test_050831
-------------------------------------------------------------------------------


  By Date           By Thread  

Current thread:
[ Nmap | Sec Tools | Mailing Lists | Site News | About/Contact | Advertising | Privacy ]
AlienVault