Re: [Cfrg] Attacker changing tag length in OCB

["David McGrew (mcgrew)" <[e-mail]>]

Hi Rene,

From: Rene Struik <[e-mail]<mailto:[e-mail]>>
Date: Thursday, May 30, 2013 3:59 PM
To: "Igoe, Kevin M." <[e-mail]<mailto:[e-mail]>>
Cc: "[e-mail]<mailto:[e-mail]>" <[e-mail]<mailto:[e-mail]>>
Subject: Re: [Cfrg] Attacker changing tag length in OCB

Dear colleagues:

The observation that some AEADs have the property that one can truncate the authentication tag without affecting AEAD unsecuring operations was brought up by Rogaway and Wagner in their 2003 "Critique on CCM" paper (Section 3.4). It is therefore kind of interesting to see that the same observation applies to the OCB mode.

Whether or not it is good to build-in protection against poor policy choices and management thereof is a matter of taste. History suggests that lots of security vulnerabilities arise precisely because of these poor choices (or lack of knowledge of impact of policy on security services offered). One could shield against poor choices in various ways, e.g., by using as key k:=Hash(K, TagLength).

Agreed.  The poor choice that is relevant here is the use of the same OCB secret key with two different tag lengths.   If someone uses a single key in both AEAD_AES_128_OCB_TAGLEN64 and AEAD_AES_128_OCB_TAGLEN128, then it is likely that they have also not satisfied the nonce uniqueness requirements, so the tag truncation and substitution attack seems like the least of their worries.   (Given two distinct nonce/associated data/plaintext tuples with identical nonces, it is easy for the attacker to craft forgeries.)

In general, it should be acceptable for an AE algorithm to require that each of the keys that are used to instantiate it are independent (as well as random or pseudorandom).

David



Best regards, Rene

Ref: CCM Mode Critique (Rogaway, Wagner,  IACR ePrint 2003-070)
http://eprint.iacr.org/2003/070

On 5/30/2013 3:21 PM, Igoe, Kevin M. wrote:
I agree with Richard.  The implicit assumption is that the length of the authentication tag has either been
agreed upon during the handshake or was a previously agreed upon parameter.  If you change the length
of the authentication tag in the live traffic, it almost certainly won’t pass authentication tag verification
by the intended recipient.  Offhand, I don’t see a security problem here.


From:[e-mail]<mailto:[e-mail]> [mailto:[e-mail]] On Behalf Of Richard Barnes
Sent: Wednesday, May 29, 2013 11:45 AM
To: Manger, James H
Cc: [e-mail]<mailto:[e-mail]>
Subject: Re: [Cfrg] Attacker changing tag length in OCB

James,

Don't most current AEAD modes have this property?  Namely, that there's no protection on the length of the authentication tag.  Off the top of my head, it seems that GCM, CCM, and SIV all have this property.

So while it might be nice for OCB to fix this issue, implementations processing AEAD messages will still have to enforce minimum tag lengths on their own.

--Richard


On Tue, May 28, 2013 at 8:47 PM, Manger, James H <[e-mail]<mailto:[e-mail]>> wrote:
>       Title           : The OCB Authenticated-Encryption Algorithm
>       Filename        : draft-irtf-cfrg-ocb-02.txt
> http://tools.ietf.org/html/draft-irtf-cfrg-ocb-02

OCB with tag lengths of 64, 96, and 128 bits are defined. 64-bit and 96-bit tags are simply truncated 128-bit tags. The tag length is not mixed into the ciphertext. It never affects any input to an AES operation.

Consequently, given a valid output from the AEAD_AES_128_OCB_TAGLEN128 algorithm it is trivial to produce a valid output from the AEAD_AES_128_OCB_TAGLEN64 algorithm -- just drop the last 8 bytes.

Is this ok?

Another consequence is that an attacker wanting to change the additional data (eg from saying "TOP SECRET" to "PUBLIC") while keeping the same plaintext only has to defeat the shortest tag a recipient accepts, regardless of the tag applied by the originator.

Is this ok? It doesn’t feel ok.

Of course if a recipient accepts 64-bit tags an attacker can forge messages with a probability of 2^-64. However, that doesn’t seem to be exactly the same as forging a message with the same plaintext as a message originally authenticated with a 128-bit tag.

Would OCB be better if the algorithms with different tag lengths couldn’t affect each other? Perhaps restricting the nonce to <126 bits (instead of <128 bits) and encoding the tag length in 2 bits.

--
James Manger
_______________________________________________
Cfrg mailing list
[e-mail]<mailto:[e-mail]>
http://www.irtf.org/mailman/listinfo/cfrg




_______________________________________________
Cfrg mailing list
[e-mail]<mailto:[e-mail]>http://www.irtf.org/mailman/listinfo/cfrg



--
email: [e-mail]<mailto:[e-mail]> | Skype: rstruik
cell: [phone] | US: [phone]