Re: [secdir] Secdir review of draft-ietf-jose-json-web-signature-31

[Richard Barnes <[e-mail]>]

"""
# Signature Algorithm Protection

In some usages of JWS, there is a risk of algorithm substitution attacks,
in which an attacker can use an existing signature value with a different
signature algorithm to make it appear that a signer has signed something
that he actually has not.  These attacks have been discussed in detail in
the context of CMS {{RFC 6211}}.  The risk arises when all of the following
are true:

* Verifiers of a signature support multiple algorithms of different
strengths
* Given an existing signature, an attacker can find another payload that
produces the same signature value with a weaker algorithm
* In particular, the payload crafted by the attacker is valid in a given
application-layer context

For example, suppose a verifier is willing to accept both "PS1" and "PS256"
as "alg" values, and a signer creates a signature using "PS256".  If the
attacker can craft a payload that has the same SHA-1 digest has as the
SHA-256 digest of the legitimate payload, then the "PS1" signature over the
bogus payload will be the same as the "PS256" signature over the legitimate
payload.

There are several ways for an application using JOSE to mitigate algorithm
substitution attacks:

* Don't accept signatures using vulnerable algorithms: Algorithm
substitution attacks do not arise for all signature algorithms.
  * Signatures using RSA PKCS#1 v1.5 ("RS1", "RS256", etc.) are not subject
to substitution attacks because the signature value itself encodes the hash
function used.
  * Signatures with HMAC algorithms ("HS1", "HS256", etc.) cannot be
substituted because the signature values have different lengths  Likewise
for signatures with ECDSA algorithms ("ES256", "ES384", etc.).
  * The only algorithms defined in JWA
{{I-D.ietf-jose-json-web-algorithms}} that is vulnerable to algorithm
substitution attacks is RSA-PSS ("PS1", "PS256", etc.).  An implementation
that does not support RSA-PSS is not vulnerable to algorithm substitution
attacks.

* Require that the "alg" parameter be carried in the protected header.
(This is the approach taken by RFC 6211.)

* Include a field reflecting the algorithm in the application payload, and
require that it be matched with the "alg" parameter during verification
(This is the approach taken by PKIX {{RFC5280}}.)

Of these mitigations, the only sure solution is the first.  Signing over
the "alg" parameter (directly or indirectly) only makes the attacker's work
more difficult, by requiring that the bogus payload also contain bogus
information about the signing algorithm.  They do not prevent attack by a
sufficiently powerful attacker.
"""

On Fri, Sep 19, 2014 at 2:49 PM, Mike Jones <[e-mail]>
wrote:

>  I would appreciate it if you would write a draft of the proposed
> security considerations text, Richard.  Perhaps title the section
> “Unsecured Algorithm Values”.
>
>
>
>                                                             Thanks!
>
>                                                             -- Mike
>
>
>
> *From:* Richard Barnes [mailto:[e-mail]]
> *Sent:* Wednesday, September 17, 2014 6:24 AM
> *To:* Tero Kivinen
> *Cc:* Mike Jones; [e-mail]; [e-mail]; [e-mail];
> [e-mail]; [e-mail]
> *Subject:* Re: Secdir review of draft-ietf-jose-json-web-signature-31
>
>
>
>
>
> On Wednesday, September 17, 2014, Tero Kivinen <[e-mail]> wrote:
>
> Richard Barnes writes:
> >     Perhaps, but is there benefits for leaving the alg without
> protection?
> >
> > Simplicity (if you omit protected headers altogether), and
> > compatibility with other signed things.  In the sense that you could
> > transform one of them into a JWS without re-signing.  This would
> > apply, for example, to an X.509 certificate -- just parse the outer
> > SEQUENCE, and re-assemble into a JWS with the tbsCertificate as
> > payload.  Same security properties that X.509 already has.
>
> Ok, having this kind of information somewhere in the draft would help
> to understand the reason. Also having text explaining that is
> possible, and that the security properties of this option (i.e. no
> problem with PKCS#1, etc... the text you had in the other email).
>
> > It's also completely unnecessary for PKCS#1 signatures, which are
> > the dominant use case today.
>
> I agree.
>
> > In general, I'm opposed to protocols baking in more
> > application-specific logic than they need to.  The point of JOSE is
> > to describe the cryptographic operation that was performed, and
> > carry the relevant bits around.  Its job is not to fix all the
> > weaknesses that every algorithm has.
>
> Yes, but this property might have security issues, so they should be
> covered by the security considerations section.
>
>
>
> I'm perfectly happy to have it documented in the Security Considerations.
>
>
>
> Mike: Should I generate some text, or do you want to take a stab?
>
>
>
> --
> [e-mail]
>
>