Re: [OAUTH-WG] Call for Adoption - AS Issuer Identifier in Authorization Response

[Warren Parad <[e-mail]>]

Since there is a potentially valid TLS case, let's shelve the non-TLS case
without the precondition. I agree a MITM attack on the authorization code
is a legitimate case which would be mitigated by resolving the ISS
parameter to determine the valid token endpoint. I would suggest to improve
the language in the *Introduction* to specifically indicate this as a
solution to the authorization code interception. (It wasn't clear to me
before this conversation that it helped solved that problem)

So +1

On Wed, Dec 9, 2020, 11:40 Karsten Meyer zu Selhausen <
[e-mail]> wrote:

> The attacker being able to manipulate the first request is an additional
> precondition for the mix-up attack variant we used as an example. The
> precondition is based on the attacker A2 defined in section 3
> <https://tools.ietf.org/html/draft-ietf-oauth-security-topics-16#section-3>
> of the security BCP.
>
> There are other mix-up variants which work without this precondition. One
> variant is described in the security BCP
> <https://tools.ietf.org/html/draft-ietf-oauth-security-topics-16#section-4.4.1>,
> for example:
>
> *Mix-Up Without Interception*: A variant of the above attack works
>       even if the first request/response pair cannot be intercepted, for
>       example, because TLS is used to protect these messages: Here, it
>       is assumed that the user wants to start the grant using A-AS (and
>       not H-AS, see Attacker A1).  After the client redirected the user
>       to the authorization endpoint at A-AS, the attacker immediately
>       redirects the user to H-AS (changing the client ID to "7ZGZldHQ").
>       Note that a vigilant user might at this point detect that she
>       intended to use A-AS instead of H-AS.
>
>
> On 09.12.2020 10:47, Warren Parad wrote:
>
> Okay, it wasn't clear that the user agent was required to be compromised
> for this to be a problem. Here's where it breaks down for me, if the
> attacker can manipulate the first request, why would they not be able to
> manipulate the AS where the Auth Response code is sent?  Unless we can
> guarantee there is an attack surface that would only affect the
> authorization request AS selection and not the auth response, the solution
> the draft lacks purpose for me.
>
> Warren Parad
>
> Founder, CTO
> Secure your user data and complete your authorization architecture.
> Implement Authress <https://bit.ly/37SSO1p>.
>
>
> On Wed, Dec 9, 2020 at 8:55 AM Karsten Meyer zu Selhausen <
> [e-mail]> wrote:
>
>> Hi Warren,
>>
>> I think there is some misunderstanding on how mix-up attacks work. I will
>> try to clear things up.
>>
>> Have a look at the following mix-up attack example (slide 4
>> <https://datatracker.ietf.org/meeting/interim-2020-oauth-17/materials/slides-interim-2020-oauth-17-sessa-as-issuer-identifier-in-authorization-response-00#page=4>
>> from the interim meeting):
>>
>> I marked the important parts:
>>
>>    - In step 1 the client stores the attacker's AS (A-AS) as the
>>    selected AS.
>>    - Step 5: The authorization response is issued by the honest *(= not
>>    compromised)* AS, not by the attacker's AS. The H-AS will use its own
>>    correct issuer identifier as the value for the AS parameter.
>>       - In a mix-up attack the attacker cannot directly influence the
>>       value of the iss parameter in the authorization response as it is issued by
>>       the H-AS.
>>    - Step 6: The client sends the token request to the token endpoint of
>>    the A-AS, because it stored the A-AS as the selected AS in step 1. This
>>    leaks the authorization code to the attacker who can use it in a code
>>    injection attack, for example.
>>
>> With an iss parameter present in step 5 the client would be able to
>> recognize that the code was issued by the H-AS, not by the A-AS. The client
>> would be able to abort the authorization grant instead of leaking the code
>> to the A-AS.
>>
>> I hope this addresses your concerns.
>>
>> Best regards,
>> Karsten
>> On 08.12.2020 20:15, Warren Parad wrote:
>>
>> As an implementer on both sides of the issue I'm struggling to understand
>> how this problem would occur. I'm finding issues with the proposed
>> problems:
>>
>>    1. Honest AS is compromised, assuming this does happen details on why
>>    adding iss to the AS response would prevent attacks is necessary for me. In
>>    other words, how would an AS be compromised in a way that would be
>>    identifiable through the issuer value? (my ignorant assumption is that a
>>    compromised AS is compromised enough that an attacker would be able to send
>>    the correct ISS)
>>    2. Attacker AS is registered. I fully support the idea that this can
>>    and will happen, however from attempting to test-implement this proposal, I
>>    can't see how the authorization would be sent to the wrong token endpoint.
>>    Since there is no information in the AS auth code response, the client must
>>    already have the knowledge of where they are going to send the token, no
>>    mix-up can be executed. I would argue, if anything, adding the ISS
>>    parameter would open a new attack surface by providing clients an
>>    opportunity to blatantly trust the ISS parameter as the honest AS and thus
>>    actually sending the code there instead of sending it to one specified in
>>    the metadata document.
>>
>> My confusion is the following:
>>
>>    - Are multi AS services utilizing authorization codes in a way where
>>    there could be a mix up attack for #2.
>>    - Is there a #3 that I'm missing which even in light of #1 & #2 I
>>    brought up that would still make this change valuable?
>>
>> Warren Parad
>>
>> Founder, CTO
>> Secure your user data and complete your authorization architecture.
>> Implement Authress <https://bit.ly/37SSO1p>.
>>
>>
>> On Tue, Dec 8, 2020 at 8:01 PM Dick Hardt <[e-mail]> wrote:
>>
>>> +1
>>> ᐧ
>>>
>>> On Tue, Dec 8, 2020 at 4:51 AM Rifaat Shekh-Yusef <
>>> [e-mail]> wrote:
>>>
>>>> All,
>>>>
>>>> This is a call for adoption for the following AS Issuer Identifier in
>>>> Authorization Response as a WG document:
>>>>
>>>> https://datatracker.ietf.org/doc/draft-meyerzuselhausen-oauth-iss-auth-resp/
>>>>
>>>> Please, provide your feedback on the mailing list by Dec 22nd.
>>>>
>>>> Regards,
>>>>  Rifaat & Hannes
>>>> _______________________________________________
>>>> OAuth mailing list
>>>> [e-mail]
>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>
>>> _______________________________________________
>>> OAuth mailing list
>>> [e-mail]
>>> https://www.ietf.org/mailman/listinfo/oauth
>>>
>> --
>> Karsten Meyer zu Selhausen
>> IT Security Consultant
>> Phone:	+49 (0)234 / 54456499
>> Web:	https://hackmanit.de | IT Security Consulting, Penetration Testing, Security Training
>>
>> Nehmen Sie an unserer nächsten Live Online-Schulung zur Sicherheit von OAuth und OpenID Connect am 27.01 + 28.01.2021 teil:https://www.hackmanit.de/de/schulungen/127-live-online-schulung-single-sign-on-sicherheit-oauth-openid-connect-am-27-01-28-01-2021
>>
>> Hackmanit GmbH
>> Universitätsstraße 60 (Exzenterhaus)
>> 44789 Bochum
>>
>> Registergericht: Amtsgericht Bochum, HRB 14896
>> Geschäftsführer: Prof. Dr. Jörg Schwenk, Prof. Dr. Juraj Somorovsky, Dr. Christian Mainka, Dr. Marcus Niemietz
>>
>> --
> Karsten Meyer zu Selhausen
> IT Security Consultant
> Phone:	+49 (0)234 / 54456499
> Web:	https://hackmanit.de | IT Security Consulting, Penetration Testing, Security Training
>
> Nehmen Sie an unserer nächsten Live Online-Schulung zur Sicherheit von OAuth und OpenID Connect am 27.01 + 28.01.2021 teil:https://www.hackmanit.de/de/schulungen/127-live-online-schulung-single-sign-on-sicherheit-oauth-openid-connect-am-27-01-28-01-2021
>
> Hackmanit GmbH
> Universitätsstraße 60 (Exzenterhaus)
> 44789 Bochum
>
> Registergericht: Amtsgericht Bochum, HRB 14896
> Geschäftsführer: Prof. Dr. Jörg Schwenk, Prof. Dr. Juraj Somorovsky, Dr. Christian Mainka, Dr. Marcus Niemietz
>
>