Re: [OAUTH-WG] A review of draft-ietf-oauth-pop-architecture-05

Erik Wahlström neXus <> Thu, 19 November 2015 16:39 UTC

Return-Path: <>
Received: from localhost ( []) by (Postfix) with ESMTP id 528571B2C4E for <>; Thu, 19 Nov 2015 08:39:44 -0800 (PST)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -2.884
X-Spam-Status: No, score=-2.884 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, MIME_8BIT_HEADER=0.3, RCVD_IN_DNSWL_LOW=-0.7, RP_MATCHES_RCVD=-0.585] autolearn=ham
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id 3AHiGtXgSj-Y for <>; Thu, 19 Nov 2015 08:39:40 -0800 (PST)
Received: from ( []) (using TLSv1 with cipher ECDHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by (Postfix) with ESMTPS id 130491B2C3D for <>; Thu, 19 Nov 2015 08:39:40 -0800 (PST)
Received: from ( by ( with Microsoft SMTP Server (TLS) id 15.0.995.29; Thu, 19 Nov 2015 17:39:37 +0100
Received: from ([fe80::1d3d:b319:f020:2bab]) by ([fe80::1d3d:b319:f020:2bab%12]) with mapi id 15.00.0995.032; Thu, 19 Nov 2015 17:39:38 +0100
From: Erik Wahlström neXus <>
To: Phil Hunt <>
Thread-Topic: [OAUTH-WG] A review of draft-ietf-oauth-pop-architecture-05
Thread-Index: AQHRIqsWCr6RmlnVq02omaMFbp3lkZ6jdZCAgAAW25Q=
Date: Thu, 19 Nov 2015 16:39:37 +0000
Message-ID: <>
References: <>, <>
In-Reply-To: <>
Accept-Language: en-US, sv-SE
Content-Language: en-US
Content-Type: multipart/alternative; boundary="_000_BE468D61B1164EB58ECBED3A21D846FBnexusgroupcom_"
MIME-Version: 1.0
Archived-At: <>
Cc: "<>" <>
Subject: Re: [OAUTH-WG] A review of draft-ietf-oauth-pop-architecture-05
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: OAUTH WG <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Thu, 19 Nov 2015 16:39:44 -0000

Just a note then. I did not see anything that prohibited the usage of pop tokens for IoT so  shipping it as is works.

Sent from my iPhone

On 19 Nov 2015, at 17:18, Phil Hunt <<>> wrote:

On the subject of making the spec(s) less JWT specific, it was a foundational assumption and (I think) in the charter. However COSE wasn't around yet.

I suppose the more generic architecture doc could be altered to cover IoT cases, but it may be problematic for the other specs that are more specific.

Another issue is I assume the COSE based tokens will have a different type (eg cpop) to differentiate between jwt and COSE web tokens (what are we calling them now?).

As this generalization change could be seen as substantial, I would like to have the chairs and AD comment. Is this a good idea?  Or is COSE better to write their own parallel arch draft?

I'm happy to bend to the will of the group(s) on this.


On Nov 19, 2015, at 01:17, Erik Wahlström neXus <<>> wrote:


I have been reviewing draft-ietf-oauth-pop-architecture-05. In ACE WG we have a draft that uses PoP tokens for IoT and the architectures defined here so my review was done with that IoT perspective. I’m a bit late with the review and some of the comments might already be mentioned by others.


3.1. Preventing Access Token Re-Use by the Resource Server

If a symmetric key is used it’s possible to re-use the key for a resource server. The section talks about the importance of scopes, but I feel it should also mention the importance for the resource server to verify the audience (“aud”) claim in the token to disable missuse.


The draft in ACE WG ( relies heavily on this work. The main reason for this is the way PoP tokens can establish key material, with the help of the authorization server, on both the client and resource server. PoP tokens is also a very good fit for constrained IoT devices that can be offline and it’s also possible to use hardware key storages to handle asymmetric pop keys.

There could be a place for a new "Use Case" under section 3 that talks about scenarios where PoP keys are a really good match for offline IoT devices. I could help out ironing out a text for that with the help of the docs authors if that’s of interest.


s/a bogus tokens/a bogus token


In the document only references are made to JSON, JWT and JOSE. More exactly in the following two sections:

   A number of the threats listed in Section 4<> demand protection of the
   access token content and a standardized solution, in form of a JSON-
   based format, is available with the JWT [RFC7519<>].

   With the JSON Web Token (JWT) [RFC7519<>] a standardized format for
   access tokens is available.  The necessary elements to bind symmetric
   or asymmetric keys to a JWT are described in

Constrained IoT devices uses other access token and messages formats (according to our draft). It does not only use signed/encrypted JWT’s but also COSE protected CBOR Web Tokens. See

I totally agree that JWT is the correct examples to have in this document due to the fact that they are RFC’s, they are well known and should be used in as many places as possible, but it would be good to open up for other types of message formats. For example like this:

   A number of the threats listed in Section 4<> demand protection of the
   access token content and a standardized solution in form of, for example, a JSON-
   based format, is available with the JWT [RFC7519<>].


 For that
   purpose the client will have to authenticate the resource server
   before transmitting the access token.

I’m missing a description about how this is handled in an end-to-end security scenario.


      The resource server queries the authorization server for the
      symmetric key.  This is an approach envisioned by the token
      introspection endpoint [I-D.ietf-oauth-introspection<>].

Not a question for this draft maybe, but in what draft is the introspection response claim defined? It’s not defined in and I don’t know in what other draft it can be defined.


In ACE WG the draft seitz-ace-oauth-authz have a profile for an access request to make it work over CoAP. CoAP is the HTTP equivalent for constrained devices, and it has limitations, for example it can’t send large tokens in options (headers in "http-speak"). This means that the draft defines a way to first send the PoP token to an new endpoint on the resource server to establish a security context. Then the real request against the resource server can be done once the security context is established. See more details here

An open question; should a flow like that be added to the architecture section? That means a new section 7.5.


Thanks for writing this. I think it’s very important work.

/ Erik

OAuth mailing list<>