Re: [OAUTH-WG] draft-parecki-oauth-browser-based-apps-00

[Torsten Lodderstedt <[e-mail]>]

Hi Brock,

> Am 15.11.2018 um 23:01 schrieb Brock Allen <[e-mail]>:
> 
> > It still lacks the ability to issue sender constraint access tokens.
> 
> So you mean at the resource server ensuring the token was really issued to the client? Isn't that an inherent limitation of all bearer tokens (modulo HTTP token binding, which is still some time off)?

Sure. That’s why the Security BCP recommends use of TLS-based methods for sender constraining access tokens (https://tools.ietf.org/html/draft-ietf-oauth-security-topics-09#section-2.2). Token Binding for OAuth (https://tools.ietf.org/html/draft-ietf-oauth-token-binding-08) as well as Mutual TLS for OAuth (https://tools.ietf.org/html/draft-ietf-oauth-mtls-12) are the options available. 


> Resource servers don't know the flow the clients might use, especially if/when they have many clients.

They don’t need to. All they need to know is how to determine whether the sender of the token is the legit client. This is achieved by comparing the hash of the token binding id or the cert of the client conveyed in the access token with the respective data from the TLS handshake. 

> 
> > The AS can bind the lifetime of the refresh tokens to the session lifetime, i.e. automatically revoke it on logout. 
> 
> Yea, I saw your other email asking about refresh token revocation relating to session management. Obviously for certain clients, this won't make sense, but for implicit/browser-based ones it's a nice feature to have.
> 
> The alternative, as you mentioned, is to not issue refresh tokens and do token renewal the "same old way" via iframe with prompt=none, while still using code flow.

yes. 

Have you ever experienced issues with the latter approach and the browser’s 3rd party cookie policy?

> 
> > The only potential „baby step“ I would see is to move towards „token id_token“. Since this requires signature/at_hash checks etc. I doubt this is really easier than moving to code and exchange the code for an access token. What’s your opinion?
> 
> Even since OIDC arrived, this is the only flow I use for JS/browser-based clients (anything less has always seemed so obviously inferior). So for me and my customers, all browser-based clients I am involved in are already there. Perhaps this is the reason for all of my questions/comments about the recent BCP doc. Given "id_token token", CSP, and using the browser history API to wipe the access token from browser history, we already have a decent set of tools to mitigate attacks. As I already conceded, the only remaining issue (IMO) is the short window of time the access token is in the URL.

There are two angles to approach access token leakage and replay from two angles, (a) preventing leakage (that’s what you can do with the browser history API, TLS, ...) or (b) detecting replay (that’s what one can do with sender constraint access tokens). 

The focus of OAuth/OIDC was on (a) but experiences have shown this is of limited effectivity, especially in dynamic/open scenarios, which we are seeing increasingly due to open banking, eHealth, eID, …. So sealing the problem from both ends seems reasonable.  
 
> 
> Given that it seems to me that OIDC and OAuth2 are typically used together (at least when a user is involved with authentication), I always wonder why the OAuth and OIDC WGs are separate. Given that so much effort of the two sets of specs overlap, it seems odd to keep adding onto the OAuth specs and ignoring the added features that OIDC provides. I don't mean to derail this thread, or step on any political toes, so apologies in advance.

No problem. As already stated, OIDC, esp. FAPI, and OAuth need to be aligned on that point. 

Thanks for the insights you shared. We will be publishing another revision of the OAuth Security BCP soon, which also adds recommendation and justification based on our discussions. 

kind regards, 
Torsten. 

> 
> 
> -Brock
>