Re: [OAUTH-WG] Cache-Control headers for Bearer URI Query Parameter method

[Mike Jones <[e-mail]>]

Could the experts on this thread please summarize the outcome of this thread?  Was the conclusion that no changes were required to either Core or Bearer?  Or if you believe that changes are required to one or both drafts, could you please propose exact wording changes for review by the working group?

				Thanks,
				-- Mike

-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On Behalf Of Torsten Lodderstedt
Sent: Monday, June 11, 2012 11:07 PM
To: Phil Hunt
Cc: [e-mail]
Subject: Re: [OAUTH-WG] Cache-Control headers for Bearer URI Query Parameter method

thanks a lot

Am 12.06.2012 01:03, schrieb Phil Hunt:
> Thanks. That makes sense.
>
> Phil
>
> On 2012-06-11, at 15:39, Amos Jeffries<[e-mail]>  wrote:
>
>> On 12.06.2012 07:23, Phil Hunt wrote:
>>> Private also seems inappropriate since no operation should be cached 
>>> for oauth as even when the same requestor.
>>>
>>> Phil
>>>
>> There is a difference in HTTP use-case between what the Bearer and core specs are covering.
>>
>> The core spec appears to be covering the request/response messages transferring credentials in the response entity. These mandate "no-store", which adds strict erasure requirements for any middleware and browser caches handling the response. Even single-user caches like a browser are not allowed to store the HTTP copy of the credentials response.
>>
>> Bearer is requiring "private" only in the specific HTTP case where the token is in query params and response is some data object (ie images or HTML page). Such that trusted proxies and other third-parties who do not implement OAuth but do relay HTTP treat the request and reply securely. With uses of Bearer via HTTP authentication framework this "private" is implicit.
>>   In these cases the response MAY be cached by a private browser cache, but not by third-party proxies. no-store is overkill and wastes bandwidth in this case.
>>
>>
>> I hope this clarifies.
>>
>>
>> AYJ
>>
>>
>>> On 2012-06-11, at 12:17, Torsten Lodderstedt wrote:
>>>
>>>> Hi all,
>>>>
>>>> I noticed a difference in usage of cache control headers between bearer and core spec.
>>>>
>>>> core -27 states:
>>>>
>>>>   " The authorization server MUST include the HTTP "Cache-Control"
>>>>    response header field [RFC2616] with a value of "no-store" in any
>>>>    response containing tokens, credentials, or other sensitive
>>>>    information, as well as the "Pragma" response header field [RFC2616]
>>>>    with a value of "no-cache"."
>>>>
>>>> So a "Pragma" response header field is required instead of the "Cache-Control" header "private".
>> Not instead of. *As well as*.  Pragma "no-cache" only tells the third-party to revalidate before using the response, it does not prevent storage and thus potential data leakage.
>>
>>
>>>> As far as I understand, both specs are nearly but not fully equivalent. Do we need to align both?
>>>>
>>>> regards,
>>>> Torsten.
>>>>
>>>> Am 09.06.2012 00:20, schrieb Mike Jones:
>>>>> Hi Amos,
>>>>>
>>>>> The OAuth Bearer specification now includes the Cache-Control language we’d discussed.
>>>>>
>>>>> See the fifth paragraph of http://tools.ietf.org/html/draft-ietf-oauth-v2-bearer-20#section-2.3.
>>>>>
>>>>>                                                             Thanks again,
>>>>>                                                             -- 
>>>>> Mike
>>>>>
>>>>>
>>>>> From: [e-mail] On Behalf Of Mike Jones
>>>>> Sent: Thursday, May 17, 2012 3:12 PM
>>>>> Subject: [OAUTH-WG] Cache-Control headers for Bearer URI Query 
>>>>> Parameter method
>>>>>
>>>>> Dear working group members:
>>>>>
>>>>> I'm going through the remaining open issues that have been raised about the Bearer spec so as to be ready to publish an updated draft once the outstanding consensus call issues are resolved.
>>>>>
>>>>> Amos Jeffries had cited this requirement in the HTTPbis spec ( http://tools.ietf.org/html/draft-ietf-httpbis-p7-auth-19#section-2.3.1):
>>>>>
>>>>>    o  The credentials carried in an Authorization header field are
>>>>>       specific to the User Agent, and therefore have the same effect on
>>>>>       HTTP caches as the "private" Cache-Control response directive,
>>>>>       within the scope of the request they appear in.
>>>>>
>>>>>       Therefore, new authentication schemes which choose not to carry
>>>>>       credentials in the Authorization header (e.g., using a newly
>>>>>       defined header) will need to explicitly disallow caching, by
>>>>>       mandating the use of either Cache-Control request directives
>>>>>       (e.g., "no-store") or response directives (e.g., "private").
>>>>>
>>>>> I propose to add the following text in order to satisfy this requirement.  I have changed Amos' MUSTs to SHOULDs because, in practice, applications that have no option but to use the URI Query Parameter method are likely to also not have control over the request's Cache-Control directives (just as they do not have the ability to use an "Authorization: Bearer" header value):
>>>>>
>>>>>     Clients using the URI Query Parameter method SHOULD also send a
>>>>>     Cache-Control header containing the "no-store" option.  Server success
>>>>>     (2XX status) responses to these requests SHOULD contain a Cache-Control
>>>>>     header with the "private" option.
>>>>>
>>>>> Comments?
>>>>>
>>>>>                                                                 -- 
>>>>> Mike
>>>>>
>>>>> -----Original Message-----
>>>>> From: Amos Jeffries
>>>>>
>>>>> On 24/04/2012 4:33 p.m., Mike Jones wrote:
>>>>>> What specific language would you suggest be added to what section(s)?
>>>>>>
>>>>>>                                                              --          Mike
>>>>>
>>>>> Perhapse the last paragraph appended:
>>>>> "
>>>>>
>>>>>     Because of the security weaknesses associated with the URI method
>>>>>     (see Section 5), including the high likelihood that the URL
>>>>>     containing the access token will be logged, it SHOULD NOT be used
>>>>>     unless it is impossible to transport the access token in the
>>>>>     "Authorization" request header field or the HTTP request entity-body.
>>>>>     Resource servers compliant with this specification MAY support this
>>>>>     method.
>>>>>
>>>>>     Clients requesting URL containing the access token MUST also send a
>>>>>     Cache-Control header containing the "no-store" option. Server success
>>>>>     (2xx status) responses to these requests MUST contain a Cache-Control
>>>>>     header with the "private" option.
>>>>>
>>>>> "
>>>>>
>>>>> I'm a little suspicious that the "SHOUDL NOT" in that top paragraph likely should be a MUST NOT to further discourage needless use.
>>>>>
>>>>>
>>>>> AYJ
>>>>>
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: [e-mail] On Behalf Of Amos Jeffries
>>>>>>
>>>>>> On 24.04.2012 13:46, [e-mail] wrote:
>>>>>>> A New Internet-Draft is available from the on-line 
>>>>>>> Internet-Drafts directories. This draft is a work item of the 
>>>>>>> Web Authorization Protocol Working Group of the IETF.
>>>>>>>
>>>>>>>            Title           : The OAuth 2.0 Authorization Protocol: Bearer
>>>>>>> Tokens
>>>>>>>            Author(s)       : Michael B. Jones
>>>>>>>                             Dick Hardt
>>>>>>>                             David Recordon
>>>>>>>            Filename        : draft-ietf-oauth-v2-bearer-19.txt
>>>>>>>            Pages           : 24
>>>>>>>            Date            : 2012-04-23
>>>>>>>
>>>>>>>      This specification describes how to use bearer tokens in HTTP
>>>>>>>      requests to access OAuth 2.0 protected resources.  Any party in
>>>>>>>      possession of a bearer token (a "bearer") can use it to get 
>>>>>>> access to
>>>>>>>      the associated resources (without demonstrating possession of a
>>>>>>>      cryptographic key).  To prevent misuse, bearer tokens need to be
>>>>>>>      protected from disclosure in storage and in transport.
>>>>>>>
>>>>>>>
>>>>>>> A URL for this Internet-Draft is:
>>>>>>> http://www.ietf.org/internet-drafts/draft-ietf-oauth-v2-bearer-1
>>>>>>> 9.txt
>>>>>>
>>>>>> The section 2.3 (URL Query Parameter) text is still lacking explicit and specific security requirements. The overarching TLS requirement is good in general, but insufficient in the presence of HTTP intermediaries on the TLS connection path as is becoming a common practice.
>>>>>>
>>>>>> The upcoming HTTPbis specs document this issue as a requirement for new auth schemes such as Bearer:
>>>>>>
>>>>>> http://tools.ietf.org/html/draft-ietf-httpbis-p7-auth-19#section-
>>>>>> 2.3.1
>>>>>> "
>>>>>>          Therefore, new authentication schemes which choose not to carry
>>>>>>          credentials in the Authorization header (e.g., using a newly
>>>>>>          defined header) will need to explicitly disallow caching, by
>>>>>>          mandating the use of either Cache-Control request directives
>>>>>>          (e.g., "no-store") or response directives (e.g., "private").
>>>>>> "
>>>>>>
>>>>>>
>>>>>> AYJ
>>>>>>
>>>>>> _______________________________________________
>>>>>> OAuth mailing list
>>>>>> [e-mail]
>>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> OAuth mailing list
>>>>> [e-mail]
>>>>> https://www.ietf.org/mailman/listinfo/oauth
>>>> _______________________________________________
>>>> OAuth mailing list
>>>> [e-mail]
>>>> https://www.ietf.org/mailman/listinfo/oauth
>> _______________________________________________
>> OAuth mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/oauth
> _______________________________________________
> OAuth mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/oauth
_______________________________________________
OAuth mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/oauth