[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [wss] Groups - WSS-SAML-09.pdf uploaded
Ron, I agree that we have 2 cases: signed and unsigned assertions (i.e. where a "signed assertion" is an assertion contains an enveloped signature applied by the assertion authority as described in saml-core). We agree on the unsigned assertion case (i.e. where the attester signs both the content and the unsigned assertion) and that is the case that is currently in both the current profile and interop specs. For the sake of continuing the discussion on the "signed assertion" case, I have this to offer in reply to your comment: I think there is a problem if the attester does not sign both the assertion and the content, because if only the content is signed then what is to stop an intruder from substituting a different assertion which will then associate the content with the subject of that other assertion. Therefore, in my opinion, even in the signed assertion case, the attester must sign both the assertion and the content. Rich -----Original Message----- From: Ron Monzillo [mailto:Ronald.Monzillo@Sun.COM] Sent: Thursday, January 29, 2004 4:54 PM To: Levinson, Richard Cc: wss@lists.oasis-open.org Subject: Re: [wss] Groups - WSS-SAML-09.pdf uploaded Thanks Rich, I'd hope others will comment on this as well. The attester is both authorizing the subject assertion and attesting for the presence of the subject. It authorizes the assertion contents , by signing the assertion, which it had previously configured to authorize it as proxy for the subject. As I mentioned previously, this only makes sense if the assertion was not signed by the assertion authority. If the assertion was signed by the assertion authority, then the same logic can be applied, except the signature done by the attesting entity need only digest the msg body. Ron Levinson, Richard wrote: >Ron, > >I will try to consolidate our 2 email threads where I was trying to >clarify how to use the V9 spec to implement sender-vouches for the >interop spec and post to the wss list for additional input from any who >are willing to delve into the details and provide further comment. > >I think we are in total agreement in your alternative paragraph (from >email 1 below): > >"As an alternative, maybe the assertion referenced from keyInfo should >be the sender-vouches confirmed assertion, and it could contain in its >subject confirmation, a reference to a key confirmed security token of >the attesting entity. The attesting entity would put this reference in >the sender-vouches confirmed assertion." > >as compared to my suggestion (from email 2 below): > > "Since we are doing sender-vouches, the subject > confirmation may be done using the sender's key. > i.e. since the subject doesn't actually sign > anything and the sender is the one confirming > that the info in the assertion can be used by > the receiver, then it makes sense that the > sender's key be in the KeyInfo of the > SubjectConfirmation." > >Also, I believe that the WS-SAML Profile spec is the only place where a >definitive explanation of how to implement sender-vouches and >holder-of-key exists. So, to some degree, that spec has flexibility to >clarify the details of these protocols. > >Also, referring back to the SAML 1.0 core spec, the relevant elements >within the assertion are defined as follows: > > "The <SubjectConfirmation> element specifies a subject by supplying > data that allows the subject to be authenticated. It contains the > following elements in order: > > <ConfirmationMethod> [One or more] > > A URI reference that identifies a protocol to be used to > authenticate the subject. URI references identifying SAML-defined > confirmation methods are currently defined with the SAML >profiles > in the SAML Binding and Profiles specification [SAMLBind]. > Additional methods may be added by defining new profiles or >by > private agreement. > > <SubjectConfirmationData> [Optional] > > Additional authentication information to be used by a specific > authentication protocol. > > <ds:KeyInfo> [Optional] > > An XML Signature [XMLSig] element that specifies a cryptographic > key held by the subject." > >As I read the above, if we want to do this with one assertion (instead >of having an in message and off message assertion in the V9 example), I >think we could put the sender (attester) cert in ds:KeyInfo, but to >keep to the letter of the spec this really would not be accurate, >because that element explicitly says the corresponding key is held by >the "subject", who is not the attester, but the attestee. > >Therefore, I think to be compliant with the above we could put a >KeyInfo element containing the attester/sender cert in the >SubjectConfirmationData element. > >I suggest that makes sense because we are defining the "sender-vouches" >protocol in this document and we could say that for the sender-vouches >protocol that the attester cert or ref to such cert MAY be placed in >the SubjectConfirmationData >element. > >It seems to me that this is a perfectly good way to implement >sender-vouches since the SubjectConfirmation element is intended to >"supply data that allows a subject to be authenticated" and that is >exactly what we are doing by supplying the sender cert which can be >used to verify the signature covering the assertion which identifies >the subject and binds the assertion to the content which is also >included in the sig. And since the receiver implicitly >trusts the sender/attester by some out of band agreement, this procedure >effectively >authenticates the subject as far as the receiver is concerned. > >Therefore, to move forward with the example in the interop document, >that is the approach I propose to take, which will allow the interop to >work on the basis of self-contained messages that do not rely on >external methods to obtain assertions or certificates. > > Rich > > > > >!*********** > > email 1: > >******************! > >-----Original Message----- >From: Ron Monzillo [mailto:Ronald.Monzillo@Sun.COM] >Sent: Thursday, January 29, 2004 11:56 AM >To: Levinson, Richard >Subject: Re: [wss] Groups - WSS-SAML-09.pdf uploaded > > >Rich, > >There may be a problem with the example, or perhaps with the >sender-vouches approach, but I think in general the example is >structured appropriately. In the example, the on-msg assertion, is >referenced by STR2, and used by the attesting entity" to sign both the >message body and a "sender-vouches confirmed assertion (referenced via >STR1). > >I changed this example to not include the sender-vouches confirmed msg >in the assertion, and such that the attesting entity uses a >holder-of-key confirmed assertion (as apposed to a key in some X509 >cert) to protect the vouched for message content and assertion (as >required by section 3.4.2.1). > >I tweaked the example in this way to give us a better look at how >sender-vouches is currently being profiled, and to bring to light some >potential weaknesses in the current use modle for sender-vouches >confirmed >assertions: > >For example, without an appropriate usage attribute value on STR1 and >without the sender-vouches confirmed msg being present in the msg, it >may not be easy for the receiver to understand the header. > >As an alternative, maybe the assertion referenced from keyInfo should >be the sender-vouches confirmed assertion, and it could contain in its >subject confirmation, a reference to a key confirmed security token of >the attesting entity. The attesting entity would put this reference in >the sender-vouches confirmed assertion. > >[I realize that statements not assertions have confirmation methods, >but I am purposely glossing over this level of detail] > >Such an approach would rekindle our discussion >about whether sender-vouches confirmed assertions must be signed by an >issuing authority, while on the other hand, it would allign better with >the chained approach suggested for using holder of key confirmed >assertions to authorize proxies; in which case the authority would sign >the top level holder-of-key assertion containing a reference in its >subject-confirmation to a key confirmed assertion authorizing a proxy. > >Ron > >Levinson, Richard wrote: > > > >>Ron, >> >>This is a more explicit comment on the sv scenario >>in section 3.4.2.3 that looks to me like it needs >>a couple things fixed. >> >> 1. line 745 the ConfrimationMethod is holder-of-key. >> shouldn't it be sender-vouches? >> >> 2. line 727, AssertionID ends with "...ebdfc", >> and on line 769 KeyIdentifier refers to >> AssertionID ending with "...ebdbe" which is >> different from line 727. (As explained below >> I think this might be more than a simple typo) >> >> 3. line 806 the STR in the Signature/KeyInfo refers >> to "...ebdfc" which is the Assertion on line 727, >> which presumably contains the key used to verify >> this signature in the SubjectConfirmation/KeyInfo >> element. >> >>A possible problem I see with this setup is that the signer is signing >>an assertion of which the signer itself appears to be the subject of >>the assertion, since the SubjectConfirmation/ KeyInfo indicates a key >>to verify the subject of the assertion, not the attester of the >>subject of the assertion. >> >>It would seem to me in this case that STR2 should point to the >>signer's key, not the assertion subject's key as it seems to now, >>although I am not sure at this point if the key in the assertion is >>intended to be the signer's or the subject's, since your original >>intention might have been to include 2 assertions which would explain >>the presence of the 2 different AssertionIDs. >> >>To locate the signer key we could either have a simple STR2 to a cert >>in the wsse header, or to a 2nd saml assertion, also in the message, >>containing the attestor's key. >> >>Please let me know what you think, since I am trying to make the >>interop example as consistent as possible with the profile. >> >> Thanks, >> >> Rich >> >> >> >>-----Original Message----- >>From: ronald.monzillo@sun.com [mailto:ronald.monzillo@sun.com] >>Sent: Tuesday, January 27, 2004 1:55 AM >>To: wss@lists.oasis-open.org >>Subject: [wss] Groups - WSS-SAML-09.pdf uploaded >> >> >>The document WSS-SAML-09.pdf has been submitted by ronald monzillo >>(ronald.monzillo@sun.com) to the OASIS Web Services Security TC >>document repository. >> >>Document Description: >>SAML token Profile >> >>Update of STR ValueTypes to use URIs as apposed to Qnames (issue 196). >>Changed use of STR to carry saml:AuthorityBinding element. Many edits >>to examples, including change to Sender-Vouches example to user STR >>Dreference Transfrom. >> >>Download Document: >>http://www.oasis-open.org/apps/org/workgroup/wss/download.php/5177/WSS >>-SAML >> >> >- > > >>09.pdf >> >>View Document Details: >>http://www.oasis-open.org/apps/org/workgroup/wss/document.php?document_ >>id=51 >>77 >> >> >>PLEASE NOTE: If the above links do not work for you, your email >>application may be breaking the link into two pieces. You may be able >>to copy and paste the entire link address into the address field of >>your web browser. >> >> >> >>To unsubscribe from this mailing list (and be removed from the roster >>of the OASIS TC), go to >>http://www.oasis-open.org/apps/org/workgroup/wss/members/leave_workgrou >>p.php >>. >> >> >> >> > > > >!********************* > > email 2: > >************************! > > >-----Original Message----- >From: Levinson, Richard >Sent: Thursday, January 29, 2004 11:37 AM >To: 'ronald.monzillo@sun.com' >Subject: RE: [wss] Groups - WSS-SAML-09.pdf - question on Sender-vouches >scenario > > >Ron, > >I have done a little more thinking on this and maybe we >can use things as they are if we interpret things in >the following context: > > Since we are doing sender-vouches, the subject > confirmation may be done using the sender's key. > i.e. since the subject doesn't actually sign > anything and the sender is the one confirming > that the info in the assertion can be used by > the receiver, then it makes sense that the > sender's key be in the KeyInfo of the > SubjectConfirmation. > >If you agree with this then the only thing I think needs >to be changed in the text is: > > 1. line 745 change "holder-of-key" to "sender-vouches" > > 2. line 769 change "ebdbe" to "ebdfc" > >This way the SignedInfo ref covers STR1 which transforms to the >assertion. And STR2 in the Signature/KeyInfo now refers to the >assertion and is interpreted as meaning: use the KeyInfo in the >SubjectConfirmation as the key to verify this signature. > >For now I will work on this assumption and proceed with updating the >interop spec. > >Please let me know if you think this all makes sense, > > Thanks, > > Rich > > >-----Original Message----- >From: Levinson, Richard >Sent: Wednesday, January 28, 2004 5:56 PM >To: 'ronald.monzillo@sun.com' >Subject: RE: [wss] Groups - WSS-SAML-09.pdf uploaded > > >Ron, > >This is a more explicit comment on the sv scenario >in section 3.4.2.3 that looks to me like it needs >a couple things fixed. > > 1. line 745 the ConfrimationMethod is holder-of-key. > shouldn't it be sender-vouches? > > 2. line 727, AssertionID ends with "...ebdfc", > and on line 769 KeyIdentifier refers to > AssertionID ending with "...ebdbe" which is > different from line 727. (As explained below > I think this might be more than a simple typo) > > 3. line 806 the STR in the Signature/KeyInfo refers > to "...ebdfc" which is the Assertion on line 727, > which presumably contains the key used to verify > this signature in the SubjectConfirmation/KeyInfo > element. > >A possible problem I see with this setup is that the signer >is signing an assertion of which the signer itself appears to be the >subject of the assertion, since the SubjectConfirmation/ KeyInfo >indicates a key to verify the subject of the assertion, not the >attester of the subject of the assertion. > >It would seem to me in this case that STR2 should point to >the signer's key, not the assertion subject's key as it seems to now, >although I am not sure at this point if the key in the assertion is >intended to be the signer's or the subject's, since your original >intention might have been to include 2 assertions which would explain >the presence of the 2 different AssertionIDs. > >To locate the signer key we could either have a simple STR2 to a cert >in the wsse header, or to a 2nd saml assertion, also in the message, >containing the attestor's key. > >Please let me know what you think, since I am trying to make >the interop example as consistent as possible with the profile. > > Thanks, > > Rich > > > >-----Original Message----- >From: ronald.monzillo@sun.com [mailto:ronald.monzillo@sun.com] >Sent: Tuesday, January 27, 2004 1:55 AM >To: wss@lists.oasis-open.org >Subject: [wss] Groups - WSS-SAML-09.pdf uploaded > > >The document WSS-SAML-09.pdf has been submitted by ronald monzillo >(ronald.monzillo@sun.com) to the OASIS Web Services Security TC >document repository. > >Document Description: >SAML token Profile > >Update of STR ValueTypes to use URIs as apposed to Qnames (issue 196). >Changed use of STR to carry saml:AuthorityBinding element. Many edits >to examples, including change to Sender-Vouches example to user STR >Dreference Transfrom. > >Download Document: >http://www.oasis-open.org/apps/org/workgroup/wss/download.php/5177/WSS-SAML - >09.pdf > >View Document Details: >http://www.oasis-open.org/apps/org/workgroup/wss/document.php?document_ >id=51 >77 > > >PLEASE NOTE: If the above links do not work for you, your email >application may be breaking the link into two pieces. You may be able >to copy and paste the entire link address into the address field of >your web browser. > > > >To unsubscribe from this mailing list (and be removed from the roster >of the OASIS TC), go to >http://www.oasis-open.org/apps/org/workgroup/wss/members/leave_workgrou >p.php >. > >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]