Internet-Draft | CSRAttrs | March 2022 |
Richardson, et al. | Expires 8 September 2022 | [Page] |
Enrollment over Secure Transport (EST) is ambiguous in specification of the CSR Attributes Response. This has resulted in implementation challenges and implementor confusion. This document updates EST and clarifies how the CSR Attributes Response can be used by an EST server to specify both CSR attribute OIDs and also CSR attribute values that the server expects the client to include in its CSR request.¶
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 https://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 8 September 2022.¶
Copyright (c) 2022 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 (https://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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
Enrollment over Secure Transport [RFC7030] (EST) has been used in a wide variety of applications. In particular, [RFC8994] and [RFC8995] describe a way to use it in order to build out an autonomic control plane (ACP) [RFC8368].¶
The ACP requires that each node be given a very specific SubjectAltName. In the ACP specification, the solution was for the EST server to use section 2.6 of [RFC7030] to convey to the EST client the actual SubjectAltName that will end up in its certificate.¶
As a result of some implementation challenges, it came to light that this particular way of using the CSR attributes was not universally agreed upon, and in fact runs contrary to section 2.6. Section 2.6 says that the CSR attributes "provide additional descriptive information that the EST server cannot access itself". This extends to specifying that a particular attribute should exist, but not to the point of having the EST server actually specify the value.¶
The way in which the CSRattributes were understood by [RFC8994] turns out to be invalid. This document, therefore, updates section 2.6 to define this behavior.¶
This document also updates section 4.5 to include revised ASN.1 that covers all uses and is backward compatible with the existing use.¶
Additional examples are provided in an appendix.¶
The ASN.1 for CSR Attributes as defined in EST section 4.5.2 is:¶
CsrAttrs ::= SEQUENCE SIZE (0..MAX) OF AttrOrOID AttrOrOID ::= CHOICE (oid OBJECT IDENTIFIER, attribute Attribute } Attribute { ATTRIBUTE:IOSet } ::= SEQUENCE { type ATTRIBUTE.&id({IOSet}), values SET SIZE(1..MAX) OF ATTRIBUTE.&Type({IOSet}{@type}) }¶
That section also states the following:¶
the values indicating the particular attributes desired to be included in the resulting certificate's extensions¶
This has been interpreted by some implementations as meaning that the CSR Attributes response can only include values for the attribute OIDs that the client should include in its CSR, and cannot include the actual values of those attributes. This is further reinforced by the example:¶
Attribute: type = extensionRequest (1.2.840.113549.1.9.14) value = macAddress (1.3.6.1.1.1.1.22)¶
This example illustrates that the 'value' specified is an attribute OID, for example the macAddress OID, and not the value (such as "10-00-00-12-23-45") of the attribute itself.¶
There is no clearly documented mechanism with supporting examples that specifies how a CSR Attributes response can include a value for a given attribute such as SubjectAltName.¶
EST section 4.5.2 also states the following:¶
The structure of the CSR Attributes Response SHOULD, to the greatest extent possible, reflect the structure of the CSR it is requesting.¶
This statement aligns closely with the goal of this document. Additionally, EST Extensions [RFC8295] Appendix A has an informative appendix that outlines how a full CSR can be included in the CSR Attributes response.¶
The WG will pick one option as part of the adoption call.¶
This ASN.1 needs fixing.¶
CsrAttrs ::= SEQUENCE SIZE (0..MAX) OF AttrOrOID AttrOrOID ::= CHOICE (oid OBJECT IDENTIFIER, attribute Attribute, value Value } Attribute { ATTRIBUTE:IOSet } ::= SEQUENCE { extType ATTRIBUTE.&id({IOSet}), extAttr SET SIZE(1..MAX) OF ATTRIBUTE.&Type({IOSet}{@type}) } Value { ATTRIBUTE:IOSet } ::= SEQUENCE { extType ATTRIBUTE.&id({IOSet}), type ATTRIBUTE.&Type({IOSet}{@type}), value OCTET STRING }¶
This would just add a value to the SEQUENCE:¶
OBJECT challengePassword SEQUENCE OBJECT subjectAltName SET OBJECT someACPgoo SEQUENCE OBJECT id-ecPublicKey SET OBJECT secp384r1 OBJECT ecdsa-with-SHA384¶
For example:¶
0 30: SEQUENCE { 2 28: SEQUENCE { 4 3: OBJECT IDENTIFIER subjectAltName (2 5 29 17) 9 21: SET { 11 19: [1] { 13 17: UTF8String 'hello@example.com' : } : } : } : }¶
The following options support complete and unambiguous specification of¶
CSR ingredients may be the subject DN, any X.509 extensions, and special attributes like a challenge password.¶
For specifying the type of keys allowed in CSRs,
they use a to-the-point KeySpec
type.
It can be defined for instance as¶
KeySpec ::= CHOICE { keyAlg AlgorithmIdentifier, rsaKeyLen INTEGER }¶
The keyAlg
type use used to specify public-key alorithms
and can include parameters, such as the name of an elliptic curve.
The rsaKeyLen
choice allows specifying the size of RSA keys,
which it is not possible using values of type AlgorithmIdentifier
.¶
The keySpec
could also be sequence of such specs,
such that the server can give several key types from which the client can choose,
e.g., EC keys on certain curves and/or RSA keys of certain sizes.¶
Stick for syntactic backward compatibility with¶
CsrAttrs ::= SEQUENCE SIZE (0..MAX) OF AttrOrOID¶
Each OID given in AttrOrOID must occur only once.¶
Plain OIDs are used mostly for challengePassword.¶
Attributes are used mostly for any X.509 extensions, subject DN, key spec, and hash alg, while defining new generally usable OIDs for¶
to be given on demand as attribute IDs of type ATTRIBUTE.&id({IOSet})
.¶
Define a new and more to-the-point type, which does not require new OIDs:¶
CsrAttrs ::= SEQUENCE { oids SEQUENCE OF OBJECT IDENTIFIER, attrs SEQUENCE OF Attribute, subject [0] Name OPTIONAL, keySpec [1] KeySpec OPTIONAL, hashAlg [2] AlgorithmIdentifier OPTIONAL }¶
Each OID given in oids or attrs must occur only once.¶
The oids are used mostly for requiring a challenge password.¶
The atttrs are used mostly for requiring certain X.509 extensions.¶
This is, typically just challengePassword and extensionRequest are used.¶
Define a new fully to-the-point type, which does not require any (direct) OIDs:¶
CsrAttrs ::= SEQUENCE { subject Name OPTIONAL, extensions SEQUENCE OF Extension, challengePassword BOOLEAN, keySpec [0] KeySpec OPTIONAL, hashAlg [1] AlgorithmIdentifier OPTIONAL }¶
There are some ways in which the new CSRattributes could co-exist with RFC7030.¶
The client can signal that it supports the new attribute format by using an Accept: header in the transaction. This acts as a signal to a server that it can/should return the attributes in the new format.¶
Clients that want to use the new format would use a new end point, such as "csrvalues" which would only support the new format. A client which supported both would have to try both "csrvalues" and then fall back "csrattrs" if the EST server did not support the new format. Some uses (such as [RFC8994]) require the new format, so if it was not suppored, that would be a protocol error.¶
ASN.1 encoding is self-describing, and some formats proposed above could possibly be parsed by legacy clients without a problem.¶
The Registrar may know which clients are which by the kind of authentication that they do. An [RFC8994] client which has just performed a [RFC8995] enrollment would be assumed to require the new format only. A client which authenticates with an LDevID for a renewal would be strongly identified, and the Registrar could be programmed whether to return new format, or legacy CSR attributes.¶
[RFC8951] clarified that the csrattrs end point was to be Base64 encoded even though the HTTP transport was 8-bit clean.¶
If this document establishes a new end point, then the new end point will not be base64 encoded according to current HTTP usage.¶
All security considertions from EST [RFC7030] section 6 are applicable.¶
An EST server may use this mechanism to instruct the EST client about the identities it should include in the CSR it sends as part of enrollment. The client may only be aware of its IDevID Subject, which includes a manufacturer serial number. The EST server can use this mechanism to tell the client to include a specific fully qualified domain name in the CSR in order to complete domain ownership proofs required by the CA. Additionally, the EST server may deem the manufacturer serial number in an IDevID as personally identifiable information, and may want to specify a new random opaque identifier that the pledge should use in its CSR. This may be desirable if the CA and EST server have different operators.¶
None.¶
TODO¶