| Internet-Draft | TLS Exported Authenticator | March 2022 |
| Sullivan | Expires 5 September 2022 | [Page] |
This document describes a mechanism that builds on Transport Layer Security (TLS) or Datagram Transport Layer Security (DTLS) and enables peers to provide a proof of ownership of an identity, such as an X.509 certificate. This proof can be exported by one peer, transmitted out-of-band to the other peer, and verified by the receiving peer.¶
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 5 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.¶
This document provides a way to authenticate one party of a Transport Layer Security (TLS) or Datagram Transport Layer Security (DTLS) connection to its peer using authentication messages created after the session has been established. This allows both the client and server to prove ownership of additional identities at any time after the handshake has completed. This proof of authentication can be exported and transmitted out-of-band from one party to be validated by its peer.¶
This mechanism provides two advantages over the authentication that TLS and DTLS natively provide:¶
Versions of TLS prior to TLS 1.3 used renegotiation as a way to enable post-handshake client authentication given an existing TLS connection. The mechanism described in this document may be used to replace the post-handshake authentication functionality provided by renegotiation. Unlike renegotiation, exported Authenticator-based post-handshake authentication does not require any changes at the TLS layer.¶
Post-handshake authentication is defined in section 4.6.3 of TLS 1.3 [RFC8446], but it has the disadvantage of requiring additional state to be stored as part of the TLS state machine. Furthermore, the authentication boundaries of TLS 1.3 post-handshake authentication align with TLS record boundaries, which are often not aligned with the authentication boundaries of the higher-layer protocol. For example, multiplexed connection protocols like HTTP/2 [RFC7540] do not have a notion of which TLS record a given message is a part of.¶
Exported Authenticators are meant to be used as a building block for application protocols. Mechanisms such as those required to advertise support and handle authentication errors are not handled by TLS (or DTLS).¶
The minimum version of TLS and DTLS required to implement the mechanisms decribed in this document are TLS 1.2 [RFC6347] and DTLS 1.2 [RFC5246].¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This document uses terminology such as client, server, connection, handshake, endpoint, peer that are defined in section 1.1 of [RFC8446]. The term "initial connection" refers to the (D)TLS connection from which the exported authenticator messages are derived.¶
There are two types of messages defined in this document: Authenticator Requests and Authenticators. These can be combined in the following three sequences:¶
Client Authentication¶
Server Authentication¶
Spontaneous Server Authentication¶
The authenticator request is a structured message that can be created by either party of a (D)TLS connection using data exported from that connection. It can be transmitted to the other party of the (D)TLS connection at the application layer. The application layer protocol used to send the authenticator request SHOULD use a secure transport channel with equivalent security to TLS, such as QUIC [RFC9001], as its underlying transport to keep the request confidential. The application MAY use the existing (D)TLS connection to transport the authenticator.¶
An authenticator request message can be constructed by either the client or the server. Server-generated authenticator requests use the CertificateRequest message from Section 4.3.2 of [RFC8446]. Client-generated authenticator requests use a new message, called the ClientCertificateRequest, which uses the same structure as CertificateRequest. (Note that the latter is not a request for a client certificate, but rather a certificate request generated by the client.) These message structures are used even if the connection protocol is TLS 1.2 or DTLS 1.2.¶
The CertificateRequest and ClientCertificateRequest messages are used to define the parameters in a request for an authenticator. These are encoded as TLS handshake messages, including length and type fields. They do not include any TLS record layer framing and are not encrypted with a handshake or application-data key.¶
The structures are defined to be:¶
struct {
opaque certificate_request_context<0..2^8-1>;
Extension extensions<2..2^16-1>;
} ClientCertificateRequest;
struct {
opaque certificate_request_context<0..2^8-1>;
Extension extensions<2..2^16-1>;
} CertificateRequest;
¶
The uniqueness requirements of the certificate_request_context apply only to CertificateRequest and ClientCertificateRequest messages that are used as part of authenticator requests, but do apply across CertificateRequest and ClientCertificateRequest messages. A certificate_request_context value used in a ClientCertificateRequest cannot be used in an authenticator CertificateRequest on the same connection, and vice versa. There is no impact if the value of a certificate_request_context used in an authenticator request matches the value of a certificate_request_context in the handshake or in a post-handshake message.¶
The authenticator is a structured message that can be exported from either party of a (D)TLS connection. It can be transmitted to the other party of the (D)TLS connection at the application layer. The application layer protocol used to send the authenticator SHOULD use a secure transport channel with equivalent security to TLS, such as QUIC [RFC9001], as its underlying transport to keep the authenticator confidential. The application MAY use the existing (D)TLS connection to transport the authenticator.¶
An authenticator message can be constructed by either the client or the server given an established (D)TLS connection, an identity, such as an X.509 certificate, and a corresponding private key. Clients MUST NOT send an authenticator without a preceding authenticator request; for servers an authenticator request is optional. For authenticators that do not correspond to authenticator requests, the certificate_request_context is chosen by the server.¶
Each authenticator is computed using a Handshake Context and Finished MAC Key derived from the (D)TLS connection. These values are derived using an exporter as described in Section 4 of [RFC5705] (for (D)TLS 1.2) or Section 7.5 of [RFC8446] (for (D)TLS 1.3). For (D)TLS 1.3, the exporter_master_secret MUST be used, not the early_exporter_master_secret. These values use different labels depending on the role of the sender:¶
The context_value used for the exporter is empty (zero length) for all four values. There is no need to include additional context information at this stage since the application-supplied context is included in the authenticator itself. The length of the exported value is equal to the length of the output of the hash function associated with the selected cipher suite (for TLS 1.3) or the hash function used for the pseudorandom function (PRF) (for (D)TLS 1.2). Exported authenticators cannot be used with (D)TLS 1.2 cipher suites that do not use the TLS PRF and with TLS 1.3 cipher suites that do not have an associated hash function. This hash is referred to as the authenticator hash.¶
To avoid key synchronization attacks, Exported Authenticators MUST NOT be generated or accepted on (D)TLS 1.2 connections that did not negotiate the extended master secret extension [RFC7627].¶
An authenticator is formed from the concatenation of TLS 1.3 [RFC8446] Certificate, CertificateVerify, and Finished messages. These messages are encoded as TLS handshake messages, including length and type fields. They do not include any TLS record layer framing and are not encrypted with a handshake or application-data key.¶
If the peer populating the certificate_request_context field in an authenticator's Certificate message has already created or correctly validated an authenticator with the same value, then no authenticator should be constructed. If there is no authenticator request, the extensions are chosen from those presented in the (D)TLS handshake's ClientHello. Only servers can provide an authenticator without a corresponding request.¶
ClientHello extensions are used to determine permissible extensions in the server's unsolicited Certificate message in order to follow the general model for extensions in (D)TLS in which extensions can only be included as part of a Certificate message if they were previously sent as part of a CertificateRequest message or ClientHello message. This ensures that the recipient will be able to process such extensions.¶
The Certificate message contains the identity to be used for authentication, such as the end-entity certificate and any supporting certificates in the chain. This structure is defined in [RFC8446], Section 4.4.2.¶
The Certificate message contains an opaque string called certificate_request_context, which is extracted from the authenticator request if present. If no authenticator request is provided, the certificate_request_context can be chosen arbitrarily but MUST be unique within the scope of the connection and be unpredictable to the peer.¶
Certificates chosen in the Certificate message MUST conform to the requirements of a Certificate message in the negotiated version of (D)TLS. In particular, the entries of certificate_list MUST be valid for the signature algorithms indicated by the peer in the "signature_algorithms" and "signature_algorithms_cert" extension, as described in Section 4.2.3 of [RFC8446] for (D)TLS 1.3 or from Sections 7.4.2 and 7.4.6 of [RFC5246] for (D)TLS 1.2.¶
In addition to "signature_algorithms" and "signature_algorithms_cert", the "server_name" [RFC6066], "certificate_authorities" (Section 4.2.4. of [RFC8446]), and "oid_filters" (Section 4.2.5. of [RFC8446]) extensions are used to guide certificate selection.¶
Only the X.509 certificate type defined in [RFC8446] is supported. Alternative certificate formats such as [RFC7250] Raw Public Keys are not supported in this version of the specification and their use in this context has not yet been analysed.¶
If an authenticator request was provided, the Certificate message MUST contain only extensions present in the authenticator request. Otherwise, the Certificate message MUST contain only extensions present in the (D)TLS ClientHello. Unrecognized extensions in the authenticator request MUST be ignored.¶
This message is used to provide explicit proof that an endpoint possesses the private key corresponding to its identity. The format of this message is taken from TLS 1.3:¶
struct {
SignatureScheme algorithm;
opaque signature<0..2^16-1>;
} CertificateVerify;
¶
The algorithm field specifies the signature algorithm used (see Section 4.2.3 of [RFC8446] for the definition of this field). The signature is a digital signature using that algorithm.¶
The signature scheme MUST be a valid signature scheme for TLS 1.3. This excludes all RSASSA-PKCS1-v1_5 algorithms and combinations of ECDSA and hash algorithms that are not supported in TLS 1.3.¶
If an authenticator request is present, the signature algorithm MUST be chosen from one of the signature schemes present in the "signature_algorithms" extensino of the authenticator request. Otherwise, with spontaneous server authentication, the signature algorithm used MUST be chosen from the "signature_algorithms" sent by the peer in the ClientHello of the (D)TLS handshake. If there are no available signature algorithms, then no authenticator should be constructed.¶
The signature is computed using the chosen signature scheme over the concatenation of:¶
The authenticator transcript is the hash of the concatenated Handshake Context, authenticator request (if present), and Certificate message:¶
Hash(Handshake Context || authenticator request || Certificate)¶
Where Hash is the authenticator hash defined in section 4.1. If the authenticator request is not present, it is omitted from this construction, i.e., it is zero-length.¶
If the party that generates the exported authenticator does so with a different connection than the party that is validating it, then the Handshake Context will not match, resulting in a CertificateVerify message that does not validate. This includes situations in which the application data is sent via TLS-terminating proxy. Given a failed CertificateVerify validation, it may be helpful for the application to confirm that both peers share the same connection using a value derived from the connection secrets (such as the Handshake Context) before taking a user-visible action.¶
An HMAC [HMAC] over the hashed authenticator transcript, which is the concatenation of the Handshake Context, authenticator request (if present), Certificate, and CertificateVerify. The HMAC is computed using the authenticator hash, using the Finished MAC Key as a key.¶
Finished = HMAC(Finished MAC Key, Hash(Handshake Context ||
authenticator request || Certificate || CertificateVerify))
¶
An endpoint constructs an authenticator by serializing the Certificate, CertificateVerify, and Finished as TLS handshake messages and concatenating the octets:¶
Certificate || CertificateVerify || Finished¶
An authenticator is valid if the CertificateVerify message is correctly constructed given the authenticator request (if used) and the Finished message matches the expected value. When validating an authenticator, constant-time comparisons SHOULD be used for signature and MAC validation.¶
If, given an authenticator request, the endpoint does not have an appropriate identity or does not want to return one, it constructs an authenticated refusal called an empty authenticator. This is a Finished message sent without a Certificate or CertificateVerify. This message is an HMAC over the hashed authenticator transcript with a Certificate message containing no CertificateEntries and the CertificateVerify message omitted. The HMAC is computed using the authenticator hash, using the Finished MAC Key as a key. This message is encoded as a TLS handshake message, including length and type field. It does not include TLS record layer framing and is not encrypted with a handshake or application-data key.¶
Finished = HMAC(Finished MAC Key, Hash(Handshake Context ||
authenticator request || Certificate))
¶
The creation and validation of both authenticator requests and authenticators SHOULD be implemented inside the (D)TLS library even if it is possible to implement it at the application layer. (D)TLS implementations supporting the use of exported authenticators SHOULD provide application programming interfaces by which clients and servers may request and verify exported authenticator messages.¶
Notwithstanding the success conditions described below, all APIs MUST fail if:¶
The following sections describe APIs that are considered necessary to implement exported authenticators. These are informative only.¶
The "request" API takes as input:¶
It returns an authenticator request, which is a sequence of octets that comprises a CertificateRequest or ClientCertificateRequest message.¶
The "get context" API takes as input:¶
It returns the certificate_request_context.¶
The "authenticate" API takes as input:¶
It returns either the exported authenticator or an empty authenticator as a sequence of octets. It is recommended that the logic for selecting the certificates and extensions to include in the exporter is implemented in the TLS library. Implementing this in the TLS library lets the implementer take advantage of existing extension and certificate selection logic and more easily remember which extensions were sent in the ClientHello.¶
It is also possible to implement this API outside of the TLS library using TLS exporters. This may be preferable in cases where the application does not have access to a TLS library with these APIs or when TLS is handled independently of the application layer protocol.¶
The "validate" API takes as input:¶
It returns a status to indicate whether the authenticator is valid or not after applying the function for validating the certificate chain to the chain contained in the authenticator. If validation is successful, it also returns the identity, such as the certificate chain and its extensions.¶
The API should return a failure if the certificate_request_context of the authenticator was used in a different authenticator that was previously validated. Well-formed empty authenticators are returned as invalid.¶
When validating an authenticator, constant-time comparison should be used.¶
IANA is requested to update the entry for server_name(0) in the registry for ExtensionType (defined in [RFC8446]) by replacing the value in the "TLS 1.3" column with the value "CH, EE, CR" and adding this document in the "Reference" column.¶
IANA is also requested to add the following note to the registry:¶
The addition of the "CR" to the "TLS 1.3" column for the server_name(0) extension only marks the extension as valid in a ClientCertificateRequest created as part of client-generated authenticator requests.¶
IANA is requested to add the following entries to the registry for Exporter Labels (defined in [RFC5705]): "EXPORTER-client authenticator handshake context", "EXPORTER-server authenticator handshake context", "EXPORTER-client authenticator handshake context", "EXPORTER-client authenticator finished key" and "EXPORTER-server authenticator finished key" with "DTLS-OK" and "Recommended" set to "Y" and this document added to the "Reference" column.¶
IANA is requested to add the following entry to the registry for HandshakeType (defined in [RFC8446]): "client_certificate_request" with "DTLS-OK" and "Recommended" set to "Y" and this document added to the "Reference" column with the following in the "Note" column: "Used in TLS versions prior to 1.3."¶
The Certificate/Verify/Finished pattern intentionally looks like the TLS 1.3 pattern which now has been analyzed several times. For example, [SIGMAC] presents a relevant framework for analysis, and section 10. of [RFC8446] contains a conprehensive set of references.¶
Authenticators are independent and unidirectional. There is no explicit state change inside TLS when an authenticator is either created or validated. The application in possession of a validated authenticator can rely on any semantics associated with data in the certificate_request_context.¶
The signatures generated with this API cover the context string "Exported Authenticator" and therefore cannot be transplanted into other protocols.¶
In TLS 1.3 the client can not explicitly learn from the TLS layer whether its Finished message was accepted. Because the application traffic keys are not dependent on the client's final flight, receiving messages from the server does not prove that the server received the client's Finished. To avoid disagreement between the client and server on the authentication status of EAs, servers MUST verify the client Finished before sending an EA or processing a received EA.¶
Comments on this proposal were provided by Martin Thomson. Suggestions for Section 9 were provided by Karthikeyan Bhargavan.¶