RFC 8737 | ACME-TLS-ALPN | February 2020 |
Shoemaker | Standards Track | [Page] |
This document specifies a new challenge for the Automated Certificate Management Environment (ACME) protocol that allows for domain control validation using TLS.¶
This is an Internet Standards Track document.¶
This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841.¶
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc8737.¶
Copyright (c) 2020 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 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.¶
The Automatic Certificate Management Environment (ACME) [RFC8555] specification describes methods for validating control of domain names via HTTP and DNS. Deployment experience has shown it is also useful to be able to validate domain control using the TLS layer alone. In particular, this allows hosting providers, Content Distribution Networks (CDNs), and TLS-terminating load balancers to validate domain control without modifying the HTTP handling behavior of their backends.¶
This document specifies a new TLS-based challenge type, tls-alpn-01. This challenge requires negotiating a new application-layer protocol using the TLS Application-Layer Protocol Negotiation (ALPN) Extension [RFC7301]. Because this protocol does not build on a pre-existing deployment base, the ability to complete tls-alpn-01 challenges requires changes by service providers, making it explicitly an opt-in process. Because service providers must proactively deploy new code in order to implement tls-alpn-01, we can specify stronger controls in that code, resulting in a stronger validation method.¶
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.¶
The TLS with Application-Layer Protocol Negotiation (TLS ALPN) validation method proves control over a domain name by requiring the ACME client to configure a TLS server to respond to specific connection attempts using the ALPN extension with identifying information. The ACME server validates control of the domain name by connecting to a TLS server at one of the addresses resolved for the domain name and verifying that a certificate with specific content is presented.¶
The tls-alpn-01 ACME challenge object has the following format:¶
The client prepares for validation by constructing a self-signed certificate that MUST contain an acmeIdentifier extension and a subjectAlternativeName extension [RFC5280]. The subjectAlternativeName extension MUST contain a single dNSName entry where the value is the domain name being validated. The acmeIdentifier extension MUST contain the SHA-256 digest [FIPS180-4] of the key authorization [RFC8555] for the challenge. The acmeIdentifier extension MUST be critical so that the certificate isn't inadvertently used by non-ACME software.¶
The acmeIdentifier extension is identified by the id-pe-acmeIdentifier object identifier (OID) in the id-pe arc [RFC5280]:¶
id-pe-acmeIdentifier OBJECT IDENTIFIER ::= { id-pe 31 }¶
The extension has the following ASN.1 [X.680] format :¶
Authorization ::= OCTET STRING (SIZE (32))¶
The extnValue of the id-pe-acmeIdentifier extension is the ASN.1 DER encoding [X.690] of the Authorization structure, which contains the SHA-256 digest of the key authorization for the challenge.¶
Once this certificate has been created, it MUST be provisioned such that it is returned during a TLS handshake where the "acme-tls/1" application-layer protocol has been negotiated and a Server Name Indication (SNI) extension [RFC6066] has been provided containing the domain name being validated.¶
A client responds by POSTing an empty JSON object ({}) to the challenge URL to acknowledge that the challenge is ready to be validated by the server. The base64url encoding of the protected headers and payload is described in Section 6.1 of [RFC8555].¶
POST /acme/authz/1234/1 Host: example.com Content-Type: application/jose+json { "protected": base64url({ "alg": "ES256", "kid": "https://example.com/acme/acct/1", "nonce": "JHb54aT_KTXBWQOzGYkt9A", "url": "https://example.com/acme/authz/1234/1" }), "payload": base64url({}), "signature": "Q1bURgJoEslbD1c5...3pYdSMLio57mQNN4" }¶
On receiving this request from a client, the server constructs and stores the key authorization from the challenge "token" value and the current client account key.¶
The server then verifies the client's control over the domain by verifying that the TLS server was configured as expected using the following steps:¶
The ACME server verifies that during the TLS handshake the application-layer protocol "acme-tls/1" was successfully negotiated (and that the ALPN extension contained only the value "acme-tls/1") and that the certificate returned contains:¶
The comparison of dNSNames MUST be case insensitive [RFC4343]. Note that as ACME doesn't support Unicode identifiers, all dNSNames MUST be encoded using the rules of [RFC3492].¶
If all of the above steps succeed, then the validation is successful. Otherwise, it fails.¶
The "acme-tls/1" protocol MUST only be used for validating ACME tls-alpn-01 challenges. The protocol consists of a TLS handshake in which the required validation information is transmitted. The "acme-tls/1" protocol does not carry application data. Once the handshake is completed, the client MUST NOT exchange any further data with the server and MUST immediately close the connection. While this protocol uses X.509 certificates, it does not use the authentication method described in [RFC5280] and, as such, does not require a valid signature on the provided certificate nor require the TLS handshake to complete successfully. An ACME server may wish to use an off-the-shelf TLS stack where it is not simple to allow these divergences in the protocol as defined. Because of this, an ACME server MAY choose to withhold authorization if either the certificate signature is invalid or the handshake doesn't fully complete.¶
ACME servers that implement "acme-tls/1" MUST only negotiate TLS 1.2 [RFC5246] or higher when connecting to clients for validation.¶
The design of this challenge relies on some assumptions centered around how an HTTPS server behaves during validation.¶
The first assumption is that when an HTTPS server is being used to serve content for multiple DNS names from a single IP address, it properly segregates control of those names to the users that own them. This means that if User A registers Host A and User B registers Host B, the HTTPS server should not allow a TLS request using an SNI value for Host A to be served by User B or a TLS connection with a server_name extension identifying Host B to be answered by User A. If the HTTPS server allows User B to serve this request, it allows them to illegitimately validate control of Host A to the ACME server.¶
The second assumption is that a server will not violate [RFC7301] by blindly agreeing to use the "acme-tls/1" protocol without actually understanding it.¶
To further mitigate the risk of users claiming domain names used by other users on the same infrastructure hosting providers, CDNs, and other service providers SHOULD NOT allow users to provide their own certificates for the TLS ALPN validation process. If providers wish to implement TLS ALPN validation, they SHOULD only generate certificates used for validation themselves and not expose this functionality to users.¶
The extensions to the ACME protocol described in this document build upon the Security Considerations and threat model defined in Section 10.1 of [RFC8555].¶
Within the "Structure of Management Information (SMI) Numbers (MIB Module Registrations)" registry, the following entry has been added to the "SMI Security for PKIX Certificate Extension" (1.3.6.1.5.5.7.1) table.¶
Decimal | Description | References |
---|---|---|
31 | id-pe-acmeIdentifier | RFC 8737 |
Within the "Transport Layer Security (TLS) Extensions" registry, the following entry has been added to the "TLS Application-Layer Protocol Negotiation (ALPN) Protocol IDs" table.¶
Protocol | Identification Sequence | Reference |
---|---|---|
acme-tls/1 | 0x61 0x63 0x6d 0x65 0x2d 0x74 0x6c 0x73 0x2f 0x31 ("acme-tls/1") | RFC 8737 |
Within the "Automated Certificate Management Environment (ACME) Protocol" registry, the following entry has been added to the "ACME Validation Methods" registry.¶
Label | Identifier Type | ACME | Reference |
---|---|---|---|
tls-alpn-01 | dns | Y | RFC 8737 |
The TLS ALPN challenge exists to iterate on the TLS SNI challenge defined in the early ACME drafts. The TLS SNI challenge was convenient for service providers who were either operating large TLS-layer load balancing systems at which they wanted to perform validation or running servers fronting large numbers of DNS names from a single host as it allowed validation purely within the TLS layer. The value provided by the TLS SNI challenge was considered large enough that this document was written in order to provide a new challenge type that addressed the existing security concerns.¶
A security issue in the TLS SNI challenge was discovered by Frans Rosen, which allowed users of various service providers to illegitimately validate control of the DNS names of other users of the provider. When the TLS SNI challenge was designed, it was assumed that a user would only be able to respond to TLS traffic via SNI for domain names they had registered with a service provider (i.e., if a user registered 'a.example', they would only be able to respond to SNI requests for 'a.example' and not for SNI requests for 'b.example'). It turns out that a number of large service providers do not honor this property. Because of this, users were able to respond to SNI requests for the names used by the TLS SNI challenge validation process. This meant that (1) if User A and User B had registered Host A and Host B, respectively, User A would be able to claim the constructed SNI challenge name for Host B, and (2) when the validation connection was made, User A would be able to answer, thereby proving 'control' of Host B. As the SNI name used was a subdomain of the domain name being validated, rather than the domain name itself, it was likely to not already be registered with the service provider for traffic routing, making it much easier for a hijack to occur.¶
The author would like to thank all those that provided design insights and editorial review of this document, including Richard Barnes, Ryan Hurst, Adam Langley, Ryan Sleevi, Jacob Hoffman-Andrews, Daniel McCarney, Marcin Walas, Martin Thomson, and especially Frans Rosen, who discovered the vulnerability in the TLS SNI method that necessitated the writing of this specification.¶