New Protocols Must Require TLS 1.3

Status of This Memo

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 12 June 2025.¶

Copyright Notice

Copyright (c) 2024 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.¶

Table of Contents

1. Introduction

TLS 1.2 [TLS12] is in use and can be configured such that it provides good security properties. However, this protocol version suffers from several deficiencies:¶

1. While application layer traffic is always encrypted, most of the handshake messages are not. Therefore, the privacy provided is suboptimal. This is a protocol issue that cannot be addressed by configuration.¶

2. The list of cryptographic primitives specified for the protocol, both in-use primitives and deprecated ones, includes several primitives that have been a source for vulnerabilities throughout the years, such as RSA key exchange, CBC cipher suites, and problematic finite-field Diffie-Hellman group negotiation. These issues could be addressed through proper configuration; however, experience shows that configuration mistakes are common, especially when deploying cryptography. See Section 6 for elaboration.¶

3. The base protocol does not provide security against some types of attacks (see Section 6); extensions are required to provide security.¶

TLS 1.3 [TLS13] is also in widespread use and fixes most known deficiencies with TLS 1.2, such as encrypting more of the traffic so that it is not readable by outsiders and removing most cryptographic primitives considered dangerous. Importantly, TLS 1.3 enjoys robust security proofs and provides excellent security without any additional configuration.¶

This document specifies that, since TLS 1.3 use is widespread, new protocols must require and assume its existence. This prescription does not pertain to DTLS (in any DTLS version); it pertains to TLS only.¶

2. Conventions and Definitions

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.¶

3. Implications for post-quantum cryptography

Cryptographically-relevant quantum computers, once available, will have a huge impact on TLS traffic. In 2016, the US National Institute of Standards and Technology (NIST) started a multi-year effort to standardize algorithms that will be "safe" once quantum computers are feasible [PQC]. The first IETF discussions happened around the same time [CFRGSLIDES].¶

In 2024 NIST released standards for [ML-KEM], [ML-DSA], and [SLH-DSA]. While industry was waiting for NIST to finish standardization, the IETF has had several efforts underway. A working group was formed in early 2023 to work on operational and transitional uses of PQC in IETF protocols, [PQUIPWG]. Several other working groups, notably LAMPS [LAMPSWG] and TLS [TLSWG], are working on drafts to support hybrid algorithms and identifiers, for use during a transition from classic to a post-quantum world.¶

For TLS it is important to note that the focus of these efforts is TLS 1.3 or later: TLS 1.2 WILL NOT be supported (see Section 7). This is one more reason for new protocols to default to TLS 1.3, where post-quantum cryptography is expected to be supported.¶

4. TLS Use by Other Protocols and Applications

Any new protocol that uses TLS MUST specify as its default TLS 1.3. For example, QUIC [QUICTLS] requires TLS 1.3 and specifies that endpoints MUST terminate the connection if an older version is used.¶

If deployment considerations are a concern, the protocol MAY specify TLS 1.2 as an additional, non-default option. As a counter example, the Usage Profile for DNS over TLS [DNSTLS] specifies TLS 1.2 as the default, while also allowing TLS 1.3. For newer specifications that choose to support TLS 1.2, those preferences are to be reversed.¶

The initial TLS handshake allows a client to specify which versions of the TLS protocol it supports and the server is intended to pick the highest version that it also supports. This is known as the "TLS version negotiation," and many TLS libraries provide a way for applications to specify the range of versions. When the API allows it, clients SHOULD specify just the minimum version they want. This MUST be TLS 1.3 or TLS 1.2, depending on the circumstances described in the above paragraphs.¶

5. Changes to RFC 9325

This document makes two changes to the recommendations in [RFC9325], Section 3.1.1:¶

• That section says that TLS 1.3 SHOULD be supported; this document says that for new protocols it MUST be supported.¶

• That section says that TLS 1.2 MUST be supported; this document says that it MAY be supported as described above.¶

Again, these changes only apply to TLS, and not DTLS.¶

6. Security Considerations

TLS 1.2 was specified with several cryptographic primitives and design choices that have, over time, weakened its security. The purpose of this section is to briefly survey several such prominent problems that have affected the protocol. It should be noted, however, that TLS 1.2 can be configured securely; it is merely much more difficult to configure it securely as opposed to using its modern successor, TLS 1.3. See [RFC9325] for a more thorough guide on the secure deployment of TLS 1.2.¶

Firstly, the TLS 1.2 protocol, without any extension points, is vulnerable to renegotiation attacks (see [RENEG1] and [RENEG2]) and the Triple Handshake attack (see [TRIPLESHAKE]). Broadly, these attacks exploit the protocol's support for renegotiation in order to inject a prefix chosen by the attacker into the plaintext stream. This is usually a devastating threat in practice, that allows e.g. obtaining secret cookies in a web setting. In light of the above problems, [RFC5746] specifies an extension that prevents this category of attacks. To securely deploy TLS 1.2, either renegotiation must be disabled entirely, or this extension must be used. Additionally, clients must not allow servers to renegotiate the certificate during a connection.¶

Secondly, the original key exchange methods specified for the protocol, namely RSA key exchange and finite field Diffie-Hellman, suffer from several weaknesses. Similarly, to securely deploy the protocol, these key exchange methods must be disabled. See [I-D.draft-ietf-tls-deprecate-obsolete-kex] for details.¶

Thirdly, symmetric ciphers which were widely-used in the protocol, namely RC4 and CBC cipher suites, suffer from several weaknesses. RC4 suffers from exploitable biases in its key stream; see [RFC7465]. CBC cipher suites have been a source of vulnerabilities throughout the years. A straightforward implementation of these cipher suites inherently suffers from the Lucky13 timing attack [LUCKY13]. The first attempt to implement the cipher suites in constant time introduced an even more severe vulnerability [LUCKY13FIX]. There have been further similar vulnerabilities throughout the years exploiting CBC cipher suites; refer to e.g. [CBCSCANNING] for an example and a survey of similar works.¶

And lastly, historically the protocol was affected by several other attacks that TLS 1.3 is immune to: BEAST [BEAST], Logjam [WEAKDH], FREAK [FREAK], and SLOTH [SLOTH].¶

7. IANA Considerations

This document makes no requests to IANA.¶

8. References

Authors' Addresses