]> Dell EMC

176 South Street Hopkinton United States Kathleen.Moriarty.ietf@gmail.com

Trinity College Dublin

Dublin 2 Ireland +353-1-896-2354 stephen.farrell@cs.tcd.ie

Security Area Internet Engineering Task Force TLS deprecate TLSv1.0 TLSv1.1 This document [if approved] formally deprecates Transport Layer Security (TLS) versions 1.0 and 1.1 and moves these documents to the historic state. These versions lack support for current and recommended cipher suites, and various government and industry profiiles of applications using TLS now mandate avoiding these old TLS versions. TLSv1.2 has been the recommended version for IETF protocols since 2008, providing sufficient time to transition away from older versions. Products having to support older versions increase the attack surface unnecessarily and increase opportunities for misconfigurations. Supporting these older versions also requires additional effort for library and product maintenance. This document updates the backward compatibility sections of TLS RFCs [[list TBD]] to prohibit fallback to TLSv1.0 and TLSv1.1. This document also updates RFC 7525.

[[Text in double-square brackets is intended to be fixed as the draft evolves. You've seen that we need to figure out the list of RFCs that this'd update in the abstract. There is a repo for this at: https://github.com/sftcd/tls-oldversions-diediedie - PRs (on the xml file) are welcome there.]] Transport Layer Security (TLS) versions 1.0 and 1.1 were superceded by TLSv1.2 in 2008, which has now itself been superceded by TLSv1.3 . It is therefore timely to further deprecate these old versions. The expection is that TLSv1.2 will continue to be used for many years alongside TLSv1.3. TLSv1.1 and TLSv1.0 are also actively being deprecated in accordance with guidance from government agencies (e.g. NIST SP 80052r2) and industry consortia such as the Payment Card Industry Association (PCI) . The primary technical reasons for deprecating these versions include: They require implementation of older cipher suites that are no longer desirable for cryptographic reasons, e.g. TLSv1.0 makes TLS_DHE_DSS_WITH_3DES_EDE_CBC_SHA mandatory to implement Lack of support for current recommended cipher suites, especially using AEAD ciphers which are not supported prior to TLS 1.2. Note: registry entries for no-longer-desirable ciphersuites remain in the registries, but many TLS registries are being updated through which denotes such entries as "not recommended." Integrity of the handshake depends on SHA-1 hash Authentication of the peers depends on SHA-1 signatures Support for four protocol versions increases the likelihood of misconfiguration At least one widely-used library has plans to drop TLSv1.1 and TLSv1.0 support in upcoming releases; products using such libraries would need to use older versions of the libraries to support TLSv1.0 and TLSv1.1, which is clearly undesirable Deprecation of these versions is intended to assist developers as additional justification to no longer support older TLS versions and to migrate to a minimum of TLSv1.2. Deprecation also assists product teams with phasing out support for the older versions to reduce the attack surface and the scope of maintenance for protocols in their offerings. [[This draft is being written now so that the TLS WG chairs can just hit the "publication requested" button as soon as there is WG consensus to deprecate these ancient versions of TLS. The authors however think that deprecation now is timely.]]

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 when, and only when, they appear in all capitals, as shown here.

Industry has actively followed guidance provided by NIST and the PCI Council to deprecate TLSv1.0 and TLSv1.1 by June 30, 2018. TLSv1.2 should remain a minimum baseline for TLS support at this time. Specific details on attacks against TLSv1.0 and TLSv1.1 as well as their mitigations are provided in NIST SP800-52r2, RFC 7457 and other referenced RFCs. Although the attacks have been mitigated, if support is dropped for future library releases for these versions, it is unlikely attacks found going forward will be mitigated in older library releases. NIST for example have provided the following rationale, copied with permission from NIST SP800-52r2, section 1.2 "History of TLS" (with references changed for RFC formating). TLS 1.1, specified in , was developed to address weaknesses discovered in TLS 1.0, primarily in the areas of initialization vector selection and padding error processing. Initialization vectors were made explicit to prevent a certain class of attacks on the Cipher Block Chaining (CBC) mode of operation used by TLS. The handling of padding errors was altered to treat a padding error as a bad message authentication code, rather than a decryption failure. In addition, the TLS 1.1 RFC acknowledges attacks on CBC mode that rely on the time to compute the message authentication code (MAC). The TLS 1.1 specification states that to defend against such attacks, an implementation must process records in the same manner regardless of whether padding errors exist. Further implementation considerations for CBC modes (which were not included in RFC4346) are discussed in Section 3.3.2. TLS 1.2, specified in RFC5246, made several cryptographic enhancements, particularly in the area of hash functions, with the ability to use or specify the SHA-2 family algorithms for hash, MAC, and Pseudorandom Function (PRF) computations. TLS 1.2 also adds authenticated encryption with associated data (AEAD) cipher suites. TLS 1.3, specified in TLSv1.3, represents a significant change to TLS that aims to address threats that have arisen over the years. Among the changes are a new handshake protocol, a new key derivation process that uses the HMAC-based Extract-and-Expand Key Derivation Function (HKDF), and the removal of cipher suites that use static RSA or DH key exchanges, the CBC mode of operation, or SHA-1. The list of extensions that can be used with TLS 1.3 has been reduced considerably. The Canadian government treasury board have mandated that these old versions of TLS not be used.

[[This section can be removed upon publication - or maybe keep it?]] Support for TLSv1.0 has been removed by the July 2018 PCI deadline from the following standards, products, and services: 3GPP 5G Amazon Elastic Load Balancing CloudFare Digicert GitHub KeyCDN PayPal Stripe [[Numerous web sites...]] Many web sites have taken the action of including the deprecation of TLSv1.1 into their plans for deprecating TLSv1.0 for the PCI council deadline. Support for TLSv1.1 has been removed by the July 2018 PCI deadline from the following standards, products, and services: 3GPP 5G Release 16 Amazon Elastic Load Balancing CloudFare GitHub PayPal Stripe [[Numerous web sites...]]

[[This section can be removed upon publication - or maybe keep it?]]

Usage statistics for TLSv1.0 and TLSv1.1 on the public web vary, but have been in general very low and declined further with the impending PCI deadline to migrate off of TLSv1.0 by June 30, 2018. As of January 2018, quoted 4 percent of browsers using TLSv1.0. The number of websites supporting TLS 1.2 is still growing (+0.4%), and has reached 92% according to sslpulse as of June 19, 2018. Deprecating TLS 1.0 and TLS 1.1 will thus not have a major impact on browser or web server implementations. presents statistics for use of TLS versions in the web.

E-Mail uses TLS for SMTP, submission (port 587), POP/POP3 and IMAP. Typically email deployments lag public web deployments in terms of the rate of adoption of new TLS versions. presents statistics for use of TLS versions in the email applications.

presents statistics for use of TLS versions in operating systems.

10.0 | ~0.3 | - | - | | Windows svr [1]| 20180709 | - | ~1.5 | ~0.0 | - | - | | Apple [2] | 20180709 | - | 0.4 | - | 99.6 | - | +----------------+----------+------+-------+-------+-------+-------+ [1] https://www.ietf.org/mail-archive/web/tls/current/msg26577.html [2] https://www.ietf.org/mail-archive/web/tls/current/msg26634.html ]]>

presents statistics for use of TLS versions in the enterprise networks. The tcd.ie numbers below were the result of a student project and need further validation.

The integrity of both TLSv1.0 and TLSv1.1 depends on a running SHA-1 hash of the exchanged messages. This makes it possible to perform a downgrade attack on the handshake by an attacker able to perform 2^77 operations, well below the acceptable modern security margin. Similarly, the authentication of the handshake depends on signatures made using SHA-1 hash or a not stronger concatenation of MD-5 and SHA-1 hashes, allowing the attacker to impersonate a server when it is able to break the severly weakened SHA-1 hash. Neither TLSv1.0 nor TLSv1.1 allow the peers to select a stronger hash for signatures in the ServerKeyExchange or CertificateVerify messages, making the only upgrade path the use of a newer protocol version. See for additional detail.

TLSv1.0 MUST NOT be used. Negotiation of TLSv1.0 from any version of TLS MUST NOT be permitted. Any other version of TLS is more secure then TLSv1.0. TLSv1.0 can be configured to prevent interception, though using the highest version available is preferable. Pragmatically, clients MUST NOT send a ClientHello with ClientHello.client_version set to {03,01}. Similarly, servers MUST NOT send a ServerHello with ServerHello.server_version set to {03,01}. Any party receiving a Hello message with the protocol version set to {03,01} MUST respond with a "protocol_version" alert message and close the connection. Historically, TLS specifications were not clear on what the record layer version number (TLSPlaintext.version) could contain when sending ClientHello. Appendix E of [RFC5246] notes that TLSPlaintext.version could be selected to maximize interoperability, though no definitive value is identified as ideal. That guidance is still applicable; therefore, TLS servers MUST accept any value {03,XX} (including {03,00}) as the record layer version number for ClientHello, but they MUST NOT negotiate TLSv1.0. [[Text here is derived (or stolen:-) from ]]

TLSv1.1 MUST NOT be used. Negotiation of TLSv1.1 from any version of TLS MUST NOT be permitted. Pragmatically, clients MUST NOT send a ClientHello with ClientHello.client_version set to {03,02}. Similarly, servers MUST NOT send a ServerHello with ServerHello.server_version set to {03,02}. Any party receiving a Hello message with the protocol version set to {03,02} MUST respond with a "protocol_version" alert message and close the connection. Any newer version of TLS is more secure then TLSv1.1. TLSv1.1 can be configured to prevent interception, though using the highest version available is preferable. Support for TLSv1.1 is dwindling in libraries and will impact security going forward if mitagations for attacks cannot be easily addressed and supported in older libraries. Historically, TLS specifications were not clear on what the record layer version number (TLSPlaintext.version) could contain when sending ClientHello. Appendix E of [RFC5246] notes that TLSPlaintext.version could be selected to maximize interoperability, though no definitive value is identified as ideal. That guidance is still applicable; therefore, TLS servers MUST accept any value {03,XX} (including {03,00}) as the record layer version number for ClientHello, but they MUST NOT negotiate TLSv1.1.

[[This section was suggested in PR#2 by Hubert Kario. We're not clear if the WG would like this draft to include this or not, so will ask the TLS WG at the appropriate time.]] SHA-1 as a signature hash MUST NOT be used. That means that clients MUST send signature_algorithms extension and that extension MUST NOT include pairs that include SHA-1 hash. In particular, values {2, 1}, {2, 2} and {2, 3} MUST NOT be present in the extension. Note: this does not affect cipher suites that use SHA-1 HMAC for data integrity as the HMAC construction is still considered secure and when they are used in TLSv1.2 SHA-256 is used for handshake integrity.

[[Since RFC7525 is BCP195, there'll probably be some process-fun to do an update of that. Formally, it may be that this document becomes a new part of BCP195 I guess, but we can figure that out with chairs and ADs.]] This documents updates Section 3.1.1 changing SHOULD NOT to MUST NOT as follows: Implementations MUST NOT negotiate TLS version 1.0 . Rationale: TLS 1.0 (published in 1999) does not support many modern, strong cipher suites. In addition, TLS 1.0 lacks a per- record Initialization Vector (IV) for CBC-based cipher suites and does not warn against common padding errors. Implementations MUST NOT negotiate TLS version 1.1 . Rationale: TLS 1.1 (published in 2006) is a security improvement over TLS 1.0 but still does not support certain stronger cipher suites. This documents updates Section 3.1.2 changing SHOULD NOT to MUST NOT as follows: Implementations MUST NOT negotiate DTLS version 1.0 . Version 1.0 of DTLS correlates to version 1.1 of TLS (see above).

This document deprecates two older protocol versions for security reasons already described. The attack surface is reduced when there are a smaller number of supported protocols and fallback options are removed.

Thanks to those that provided usage data, reviewed and/or improved this document, including: David Benjamin, David Black, Viktor Dukhovni, Alessandro Ghedini, Jeremy Harris, Russ Housley, Hubert Kario, Loganaden Velvindron, Eric Mill, Yoav Nir, Andrei Popov, Eric Rescorla, and Yaron Sheffer.

[[This memo includes no request to IANA.]]

StackExchange - will be deleted before publication SSLpulse - will be deleted before publication National Institute of Standards and Technology PCI Security Standards Council Stripe Paypal GitHub CloudFlare Treasury Board of Canada Secretariat Digicert KeyCDN Amazon INRIA INRIA

[[RFC editor: please remove this before publication.]] From -00 to -01: Added stats sent to list so far PR's #2,3 a few more references added section on email