No MTI Crypto without Public Review
draft-rsalz-drbg-speck-wap-wep-00

Abstract

Cryptography is becoming more important to the IETF and its protocols, and more IETF protocols are using, or looking at, cryptography to increase privacy on the Internet [RFC7258].

This document specifies a proposed best practice for any protocol (or data format) that uses cryptography. Namely, that such RFCs cannot specify an algorithm as mandatory-to-implement (MTI) unless that algorithm has had reasonable public review. This document also “sketches out” a rough definition around what such a review would look like.

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 http://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 September 22, 2016.

Copyright Notice

Copyright (c) 2016 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 (http://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.

Table of Contents

• 1. Terminology
• 2. Introduction
• 3. Why is Cryptography Hard?
• 4. Things to avoid
• 5. How to Do it Right
• 6. Acknowledgements
• 7. References
• 7.1. Normative References
• 7.2. Informative References
• Author's Address

1. Terminology

The key words “MUST”, “MUST NOT”, “REQUIRED”, “SHALL”, “SHALL NOT”, “SHOULD”, “SHOULD NOT”, “RECOMMENDED”, “MAY”, and “OPTIONAL” in this document are to be interpreted as described in RFC 2119 [RFC2119].

The term “snake oil” is used as a pejorative for something which appears to do its job acceptably, but actually does not; see https://en.wikipedia.org/wiki/Snake_oil_%28cryptography%29. It is a goal of the IETF that we never be mislead into being, or mistakenly taken as, snake oil salesman.

2. Introduction

Cryptography is becoming more important to the IETF and its protocols, and more IETF protocols are using, or looking at, cryptography to increase privacy on the Internet [RFC7258].

This document specifies a proposed best practice for any protocol (or data format) that uses cryptography. Namely, that such RFCs cannot specify an algorithm as mandatory-to-implement (MTI) unless that algorithm has had reasonable public review. This document also “sketches out” a rough definition around what such a review would look like.

3. Why is Cryptography Hard?

Cryptography is hard because it is not like traditional IETF protocol deployments. In this classic situation, if one one party implements a protocol incorrectly, it usually becomes obvious as interoperability suffers or completely fails. But with cryptography, one party can have implementation defects, or known exploitable weaknesses, that expose the entire communication stream to an attacker. Open source and code reviews are not a panacea here, but using only widely-accepted cryptographic mechanisms (e.g., avoiding facilities like https://en.wikipedia.org/wiki/Dual_EC_DRBG) will reduce the attack surface.

Cryptography is hard because subtle design characteristics can have disastrous consequences. For example, the US Digital Signatiure Algorithm requires the random nonce to be protected and never re-used. If those requirements are not met, the private key can be leaked.

Cryptography is hard because adversaries design new attacks and refine existing ones. Attacks get better over time; they never get worse. For example, it is now de riguer to protect against CPU timing attacks, even when the device is only viewable over a network. A recent paper (XXX reference) can identify a private key if your smartphone is just laid next to an innocuous charging device. We understand power differential attacks, timing attacks, and perhaps cache line attacks; we now have to think about RFI emissions from our phone.

Cryptography is hard because the order of operations can matter. It is not intuitively obvious to most developers, which should come first among signing, compression and encryption. This issues was first raised in Spring of 2001 (“Defective Sign & Encrpyt in S/MIME, PKCS#7, MOSS, PEM, and XML” by Don Davis) but was only addressed in TLS by [RFC7366] more than a dozen years later.

Getting the cryptography right is important because the Internet, and therefore the work of the IETF, has become a tempting target for all types of attackers, from individual “script kiddies,” through criminal commercial botnet and phishing ventures, up to national-scale adversaries operating on behalf of their nation-state.

4. Things to avoid

“Sunlight is said to be the best of disinfectants; electric light the most efficient policeman.” – Louis Brandeis, Other People’s Money and How Bankers Use it, first published as a set of articles in Harper’s Weekly.

Cryptography that is developed in private, such as among an industry consortium is a bad idea. Notable examples of this include:

• A5/1 and A5/2 for GSM-based mobile phones.
• WEP and WPA for WiFi access.

It is hard to get good public review of patented cryptography, unless there is a strongly compelling need. For example, decades ago RSA was the only practial public-key mechanism available. Part of the concern about patented cryptography is that the patent-holder has every incentive to provide that their system is good, while the rest of the world generally has little interest in proving that their commercial venture is bad. Examples of this include:

• Algebraic Eraser, presented at IETF-xx (XXX reference), and since then has had a number of effective attacks.
• XXX MORE NEEDED

5. How to Do it Right

Cryptographic agility, [RFC7696], is probably a MUST. While it has its detractors, there are no known (to the author) practical considerations to evolving a deployed based to stronger crypto, while still maintaining interoperability with existing entities. This requires being able to make informed choices about when to use old weak crypto, and when to use the “latest and greatest,” and while not much software, and essentially no end-users, are capable of making that choice, it seems sadly the best we can do.

The second consideration is to avoid temptation and premature optimization. Do not adopt an algorithm just because it seems “small and fast” or comes from “someone I respect.”

What constitutes sufficient public review? It is hard to say. An open competition, such as those that led to AES (XXX ref) and SHA-3 (XXX ref) seem to be good, even when they come from sources that are under widespread suspicion, like the US Government. These efforts, like the Password Hashing Competition https://password-hashing.net/, had wide international participation and analysis by many noted exports.

Papers presented in the various Crypto conferences (XXX need list) are good. Same for various Usenix workshops.

Proof by contest – “Nobody’s Claimed my $200 reward” – are useless for a number of reasons. They tend to be promoted by amateur cryptographers as a way to get attention, and if someone actually looks at them they are always cracked. Numerical analysis is a better approach, albeit much harder work.

6. Acknowledgements

Thanks to Stephen Farrell for instigating this.

7. References

7.1. Normative References

7.2. Informative References

Author's Address