International Association
for Cryptologic Research

We present a formally verified proof of the correctness and IND-CCA security of ML-KEM, the Kyber-based Key Encapsulation Mechanism (KEM) undergoing standardization by NIST. The proof is machine-checked in EasyCrypt and it includes: 1) A formalization of the correctness (decryption failure probability) and IND-CPA security of the Kyber base public-key encryption scheme, following Bos et al. at Euro S&P 2018; 2) A formalization of the relevant variant of the Fujisaki-Okamoto transform in the Random Oracle Model (ROM), which follows closely (but not exactly) Hofheinz, Hovelmanns and Kiltz at TCC 2017; 3) A proof that the IND-CCA security of the ML-KEM specification and its correctness as a KEM follows from the previous results; 4) Two formally verified implementations of ML-KEM written in Jasmin that are provably constant-time, functionally equivalent to the ML-KEM specification and, for this reason, inherit the provable security guarantees established in the previous points. The top-level theorems give self-contained concrete bounds for the correctness and security of ML-KEM down to (a variant of) Module-LWE. We discuss how they are built modularly by leveraging various EasyCrypt features.

In this paper we revisit the problem of erasing sensitive data from memory and registers during return from a cryptographic routine. While the problem and related attacker model is fairly easy to phrase, it turns out to be surprisingly hard to guarantee security in this model when implementing cryptography in common languages such as C/C++ or Rust. We revisit the issues surrounding zeroization and then present a principled solution in the sense that it guarantees that sensitive data is erased and it clearly defines when this happens. We implement our solution as extension to the formally verified Jasmin compiler and extend the correctness proof of the compiler to cover zeroization. We show that the approach seamlessly integrates with state-of-the-art protections against microarchitectural attacks by integrating zeroization into Libjade, a cryptographic library written in Jasmin with systematic protections against timing and Spectre-v1 attacks. We present benchmarks showing that in many cases the overhead of zeroization is barely measurable and that it stays below 2% except for highly optimized symmetric crypto routines on short inputs.