Proving the TLS handshake secure (as it is)

The TLS Internet Standard features a mixed bag of cryptographic algorithms and constructions, letting clients and servers negotiate their use for each run of the handshake. Although many ciphersuites are now well-understood in isolation, their composition remains problematic, and yet it is critical to obtain practical security guarantees for TLS, as all mainstream implementations support multiple related runs of the handshake and share keys between algorithms. We study the provable security of the TLS handshake, as it is implemented and deployed. To capture the details of the standard and its main extensions, we rely on miTLS, a verified reference implementation of the protocol. We propose new agile security definitions and assumptions for the signatures, key encapsulation mechanisms (KEM), and key derivation algorithms used by the TLS handshake. To validate our model of key encapsulation, we prove that both RSA and Diffie-Hellman ciphersuites satisfy our definition for the KEM. In particular, we formalize the use of PKCS#1v1.5 and build a 3,000-line EasyCrypt proof of the security of the resulting KEM against replayable chosen-ciphertext attacks under the assumption that ciphertexts are hard to re-randomize. Based on our new agile definitions, we construct a modular proof of security for the miTLS reference implementation of the handshake, including ciphersuite negotiation, key exchange, renegotiation, and resumption, treated as a detailed 3,600-line executable model. We present our main definitions, constructions, and proofs for an abstract model of the protocol, featuring series of related runs of the handshake with different ciphersuites. We also describe its refinement to account for the whole reference implementation, based on automated verification tools.