CipherH: Automated Detection of Ciphertext Side-channel Vulnerabilities in Cryptographic Implementations

The ciphertext side channel is a new type of side channels that exploits deterministic memory encryption of trusted execution environments (TEE). It enables the adversary with read accesses to the ciphertext of the encrypted memory, either logically or physically, to compromise cryptographic implementations protected by TEEs with high fidelity. Prior studies have concluded that the ciphertext side channel is a severe threat to not only AMD SEV-SNP, where the vulnerability was first discovered, but to all TEEs with deterministic memory encryption.

In this paper, we propose CipherH, a practical framework for automating the analysis of cryptographic software and detecting program points vulnerable to ciphertext side channels. CipherH is designed to perform a practical hybrid analysis in production cryptographic software, with a speedy dynamic taint analysis to track the usage of secrets throughout the entire program and a static symbolic execution procedure on each “tainted” function to reason about ciphertext side channel vulnerabilities using symbolic constraint. Empirical evaluation has led to the discovery of over 200 vulnerable program points from the state-of-the-art RSA and ECDSA/ECDH implementations from OpenSSL, MbedTLS, and WolfSSL. Representative cases have been reported to and confirmed or patched by the developers.