Interpretable noninterference measurement and its application to processor designs

Abstract

Noninterference measurement quantifies the secret information that might leak to an adversary from what the adversary can observe and influence about the computation. Static and high-fidelity noninterference measurement has been difficult to scale to complex computations, however. This paper scales a recent framework for noninterference measurement to the open-source RISC-V BOOM core as specified in Verilog, through three key innovations: logically characterizing the core's execution incrementally, applying specific optimizations between each cycle; permitting information to be declassified, to focus leakage measurement to only secret information that cannot be inferred from the declassified information; and interpreting leakage measurements for the analyst in terms of simple rules that characterize when leakage occurs. Case studies on cache-based side channels generally, and on specific instances including Spectre attacks, show that the resulting toolchain, called DINoMe, effectively scales to this modern processor design.