Model checkers and sequential equivalence checkers have become essential tools for the semiconductor industry in recent years. The Hardware Model Checking Competition (HWMCC) was founded in 2006 with the purpose of intensifying research interest in t
The article presents a case study comparing two types of properties for formal verification of security requirements in embedded systems. —Wen Chen, NXP
Nowadays embedded devices collect various kinds of information and provide it to communication networks for further processing. These devices often provide critical functionalities that could be exploited by malicious parties. Using formal techniques is a natural way to increase the confidence in the overall embedded system security. However, the major research focus is on verifying only the correctness of encryption algorithms and their implementation in software and hardware and not the whole security process. Following novel research studies, many security requirements of an embedded architecture can be specified as Taint Properties, expressing properties related to information flow and access control. In this paper we define Taint Properties as a way to find out whether there is a path from src to dest, where src is an RTL signal seeded with the Taint and dest is a signal not to be reached by the Taint in order to satisfy the security requirements. In our scenario a Taint is an untrusted code following an illegal path from src to dest. We present a systematic approach to formalize generic security requirements, referring to a model abstraction, and their related Taint Properties of an embedded architecture. First, we present our model abstraction of two selected embedded secure architectures, then we define a portfolio of Taint Properties to verify key secrecy, isolation, attestation, confidentiality and availability features. We finally perform verification of previously defined formal security properties, hence presenting results on two selected embedded architectures proving the effectiveness of our approach.
This paper introduces a new technique for a fast computation of the Cone-Of-Influence (COI) of multiple properties. It specifically addresses frameworks where multiple properties belongs to the same model, and they partially or fully share their COI. In order to avoid multiple repeated visits of the same circuit sub-graph representation, it proposes a new algorithm, which performs a single topological visit of the variable dependency graph. It also studies mutual relationships among different properties, based on the overlapping of their COIs. It finally considers state variable scoring, based on their own COIs and/or their appearance in multiple COIs, as a new statistic for variable sorting and grouping/clustering in various Model Checking algorithms. Preliminary results show the advantages, and potential applications of these idea
Though modern microprocessors embed several hardware security mechanisms, aimed at guaranteeing confidentiality and integrity of sensible data, recently disclosed attacks such as Spectre and Meltdown witness weaknesses with potentially great impact on CPU security. Both vulnerabilities exploit speculative execution of modern high-performance micro-architectures, allowing the attacker to observe data leaked via a memory side channel, during speculated and mispredicted instructions. In this paper we present a methodology to formally verify, by means of a model checker, speculative vulnerabilities, such as the class of Spectre/Meltdown attacks, in microprocessors based on speculative execution. In detail, we discuss the problem of formally verifying confidentiality violations, since we deem it will help preventing new vulnerabilities of the same typology. We describe our methodology on a pipelined CPU inspired by the DLX RISC processor architecture. Due to scalability issues, and following related approaches in formal verification of correctness, our approach simplifies the model under verification by proper abstraction and reduction steps. The approach is based on flushing the pipeline, abstracting data and most of the speculative execution logic, keeping a subset of control data, plus speculated data state and tainting logic. Illegal propagation (data leakage) is encoded in terms of taint propagation, from a protected/invalid memory address to the address bus on a subsequent memory read, affecting the cache. We introduce the theoretical flow, relying on known theoretical results combined and exploited to prove soundness and completeness. Finally, using a state-of-the-art model checking tool, we present preliminary data on formal verification based on Bounded Model Checking, that to support our claims and highlight the feasibility of the approach.
Many embedded systems, like medical, sensing, automotive, military, require basic security functions, often referred to as "secure communications". Nowadays, interest has been growing around defining new security related properties, expressing relationships with information flow and access control. In particular, novel research works are focused on formalizing generic security requirements as propagation properties. These kinds of properties, we name them Path properties, are used to see whether it is possible to leak secure data via unexpected paths. In this paper we compare Path properties, described above, with formal security properties expressed in CTL Logic, named Taint properties. We also compare two verification techniques used to verify Path and Taint properties considering an abstraction of a Secure Embedded Architecture discussing the advantages and drawbacks of each approach.