Independently Checkable Proofs from Decision Procedures: Issues and Progress

Bitvectors are foundational for automated reasoning. A few interactive theorem provers (ITP), such as Lean, have strong support for deciding fixed-width bitvector predicates by means of bitblasting. However, even these ITPs provide little automation for width-independent bitvector predicates. To fill this gap, we contribute novel, mechanized decision procedures for width-independent bitvector predicates in Lean. Classical algorithms to decide fragments of width-independent bitvector theory can be viewed from the lens of model checking, where the formula corresponds to an automaton and the correctness of the formula is a safety property. However, we cannot currently use this lens in mechanized proofs, as there are no executable, fast, and formally verified model checking algorithms that can be used interactively from within ITPs. To fill this gap, we mechanize key algorithms in the model checking literature: k -induction, automata reachability, automata emptiness checking, and automata minimization. Using these mechanized algorithms, we contribute scalable, mechanized, decision procedures for width-independent bitvector predicates. Furthermore, for controlled fragments of mixtures of arithmetic and bitwise operations which occur in the deobfuscation literature, we mechanize a recent fast algorithm (MBA-Blast), which outperforms the more general procedures on this fragment. Finally, we evaluate our decision procedures on benchmarks from classical compiler problems such as Hacker’s Delight and the LLVM peephole optimizer, as well as on equivalence checking problems for program obfuscation. Our tools solve 100% of Hacker’s Delight, two of our tools solve 100% of the deobfuscation dataset, and up to 27% of peephole rewrites extracted from LLVM’s peephole rewriting test suite. Our new decision procedures provide a push-button experience for width-independent bitvector reasoning in interactive theorem provers, and, more broadly, pave the way for foundational algorithms for fast, formally verified model checking.

Modern proof assistants such as Coq and Isabelle provide high degrees of expressiveness and assurance because they support formal reasoning in higher-order logic and supply explicit machine-checkable proof objects. Unfortunately, large scale proof development in these proof assistants is still an extremely difficult and time-consuming task. One major weakness of these proof assistants is the lack of a single language where users can develop complex tactics and decision procedures using a rich programming model and in a typeful manner. This limits the scalability of the proof development process, as users avoid developing domain-specific tactics and decision procedures. In this paper, we present VeriML - a novel language design that couples a type-safe effectful computational language with first-class support for manipulating logical terms such as propositions and proofs. The main idea behind our design is to integrate a rich logical framework - similar to the one supported by Coq - inside a computational language inspired by ML. The language design is such that the added features are orthogonal to the rest of the computational language, and also do not require significant additions to the logic language, so soundness is guaranteed. We have built a prototype implementation of VeriML including both its type-checker and an interpreter. We demonstrate the effectiveness of our design by showing a number of type-safe tactics and decision procedures written in VeriML.

Proof Certificates for Equality Reasoning

Nowadays, formal methods rely on tools of different kinds: proof assistants with which the user interacts to discover a proof step by step; and fully automated tools which make use of (intricate) decision procedures. But while some proof assistants can check the soundness of a proof, they lack automation. Regarding automated tools, one still has to be satisfied with their answers Yes/No/Do not know, the validity of which can be subject to question, in particular because of the increasing size and complexity of these tools.In the context of rewriting techniques, we aim at bridging the gap between proof assistants that yield formal guarantees of reliability and highly automated tools one has to trust. We present an approach making use of both shallow and deep embeddings. We illustrate this approach with a prototype based on the CiME rewriting toolbox, which can discover involved termination proofs that can be certified by the Coq proof assistant, using the Coccinelle library for rewriting.KeywordsModule TypeAutomate ToolProof AssistantGraph CriterionStrongly Connect ComponentThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Many applications from artificial intelligence and formal methods use decision procedures as their core solving engines. In this context, automated reasoning based on Satisfiability (SAT) or Satisfiability Modulo Theories (SMT) is very effective. For a given application, however, selecting the best reasoning engine is a daunting task requiring first-hand experience and insight into engine-specific implementation details. Developers have to decide which concrete engine to use and how to integrate the engine into an application. Although file formats, e.g., DIMACS CNF or SMT-LIB, standardize the input of SAT and SMT solvers, not all engines provide input interfaces compliant with these standards. When following the standard, advanced (and not standardized) features of the solvers remain unused and their integration is left to the users. This work presents metaSMT, a framework that eases the integration of existing reasoning engines into applications. Inspired by SMT-LIB, metaSMT provides a domain-specific language that allows for engine-independent programming and offers a generic interface to advanced features as an extra abstraction layer. State-of-the-art solvers for satisfiability and other theories are available via metaSMT with little programming effort. Language bindings for C++ and Python are provided. We show how metaSMT can be used as a portfolio consistency checker for SMT-LIB2 instances. The benchmark set of the category quantifier-free bit-vector theory from SMT-LIB (1.6 GB) is used for these experiments.

Modern proof assistants such as Coq and Isabelle provide high degrees of expressiveness and assurance because they support formal reasoning in higher-order logic and supply explicit machine-checkable proof objects. Unfortunately, large scale proof development in these proof assistants is still an extremely difficult and time-consuming task. One major weakness of these proof assistants is the lack of a single language where users can develop complex tactics and decision procedures using a rich programming model and in a typeful manner. This limits the scalability of the proof development process, as users avoid developing domain-specific tactics and decision procedures.In this paper, we present VeriML - a novel language design that couples a type-safe effectful computational language with first-class support for manipulating logical terms such as propositions and proofs. The main idea behind our design is to integrate a rich logical framework - similar to the one supported by Coq - inside a computational language inspired by ML. The language design is such that the added features are orthogonal to the rest of the computational language, and also do not require significant additions to the logic language, so soundness is guaranteed. We have built a prototype implementation of VeriML including both its type-checker and an interpreter. We demonstrate the effectiveness of our design by showing a number of type-safe tactics and decision procedures written in VeriML.

International audience

TLAPS, the TLA + proof system, is a platform for the development and mechanical verification of TLA + proofs. The TLA + proof language is declarative, and understanding proofs requires little background beyond elementary mathematics. The language supports hierarchical and non-linear proof construction and verification, and it is independent of any verification tool or strategy. Proofs are written in the same language as specifications; engineers do not have to translate their high-level designs into the language of a particular verification tool. A proof manager interprets a TLA + proof as a collection of proof obligations to be verified, which it sends to backend verifiers that include theorem provers, proof assistants, SMT solvers, and decision procedures.The first public release of TLAPS is available from [1], distributed with a BSD-like license. It handles almost all the non-temporal part of TLA + as well as the temporal reasoning needed to prove standard safety properties, in particular invariance and step simulation, but not liveness properties. Intuitively, a safety property asserts what is permitted to happen; a liveness property asserts what must happen; for a more formal overview, see [3,10].KeywordsTheorem ProverProof SystemSafety PropertyProof ObligationProof AssistantThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Intermediate verification languages like Why3 and Boogie have made it much easier to build program verifiers, transforming the process into a logic compilation problem rather than a proof automation one. Why3 in particular implements a rich logic for program specification with polymorphism, algebraic data types, recursive functions and predicates, and inductive predicates; it translates this logic to over a dozen solvers and proof assistants. Accordingly, it serves as a backend for many tools, including Frama-C, EasyCrypt, and GNATProve for Ada SPARK. But how can we be sure that these tools are correct? The alternate foundational approach, taken by tools like VST and CakeML, provides strong guarantees by implementing the entire toolchain in a proof assistant, but these tools are harder to build and cannot directly take advantage of SMT solver automation. As a first step toward enabling automated tools with similar foundational guarantees, we give a formal semantics in Coq for the logic fragment of Why3. We show that our semantics are useful by giving a correct-by-construction natural deduction proof system for this logic, using this proof system to verify parts of Why3’s standard library, and proving sound two of Why3’s transformations used to convert terms and formulas into the simpler logics supported by the backend solvers.

Preface: Volume 58, Issue 2

Recursive algebraic data types (term algebras, ADTs) are one of the most well-studied theories in logic, and find application in contexts including functional programming, modelling languages, proof assistants, and verification. At this point, several state-of-the-art theorem provers and SMT solvers include tailor-made decision procedures for ADTs, and version~2.6 of the SMT-LIB standard includes support for ADTs. We study an extremely simple approach to decide satisfiability of ADT constraints, the reduction of ADT constraints to equisatisfiable constraints over uninterpreted functions (EUF) and linear integer arithmetic (LIA). We show that the reduction approach gives rise to both decision and Craig interpolation procedures in (extensions of) ADTs.

We present a mechanically-verified proof checker developed with the ACL2 theorem-proving system that is general enough to support the growing variety of increasingly complex satisfiability (SAT) solver techniques, including those based on extended resolution. A common approach to assure the correctness of SAT solvers is to emit a proof of unsatisfiability when no solution is reported to exist. Contemporary proof checkers only check logical equivalence using resolution-style inference. However, some state-of-the-art, conflict-driven clause-learning SAT solvers use preprocessing, inprocessing, and learning techniques, that cannot be checked solely by resolution-style inference. We have developed a mechanically-verified proof checker that assures refutation clauses preserve satisfiability. We believe our approach is sufficiently expressive to validate all known SAT-solver techniques.KeywordsUnit PropagationConjunctive Normal FormUnit ClauseConjunctive Normal Form FormulaEmpty ClauseThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

We formalize a Hoare logic for the partial correctness of while programs in PVS and prove its soundness and relative completeness. We use the PVS higher-order logic to define the syntax and semantics of a small imperative programming language, and describe a proof system for Hoare triples involving programs in this language. We prove the soundness of the proof system by demonstrating that only valid triples are provable. We also demonstrate the relative completeness of the proof system by defining the weakest liberal precondition operator and using it to prove that all valid Hoare triples are provable modulo the assertion logic. Finally, we verify a verification condition generator for Hoare logic Variants of Hoare logic have been formalized before in PVS and using other interactive proof assistants. We use Hoare logic as a tutorial exercise to illustrate the effective use of PVS in capturing the syntax and semantics of embedded logics. The embedding of Hoare logic is simple enough to be easily reproduced by the reader, but it also illustrates some of the nuances of formalization and proof using PVS, in particular, and higher-order logic, in general.

Separation logic (SL) is a logical formalism for reasoning about programs that use pointers to mutate data structures. It is successful for program verification as an assertion language to state properties about memory heaps using Hoare triples. Most of the proof systems and verification tools for ${\textrm{SL}}$ focus on the decidable but rather restricted symbolic heaps fragment. Moreover, recent proof systems that go beyond symbolic heaps are purely syntactic or labelled systems dedicated to some fragments of ${\textrm{SL}}$ and they mainly allow either the full set of connectives, or the definition of arbitrary inductive predicates, but not both. In this work, we present a labelled proof system, called ${\textrm{G}_{\textrm{SL}}}$, that allows both the definition of cyclic proofs with arbitrary inductive predicates and the full set of SL connectives. We prove its soundness and show that we can derive in ${\textrm{G}_{\textrm{SL}}}$ the built-in rules for data structures of another non-cyclic labelled proof system and also that ${\textrm{G}_{\textrm{SL}}}$ is strictly more powerful than that system.

Modern SAT solvers can emit independently-checkable proof certificates to validate their results. The state-of-the-art proof system that allows for compact proof certificates is propagation redundancy (textsf{PR}). However, the only existing method to validate proofs in this system with a formally verified tool requires a transformation to a weaker proof system, which can result in a significant blowup in the size of the proof and increased proof validation time. This article describes the first approach to formally verify textsf{PR} proofs on a succinct representation. We present (i) a new Linear PR (LPR) proof format, (ii) an extension of the DPR-trim tool to efficiently convert textsf{PR} proofs into LPR format, and (iii) cake_lpr, a verified LPR proof checker developed in CakeML. We also enhance these tools with (iv) a new compositional proof format designed to enable separate (parallel) proof checking. The LPR format is backwards compatible with the existing LRAT format, but extends LRAT with support for the addition of textsf{PR} clauses. Moreover, cake_lpr is verified using CakeML ’s binary code extraction toolchain, which yields correctness guarantees for its machine code (binary) implementation. This further distinguishes our clausal proof checker from existing checkers because unverified extraction and compilation tools are removed from its trusted computing base. We experimentally show that: LPR provides efficiency gains over existing proof formats; cake_lpr ’s strong correctness guarantees are obtained without significant sacrifice in its performance; and the compositional proof format enables scalable parallel proof checking for large proofs.