The complexity of browsers has steadily increased over the years, driven by the continuous introduction and update of Web platform components, such as novel Web APIs and security mechanisms. Their specifications are manually reviewed by experts to identify potential security issues. However, this process has proved to be error-prone due to the extensiveness of modern browser specifications and the interplay between new and existing Web platform components. To tackle this problem, we developed WebSpec, the first formal security framework for the analysis of browser security mechanisms, which enables both the automatic discovery of logical flaws and the development of machine-checked security proofs. WebSpec, in particular, includes a comprehensive semantic model of the browser in the Coq proof assistant, a formalization in this model of ten Web security invariants, and a toolchain turning the Coq model and the Web invariants into SMT-lib formulas to enable model checking with the Z3 theorem prover. If a violation is found, the toolchain automatically generates executable tests corresponding to the discovered attack trace, which is validated across major browsers.We showcase the effectiveness of WebSpec by discovering two new logical flaws caused by the interaction of different browser mechanisms and by identifying three previously discovered logical flaws in the current Web platform, as well as five in old versions. Finally, we show how WebSpec can aid the verification of our proposed changes to amend the reported inconsistencies affecting the current Web platform.

Abstract This paper introduces a new property calledrobust reachabilitywhich refines the standard notion of reachability in order to take replicability into account. A bug is robustly reachable if acontrolled inputcan make it so the bug is reached whatever the value ofuncontrolled input. Robust reachability is better suited than standard reachability in many realistic situations related to security (e.g., criticality assessment or bug prioritization) or software engineering (e.g., replicable test suites and flakiness). We propose a formal treatment of the concept, and we revisit existing symbolic bug finding methods through this new lens. Remarkably, robust reachability allows differentiating bounded model checking from symbolic execution while they have the same deductive power in the standard case. Finally, we propose the first symbolic verifier dedicated to robust reachability: we use it for criticality assessment of 4 existing vulnerabilities, and compare it with standard symbolic execution.

We focus in this paper on generating models of quantified first-order formulas over built-in theories, which is paramount in software verification and bug finding. While standard methods are either geared toward proving the absence of a solution or targeted to specific theories, we propose a generic and radically new approach based on a reduction to the quantifier-free case. Our technique thus allows to reuse all the efficient machinery developed for that context. Experiments show a substantial improvement over state-of-the-art methods.

The theory of arrays has a central place in software verification due to its ability to model memory or data structures. Yet, this theory is known to be hard to solve in both theory and practice, especially in the case of very long formulas coming from unrolling-based verification methods. Standard simplification techniques à la read-over-write suffer from two main drawbacks: they do not scale on very long sequences of stores and they miss many simplification opportunities because of a crude syntactic (dis-)equality reasoning. We propose a new approach to array formula simplification based on a new dedicated data structure together with original simplifications and low-cost reasoning. The technique is efficient, scalable and it yields significant simplification. The impact on formula resolution is always positive, and it can be dramatic on some specific classes of problems of interest, e.g. very long formula or binary-level symbolic execution. While currently implemented as a preprocessing, the approach would benefit from a deeper integration in an array solver.