29 August 2022 in Proceedings of the ACM on Programming Languages
Session types have recently been integrated with functional languages, bringing message-passing concurrency to functional programming. Channel endpoints then become first-class and can be stored in data structures, captured in closures, and sent along channels. Representatives of the GV (Wadler's "Good Variation") session type family are of particular appeal because they not only assert session fidelity but also deadlock freedom, inspired by a Curry-Howard correspondence to linear logic. A restriction of current versions of GV, however, is the focus on binary sessions, limiting concurrent interactions within a session to two participants. This paper introduces Multiparty GV (MPGV), a functional language with multiparty session types, allowing concurrent interactions among several participants. MPGV upholds the strong guarantees of its ancestor GV, including deadlock freedom, despite session interleaving and delegation. MPGV has a novel redirecting construct for modular programming with first-class endpoints, thanks to which we give a type-preserving translation from binary session types to MPGV to show that MPGV is strictly more general than binary GV. All results in this paper have been mechanized using the Coq proof assistant.
9 January 2023 in Proceedings of the ACM on Programming Languages
Reasoning about concurrent programs is challenging, especially if data is shared among threads. Program correctness can be violated by the presence of data races—whose prevention has been a topic of concern both in research and in practice. The Rust programming language is a prime example, putting the slogan fearless concurrency in practice by not only employing an ownership-based type system for memory management, but also using its type system to enforce mutual exclusion on shared data. Locking, unfortunately, not only comes at the price of deadlocks but shared access to data may also cause memory leaks . This paper develops a theory of deadlock and leak freedom for higher-order locks in a shared memory concurrent setting. Higher-order locks allow sharing not only of basic values but also of other locks and channels, and are themselves first-class citizens. The theory is based on the notion of a sharing topology , administrating who is permitted to access shared data at what point in the program. The paper first develops higher-order locks for acyclic sharing topologies, instantiated in a λ-calculus with higher-order locks and message-passing concurrency. The paper then extends the calculus to support circular dependencies with dynamic lock orders, which we illustrate with a dynamic version of Dijkstra’s dining philosophers problem. Well-typed programs in the resulting calculi are shown to be free of deadlocks and memory leaks, with proofs mechanized in the Coq proof assistant.
2 January 2024 in Proceedings of the ACM on Programming Languages
Disentanglement is a run-time property of parallel programs that facilitates task-local reasoning about the memory footprint of parallel tasks. In particular, it ensures that a task does not access any memory locations allocated by another concurrently executing task. Disentanglement can be exploited, for example, to implement a high-performance parallel memory manager, such as in the MPL (MaPLe) compiler for Parallel ML. Prior research on disentanglement has focused on the design of optimizations, either trusting the programmer to provide a disentangled program or relying on runtime instrumentation for detecting and managing entanglement. This paper provides the first static approach to verify that a program is disentangled: it contributes DisLog, a concurrent separation logic for disentanglement. DisLog enriches concurrent separation logic with the notions necessary for reasoning about the fork-join structure of parallel programs, allowing the verification that memory accesses are effectively disentangled. A large class of programs, including race-free programs, exhibit memory access patterns that are disentangled "by construction". To reason about these patterns, the paper distills from DisLog an almost standard concurrent separation logic, called DisLog+. In this high-level logic, no specific reasoning about memory accesses is needed: functional correctness proofs entail disentanglement. The paper illustrates the use of DisLog and DisLog+ on a range of case studies, including two different implementations of parallel deduplication via concurrent hashing. All our results are mechanized in the Coq proof assistant using Iris.