[1] Arthur Charguéraud, Alan Schmitt, and Thomas Wood. JSExplain: a double debugger for JavaScript. In WWW '18 Companion: The 2018 Web Conference Companion, April 23-27, 2018, Lyon, France. ACM, April 2018. [ bib | DOI | .pdf ]
We present JSExplain, a reference interpreter for JavaScript that closely follows the specification and that produces execution traces. These traces may be interactively investigated in a browser, with an interface that displays not only the code and the state of the interpreter, but also the code and the state of the interpreted program. Conditional breakpoints may be expressed with respect to both the interpreter and the interpreted program. In that respect, JSExplain is a double-debugger for the specification of JavaScript.

[2] Christopher Pulte, Shaked Flur, Will Deacon, Jon French, Susmit Sarkar, and Peter Sewell. Simplifying ARM Concurrency: Multicopy-atomic Axiomatic and Operational Models for ARMv8. In POPL 2018, pages 19:1-19:29, January 2018. [ bib | DOI | http | .pdf ]
ARM has a relaxed memory model, previously specified in informal prose for ARMv7 and ARMv8. Over time, and partly due to work building formal semantics for ARM concurrency, it has become clear that some of the complexity of the model is not justified by the potential benefits. In particular, the model was originally non-multicopy-atomic: writes could become visible to some other threads before becoming visible to all - but this has not been exploited in production implementations, the corresponding potential hardware optimisations are thought to have insufficient benefits in the ARM context, and it gives rise to subtle complications when combined with other ARMv8 features. The ARMv8 architecture has therefore been revised: it now has a multicopy-atomic model. It has also been simplified in other respects, including more straightforward notions of dependency, and the architecture now includes a formal concurrency model.

In this paper we detail these changes and discuss their motivation. We define two formal concurrency models: an operational one, simplifying the Flowing model of Flur et al., and the axiomatic model of the revised ARMv8 specification. The models were developed by an academic group and by ARM staff, respectively, and this extended collaboration partly motivated the above changes. We prove the equivalence of the two models. The operational model is integrated into an executable exploration tool with new web interface, demonstrated by exhaustively checking the possible behaviours of a loop-unrolled version of a Linux kernel lock implementation, a previously known bug due to unprevented speculation, and a fixed version.

[3] Conrad Watt. Mechanising and verifying the webassembly specification. In CPP 2018: Proceedings of the 7th ACM SIGPLAN International Conference on Certified Programs and Proofs, CPP 2018, pages 53-65, New York, NY, USA, 2018. ACM. [ bib | DOI | http | .pdf ]
WebAssembly is a new low-level language currently being implemented in all major web browsers. It is designed to become the universal compilation target for the web, obsoleting existing solutions in this area, such as asm.js and Native Client. The WebAssembly working group has incorporated formal techniques into the development of the language, but their efforts so far have focussed on pen and paper formal specification.

We present a mechanised Isabelle specification for the WebAssembly language, together with a verified executable interpreter and type checker. Moreover, we present a fully mechanised proof of the soundness of the WebAssembly type system, and detail how our work on this proof has exposed several issues with the official WebAssembly specification, influencing its development. Finally, we give a brief account of our efforts in performing differential fuzzing of our interpreter against industry implementations.

Keywords: bytecode, reduction, soundness, stack machine
[4] José Fragoso Santos, Petar Maksimović, Daiva Naudžiunienė, Thomas Wood, and Philippa Gardner. JaVerT: JavaScript verification toolchain. PACMPL, 2(POPL):50:1-50:33, 2018. [ bib | DOI | .pdf ]
The dynamic nature of JavaScript and its complex semantics make it a difficult target for logic-based verification. We introduce JaVerT, a semi-automatic JavaScript Verification Toolchain, based on separation logic and aimed at the specialist developer wanting rich, mechanically verified specifications of critical JavaScript code. To specify JavaScript programs, we design abstractions that capture its key heap structures (for example, prototype chains and function closures), allowing the developer to write clear and succinct specifications with minimal knowledge of the JavaScript internals. To verify JavaScript programs, we develop JaVerT, a verification pipeline consisting of: JS-2-JSIL, a well-tested compiler from JavaScript to JSIL, an intermediate goto language capturing the fundamental dynamic features of JavaScript; JSIL Verify, a semi-automatic verification tool based on a sound JSIL separation logic; and verified axiomatic specifications of the JavaScript internal functions. Using JaVerT, we verify functional correctness properties of: data-structure libraries (key-value map, priority queue) written in an object-oriented style; operations on data structures such as binary search trees (BSTs) and lists; examples illustrating function closures; and test cases from the official ECMAScript test suite. The verification times suggest that reasoning about larger, more complex code using JaVerT is feasible.

[5] Kasper Svendsen, Jean Pichon-Pharabod, Marko Doko, Ori Lahav, and Viktor Vafeiadis. A separation logic for a promising semantics. In ESOP 2018, 2018. [ bib ]
[6] Thomas Bauereiß, Armando Pesenti Gritti, Andrei Popescu, and Franco Raimondi. Cosmed: A confidentiality-verified social media platform. Journal of Automated Reasoning, December 2017. [ bib | DOI | http ]
This paper describes progress with our agenda of formal verification of information flow security for realistic systems. We present CoSMed, a social media platform with verified document confidentiality. The system's kernel is implemented and verified in the proof assistant Isabelle/HOL. For verification, we employ the framework of Bounded-Deducibility (BD) Security, previously introduced for the conference system CoCon. CoSMed is a second major case study in this framework. For CoSMed, the static topology of declassification bounds and triggers that characterized previous instances of BD Security has to give way to a dynamic integration of the triggers as part of the bounds. We also show that, from a theoretical viewpoint, the removal of triggers from the notion of BD Security does not restrict its expressiveness.

[7] Alexandre Joannou, Jonathan Woodruff, Robert Kovacsics, Simon W. Moore, Alex Bradbury, Hongyan Xia, Robert N. M. Watson, David Chisnall, Michael Roe, Brooks Davis, Edward Napierala, John Baldwin, Khilan Gudka, Peter G. Neumann, Alfredo Mazzinghi, Alex Richardson, Stacey Son, and A. Theodore Markettos. Efficient tagged memory. In ICCD 2017: IEEE 35th International Conference on Computer Design, November 2017. [ bib | DOI | www: | .pdf ]
We characterize the cache behavior of an in-memory tag table and demonstrate that an optimized implementation can typically achieve a near-zero memory traffic overhead. Both industry and academia have repeatedly demonstrated tagged memory as a key mechanism to enable enforcement of power- ful security invariants, including capabilities, pointer integrity, watchpoints, and information-flow tracking. A single-bit tag shadowspace is the most commonly proposed requirement, as one bit is the minimum metadata needed to distinguish between an untyped data word and any number of new hardware- enforced types. We survey various tag shadowspace approaches and identify their common requirements and positive features of their implementations. To avoid non-standard memory widths, we identify the most practical implementation for tag storage to be an in-memory table managed next to the DRAM controller. We characterize the caching performance of such a tag table and demonstrate a DRAM traffic overhead below 5% for the vast majority of applications. We identify spatial locality on a page scale as the primary factor that enables surprisingly high table cache-ability. We then demonstrate tag-table compression for a set of common applications. A hierarchical structure with elegantly simple optimizations reduces DRAM traffic overhead to below 1% for most applications. These insights and optimizations pave the way for commercial applications making use of single-bit tags stored in commodity memory.

[8] Cristina Matache, Victor B. F. Gomes, and Dominic P. Mulligan. Programming and proving with classical types. In APLAS 2017: Proceedings of the 15th Asian Symposium on Programming Languages and Systems, November 2017. [ bib | DOI | http ]
The propositions-as-types correspondence is ordinarily presented as linking the metatheory of typed λ-calculi and the proof theory of intuitionistic logic. Griffin observed that this correspondence could be extended to classical logic through the use of control operators. This observation set off a flurry of further research, leading to the development of Parigot’s λμ-calculus. In this work, we use the λμ-calculus as the foundation for a system of proof terms for classical first-order logic. In particular, we define an extended call-by-value λμ-calculus with a type system in correspondence with full classical logic. We extend the language with polymorphic types, add a host of data types in ‘direct style’, and prove several metatheoretical properties. All of our proofs and definitions are mechanised in Isabelle/HOL, and we automatically obtain an interpreter for a system of proof terms cum programming language—called μML—using Isabelle’s code generation mechanism. Atop our proof terms, we build a prototype LCF-style interactive theorem prover—called μTP—for classical first-order logic, capable of synthesising μML programs from completed tactic-driven proofs. We present example closed μML programs with classical tautologies for types, including some inexpressible as closed programs in the original λμ-calculus, and some example tactic-driven μTP proofs of classical tautologies.

[9] Stephen Kell. Some were meant for C: the endurance of an unmanageable language. In Onward! 2017: Proceedings of the 2017 ACM International Symposium on New Ideas, New Paradigms, and Reflections on Programming & Software, Onward! 2017, New York, NY, USA, October 2017. ACM. [ bib | DOI | .pdf ]
The C language leads a double life: as an application programming language of yesteryear, perpetuated by circumstance, and as a systems programming language which remains a weapon of choice decades after its cre- ation. This essay is a C programmer's reaction to the call to abandon ship. It questions several properties com- monly held to define the experience of using C; these include unsafety, undefined behaviour, and the moti- vation of performance. It argues all these are in fact inessential; rather, it traces C's ultimate strength to a communicative design which does not fit easily within the usual conception of “a programming language”, but can be seen as a counterpoint to so-called “managed languages”. This communicativity is what facilitates the essential aspect of system-building: creating parts which interact with other, remote parts-being “alongside” not “within”.

Keywords: systems programming, virtual machine, managed languages, safety, undefined behavior
[10] Victor B. F. Gomes, Martin Kleppmann, Dominic P. Mulligan, and Alastair R. Beresford. Verifying strong eventual consistency in distributed systems. In OOPSLA 2017: Proceedings of the ACM SIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications (OOPSLA), pages 109:1-109:28, October 2017. Distinguished Paper award. [ bib | DOI | http ]
Data replication is used in distributed systems to maintain up-to-date copies of shared data across multiple computers in a network. However, despite decades of research, algorithms for achieving consistency in replicated systems are still poorly understood. Indeed, many published algorithms have later been shown to be incorrect, even some that were accompanied by supposed mechanised proofs of correctness. In this work, we focus on the correctness of Conflict-free Replicated Data Types (CRDTs), a class of algorithm that provides strong eventual consistency guarantees for replicated data. We develop a modular and reusable framework in the Isabelle/HOL interactive proof assistant for verifying the correctness of CRDT algorithms. We avoid correctness issues that have dogged previous mechanised proofs in this area by including a network model in our formalisation, and proving that our theorems hold in all possible network behaviours. Our axiomatic network model is a standard abstraction that accurately reflects the behaviour of real-world computer networks. Moreover, we identify an abstract convergence theorem, a property of order relations, which provides a formal definition of strong eventual consistency. We then obtain the first machine-checked correctness theorems for three concrete CRDTs: the Replicated Growable Array, the Observed-Remove Set, and an Increment-Decrement Counter. We find that our framework is highly reusable, developing proofs of correctness for the latter two CRDTs in a few hours and with relatively little CRDT-specific code.

[11] Andrea Cerone, Alexey Gotsman, and Hongseok Yang. Algebraic laws for weak consistency. In CONCUR 2017: Proceedings of 28th International Conference on Concurrency Theory, September 2017. [ bib | DOI | .html ]
Modern distributed systems often rely on so called weakly-consistent databases, which achieve scalability by sacrificing the consistency guarantee of distributed transaction processing. Such databases have been formalised in two different styles, one based on abstract executions and the other based on dependency graphs. The choice between these styles has been made according to intended applications. The former has been used for specifying and verifying the implementation of these databases, while the latter for proving properties of client programs of the databases. In this paper, we present a set of novel algebraic laws (i.e. inequations) that connect these two styles of specifications. The laws relate binary relations used in a specification based on abstract executions, to those used in a specification based on dependency graphs. We then show that this algebraic connection gives rise to so called robustness criteria, conditions which ensure that a client program of a weakly-consistent database does not exhibit anomalous behaviours due to weak consistency. These criteria make it easy to reason about these client programs, and may become a basis for dynamic or static program analyses. For a certain class of consistency models specifications, we prove a full abstraction result that connects the two styles of specifications.

[12] Philippa Gardner. Verified trustworthy software systems. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 375(2104), September 2017. [ bib | DOI | .pdf ]
Modern society is faced with a fundamental problem: the reliability of complex, evolving software systems on which society critically depends cannot be guaranteed by the established, non-mathematical computer engineering techniques such as informal prose specification and ad hoc testing. The situation is worsening: modern companies are moving fast, leaving little time for code analysis and testing; the behaviour of concurrent and distributed programs cannot be adequately assessed using traditional testing methods; users of mobile applications often neglect to apply software fixes; and malicious users increasingly exploit even simple programming errors, causing major security disruptions. Building trustworthy, reliable software is becoming harder and harder to achieve, while new business and cybersecurity challenges make it of escalating critical importance

[13] José Fragoso Santos, Philippa Gardner, Petar Maksimovic, and Daiva Naudziuniene. Towards logic-based verification of JavaScript programs. In Leonardo de Moura, editor, CADE 2017: 26th International Conference on Automated Deduction, volume 10395 of Lecture Notes in Computer Science, pages 8-25. Springer, August 2017. [ bib | DOI | .html ]
[14] Robert N. M. Watson, Peter G. Neumann, Jonathan Woodruff, Michael Roe, Jonathan Anderson, John Baldwin, David Chisnall, Brooks Davis, Alexandre Joannou, Ben Laurie, Simon W. Moore, Steven J. Murdoch, Robert Norton, Stacey Son, and Hongyan Xia. Capability Hardware Enhanced RISC Instructions: CHERI Instruction-Set Architecture (Version 6). Technical Report UCAM-CL-TR-907, University of Cambridge, Computer Laboratory, April 2017. [ bib | .pdf ]
[15] David Chisnall, Brooks Davis, Khilan Gudka, David Brazdil, Alexandre Joannou, Jonathan Woodruff, A. Theodore Markettos, J. Edward Maste, Robert Norton, Stacey Son, Michael Roe, Simon W. Moore, Peter G. Neumann, Ben Laurie, and Robert N.M. Watson. CHERI JNI: Sinking the Java Security Model into the C. In ASPLOS 2017: Proceedings of the Twenty-Second International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS '17, pages 569-583, New York, NY, USA, April 2017. ACM. [ bib | DOI | http ]
Java provides security and robustness by building a high-level security model atop the foundation of memory protection. Unfortunately, any native code linked into a Java program - including the million lines used to implement the standard library - is able to bypass both the memory protection and the higher-level policies. We present a hardware-assisted implementation of the Java native code interface, which extends the guarantees required for Java's security model to native code.

Our design supports safe direct access to buffers owned by the JVM, including hardware-enforced read-only access where appropriate. We also present Java language syntax to declaratively describe isolated compartments for native code.

We show that it is possible to preserve the memory safety and isolation requirements of the Java security model in C code, allowing native code to run in the same process as Java code with the same impact on security as running equivalent Java code. Our approach has a negligible impact on performance, compared with the existing unsafe native code interface. We demonstrate a prototype implementation running on the CHERI microprocessor synthesized in FPGA.

Keywords: architecture, capability systems, cheri, compartmentalization, compilers, hardware security, java, jni, language security, memory protection, sandboxing
[16] Jeremy Yallop, David Sheets, and Anil Madhavapeddy. A modular foreign function interface. Science of Computer Programming, April 2017. [ bib | DOI | http | .pdf ]
Abstract foreign function interfaces are typically organised monolithically, tying together the specification of each foreign function with the mechanism used to make the function available in the host language. This leads to inflexible systems, where switching from one binding mechanism to another (say from dynamic binding to static code generation) often requires changing tools and rewriting large portions of code. We show that ML-style module systems support exactly the kind of abstraction needed to separate these two aspects of a foreign function binding, leading to declarative foreign function bindings that support switching between a wide variety of binding mechanisms — static and dynamic, synchronous and asynchronous, etc. — with no changes to the function specifications. Note. This is a revised and expanded version of an earlier paper, Declarative Foreign Function Binding Through Generic Programming. This paper brings a greater focus on modularity, and adds new sections on error handling, and on the practicality of the approach we describe.

Keywords: Foreign functions
[17] Thomas Dinsdale-Young, Pedro da Rocha Pinto, Kristoffer Just Andersen, and Lars Birkedal. Caper: Automatic Verification for Fine-grained Concurrency. In Proc. ESOP 2017: 26th European Symposium on Programming, Lecture Notes in Computer Science. Springer, April 2017. [ bib | .pdf ]
Recent program logics based on separation logic emphasise a modular approach to proving functional correctness for fine-grained concurrent programs. However, these logics have no automation support. In this paper, we present Caper, a prototype tool for automated reasoning in such a logic. Caper is based on symbolic execution, integrating reasoning about interference on shared data and about ghost resources that are used to mediate this interference. This enables Caper to verify the functional correctness of fine-grained concurrent algorithms.

[18] Shale Xiong, Pedro da Rocha Pinto, Gian Ntzik, and Philippa Gardner. Abstract Specifications for Concurrent Maps. In Proc. ESOP 2017: 26th European Symposium on Programming, Lecture Notes in Computer Science. Springer, April 2017. [ bib | .pdf ]
Despite recent advances in reasoning about concurrent data structure libraries, the largest implementations in java.util.concurrent have yet to be verified. The key issue lies in the development of modular specifications, which provide clear logical boundaries between clients and implementations. A solution is to use recent advances in fine-grained concurrency reasoning, in particular the introduction of abstract atomicity to concurrent separation logic reasoning. We present two specifications of concurrent maps, both providing the clear boundaries we seek. We show that these specifications are equivalent, in that they can be built from each other. We show how we can verify client programs, such as a concurrent set and a producer-consumer client. We also give a substantial first proof that the main operations of ConcurrentSkipListMap in java.util.concurrent satisfy the map specification. This work demonstrates that we now have the technology to verify the largest implementations in java.util.concurrent.

[19] Lars Birkedal, Thomas Dinsdale-Young, Guilhem Jaber, Kasper Svendsen, and Nikos Tzevelekos. Trace properties from separation logic specifications. CoRR, abs/1702.02972, February 2017. [ bib | http | http ]
We propose a formal approach for relating abstract separation logic library specifications with the trace properties they enforce on interactions between a client and a library. Separation logic with abstract predicates enforces a resource discipline that constrains when and how calls may be made between a client and a library. Intuitively, this can enforce a protocol on the interaction trace. This intuition is broadly used in the separation logic community but has not previously been formalised. We provide just such a formalisation. Our approach is based on using wrappers which instrument library code to induce execution traces for the properties under examination. By considering a separation logic extended with trace resources, we prove that when a library satisfies its separation logic specification then its wrapped version satisfies the same specification and, moreover, maintains the trace properties as an invariant. Consequently, any client and library implementation that are correct with respect to the separation logic specification will satisfy the trace properties.

[20] Shaked Flur, Susmit Sarkar, Christopher Pulte, Kyndylan Nienhuis, Luc Maranget, Kathryn E. Gray, Ali Sezgin, Mark Batty, and Peter Sewell. Mixed-size concurrency: ARM, POWER, C/C++11, and SC. In POPL 2017: The 44st Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, Paris, France, pages 429-442, January 2017. [ bib | DOI | http | .pdf ]
Previous work on the semantics of relaxed shared-memory concurrency has only considered the case in which each load reads the data of exactly one store. In practice, however, multiprocessors support mixed-size accesses, and these are used by systems software and (to some degree) exposed at the C/C++ language level. A semantic foundation for software, therefore, has to address them.

We investigate the mixed-size behaviour of ARMv8 and IBM POWER architectures and implementations: by experiment, by developing semantic models, by testing the correspondence between these, and by discussion with ARM and IBM staff. This turns out to be surprisingly subtle, and on the way we have to revisit the fundamental concepts of coherence and sequential consistency, which change in this setting. In particular, we show that adding a memory barrier between each instruction does not restore sequential consistency. We go on to extend the C/C++11 model to support non-atomic mixed-size memory accesses.

This is a necessary step towards semantics for real-world shared-memory concurrent code, beyond litmus tests.

[21] Anthony C. J. Fox, Magnus O. Myreen, Yong Kiam Tan, and Ramana Kumar. Verified compilation of CakeML to multiple machine-code targets. In Yves Bertot and Viktor Vafeiadis, editors, Proc. CPP 2017: Certified Programs and Proofs. ACM, January 2017. [ bib | DOI | .pdf ]
This paper describes how the latest CakeML compiler supports verified compilation down to multiple realistically modelled target architectures. In particular, we describe how the compiler definition, the various language semantics, and the correctness proofs were organised to minimize target-specific overhead. With our setup we have incorporated compilation to four 64-bit architectures, ARMv8, x86-64, MIPS-62, RISC-V, and one 32-bit architecture, ARMv6. Our correctness theorem allows interference from the environment: the top-level correctness statement takes into account execution of foreign code and per-instruction interference from external processes, such as interrupt handlers in operating systems. The entire CakeML development is formalised in the HOL4 theorem prover.

[22] Morten Krogh-Jespersen, Kasper Svendsen, and Lars Birkedal. A relational model of types-and-effects in higher-order concurrent separation logic. In POPL 2017: The 44st Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, Paris, France, pages 218-231, New York, NY, USA, January 2017. ACM. [ bib | DOI | http ]
Recently we have seen a renewed interest in programming languages that tame the complexity of state and concurrency through refined type systems with more fine-grained control over effects. In addition to simplifying reasoning and eliminating whole classes of bugs, statically tracking effects opens the door to advanced compiler optimizations.

In this paper we present a relational model of a type-and-effect system for a higher-order, concurrent programming language. The model precisely captures the semantic invariants expressed by the effect annotations. We demonstrate that these invariants are strong enough to prove advanced program transformations, including automatic parallelization of expressions with suitably disjoint effects. The model also supports refinement proofs between abstract data types implementations with different internal data representations, including proofs that fine-grained concurrent algorithms refine their coarse-grained counterparts. This is the first model for such an expressive language that supports both effect-based optimizations and data abstraction.

The logical relation is defined in Iris, a state-of-the-art higher-order concurrent separation logic. This greatly simplifies proving well-definedness of the logical relation and also provides us with a powerful logic for reasoning in the model.

[23] Stephen Kell. Dynamically diagnosing type errors in unsafe code. In OOPSLA 2016: Proceedings of the ACM SIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA 2016, New York, NY, USA, November 2016. ACM. [ bib | .pdf ]
Existing approaches for detecting type errors in unsafe languages are limited. Static analysis methods are imprecise, and often require source-level changes, while most dynamic methods check only memory properties (bounds, liveness, etc.), owing to a lack of run-time type information. This paper describes libcrunch, a system for binary-compatible run-time type checking of unmodified unsafe code, currently focusing on C. Practical experience shows that our prototype implementation is easily applicable to many real codebases without source-level modification, correctly flags programmer errors with a very low rate of false positives, offers a very low run-time overhead, and covers classes of error caught by no previously existing tool.

Keywords: C, unsafe code, pointers, type information
[24] Stephen Kell, Dominic P. Mulligan, and Peter Sewell. The missing link: explaining ELF static linking, semantically. In OOPSLA 2016: Proceedings of the ACM SIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA 2016, New York, NY, USA, November 2016. ACM. [ bib | .pdf ]
Beneath the surface, software usually depends on complex linker behaviour to work as intended. Even linking hello_world.c is surprisingly involved, and systems software such as libc and operating system kernels rely on a host of linker features. But linking is poorly understood by working programmers and has largely been neglected by language researchers.

In this paper we survey the many use-cases that linkers support and the poorly specified linker speak by which they are controlled: metadata in object files, command-line options, and linker-script language. We provide the first validated formalisation of a realistic executable and linkable format (ELF), and capture aspects of the Application Binary Interfaces for four mainstream platforms (AArch64, AMD64, Power64, and IA32). Using these, we develop an executable specification of static linking, covering (among other things) enough to link small C programs (we use the example of bzip2) into a correctly running executable. We provide our specification in Lem and Isabelle/HOL forms. This is the first formal specification of mainstream linking. We have used the Isabelle/HOL version to prove a sample correctness property for one case of AMD64 ABI relocation, demonstrating that the specification supports formal proof, and as a first step towards the much more ambitious goal of verified linking. Our work should enable several novel strands of research, including linker-aware verified compilation and program analysis, and better languages for controlling linking.

Keywords: Linking, formal specification, Executable and Linkable Format (ELF), theorem-proving
[25] Kyndylan Nienhuis, Kayvan Memarian, and Peter Sewell. An operational semantics for C/C++11 concurrency. In OOPSLA 2016: Proceedings of the ACM SIGPLAN International Conference on Object-Oriented Programming, Systems, Languages, and Applications, New York, NY, USA, November 2016. ACM. [ bib | .pdf ]
The C/C++11 concurrency model balances two goals: it is relaxed enough to be efficiently implementable and (leaving aside the “thin-air” problem) it is strong enough to give useful guarantees to programmers. It is mathematically precise and has been used in verification research and compiler testing. However, the model is expressed in an axiomatic style, as predicates on complete candidate executions. This suffices for computing the set of allowed executions of a small litmus test, but it does not directly support the incremental construction of executions of larger programs. It is also at odds with conventional operational semantics, as used implicitly in the rest of the C/C++ standards.

Our main contribution is the development of an operational model for C/C++11 concurrency. This covers all the features of the previous formalised axiomatic model, and we have a mechanised proof that the two are equivalent, in Isabelle/HOL. We also integrate this semantics with an operational semantics for sequential C (described elsewhere); the combined semantics can incrementally execute programs in a small fragment of C.

Doing this uncovered several new aspects of the C/C++11 model: we show that one cannot build an equivalent operational model that simply follows program order, sequential consistent order, or the synchronises-with order. The first negative result is forced by hardware-observable behaviour, but the latter two are not, and so might be ameliorated by changing C/C++11. More generally, we hope that this work, with its focus on incremental construction of executions, will inform the future design of new concurrency models.

[26] Victor B. F. Gomes and Georg Struth. Modal Kleene algebra applied to program correctness. In John S. Fitzgerald, Stefania Gnesi, and Constance L. Heitmeyer, editors, FM 2016: Formal Methods - 21st International Symposium, Limassol, Lecture Notes in Computer Science, pages 310-325. Springer, November 2016. [ bib | DOI | http | .pdf ]
Modal Kleene algebras are relatives of dynamic logics that support program construction and verification by equational reasoning. We describe their application in implementing versatile program correctness components in interactive theorem provers such as Isabelle/HOL. Starting from a weakest precondition based component with a simple relational store model, we show how variants for Hoare logic, strongest postconditions and program refinement can be built in a principled way. Modularity of the approach is demonstrated by variants that capture program termination and recursion, memory models for programs with pointers, and program trace semantics.

[27] Azalea Raad, Aquinas Hobor, Jules Villard, and Philippa Gardner. Verifying concurrent graph algorithms. In Proc. APLAS 2016: 14th Asian Symposium on Programming Languages and Systems, pages 314-334, November 2016. [ bib | DOI | http ]
We show how to verify four challenging concurrent finegrained graph-manipulating algorithms, including graph copy, a speculatively parallel Dijkstra, graph marking and spanning tree. We develop a reasoning method for such algorithms that dynamically tracks the contributions and responsibilities of each thread operating on a graph, even in cases of arbitrary recursive thread creation. We demonstrate how to use a logic without abstraction (CoLoSL) to carry out abstract reasoning in the style of iCAP, by building the abstraction into the proof structure rather than incorporating it into the semantic model of the logic.

[28] Azalea Raad, José Fragoso Santos, and Philippa Gardner. DOM: specification and client reasoning. In Proc. APLAS 2016: 14th Asian Symposium on Programming Languages and Systems, pages 401-422, November 2016. [ bib | DOI | http ]
We present an axiomatic specification of a key fragment of DOM using structural separation logic. This specfication allows us to develop modular reasoning about client programs that call the DOM.

[29] Brian Campbell and Ian Stark. Extracting behaviour from an executable instruction set model. In Ruzica Piskac and Muralidhar Talupur, editors, FMCAD 2016: Formal Methods in Computer-Aided Design, pages 33-40, October 2016. Full proceedings http://www.cs.utexas.edu/users/hunt/FMCAD/FMCAD16/proceedings/fmcad-2016-proceedings.pdf. [ bib | .pdf ]
Presenting large formal instruction set models as executable functions makes them accessible to engineers and useful for less formal purposes such as simulation. However, it is more difficult to extract information about the behaviour of individual instructions for reasoning. We present a method which combines symbolic evaluation and symbolic execution techniques to provide a rule-based view of instruction behaviour, with particular application to automatic test generation for large MIPS-like models.

[30] M. Naylor, S. W. Moore, and A. Mujumdar. A consistency checker for memory subsystem traces. In FMCAD 2016: Formal Methods in Computer-Aided Design, September 2016. [ bib | DOI | .pdf ]
Verifying the memory subsystem in a modern shared-memory multiprocessor is a big challenge. Optimized implementations are highly sophisticated, yet must provide subtle consistency and liveness guarantees for the correct execution of concurrent programs. We present a tool that supports efficient specification-based testing of the memory subsystem against a range of formally specified consistency models. Our tool operates directly on the memory subsystem interface, promoting a compositional approach to system-on-chip verification, and can be used to search for simple failure cases – assisting rapid debug. It has recently been incorporated into the development flows of two open-source implementations - Berkeley’s Rocket Chip (RISCV) and Cambridge’s BERI (MIPS) - where it has uncovered a number of serious bugs.

[31] Robert N.M. Watson, Robert M. Norton, Jonathan Woodruff, Simon W. Moore, Peter G. Neumann, Jonathan Anderson, David Chisnall, Brooks Davis, Ben Laurie, Michael Roe, Nirav H. Dave, Khilan Gudka, Alexandre Joannou, A. Theodore Markettos, Ed Maste, Steven J. Murdoch, Colin Rothwell, Stacey D. Son, and Munraj Vadera. Fast protection-domain crossing in the CHERI capability-system architecture. IEEE Micro, 36(5):38-49, September 2016. [ bib | DOI | http ]
Capability Hardware Enhanced RISC Instructions (CHERI) supplement the conventional memory management unit (MMU) with instruction-set architecture (ISA) extensions that implement a capability system model in the address space. CHERI can also underpin a hardware-software object-capability model for scalable application compartmentalization that can mitigate broader classes of attack. This article describes ISA additions to CHERI that support fast protection-domain switching, not only in terms of low cycle count, but also efficient memory sharing with mutual distrust. The authors propose ISA support for sealed capabilities, hardware-assisted checking during protection-domain switching, a lightweight capability flow-control model, and fast register clearing, while retaining the flexibility of a software-defined protection-domain transition model. They validate this approach through a full-system experimental design, including ISA extensions, a field-programmable gate array prototype (implemented in Bluespec SystemVerilog), and a software stack including an OS (based on FreeBSD), compiler (based on LLVM), software compartmentalization model, and open-source applications.

[32] Kayvan Memarian and Peter Sewell. Clarifying unspecified values (draft defect report or proposal for c2x). ISO SC22 WG14 N2089, http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2089.htm, September 2016. [ bib ]
[33] Kayvan Memarian and Peter Sewell. Clarifying pointer provenance (draft defect report or proposal for c2x). ISO SC22 WG14 N2090, http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2090.htm, September 2016. [ bib ]
[34] Kayvan Memarian and Peter Sewell. Clarifying trap representations (draft defect report or proposal for c2x). ISO SC22 WG14 N2091, http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2091.htm, September 2016. [ bib ]
[35] Yong Kiam Tan, Magnus O. Myreen, Ramana Kumar, Anthony Fox, Scott Owens, and Michael Norrish. A new verified compiler backend for CakeML. In Proc. ICFP 2016: International Conference on Functional Programming. ACM Press, September 2016. [ bib | DOI | .pdf ]
We have developed and mechanically verified a new compiler backend for CakeML. Our new compiler features a sequence of intermediate languages that allows it to incrementally compile away high-level features and enables verification at the right levels of semantic detail. In this way, it resembles mainstream (unverified) compilers for strict functional languages. The compiler supports efficient curried multi-argument functions, configurable data representations, exceptions that unwind the call stack, register allocation, and more. The compiler targets several architectures: x86-64, ARMv6, ARMv8, MIPS-64, and RISC-V.

In this paper, we present the overall structure of the compiler, including its 12 intermediate languages, and explain how everything fits together. We focus particularly on the interaction between the verification of the register allocator and the garbage collector, and memory representations. The entire development has been carried out within the HOL4 theorem prover.

[36] Hannes Mehnert and Louis Gesbert. Conex - establishing trust into data repositories. In Proceedings of the ACM OCaml 2016 Workshop, September 2016. [ bib | .pdf ]
Opam is a software update system, responsible for discovering, downloading, building, and installing packages from community repositories. The data and metadata at each step should be authenticated to originate from a (transitively) trusted author. Opam does not include any mechanism to authenticate any step at the moment. We propose conex, which establishes digital signatures from the package author to the user. Using conex, neither the distribution server nor the transport layer need to be trusted. Clients will install authenticated packages, and notice conex only if signatures cannot be verified. Authors need to adapt their workflow slightly, an updated opam publish will be developed. Janitors (repository maintainers) need to adapt their workflow to emit cryptographic signatures to the repository (in addition to pressing merge PR buttons).

[37] Enguerrand Decorne, Jeremy Yallop, and David Meršinjak. Ocaml inside: a drop-in replacement for libtls (extended abstract). In Proceedings of the ACM OCaml 2016 Workshop, September 2016. [ bib | .pdf ]
The C programming language pervades systems software. An operating system in the Unix tradition consists of a kernel, written in C, and a collection of libraries and executables, also written in C, which communicate in large part via APIs defined as C types and functions. Systems built in C typically suffer from a number of problems, ranging from buffer overflows and other violations that follow inevitably from unrestricted access to memory, to awkward APIs that result from an inexpressive type system and a lack of automatic memory management. The openssl library, which implements the cryptographic protocols TLS and SSL, suffers from both these problems. The lack of bounds checking in C led to the notorious Heartbleed bug in 2014; a study two years earlier found that almost no clients of openssl use the library correctly, apparently due to its unhelpful interface (Georgiev et al. 2012). In response to the Heartbleed bug, the OpenBSD team created libressl, a fork of openssl with the aim of correcting the deficiencies. One early fruit of the libressl project is libtls, a much simpler, more disciplined interface to the TLS protocol implementation in libressl. However, libtls is still built in C, and so is still vulnerable to potential buffer overflows, type errors, and similar defects. In this talk we describe one approach to avoiding these problems, namely replacing libtls with a fully compatible library written in OCaml. Our library, libnqsb-tls1, matches the libtls function-for-function, but the implementation contains no C; instead, it it wraps the pure OCaml TLS implementation ocaml-tls (Kaloper-Meršinjak et al. 2015)

[38] David Kaloper-Meršinjak and Jeremy Yallop. Generic partially-static data (extended abstract). In Proceedings of the 1st International Workshop on Type-Driven Development, TyDe 2016, pages 39-40, New York, NY, USA, September 2016. ACM. [ bib | DOI | http | http ]
We describe a generic approach to defining partially-static data and corresponding operations.

Keywords: generic programming, partial evaluation, staging
[39] Gian Ntzik. Reasoning About POSIX File Systems. PhD thesis, Imperial College London, September 2016. [ bib | .pdf ]
POSIX is a standard for operating systems, with a substantial part devoted to specifying file-system operations. File-system operations exhibit complex concurrent behaviour, comprising multiple actions affecting different parts of the state: typically, multiple atomic reads followed by an atomic update. However, the standard's description of concurrent behaviour is unsatisfactory: it is fragmented; contains ambiguities; and is generally under-specified. We provide a formal concurrent specification of POSIX file systems and demonstrate scalable reasoning for clients. Our speciation is based on a concurrent specification language, which uses a modern concurrent separation logic for reasoning about abstract atomic operations, and an associated refinement calculus. Our reasoning about clients highlights an important difference between reasoning about modules built over a heap, where the interference on the shared state is restricted to the operations of the module, and modules built over a file system, where the interference cannot be restricted as the file system is a public namespace. We introduce specifications conditional on context invariants used to restrict the interference, and apply our reasoning to lock files and named pipes. Program reasoning based on separation logic has been successful at verifying that programs do not crash due to illegal use of resources, such invalid memory accesses. The underlying assumption of separation logics, however, is that machines do not fail. In practice, machines can fail unpredictably for various reasons, such as power loss, corrupting resources or resulting in permanent data loss. Critical software, such as file systems and databases, employ recovery methods to mitigate these effects. We introduce an extension of the Views framework to reason about programs in the presence of such events and their associated recovery methods. We use concurrent separation logic as an instance of the framework to illustrate our reasoning, and explore programs using write-ahead logging, such a stylised ARIES recovery algorithm.

[40] Pedro da Rocha Pinto. Reasoning with Time and Data Abstractions. PhD thesis, Imperial College London, September 2016. [ bib | .pdf ]
In this thesis, we address the problem of verifying the functional correctness of concurrent programs, with emphasis on fine-grained concurrent data structures. Reasoning about such programs is challenging since data can be concurrently accessed by multiple threads: the reasoning must account for the interference between threads, which is often subtle. To reason about interference, concurrent operations should either be at distinct times or on distinct data. We present TaDA, a sound program logic for verifying clients and implementations that use abstract specifications that incorporate both abstract atomicity-the abstraction that operations take effect at a single, discrete instant in time-and abstract disjointness-the abstraction that operations act on distinct data resources. Our key contribution is the introduction of atomic triples, which offer an expressive approach for specifying program modules. We also present Total-TaDA, a sound extension of TaDA with which we can verify total correctness of concurrent programs, i.e. that such programs both produce the correct result and terminate. With Total-TaDA, we can specify constraints on a thread=92s concurrent environment that are necessary to guarantee termination. This allows us to verify total correctness for nonblocking algorithms and express lock- and wait-freedom. More generally, the abstract specifications can express that one operation cannot impede the progress of another, a new non-blocking property that we call non-impedance. Finally, we describe how to extend TaDA for proving abstract atomicity for data structures that make use of helping-where one thread is performing an abstract operation on behalf of another-and speculation-where an abstract operation is determined by future behaviour.

[41] Azalea Raad. Abstraction, Refinement and Concurrent Reasoning. PhD thesis, Imperial College London, September 2016. [ bib | .pdf ]
This thesis explores the challenges in abstract library specification, library refinement and reasoning about fine-grained concurrent programs.For abstract library specification, this thesis applies structural separation logic (SSL) to formally specify the behaviour of several libraries in an abstract, local and compositional manner. This thesis further generalises the theory of SSL to allow for library specifications that are language independent. Most notably, we specify a fragment of the Document Object Model (DOM) library. This result is compelling as it significantly improves upon existing DOM formalisms in that the specifications produced are local, compositional and language-independent. Concerning library refinement, this thesis explores two existing approaches to library refinement for separation logic, identifying their advantages and limitations in different settings. This thesis then introduces a hybrid approach to refinement, combining the strengths of both techniques for simple scalable library refinement. These ideas are then adapted to refinement for SSL by presenting a JavaScript implementation of the DOM fragment studied and establishing its correctness with respect to its specification using the hybrid refinement approach.As to concurrent reasoning, this thesis introduces concurrent local subjective logic (CoLoSL) for compositional reasoning about fine-grained concurrent programs. CoLoSL introduces subjective views, where each thread is verified with respect to a customised local view of the state, as well as the general composition and framing of interference relations, allowing for better proof reuse.

[42] Andreas Haas, Thomas A. Henzinger, Andreas Holzer, Christoph M. Kirsch, Michael Lippautz, Hannes Payer, Ali Sezgin, Ana Sokolova, and Helmut Veith. Local Linearizability for Concurrent Container-Type Data Structures. In 27th International Conference on Concurrency Theory (CONCUR 2016), volume 59 of Leibniz International Proceedings in Informatics (LIPIcs), pages 6:1-6:15, Dagstuhl, Germany, August 2016. Schloss Dagstuhl-Leibniz-Zentrum fuer Informatik. [ bib | DOI | .pdf ]
The semantics of concurrent data structures is usually given by a sequential specification and a consistency condition. Linearizability is the most popular consistency condition due to its simplicity and general applicability. Nevertheless, for applications that do not require all guarantees offered by linearizability, recent research has focused on improving performance and scalability of concurrent data structures by relaxing their semantics. In this paper, we present local linearizability, a relaxed consistency condition that is applicable to container-type concurrent data structures like pools, queues, and stacks. While linearizability requires that the effect of each operation is observed by all threads at the same time, local linearizability only requires that for each thread T, the effects of its local insertion operations and the effects of those removal operations that remove values inserted by T are observed by all threads at the same time. We investigate theoretical and practical properties of local linearizability and its relationship to many existing consistency conditions. We present a generic implementation method for locally linearizable data structures that uses existing linearizable data structures as building blocks. Our implementations show performance and scalability improvements over the original building blocks and outperform the fastest existing container-type implementations.

[43] Kayvan Memarian, Justus Matthiesen, James Lingard, Kyndylan Nienhuis, David Chisnall, Robert N.M. Watson, and Peter Sewell. Into the depths of C: elaborating the de facto standards. In PLDI 2016, June 2016. PLDI 2016 Distinguished Paper award. [ bib | .pdf ]
C remains central to our computing infrastructure. It is notionally defined by ISO standards, but in reality the properties of C assumed by systems code and those implemented by compilers have diverged, both from the ISO standards and from each other, and none of these are clearly understood.

We make two contributions to help improve this error-prone situation. First, we describe an in-depth analysis of the design space for the semantics of pointers and memory in C as it is used in practice. We articulate many specific questions, build a suite of semantic test cases, gather experimental data from multiple implementations, and survey what C experts believe about the de facto standards. We identify questions where there is a consensus (either following ISO or differing) and where there are conflicts. We apply all this to an experimental C implemented above capability hardware. Second, we describe a formal model, Cerberus, for large parts of C. Cerberus is parameterised on its memory model; it is linkable either with a candidate de facto memory object model, under construction, or with an operational C11 concurrency model; it is defined by elaboration to a much simpler Core language for accessibility, and it is executable as a test oracle on small examples.

This should provide a solid basis for discussion of what mainstream C is now: what programmers and analysis tools can assume and what compilers aim to implement. Ultimately we hope it will be a step towards clear, consistent, and accepted semantics for the various use-cases of C.

[44] Victor B. F. Gomes and Georg Struth. Program construction and verification components based on Kleene algebra. Archive of Formal Proofs, 2016, June 2016. [ bib | http ]
Variants of Kleene algebra support program construction and verification by algebraic reasoning. This entry provides a verification component for Hoare logic based on Kleene algebra with tests, verification components for weakest preconditions and strongest postconditions based on Kleene algebra with domain and a component for step-wise refinement based on refinement Kleene algebra with tests. In addition to these components for the partial correctness of while programs, a verification component for total correctness based on divergence Kleene algebras and one for (partial correctness) of recursive programs based on domain quantales are provided. Finally we have integrated memory models for programs with pointers and a program trace semantics into the weakest precondition component.

[45] Pedro da Rocha Pinto, Thomas Dinsdale-Young, Philippa Gardner, and Julian Sutherland. Modular termination verification for non-blocking concurrency. In ESOP 2016: Proceedings of the 25th European Symposium on Programming, pages 176-201, April 2016. [ bib | DOI | http | .pdf ]
We present Total-TaDA, a program logic for verifying the total correctness of concurrent programs: that such programs both terminate and produce the correct result. With Total-TaDA, we can specify constraints on a thread's concurrent environment that are necessary to guarantee termination. This allows us to verify total correctness for non-blocking algorithms, e.g. a counter and a stack. Our specifications can express lock- and wait-freedom. More generally, they can express that one operation cannot impede the progress of another, a new non-blocking property we call non-impedance. Moreover, our approach is modular. We can verify the operations of a module independently, and build up modules on top of each other.

[46] Kasper Svendsen, Filip Sieczkowski, and Lars Birkedal. Transfinite step-indexing: Decoupling concrete and logical steps. In ESOP 2016, pages 727-751, April 2016. [ bib | DOI | http | .pdf ]
Step-indexing has proven to be a powerful technique for defining logical relations for languages with advanced type systems and models of expressive program logics. In both cases, the model is stratified using natural numbers to solve a recursive equation that has no naive solutions. As a result of this stratification, current models require that each unfolding of the recursive equation - each logical step - must coincide with a concrete reduction step. This tight coupling is problematic for applications where the number of logical steps cannot be statically bounded.

In this paper we demonstrate that this tight coupling between logical and concrete steps is artificial and show how to loosen it using transfinite step-indexing. We present a logical relation that supports an arbitrary but finite number of logical steps for each concrete step.

[47] Victor B. F. Gomes, Walter Guttmann, Peter Höfner, Georg Struth, and Tjark Weber. Kleene algebras with domain. Archive of Formal Proofs, 2016, April 2016. [ bib | http ]
Kleene algebras with domain are Kleene algebras endowed with an operation that maps each element of the algebra to its domain of definition (or its complement) in abstract fashion. They form a simple algebraic basis for Hoare logics, dynamic logics or predicate transformer semantics. We formalise a modular hierarchy of algebras with domain and antidomain (domain complement) operations in Isabelle/HOL that ranges from domain and antidomain semigroups to modal Kleene algebras and divergence Kleene algebras. We link these algebras with models of binary relations and program traces. We include some examples from modal logics, termination and program analysis.

[48] Brian Campbell and Ian Stark. Randomised testing of a microprocessor model using SMT-solver state generation. Sci. Comput. Program., 118:60-76, March 2016. [ bib | DOI | .pdf ]
We validate a HOL4 model of the ARM Cortex-M0 microcontroller core by testing the model’s behaviour on randomly chosen instructions against real chips from several manufacturers. The model and our intended application involve precise timing information about instruction execution, but the implementations are pipelined, so checking the behaviour of single instructions would not give us sufficient confidence in the model. Thus we test the model using sequences of randomly chosen instructions. The main challenge is to meet the constraints on the initial and intermediate execution states: we must ensure that memory accesses are in range and that we respect restrictions on the instructions. By careful transformation of these constraints an off-the-shelf SMT solver can be used to find suitable states for executing test sequences. We also use additional constraints to test our hypotheses about the timing anomalies encountered.

[49] Kayvan Memarian and Peter Sewell. Clarifying the C memory object model. ISO SC22 WG14 N2012, http://www.cl.cam.ac.uk/~pes20/cerberus/notes64-wg14.html, March 2016. [ bib ]
[50] David Chisnall, Justus Matthiesen, Kayvan Memarian, Kyndylan Nienhuis, Peter Sewell, and Robert N. M. Watson. C memory object and value semantics: the space of de facto and ISO standards. ISO SC22 WG14 N2013, http://www.cl.cam.ac.uk/~pes20/cerberus/notes30.pdf, March 2016. [ bib ]
[51] Kayvan Memarian and Peter Sewell. What is C in practice? (Cerberus survey v2): Analysis of responses. ISO SC22 WG14 N2014, http://www.cl.cam.ac.uk/~pes20/cerberus/notes50-survey-discussion.html, March 2016. [ bib ]
[52] Kayvan Memarian and Peter Sewell. What is C in practice? (Cerberus survey v2): Analysis of responses - with comments. ISO SC22 WG14 N2015, http://www.cl.cam.ac.uk/~pes20/cerberus/analysis-2016-02-05-anon.txt, March 2016. [ bib ]
[53] David Kaloper-Meršinjak and Hannes Mehnert. Not-quite-so-broken TLS 1.3 mechanised conformance checking. In TLSv1.3 - Ready or Not? (TRON) workshop, February 2016. [ bib | .pdf ]
We present a set of tools to aid TLS 1.3 implementors, all derived from a single TLS implementation/model. These include an automated offline TLS protocol conformance checker, a test oracle validating recorded sessions, a tool replicating recorded sessions with other implementations, and an interactive online handshake visualisation.

The conformance checker repeatedly runs a user-provided TLS implementation, attempting to establish TLS sessions with it; the checker explores the TLS parameter space to determine feature coverage of the provided implementation. The test oracle takes a recorded session between two endpoints and decides whether the session was conformant with the specification. The replication utility re-runs one side of a recorded session against another TLS implementation, and observes its behaviour. The online visualisation accepts connections from clients and presents the TLS session as an interactive sequence diagram.

All of these tools are based on our clean-slate nqsb-TLS implementation/model. It already supports TLS 1.0-1.2, and interoperates with a broad range of other TLS implementations. We are currently extending nqsb-TLS with TLS 1.3 support, and tracking the progress of the TLS 1.3 draft, adapting our implementation/model accordingly.

We invite the community to use our tools while implementing the TLS 1.3 RFC, and provide feedback on deviations in the interpretation thereof. This process enables the community to converge to a single, mechanically checkable TLS 1.3 model, as implemented by nqsb-TLS.

[54] Shaked Flur, Kathryn E. Gray, Christopher Pulte, Susmit Sarkar, Ali Sezgin, Luc Maranget, Will Deacon, and Peter Sewell. Modelling the ARMv8 architecture, operationally: concurrency and ISA. In POPL 2016: Proceedings of the 43rd Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, St. Petersburg, FL, USA, pages 608-621, January 2016. [ bib | DOI | http | .pdf ]
In this paper we develop semantics for key aspects of the ARMv8 multiprocessor architecture: the concurrency model and much of the 64-bit application-level instruction set (ISA). Our goal is to clarify what the range of architecturally allowable behaviour is, and thereby to support future work on formal verification, analysis, and testing of concurrent ARM software and hardware.

Establishing such models with high confidence is intrinsically difficult: it involves capturing the vendor's architectural intent, aspects of which (especially for concurrency) have not previously been precisely defined. We therefore first develop a concurrency model with a microarchitectural flavour, abstracting from many hardware implementation concerns but still close to hardware-designer intuition. This means it can be discussed in detail with ARM architects. We then develop a more abstract model, better suited for use as an architectural specification, which we prove sound w.r.t. the first.

The instruction semantics involves further difficulties, handling the mass of detail and the subtle intensional information required to interface to the concurrency model. We have a novel ISA description language, with a lightweight dependent type system, letting us do both with a rather direct representation of the ARM reference manual instruction descriptions.

We build a tool from the combined semantics that lets one explore, either interactively or exhaustively, the full range of architecturally allowed behaviour, for litmus tests and (small) ELF executables. We prove correctness of some optimisations needed for tool performance.

We validate the models by discussion with ARM staff, and by comparison against ARM hardware behaviour, for ISA single- instruction tests and concurrent litmus tests.

[55] Jean Pichon-Pharabod and Peter Sewell. A concurrency semantics for relaxed atomics that permits optimisation and avoids thin-air executions. In POPL 2016: Proceedings of the 43rd Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, St. Petersburg, FL, USA, January 20 - 22, 2016, pages 622-633, January 2016. [ bib | DOI | http | .pdf ]
Despite much research on concurrent programming languages, especially for Java and C/C++, we still do not have a satisfactory definition of their semantics, one that admits all common optimisations without also admitting undesired behaviour. Especially problematic are the “thin-air” examples involving high-performance concurrent accesses, such as C/C++11 relaxed atomics. The C/C++11 model is in a per-candidate-execution style, and previous work has identified a tension between that and the fact that compiler optimisations do not operate over single candidate executions in isolation; rather, they operate over syntactic representations that represent all executions.

In this paper we propose a novel approach that circumvents this difficulty. We define a concurrency semantics for a core calculus, including relaxed-atomic and non-atomic accesses, and locks, that admits a wide range of optimisation while still forbidding the classic thin-air examples. It also addresses other problems relating to undefined behaviour.

The basic idea is to use an event-structure representation of the current state of each thread, capturing all of its potential executions, and to permit interleaving of execution and transformation steps over that to reflect optimisation (possibly dynamic) of the code. These are combined with a non-multi-copy-atomic storage subsystem, to reflect common hardware behaviour.

The semantics is defined in a mechanised and executable form, and designed to be implementable above current relaxed hardware and strong enough to support the programming idioms that C/C++11 does for this fragment. It offers a potential way forward for concurrent programming language semantics, beyond the current C/C++11 and Java models.

[56] Kathryn E. Gray, Gabriel Kerneis, Dominic P. Mulligan, Christopher Pulte, Susmit Sarkar, and Peter Sewell. An integrated concurrency and core-ISA architectural envelope definition, and test oracle, for IBM POWER multiprocessors. In MICRO 2015: Proceedings of the 48th International Symposium on Microarchitecture, , Waikiki, HI, USA, pages 635-646, December 2015. [ bib | DOI | http | .pdf ]
Weakly consistent multiprocessors such as ARM and IBM POWER have been with us for decades, but their subtle programmer-visible concurrency behaviour remains challenging, both to implement and to use; the traditional architecture documentation, with its mix of prose and pseudocode, leaves much unclear.

In this paper we show how a precise architectural envelope model for such architectures can be defined, taking IBM POWER as our example. Our model specifies, for an arbitrary test program, the set of all its allowable executions, not just those of some particular implementation. The model integrates an operational concurrency model with an ISA model for the fixed-point non-vector user-mode instruction set (largely automatically derived from the vendor pseudocode, and expressed in a new ISA description language). The key question is the interface between these two: allowing all the required concurrency behaviour, without overcommitting to some particular microarchitectural implementation, requires a novel abstract structure.

Our model is expressed in a mathematically rigorous language that can be automatically translated to an executable test-oracle tool; this lets one either interactively explore or exhaustively compute the set of all allowed behaviours of intricate test cases, to provide a reference for hardware and software development.

[57] Gian Ntzik, Pedro da Rocha Pinto, and Philippa Gardner. Fault-tolerant Resource Reasoning. In APLAS 2015: Proceedings of the 13th Asian Symposium on Programming Languages and Systems, pages 169-188, December 2015. [ bib | DOI | http | .pdf ]
Separation logic has been successful at verifying that programs do not crash due to illegal use of resources. The underlying assumption, however, is that machines do not fail. In practice, machines can fail unpredictably for various reasons, e.g. power loss, corrupting resources. Critical software, e.g. file systems, employ recovery methods to mitigate these effects. We introduce an extension of the Views framework to reason about such methods. We use concurrent separation logic as an instance of the framework to illustrate our reasoning, and explore programs using write-ahead logging, e.g. an ARIES recovery algorithm.

[58] Khilan Gudka, Robert N. M. Watson, Jonathan Anderson, David Chisnall, Brooks Davis, Ben Laurie, Ilias Marinos, Peter G. Neumann, and Alex Richardson. Clean Application Compartmentalization with SOAAP. In CCS 2015: Proceedings of the 22nd ACM Conference on Computer and Communications Security, October 2015. [ bib | .pdf ]
Application compartmentalization, a vulnerability mitigation technique employed in programs such as OpenSSH and the Chromium web browser, decomposes software into isolated components to limit privileges leaked or otherwise available to attackers. However, compartmentalizing applications - and maintaining that compartmentalization - is hindered by ad hoc methodologies and significantly increased programming effort. In practice, programmers stumble through (rather than overtly reason about) compartmentalization spaces of possible decompositions, unknowingly trading off correctness, security, complexity, and performance. We present a new conceptual framework embodied in an LLVM-based tool: the Security-Oriented Analysis of Application Programs (SOAAP) that allows programmers to reason about co mpartmentalization using source-code annotations (compartmentalization hypotheses). We demonstrate considerable benefit when creating new compartmentalizations for complex applications, and analyze existing compartmentalized applications to discover design faults and maintenance issues arising from application evolution.

[59] Stephen Kell. Towards a dynamic object model within Unix processes. In Onward! 2015: 2015 ACM International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software, Pittsburgh, PA, USA, pages 224-239, October 2015. [ bib | DOI | http | .pdf ]
Programmers face much complexity from the co-existence of “native” (Unix-like) and virtual machine (VM) “managed” run-time environments. Rather than having VMs replace Unix processes, we investigate whether it makes sense for the latter to “become VMs”, by evolving Unix's user-level services to subsume those of VMs. We survey the (little-understood) VM-like features in modern Unix, noting common shortcomings: a lack of semantic metadata (“type information”) and the inability to bind from objects “back” to their metadata. We describe the design and implementation of a system, liballocs, which adds these capabilities in a highly compatible way, and explore its consequences.

[60] Tom Ridge, David Sheets, Thomas Tuerk, Andrea Giugliano, Anil Madhavapeddy, and Peter Sewell. SibylFS: formal specification and oracle-based testing for POSIX and real-world file systems. In SOSP 2015: Proceedings of the 25th Symposium on Operating Systems Principles, Monterey, CA, USA, pages 38-53, October 2015. [ bib | DOI | http | .pdf ]
Systems depend critically on the behaviour of file systems, but that behaviour differs in many details, both between implementations and between each implementation and the POSIX (and other) prose specifications. Building robust and portable software requires understanding these details and differences, but there is currently no good way to systematically describe, investigate, or test file system behaviour across this complex multi-platform interface.

In this paper we show how to characterise the envelope of allowed behaviour of file systems in a form that enables practical and highly discriminating testing. We give a mathematically rigorous model of file system behaviour, SibylFS, that specifies the range of allowed behaviours of a file system for any sequence of the system calls within our scope, and that can be used as a test oracle to decide whether an observed trace is allowed by the model, both for validating the model and for testing file systems against it. SibylFS is modular enough to not only describe POSIX, but also specific Linux, OSX and FreeBSD behaviours. We complement the model with an extensive test suite of over 21000 tests; this can be run on a target file system and checked in less than 5 minutes, making it usable in practice. Finally, we report experimental results for around 40 configurations of many file systems, identifying many differences and some serious flaws.

[61] Gian Ntzik and Philippa Gardner. Reasoning about the POSIX File System: Local Update and Global Pathnames. In OOPSLA 2015: Proceedings of the 30th Annual ACM SIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications, October 2015. [ bib ]
We introduce a program logic for specifying a core sequential subset of the POSIX file system and for reasoning abstractly about client programs working with the file system. The challenge is to reason about the combination of local directory update and global pathname traversal (including '..' and symbolic links) which may overlap the directories being updated. Existing reasoning techniques are either based on first-order logic and do not scale, or on separation logic and can only handle linear pathnames (no '..' or symbolic links). We introduce fusion logic for reasoning about local update and global pathname traversal, introducing a novel effect frame rule to propagate the effect of a local update on overlapping pathnames. We apply our reasoning to the standard recursive remove utility (rm -r), discovering bugs in well-known implementations.

[62] Matthew Naylor and Simon W. Moore. A generic synthesisable test bench. In MEMOCODE 2015: ACM/IEEE International Conference on Formal Methods and Models for Codesign, Austin, TX, USA, pages 128-137, September 2015. [ bib | DOI | http | .pdf ]
Writing test benches is one of the most frequently-performed tasks in the hardware development process. The ability to reuse common test bench features is therefore key to productivity. In this paper, we present a generic test bench, parameterised by a specification of correctness, which can be used to test any design. Our test bench provides several important features, including automatic test-sequence generation and shrinking of counter-examples, and is fully synthesisable, allowing rigorous testing on FPGA as well as in simulation. The approach is easy to use, cheap to implement, and encourages the formal specification of hardware components through the reward of automatic testing and simple failure cases.

[63] Robert N. M. Watson, Peter G. Neumann, Jonathan Woodruff, Michael Roe, Jonathan Anderson, David Chisnall, Brooks Davis, Alexandre Joannou, Ben Laurie, Simon W. Moore, Steven J. Murdoch, Robert Norton, and Stacey Son. Capability Hardware Enhanced RISC Instructions: CHERI Instruction-Set Architecture. Technical Report UCAM-CL-TR-876, University of Cambridge, Computer Laboratory, September 2015. [ bib | .pdf ]
This technical report describes CHERI ISAv4, the fourth version of the Capability Hardware Enhanced RISC Instructions (CHERI) Instruction-Set Architecture (ISA). CHERI is being developed by SRI International and the University of Cambridge. This design captures four years of research, development, refinement, formal analysis, and testing, and is a substantial enhancement to the ISA version described in UCAM-CL-TR-850. Key improvements lie in tighter C-language integration, and more mature support for software object-capability models; these changes result from experience gained in adapting substantial software stacks to run on prototype hardware.

The CHERI instruction set is based on a hybrid capability-system architecture that adds new capability-system primitives to a commodity 64-bit RISC ISA enabling software to efficiently implement fine-grained memory protection and a hardware-software object-capability security model. These extensions support incrementally adoptable, high-performance, formally based, programmer-friendly underpinnings for fine-grained software decomposition and compartmentalization, motivated by and capable of enforcing the principle of least privilege.

The CHERI system architecture purposefully addresses known performance and robustness gaps in commodity ISAs that hinder the adoption of more secure programming models centered around the principle of least privilege. To this end, CHERI blends traditional paged virtual memory with a per-address-space capability model that includes capability registers, capability instructions, and tagged memory that have been added to the 64-bit MIPS ISA via a new capability coprocessor. CHERI also learns from the C-language fat-pointer literature: CHERI capabilities can describe not only regions of memory, but can also capture C pointer semantics allowing capabilities to be substituted for pointers in generated code.

CHERI’s hybrid system approach, inspired by the Capsicum security model, allows incremental adoption of capability-oriented software design: software implementations that are more robust and resilient can be deployed where they are most needed, while leaving less critical software largely unmodified, but nevertheless suitably constrained to be incapable of having adverse effects. For example, we are focusing conversion efforts on low-level TCB components of the system: separation kernels, hypervisors, operating system kernels, language runtimes, and userspace TCBs such as web browsers. Likewise, we see early-use scenarios (such as data compression, protocol parsing, image processing, and video processing) that relate to particularly high-risk software libraries, which are concentrations of both complex and historically vulnerability-prone code combined with untrustworthy data sources, while leaving containing applications unchanged.

This report describes the CHERI Instruction-Set Architecture (ISA) and design, and provides reference documentation and potential memory models, along with their requirements. It also briefly addresses the CHERI system hardware-software architecture, documenting our current thinking on integrating programming languages and operating systems with the CHERI hardware.

[64] Robert N. M. Watson, David Chisnall, Brooks Davis, Wojciech Koszek, Simon W. Moore, Steven J. Murdoch, Peter G. Neumann, and Jonathan Woodruff. Capability Hardware Enhanced RISC Instructions: CHERI Programmer's Guide. Technical Report UCAM-CL-TR-877, University of Cambridge, Computer Laboratory, September 2015. [ bib | .pdf ]
The CHERI Programmer's Guide documents the software environment for the Capability Hardware Enhanced RISC Instructions (CHERI) prototype developed by SRI International and the University of Cambridge. The Guide is targeted at hardware and software developers working with capability-enhanced software. It describes how to use the CHERI Clang/LLVM compiler suite and CheriBSD operating system - versions of the off-the-shelf LLVM suite and FreeBSD operating system adapted to use CHERI’s protection features - as well as implementation details of both.

[65] Anthony C. J. Fox. Improved tool support for machine-code decompilation in HOL4. In ITP 2015:Interactive Theorem Proving - 6th International Conference, Nanjing, China, pages 187-202, August 2015. [ bib | DOI | http | .pdf ]
The HOL4 interactive theorem prover provides a sound logical environment for reasoning about machine-code programs. The rigour of HOL's LCF-style kernel naturally guarantees very high levels of assurance, but it does present challenges when it comes implementing efficient proof tools. This paper presents improvements that have been made to our methodology for soundly decompiling machine-code programs to functions expressed in HOL logic. These advancements have been facilitated by the development of a domain specific language, called L3, for the specification of Instruction Set Architectures (ISAs). As a result of these improvements, decompilation is faster (on average by one to two orders of magnitude), the instruction set specifications are easier to write, and the proof tools are easier to maintain.

[66] David Kaloper-Mersinjak, Hannes Mehnert, Anil Madhavapeddy, and Peter Sewell. Not-quite-so-broken TLS: lessons in re-engineering a security protocol specification and implementation. In USENIX Security 2015: 24th USENIX Security Symposium, Washington, D.C., USA, pages 223-238, August 2015. [ bib | http | .pdf ]
Transport Layer Security (TLS) implementations have a history of security flaws. The immediate causes of these are often programming errors, e.g. in memory management, but the root causes are more fundamental: the challenges of interpreting the ambiguous prose specification, the complexities inherent in large APIs and code bases, inherently unsafe programming choices, and the impossibility of directly testing conformance between implementations and the specification.

We present nqsb-TLS, the result of our re-engineered approach to security protocol specification and implementation that addresses these root causes. The same code serves dual roles: it is both a specification of TLS, executable as a test oracle to check conformance of traces from arbitrary implementations, and a usable implementation of TLS; a modular and declarative programming style provides clean separation between its components. Many security flaws are thus excluded by construction.

nqsb-TLS can be used in standalone applications, which we demonstrate with a messaging client, and can also be compiled into a Xen unikernel (a specialised virtual machine image) with a TCB that is 4% of a standalone system running a standard Linux/OpenSSL stack, with all network traffic being handled in a memory-safe language; this supports applications including HTTPS, IMAP, Git, and Websocket clients and servers.

[67] Ali Sezgin and Serdar Tasiran. Moving around: Lipton's reduction for TSO - (regular submission). In VSTTE 2015: Verified Software: Theories, Tools, and Experiments - 7th International Conference, San Francisco, CA, USA. Revised Selected Papers, pages 165-182, July 2015. [ bib | DOI | http ]
We generalize Lipton’s reduction theory, hitherto limited to SC, for TSO programs. We demonstrate the use of our theory by specifying the conditions under which a particular write is SC-like (i.e. placing a fence immediately after the write does not constrain the behavior of the overall program) and a library implementation can be safely used (i.e. compositionality). Our theory is complete: a program has only SC behaviors iff there is a proof that establishes that every write in the program is SC-like. We adapt the notion of program abstraction to TSO analysis via our theory. We define precisely what is meant by abstraction, and propose a methodology by which one can obtain via abstraction SC summaries of a program which may have non-SC behaviors. Finally, we show how checking whether a write instruction is SC-like can be mechanized. We describe a transformation in which the execution of each thread of the original program (under TSO) is simulated by the execution of two tightly coupled threads in the new program (under SC).

[68] Pedro da Rocha Pinto, Thomas Dinsdale-Young, and Philippa Gardner. Steps in Modular Specifications for Concurrent Modules (Invited Tutorial Paper). In MFPS 2015: Proceedings of the 31st Conference on the Mathematical Foundations of Programming Semantics, pages 3-18, June 2015. [ bib | DOI | http | .pdf ]
The specification of a concurrent program module is a difficult problem. The specifications must be strong enough to enable reasoning about the intended clients without reference to the underlying module implementation. We survey a range of verification techniques for specifying concurrent modules, in particular highlighting four key concepts: auxiliary state, interference abstraction, resource ownership and atomicity. We show how these concepts combine to provide powerful approaches to specifying concurrent modules.

[69] S. Flur, K. Gray, G. Kerneis, D. Mulligan, C. Pulte, S. Sarkar, and P. Sewell. Rigorous architectural modelling for production multiprocessors, May 2015. Presentation at HCSS 2015: the Fifteenth Annual High Confidence Software and Systems Conference. [ bib ]
Processor architectures are critical interfaces in computing, but they are typically defined only by prose and pseudocode documentation. This is especially problematic for the subtle concurrency behaviour of weakly consistent multiprocessors such as ARM and IBM POWER: the traditional documentation does not define precisely what programmer-observable behaviour is (and is not) allowed for concurrent code; it is not executable as a test oracle for pre-Silicon or post-Silicon hardware testing; it is not executable as an emulator for software testing; and it is not mathematically rigorous enough to serve as a foundation for software verification.

We present a rigorous architectural envelope model for IBM POWER multiprocessors, and similar work in progress for ARM, that aims to support all of these for small-but-intricate test cases, integrating an operational concurrency model with an ISA model for the sequential behaviour of a substantial fragment of the user-mode instruction set (largely automatically derived from the vendor pseudocode, and expressed in a new ISA description language). The key question is the interface between these two: we have to allow all the required relaxed-memory behaviour, without overcommitting to some particular microarchitectural model. Our models can be automatically translated into executable code, which, combined with front-ends for concurrency litmus tests and ELF executables, can interactively or exhaustively explore all the allowed behaviours of small test cases.

[70] Robert N. M. Watson, Jonathan Woodruff, Peter G. Neumann, Simon W. Moore, Jonathan Anderson, David Chisnall, Nirav Dave, Brooks Davis, Khilan Gudka, Ben Laurie, Steven J. Murdoch, Robert Norton, Michael Roe, Stacey Son, and Munraj Vadera. CHERI: A hybrid capability-system architecture for scalable software compartmentalization. In Security and Privacy 2015: Proceedings of the 36th IEEE Symposium on Security and Privacy (“Oakland”), May 2015. [ bib | DOI | .pdf ]
CHERI extends a conventional RISC Instruction-Set Architecture, compiler, and operating system to support fine-grained, capability-based memory protection to mitigate memory-related vulnerabilities in C-language TCBs. We describe how CHERI capabilities can also underpin a hardware-software object-capability model for application compartmentalization that can mitigate broader classes of attack. Prototyped as an extension to the open-source 64-bit BERI RISC FPGA softcore processor, FreeBSD operating system, and LLVM compiler, we demonstrate multiple orders-of-magnitude improvement in scalability, simplified programmability, and resulting tangible security benefits as compared to compartmentalization based on pure Memory-Management Unit (MMU) designs. We evaluate incrementally deployable CHERI-based compartmentalization using several real-world UNIX libraries and applications.

[71] Kayvan Memarian, Kyndylan Nienhuis, Justus Matthiesen, James Lingard, and Peter Sewell. Cerberus: towards an executable semantics for sequential and concurrent C11, May 2015. Presentation at HCSS 2015: the Fifteenth Annual High Confidence Software and Systems Conference. [ bib ]
C remains central to our computing infrastructure but still lacks a clear and complete semantics. Programmers lack tools to explore the range of behaviours they should expect; compiler development lacks test oracles; and formal verification and analysis must make (explicitly or implicitly) many choices about the specific C they target.

We describe Cerberus, a semantics for a substantial fragment of C11. Its thread-local semantics is factored via an elaboration into a simpler Core language, to make it conceptually and mathematically tractable. This is integrated with an operational model for C11 concurrency, with a mechanised proof of equivalence to the axiomatic C11 model of Batty et al. The front-end includes a parser that closely follows the C11 standard grammar and a typechecker. Cerberus is executable, to explore all behaviours or single paths of test programs, and it supports proof, as shown by a preliminary experiment in translation validation for the front-end of Clang, for very simple programs. This is a step towards a clear, consistent, and unambiguous semantics for C.

[72] David Kaloper Meršinjak, Hannes Mehnert, Anil Madhavapeddy, and Peter Sewell. Not-quite-so-broken TLS: lessons in re-engineering a security protocol specification and implementation, May 2015. Presentation at HCSS 2015: the Fifteenth Annual High Confidence Software and Systems Conference. [ bib ]
Transport Layer Security (TLS) implementations have a history of security flaws. The immediate causes of these range from simple programming errors, such as memory management, to subtle violations of the protocol logic. Deeper causes can be seen in the challenges of interpreting the ambiguous prose specification, the complexities inherent in large APIs and code bases, unsafe performance-oriented programming choices, and the impossibility of directly testing conformance between implementations and the specification.

We present nqsb-TLS, the result of our re-engineering approach to improve the quality of security protocol implementations. The same code serves dual roles: it is both a specification of TLS, executable as a test oracle to check conformance of traces from arbitrary implementations, and a secure and usable executable implementation of TLS. nqsb-TLS employs a modular and declarative programming style that also allows it to be compiled into a Xen unikernel (a specialised virtual machine image) with a TCB that is 2.5% of a standalone system running a standard Linux/OpenSSL stack, with all network traffic being handled in a memory-safe language.

nqsb-TLS focuses on protocol-level interoperability, and makes no effort to be compatible with existing (and often poorly designed) library APIs such as OpenSSL. The higher-level API in nqsb-TLS makes it harder to misuse the library, and is demonstrated via several unikernel applications ranging over HTTPS, IMAP, Git, Websocket clients and servers.

[73] A. M. Pitts, J. Matthiesen, and J. Derikx. A dependent type theory with abstractable names. In I. Mackie and M. Ayala-Rincon, editors, LSFA 2014: Proceedings of the 9th Workshop on Logical and Semantic Frameworks, with Applications, volume 312 of Electronic Notes in Theoretical Computer Science, pages 19-50. Elsevier, April 2015. [ bib | DOI | http ]
This paper describes a version of Martin-Löf's dependent type theory extended with names and constructs for freshness and name-abstraction derived from the theory of nominal sets. We aim for a type theory for computing and proving (via a Curry-Howard correspondence) with syntactic structures which captures familiar, but informal, "nameful" practices when dealing with binders.

[74] Mark Batty, Kayvan Memarian, Kyndylan Nienhuis, Jean Pichon-Pharabod, and Peter Sewell. The problem of programming language concurrency semantics. In ESOP 2015: Programming Languages and Systems - 24th European Symposium on Programming, Held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2015, London, UK, April 11-18, 2015., pages 283-307, April 2015. [ bib | DOI | http | .pdf ]
Despite decades of research, we do not have a satisfactory concurrency semantics for any general-purpose programming language that aims to support concurrent systems code. The Java Memory Model has been shown to be unsound with respect to standard compiler optimisations, while the C/C++11 model is too weak, admitting undesirable thin-air executions. Our goal in this paper is to articulate this major open problem as clearly as is currently possible, showing how it arises from the combination of multiprocessor relaxed-memory behaviour and the desire to accommodate current compiler optimisations. We make several novel contributions that each shed some light on the problem, constraining the possible solutions and identifying new difficulties.

[75] Filip Sieczkowski, Kasper Svendsen, Lars Birkedal, and Jean Pichon-Pharabod. A separation logic for fictional sequential consistency. In ESOP 2015, April 2015. [ bib | DOI ]
To improve performance, modern multiprocessors and programming languages typically implement relaxed memory models that do not require all processors/threads to observe memory operations in the same order. To relieve programmers from having to reason directly about these relaxed behaviors, languages often provide efficient synchronization primitives and concurrent data structures with stronger high-level guarantees about memory reorderings. For instance, locks usually ensure that when a thread acquires a lock, it can observe all memory operations of the releasing thread, prior to the release. When used correctly, these synchronization primitives and data structures allow clients to recover a fiction of a sequentially consistent memory model.

In this paper we propose a new proof system, iCAP-TSO, that captures this fiction formally, for a language with a TSO memory model. The logic supports reasoning about libraries that directly exploit the relaxed memory model to achieve maximum efficiency. When these libraries provide sufficient guarantees, the logic hides the underlying complexity and admits standard separation logic rules for reasoning about their more high-level clients.

[76] Azalea Raad, Jules Villard, and Philippa Gardner. CoLoSL: Concurrent local subjective logic. In ESOP 2015, April 2015. [ bib | DOI | .pdf ]
A key difficulty in verifying shared-memory concurrent programs is reasoning compositionally about each thread in isolation. Existing verification techniques for fine-grained concurrency typically require reasoning about either the entire shared state or disjoint parts of the shared state, impeding compositionality. This paper introduces the program logic CoLoSL, where each thread is verified with respect to its subjective view of the global shared state. This subjective view describes only that part of the state accessed by the thread. Subjective views may arbitrarily overlap with each other, and expand and contract depending on the resource required by the thread. This flexibility gives rise to small specifications and, hence, more compositional reasoning for concurrent programs. We demonstrate our reasoning on a range of examples, including a concurrent computation of a spanning tree of a graph.

[77] Soham Chakraborty, Thomas A. Henzinger, Ali Sezgin, and Viktor Vafeiadis. Aspect-oriented linearizability proofs. Logical Methods in Computer Science, 11(1):1-33, April 2015. [ bib | DOI | http ]
Linearizability of concurrent data structures is usually proved by monolithic simulation arguments relying on the identification of the so-called linearization points. Regrettably, such proofs, whether manual or automatic, are often complicated and scale poorly to advanced non-blocking concurrency patterns, such as helping and optimistic updates. In response, we propose a more modular way of checking linearizability of concurrent queue algorithms that does not involve identifying linearization points. We reduce the task of proving linearizability with respect to the queue specification to establishing four basic properties, each of which can be proved independently by simpler arguments. As a demonstration of our approach, we verify the Herlihy and Wing queue, an algorithm that is challenging to verify by a simulation proof.

[78] Jade Alglave, Mark Batty, Alastair F. Donaldson, Ganesh Gopalakrishnan, Jeroen Ketema, Daniel Poetzl, Tyler Sorensen, and John Wickerson. GPU concurrency: Weak behaviours and programming assumptions. In ASPLOS 2015: Proceedings of the Twentieth International Conference on Architectural Support for Programming Languages and Operating Systems, Istanbul, Turkey, pages 577-591, March 2015. [ bib | DOI | http ]
Concurrency is pervasive and perplexing, particularly on graphics processing units (GPUs). Current specifications of languages and hardware are inconclusive; thus programmers often rely on folklore assumptions when writing software.

To remedy this state of affairs, we conducted a large empirical study of the concurrent behaviour of deployed GPUs. Armed with litmus tests (i.e. short concurrent programs), we questioned the assumptions in programming guides and vendor documentation about the guarantees provided by hardware. We developed a tool to generate thousands of litmus tests and run them under stressful workloads. We observed a litany of previously elusive weak behaviours, and exposed folklore beliefs about GPU programming-often supported by official tutorials-as false.

As a way forward, we propose a model of Nvidia GPU hardware, which correctly models every behaviour witnessed in our experiments. The model is a variant of SPARC Relaxed Memory Order (RMO), structured following the GPU concurrency hierarchy.

[79] David Chisnall, Colin Rothwell, Brooks Davis, Robert N.M. Watson, Jonathan Woodruff, Munraj Vadera, Simon W. Moore, Peter G. Neumann, and Michael Roe. Beyond the PDP-11: Processor support for a memory-safe C abstract machine. In ASPLOS 2015: Proceedings of the Fifteenth Edition of ASPLOS on Architectural Support for Programming Languages and Operating Systems, ASPLOS XX, New York, NY, USA, March 2015. ACM. [ bib | DOI | .pdf ]
We propose a new memory-safe interpretation of the C abstract machine that provides stronger protection to benefit security and debugging. Despite ambiguities in the specification intended to provide implementation flexibility, contemporary implementations of C have converged on a memory model similar to the PDP-11, the original target for C. This model lacks support for memory safety despite well-documented impacts on security and reliability.

Attempts to change this model are often hampered by assumptions embedded in a large body of existing C code, dating back to the memory model exposed by the original C compiler for the PDP-11. Our experience with attempting to implement a memory-safe variant of C on the CHERI experimental microprocessor led us to identify a number of problematic idioms. We describe these as well as their interaction with existing memory safety schemes and the assumptions that they make beyond the requirements of the C specification. Finally, we refine the CHERI ISA and abstract model for C, by combining elements of the CHERI capability model and fat pointers, and present a softcore CPU that implements a C abstract machine that can run legacy C code with strong memory protection guarantees.

[80] David Kaloper Meršinjak and Hannes Mehnert. Trustworthy secure modular operating system engineering. Invited talk at 31st Chaos Communication Congress (31C3), January 2015. [ bib ]
We present Mirage OS, a modular library operating system developed from scratch in the functional programming language OCaml. Each service, called unikernel, is an OCaml application using libraries such as a TCP/IP stack, DNS. It is either compiled to a Xen virtual machine image or to a Unix binary (for development). State in 2014 is that it runs on x86 and arm, we implemented a clean-slate TLS (1.0, 1.1, 1.2), X.509, ASN.1 stack, crypto primitives, Off-the-record. We also have TCP/IP, HTTP, a persistent branchable store (similar to git) - all implemented in OCaml. A virtual machine serving data via https is roughly 2MB in size - no libc inside :)

[81] Mark John Batty. The C11 and C++11 Concurrency Model. PhD thesis, University of Cambridge Computer Laboratory, November 2014. Winner of 2015 ACM SIGPLAN John C. Reynolds Doctoral Dissertation Award and 2015 CPHC/BCS Distinguished Dissertation competition. [ bib ]
[82] Stephen Kell. In search of types. In Onward! 2014: Proceedings of the 2014 ACM International Symposium on New Ideas, New Paradigms, and Reflections on Programming & Software, Onward! 2014, pages 227-241, New York, NY, USA, October 2014. ACM. [ bib | DOI | http ]
The concept of "type" has been used without a consistent, precise definition in discussions about programming languages for 60 years. In this essay I explore various concepts lurking behind distinct uses of this word, highlighting two traditions in which the word came into use largely independently: engineering traditions on the one hand, and those of symbolic logic on the other. These traditions are founded on differing attitudes to the nature and purpose of abstraction, but their distinct uses of "type" have never been explicitly unified. One result is that discourse across these traditions often finds itself at cross purposes, such as overapplying one sense of "type" where another is appropriate, and occasionally proceeding to draw wrong conclusions. I illustrate this with examples from well-known and justly well-regarded literature, and argue that ongoing developments in both the theory and practice of programming make now a good time to resolve these problems.

[83] Dominic P. Mulligan, Scott Owens, Kathryn E. Gray, Tom Ridge, and Peter Sewell. Lem: Reusable engineering of real-world semantics. In ICFP 2014: Proceedings of the 19th ACM SIGPLAN International Conference on Functional Programming, pages 175-188, New York, NY, USA, September 2014. ACM. [ bib | DOI | http | .pdf ]
Recent years have seen remarkable successes in rigorous engineering: using mathematically rigorous semantic models (not just idealised calculi) of real-world processors, programming languages, protocols, and security mechanisms, for testing, proof, analysis, and design. Building these models is challenging, requiring experimentation, dialogue with vendors or standards bodies, and validation; their scale adds engineering issues akin to those of programming to the task of writing clear and usable mathematics. But language and tool support for specification is lacking. Proof assistants can be used but bring their own difficulties, and a model produced in one, perhaps requiring many person-years effort and maintained over an extended period, cannot be used by those familiar with another.

We introduce Lem, a language for engineering reusable large-scale semantic models. The Lem design takes inspiration both from functional programming languages and from proof assistants, and Lem definitions are translatable into OCaml for testing, Coq, HOL4, and Isabelle/HOL for proof, and LaTeX and HTML for presentation. This requires a delicate balance of expressiveness, careful library design, and implementation of transformations - akin to compilation, but subject to the constraint of producing usable and human-readable code for each target. Lem's effectiveness is demonstrated by its use in practice.

[84] Brian Campbell and Ian Stark. Randomised testing of a microprocessor model using SMT-solver state generation. In Frédéric Lang and Francesco Flammini, editors, FMICS 2014: Formal Methods for Industrial Critical Systems, volume 8718 of Lecture Notes in Computer Science, pages 185-199. Springer, September 2014. [ bib | DOI | .pdf ]
We validate a HOL4 model of the ARM Cortex-M0 microcontroller core by testing the model's behaviour on randomly chosen instructions against a real chip. The model and our intended application involve precise timing information about instruction execution, but the implementations are pipelined, so checking the behaviour of single instructions would not give us sufficient confidence in the model. Thus we test the model using sequences of randomly chosen instructions. The main challenge is to meet the constraints on the initial and intermediate execution states: we must ensure that memory accesses are in range and that we respect restrictions on the instructions. By careful transformation of these constraints an off-the-shelf SMT solver can be used to find suitable states for executing test sequences.

[85] Hannes Mehnert and David Kaloper Mersinjak. Transport Layer Security purely in OCaml. In Proceedings of the ACM OCaml 2014 Workshop, September 2014. [ bib | .pdf ]
Transport Layer Security (TLS) is probably the most widely deployed security protocol on the Internet. It is used to setup virtual private networks, secure various services such as web and email, etc. In this paper we describe our clean slate TLS implementation developed in OCaml. Our motivating goals are reliability, robustness, and API conciseness. While our implementation is still a work in progress and lacks some of the protocol features (such as client authentication and session resumption), it already interoperates with other existing TLS implementations. Preliminary performance evaluation shows that our library is roughly five times slower compared to OpenSSL, but we expect to improve our performance.

[86] Basile Clement. Running programming language specifications, August 2014. MPRI report. [ bib ]
[87] Pedro da Rocha Pinto, Thomas Dinsdale-Young, and Philippa Gardner. TaDA: A logic for time and data abstraction. In Richard Jones, editor, ECOOP 2014, volume 8586 of Lecture Notes in Computer Science, pages 207-231. Springer Berlin Heidelberg, July 2014. [ bib | DOI | http ]
To avoid data races, concurrent operations should either be at distinct times or on distinct data. Atomicity is the abstraction that an operation takes effect at a single, discrete instant in time, with linearisability being a well-known correctness condition which asserts that concurrent operations appear to behave atomically. Disjointness is the abstraction that operations act on distinct data resource, with concurrent separation logics enabling reasoning about threads that appear to operate independently on disjoint resources.

We present TaDA, a program logic that combines the benefits of abstract atomicity and abstract disjointness. Our key contribution is the introduction of atomic triples, which offer an expressive approach to specifying program modules. By building up examples, we show that TaDA supports elegant modular reasoning in a way that was not previously possible.

[88] Philippa Gardner, Azalea Raad, Mark J. Wheelhouse, and Adam Wright. Abstract local reasoning for concurrent libraries: Mind the gap. MFPS 2014: Electr. Notes Theor. Comput. Sci., 308:147-166, June 2014. [ bib | DOI | http ]
We study abstract local reasoning for concurrent libraries. There are two main approaches: provide a specification of a library by abstracting from concrete reasoning about an implementation; or provide a direct abstract library specification, justified by refining to an implementation. Both approaches have a significant gap in their reasoning, due to a mismatch between the abstract connectivity of the abstract data structures and the concrete connectivity of the concrete heap representations. We demonstrate this gap using structural separation logic (SSL) for specifying a concurrent tree library and concurrent abstract predicates (CAP) for reasoning about a concrete tree implementation. The gap between the abstract and concrete connectivity emerges as a mismatch between the SSL tree predicates and CAP heap predicates. This gap is closed by an interface function I which links the abstract and concrete connectivity. In the accompanying technical report, we generalise our SSL reasoning and results to arbitrary concurrent data libraries.

[89] Reinoud Elhorst, Mark Batty, and David Chisnall. Efficient code generation for weakly ordered architectures. Presentation at the 4th European LLVM conference (EuroLLVM), April 2014. Slides and report available at http://llvm.org/devmtg/2014-04/. [ bib ]
This report explores the way LLVM generates the memory barriers needed to support the C11/C++11 atomics for ARM. I measure the influence of memory barriers on performance, and I show that in some cases LLVM generates too many barriers. By leaving these barriers out, performance increases significantly. I introduce two LLVM passes, which will remove these extra barriers, improving performance in my test by 40%. I believe one of these passes is ready to be upstreamed to LLVM, while the other will need more testing.

[90] Philippa Gardner, Gian Ntzik, and Adam Wright. Local reasoning for the POSIX file system. In ESOP 2014: Programming Languages and Systems - 23rd European Symposium on Programming, Held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2014, Grenoble, France, pages 169-188, April 2014. [ bib | DOI | http ]
We provide a program logic for specifying a core subset of the sequential POSIX file system, and for reasoning abstractly about client programs working with the file system.

[91] Tony Hoare, Stephan van Staden, Bernhard Möller, Georg Struth, Jules Villard, Huibiao Zhu, and Peter W. O'Hearn. Developments in concurrent Kleene algebra. In RAMiCS 2014: Relational and Algebraic Methods in Computer Science - 14th International Conference, Marienstatt, Germany, pages 1-18, April 2014. [ bib | DOI | http ]
This report summarises recent progress in the research of its co-authors towards the construction of links between algebraic presentations of the principles of programming and the exploitation of concurrency in modern programming practice. The research concentrates on the construction of a realistic family of partial order models for Concurrent Kleene Algebra (aka, the Laws of Programming). The main elements of the model are objects and the events in which they engage. Further primitive concepts are traces, errors and failures, and transferrable ownership. In terms of these we can define other concepts which have proved useful in reasoning about concurrent programs, for example causal dependency and independence, sequentiality and concurrency, allocation and disposal, synchrony and asynchrony, sharing and locality, input and output.

[92] Reinoud Elhorst. Lowering C11 atomics for ARM in LLVM, March 2014. [ bib | .pdf ]
This report explores the way LLVM generates the memory barriers needed to support the C11/C++11 atomics for ARM. I measure the influence of memory barriers on performance, and I show that in some cases LLVM generates too many barriers. By leaving these barriers out, performance increases significantly. I introduce two LLVM passes, which will remove these extra barriers, improving performance in my test by 40%. I believe one of these passes is ready to be upstreamed to LLVM, while the other will need more testing.

[93] Gabriel Kerneis, Charlie Shepherd, and Stefan Hajnoczi. QEMU/CPC: static analysis and CPS conversion for safe, portable, and efficient coroutines. In PEPM 2014: Proceedings of the ACM SIGPLAN 2014 workshop on Partial evaluation and program manipulation, San Diego, California, USA, pages 83-94, January 2014. [ bib | DOI | http ]

Coroutines and events are two common abstractions for writing concurrent programs. Because coroutines are often more convenient, but events more portable and efficient, it is natural to want to translate the former into the latter. CPC is such a source-to-source translator for C programs, based on a partial conversion into continuation-passing style (CPS conversion) of functions annotated as cooperative.

In this article, we study the application of the CPC translator to QEMU, an open-source machine emulator which also uses annotated coroutine functions for concurrency. We first propose a new type of annotations to identify functions which never cooperate, and we introduce CoroCheck, a tool for the static analysis and inference of cooperation annotations. Then, we improve the CPC translator, defining CPS conversion as a calling convention for the C language, with support for indirect calls to CPS-converted function through function pointers. Finally, we apply CoroCheck and CPC to QEMU (750 000 lines of C code), fixing hundreds of missing annotations and comparing performance of the translated code with existing implementations of coroutines in QEMU.

Our work shows the importance of static annotation checking to prevent actual concurrency bugs, and demonstrates that CPS conversion is a flexible, portable, and efficient compilation technique, even for very large programs written in an imperative language.

[94] Ramana Kumar, Magnus O. Myreen, Michael Norrish, and Scott Owens. CakeML: a verified implementation of ML. In POPL 2014: The 41st Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, San Diego, CA, USA, pages 179-192, January 2014. [ bib | DOI | http ]
We have developed and mechanically verified an ML system called CakeML, which supports a substantial subset of Standard ML. CakeML is implemented as an interactive read-eval-print loop (REPL) in x86-64 machine code. Our correctness theorem ensures that this REPL implementation prints only those results permitted by the semantics of CakeML. Our verification effort touches on a breadth of topics including lexing, parsing, type checking, incremental and dynamic compilation, garbage collection, arbitrary-precision arithmetic, and compiler bootstrapping.

Our contributions are twofold. The first is simply in building a system that is end-to-end verified, demonstrating that each piece of such a verification effort can in practice be composed with the others, and ensuring that none of the pieces rely on any over-simplifying assumptions. The second is developing novel approaches to some of the more challenging aspects of the verification. In particular, our formally verified compiler can bootstrap itself: we apply the verified compiler to itself to produce a verified machine-code implementation of the compiler. Additionally, our compiler proof handles diverging input programs with a lightweight approach based on logical timeout exceptions. The entire development was carried out in the HOL4 theorem prover.

[95] Magnus O. Myreen and Scott Owens. Proof-producing translation of higher-order logic into pure and stateful ML. J. Funct. Program., 24(2-3):284-315, January 2014. [ bib | DOI | http ]
The higher-order logic found in proof assistants such as Coq and various HOL systems provides a convenient setting for the development and verification of functional programs. However, to efficiently run these programs, they must be converted (or `extracted') to functional programs in a programming language such as ML or Haskell. With current techniques, this step, which must be trusted, relates similar looking objects that have very different semantic definitions, such as the set-theoretic model of a logic and the operational semantics of a programming language. In this paper, we show how to increase the trustworthiness of this step with an automated technique. Given a functional program expressed in higher-order logic, our technique provides the corresponding program for a functional language defined with an operational semantics, and it provides a mechanically checked theorem relating the two. This theorem can then be used to transfer verified properties of the logical function to the program. We have implemented our technique in the HOL4 theorem prover, translating functions to a subset of Standard ML, and have applied the implementation to examples including functional data structures, a parser generator, cryptographic algorithms, a garbage collector and the 500-line kernel of the HOL light theorem prover. This paper extends our previous conference publication with new material that shows how functions defined in terms of a state-and-exception monad can be translated, with proofs, into stateful ML code. The HOL light example is also new.

This file was generated by bibtex2html 1.96.