Papers

Security is among the top concerns of operating system (OS) developers. A secure runtime environment relies on the OS to correctly configure the memory hardware on which it runs. This is mission-critical as it provides essential security-relevant features and abstractions that ensure the integrity and isolation of untrusted applications running alongside each other. Configuring a platform's memory hardware is not a one-off effort as designers constantly develop new mechanisms for translation and protection with different features and means of configuration. Adapting the OS code to the new hardware is not only a manual, repetitive and time consuming task, it may also introduce subtle, but security critical bugs that break security and isolation guarantees. We present Velosiraptor, a system that automatically generates correct, low-level OS code that programs the memory hardware of a machine. Velosiraptor leverages software synthesis techniques and exploits the domain specificity of the problem to make the synthesis process efficient. With Velosiraptor, developers write only a high-level description of the memory hardware's mapping behavior and OS environment. The Velosiraptor toolchain transforms this specification into a verified implementation that can be linked directly with the rest of the operating system. Incorporating the OS environment into this process allows porting an OS to new hardware platforms without worrying about writing code to configure the memory hardware. We can also use the same specification to generate hardware components. This enables research in new translation mechanisms, freeing up OS developers from manually writing OS code.

Recent advances have highlighted the potential of machine learning, particularly Large Language Models (LLMs), for analyzing and optimizing programs. We present the first application of program embeddings from LLMCompiler—an LLM massively pretrained on intermediate representation (IR) code—to representative program analysis and optimization tasks. We generate program embeddings directly from source and IR code using a simple approach: split programs into chunks, independently embed each chunk with pretrained LLMs, and then aggregate the chunk embeddings into a single program embedding. Our experiments show that combining source and IR code embeddings achieves an error rate of 1.54% in algorithm classification, a 12% improvement over the current state-of-the-art, and a competitive accuracy on heterogeneous device mapping. These findings suggest that training a performance-aware LLM for embedding IR code might yield state-of-the-art results in code optimization tasks.

MetaML-style multi-stage programming (MSP) supports quasi-quotation-based code generation, runtime execution of generated code, and cross-stage persistence (CSP). However, its interaction with computational effects is subtle: mutable state can cause scope extrusion, where generated code escapes the scope of variables on which it depends. This paper presents a type system for MetaML-style MSP with mutable state that statically rules out harmful scope extrusion while supporting multi-level code generation, runtime execution, and a variant of CSP. Our system builds on refined environment classifiers (RECs), a discipline that annotates code types with the variable scopes on which generated code depends. To scale RECs to the MetaML-style setting, we refine classifiers so that they track not only variable scopes, but also the scopes of classifiers themselves. Further, we integrated polymorphism over classifiers, enabling more general and reusable code generation patterns in a multi-level setting. For the resulting system, we define an operational semantics via a definitional interpreter and prove type soundness and safety of offline code generation, showing that generated code can be extracted as standalone well-typed programs. We provide working implementations and mechanized proofs in Rocq.

We present a mathematically rigorous, systematic approach for the verification of database isolation guarantees, which (i) supports a spectrum of seven isolation levels, (ii) uncovers a fundamental dichotomy among isolation levels: stronger levels can be verified via refinement alone, whereas weaker levels additionally require reduction, and (iii) provides a hierarchy of abstract models that substantially simplifies proofs by factoring out their most labor-intensive parts. In particular, we eliminate the need for per-protocol reduction proofs for the weaker class of isolation levels by performing a once-and-for-all reduction at a high level of abstraction in our hierarchy. To achieve this, we develop and apply a generic theory of reduction, which is also of more general interest. Overall, our approach minimizes the user's proof effort to a single, simpler refinement of the most concrete model in our hierarchy. All our results are formalized in Isabelle/HOL.

Static program analysis infers program properties automatically. Yet precise interprocedural analysis remains challenging, and dynamically typed languages amplify the difficulty. Python is particularly problematic: dynamic dispatch, first-class functions, metaprogramming, pervasive exceptions, and an object model based on descriptors and attribute-driven lookup collectively impede precise reasoning. We present PyFlow, a generic IFDS-based static-analysis framework for Python. PyFlow provides a multi-stage intermediate-representation pipeline and a generic IFDS solver parameterized by abstract domains. Analysis developers implement only the dataflow semantics; the framework constructs the supergraph, performs fixed-point iteration, and caches summaries. We implement a taint analysis in \pyflow and evaluate it against eight Python SAST tools (DevSkim, Dlint, Bandit, Bearer, CodeQL, Pysa, Semgrep, and Snyk) on the synthetic and real-world benchmarks from a recent ICSE~'26 study. On the synthetic benchmark, PyFlow achieves the best aggregate recall and F1 score among all nine tools. On the real-world benchmark, it attains the highest recall and F1 score while maintaining precision competitive with taint-based engines. We conclude with lessons learned from building IFDS analyses for Python.

Large Language Models (LLMs) are increasingly used to generate Object Constraint Language (OCL) constraints from natural language specifications and UML class diagrams. However, existing work mainly focuses on improving accuracy, with limited understanding of why these models fail. Aims. This study investigates the underlying causes of LLM failures in OCL generation, framing the task as a graph reasoning problem over UML class diagrams. Method. We conduct an empirical evaluation using the PathOCL dataset across six state-of-the-art LLMs. We analyze the impact of UML structural properties (e.g., navigation depth and model complexity), lexical similarity, prompt ordering strategies, and graph-aware prompting on OCL correctness. Results. We find that OCL generation performance significantly degrades with increasing navigation depth and structural complexity. Lexical similarity has limited influence, while textual ordering of UML elements affects performance. Graph-based prompting yields partial improvements but does not eliminate structural reasoning errors. Conclusions. OCL generation is primarily constrained by graph reasoning limitations rather than purely linguistic factors. These results highlight structural reasoning as a key bottleneck for current LLMs in model-driven engineering tasks.

Few grammars are designed in a straight line. During grammar prototyping, authors often want to run a partially settled rule before refactoring overlapping alternatives or regenerating parser code. Predictive parsers are effective for stable grammars, but their generation and compilation loop can slow down interactive grammar development. This paper presents MSLL, a lightweight runtime extension of recursive-descent LL parsing for edit-time grammar exploration. When a FIRST/FIRST conflict appears, MSLL keeps multiple live parsing stacks, lets each stack follow a different candidate production, and prunes a stack as soon as the input contradicts it. Ambiguity is therefore handled as a runtime state-management problem rather than as a condition that must be removed before execution. The prototype targets an ANTLR-style grammar workflow: run and inspect an evolving grammar directly, collect conflict traces, and later translate or adapt the stabilized grammar for a generated parser. Experiments on DSL-style workloads show that MSLL finishes three 1k-token benchmarks in 48, 49, and 59 ms; stays under one second on two ambiguous 50k-token workloads; and reaches 5.395 s on a 500k-token deeply nested stress case. A real-world JSON grammar case confirms interactive behavior on small and medium inputs while exposing prototype-level super-linear overhead at 500k tokens. The results position MSLL as a development-time companion to existing LL-based parser generators rather than a replacement for production parsers.

While most research on quantum programming considers an idealized, noise-free semantics for quantum programs, we reason about quantum programs that are executed on real, noisy hardware. We consider the error models published by quantum hardware vendors to give a hardware-dependent semantics to quantum programs. This work presents a comprehensive study of noise-aware quantum programming, ranging from logical foundations to automated verification and synthesis. We develop a noise-aware quantum Hoare logic, and use it to derive algorithmic methods for the bounded verification of quantum programs on specific hardware, and for the automatic synthesis of noise-optimal loop-free quantum programs. In this way, we synthesize hardware-dependent subroutines that commonly occur in quantum algorithms, such as parity checks, quantum state preparation, and quantum state discrimination. We evaluate our method on the hardware specifications provided by the IBM Qiskit toolkit. Besides finding different optimal subroutines for different noise models, our synthesis tool also shows that classical probabilistic branching is needed for optimality in quantum programming.

Language models (LMs) are increasingly used to interact with external services via programs written in domain-specific languages (DSLs). Unfortunately, since DSLs are often low-resource and esoteric, LMs frequently produce syntactically invalid programs in these languages. Grammar-constrained decoding can eliminate such failures, but requires syntactic constraints. These are usually in the form of a context-free grammar for the target language, an artifact that is hard to come by for third-party DSLs. In this work, we define an agent, called Autogrammar, that automatically learns context-free grammars from documentation and execution data. Autogrammar is formalized as a Kripke structure whose nondeterministic choices are resolved by a language model, enabling declarative control of agent behavior via linear temporal logic constraints. We evaluate four versions of Autogrammar on three DSLs (i.e., Amazon CloudWatch Logs Insights, Dynatrace Query Language, and Datadog Search Syntax) and find that it generates grammars that achieve near perfect precision on unseen data; that temporal restrictions reduce execution time by 3.8x without incurring statistically-significant loss in precision; that execution data is crucial while documentation is dispensable; and that grammar-constrained decoding using Autogrammar-generated grammars significantly improves end-to-end LM performance on eight out of ten real tasks, matching or exceeding the performance of a professionally-maintained grammar. In comparison, the context-free grammars generated by existing LM baselines and a state-of-the-art formal technique perform significantly worse over the same evaluation.

Dataflow architectures have gained renewed interest due to their balance between power efficiency and performance. In (spatial) dataflow architectures, a program is represented as a set of entirely distributed and dynamically scheduled dataflow operators that communicate through asynchronous channels, which greatly improves data locality and parallelism. However, compiling to dataflow architectures remains an error-prone process, in order to maintain determinacy while enabling pipelining. Determinacy means that the result of a dataflow program is deterministic and independent of the schedule of operator execution, and pipelining is an important optimization in spatial dataflow that enables parallelism across loop iterations. In this work, we present Wavelet, the first effort to formally verify a compiler for asynchronous dataflow. We use a mix of techniques to achieve this goal. Our frontend uses a novel capability type system with fences to synchronize conflicting memory accesses and enable pipelining. We then verify a Lean formalization of two core passes of our compiler that translates elaborated programs from the type checker to dataflow graphs, proving important properties of forward simulation and determinacy. Notably, our formalization semantically propagates the soundness guarantees of the frontend type system, ensuring modularity between simulation and determinacy proofs. In evaluation, we show that dataflow graphs compiled by Wavelet have comparable sizes to those produced by an unverified optimizing compiler for the RipTide spatial dataflow architecture.

Deciding satisfiability of Constrained Horn Clauses (CHCs) modulo the theory of fixed-size bit-vectors ($\mathcal{T}_B$) is fundamental to bit-precise program verification. However, state-of-the-art CHC-solvers often struggle with $\mathcal{T}_B$, limiting scalability in bit-precise reasoning. We present Mosaic, a theory-modular framework for deciding satisfiability of CHCs modulo $\mathcal{T}_B$ by combining reasoning in $\mathcal{T}_B$ and the theory of Integer Arithmetic ($\mathcal{T}_I$). Given a CHC set modulo $\mathcal{T}_B$, Mosaic partitions it into two fragments interpreted over $\mathcal{T}_B$ and $\mathcal{T}_I$. Moreover, it implements an algorithm that reasons about the fragments in a modular fashion, exchanges information between them via sound translations across theories, and determines satisfiability w.r.t. the original CHC set. We implemented a prototype of Mosaic using Z3 and Spacer and evaluated it on bit-manipulating benchmarks. Our evaluation shows that Mosaic significantly outperforms Spacer on these benchmarks.

Building reliable applications that leverage large language models (LLMs) remains a significant challenge. While LLMs offer impressive capabilities across diverse tasks, their outputs often lack accuracy and provide no clear measure of confidence. This uncertainty compounds in flows of multiple calls to LLMs and other tools, making it difficult for developers and end-users to trust the results. This paper introduces a probabilistic language for programming LLM-based flows. It enables developers to quantify and propagate uncertainty throughout the application's flow, and experiment with different inference scaling techniques without adding a single line of code beyond the flow's logic. We present an experimental study to demonstrate this capability, and a case study building a theorem proving agent for the Rocq theorem prover.

Quantum annealers solve problems by finding the lowest-energy (ground) state of a programmable physical system, a 2-local Ising model, whose energy function is the Hamiltonian. We compile recursive Datalog programs into such models so that the ground state projects onto the program's minimal Herbrand model. The compiler has four stages: binarization, grounding, reduction to a Min-Ones SAT formula, and Ising encoding. Each rule becomes an energy penalty on the one assignment that violates it, and a small uniform cost on every true atom selects the minimal model. We contribute both in theory and in practice with per-stage correctness lemmas and a correspondence theorem, verified in Lean 4, establishing that the ground state of the compiled model projects onto the program's minimal Herbrand model. We map the compiled models onto the topologies of commercial annealers and characterize, under classical and simulated-quantum annealing, whether and when that certified ground state is attained.

Fault-tolerant quantum computation enables the deployment of practical quantum algorithms but incurs substantial overhead from error correction, making resource estimation a central concern. Beyond case-by-case analyses, existing quantum programming languages either require programmers to manipulate low-level hardware details, rendering fault-tolerant implementations cumbersome, or abstract away the underlying error-correction schemes, reducing the effectiveness of resource utilization and estimation. To address these limitations while preserving programmability, we present a quantum programming language that enables efficient resource utilization, together with a resource-estimation framework for comprehensive resource analysis. Our framework features programmer-visible abstractions of error-correction schemes and cross-layer program-hardware analysis, allowing systematic exploration of resource trade-offs. We evaluate our approach on detailed fault-tolerant implementations of practical large-scale quantum algorithms, including components typically treated as black boxes in existing frameworks. The results demonstrate that our framework enables substantial resource savings while delivering detailed, fine-grained, and accurate resource estimates for fault-tolerant quantum programs.

Pointer analysis is a cornerstone of numerous static analysis applications, including compiler optimizations, slicing, bug detection, and verification. While offline simplification is a common approach to boosting performance, existing methods are often tightly coupled to specific analysis algorithms and limited to a set of simplification rules. This paper explores a new perspective: applying semantic-preserving compiler optimizations directly to intermediate representation (IR) before pointer analysis. This strategy is modular, analysis-agnostic, and easily integrates with existing tools. We conduct an empirical study using diverse programs and three pointer analyses. The results show substantial performance gains—up to 3.14x speedup and 1.94x memory reduction—while precision remains largely unchanged. We also analyze the trade-offs between optimization overhead and analysis speedup, quantify changes in IR structure, assess the characteristics of optimization configurations, and identify promising directions for future research.

Emfrp is a functional reactive programming (FRP) language designed for small-scale embedded systems. Time-varying values are the primary abstraction mechanism in FRP and enable concise descriptions of reactive behavior. In practice, however, Emfrp programs are compiled into C and combined with platform-dependent input/output components written in C or C++. Consequently, developers must debug the resulting mixed C/C++ program using conventional debuggers such as GDB, even though the application logic is written in Emfrp. This situation creates an abstraction gap between the source-level FRP program and the executable system. This paper presents a multi-mode debugging framework for Emfrp-based embedded applications. The framework supports debugging at the level of Emfrp abstractions while also allowing inspection of platform-specific C/C++ I/O code. Our approach uses a source code mapping technique that relates Emfrp constructs to corresponding locations in the compiled program. A case study on an ESP32 microcontroller using representative debugging scenarios demonstrates improved debugging efficiency.

Optimizing compilers miss profitable transformations when their enabling semantics are absent from the analyzed program representation. We ask whether large language models (LLMs) can recover such semantics from heterogeneous C/C++ context and realize them as validated, contract-preserving artifacts. We introduce SeGaBench, an executable benchmark containing 100 synthetic and 20 source-backed cases spanning low-level assumptions, data-structure invariants, and high-level semantic lifting. Each case includes hidden enabling semantics, an oracle artifact, correctness and semantic validators, and a reproducible performance protocol. We evaluate five LLMs using five independent responses per case. The strongest model produces correct artifacts in 94.8% of responses, achieves at least 1.05x speedup in 83.3%, and obtains a performance success on 93.3% of cases. Nevertheless, correct artifacts often close only part of the oracle gap. These results show that LLMs can complement compiler analysis as speculative semantic proposers, provided that their artifacts are validated and evaluated.

AI coding agents are stochastic workflows: prompts are interpreted, artifacts are sampled, validators produce observations, and orchestrators commit or repair. Small prompt or specification changes can therefore alter program-behavior distributions even when the texts appear synonymous. Existing systems evaluate correctness, but lack a compositional criterion for safely replacing a prompt, contract, generator, or program inside a complete agentic workflow. We introduce GenOS, a probabilistic operational semantics for this replacement problem. Each layer is modeled as a Markov kernel, and each interface carries an observer-relative equivalence. We prove that equivalence-compatible kernels descend to quotient classes and that quotienting commutes with distributional extension and sequential composition. Hence, equivalent prompts induce equal probabilities for all downstream equivalence-closed events, including verified commit. We also establish workflow bisimulation, guarded-commit safety under sound validation, total-variation non-expansiveness, and an additive robustness bound that attributes approximation error to individual pipeline layers. An executable insertion-sort audit instantiates the theory with natural-language paraphrases, a formal contract, six programs, two observers, and exhaustive execution on 121 inputs. Equivalent prompts yield identical code-class and commit distributions; a prompt assigning 5% probability to an in-place contract is distinguished by a mutation observer, while downstream distances remain within the predicted bound. Across 20,000 randomized finite-kernel trials, no exact or approximate law is violated. GenOS is model-parametric: compatibility is a measurable property to test, not an assumption about language-model behavior.

Table-driven DFA lexing is sequential: each transition depends on the previous byte's state. Scanning one input in parallel needs each chunk's entry state, which existing methods recover by simulation, speculation, prescanning, or overlap. We give two conditions under which none is needed. For a longest-match scanner restarting from q0 at every token boundary, a byte b is a certified split symbol when no reachable state other than q0 has a b-transition whose target can reach acceptance, and q0 is not re-entrant if it has one. Every occurrence of such a byte in completely tokenizable input begins a token, so chunks starting there reproduce the serial sequence of kinds and lengths by ordered concatenation. The condition is necessary as well as sufficient, and fragile: one string, comment, or whitespace run can eliminate every useful certificate, and comments and whitespace are usually discarded. We therefore weaken the guarantee to equality after deleting a declared discarded set, and give a second condition, sound and more permissive, coinciding with the first when the discarded set is empty and strictly gaining on suitable pairs of token set and discarded set, but conservative rather than exact, decided from the same tables, answered by a second constant-time one-bit query. It recovers newline for a conventional C-like tokenization and tab, newline and carriage return for JSON, without altering their token definitions, and refuses it where block comments are unrestricted. Splitting at exact certificates in the munch library reaches 92.6-95.3% parallel efficiency at eight threads on a restricted CPU set, on a 512 MiB dense corpus beyond last-level cache, and a 3.46-3.94x end-to-end speedup at four threads, across two benchmark revisions on one machine. It turns delimiter-based parallel lexing from a language-specific assumption into a property a compiler checks.

Most of the conventional implementations of regular expressions are based on backtracking. Such implementations are slow in the worst case, and thus, we would like to develop a better matching algorithm. However, it is nontrivial to provide an efficient matching algorithm that can deal with practical extensions including submatch addressing. This paper studies regular expression with lookaheads and negative lookaheads, abbreviated to REwLA. First, we propose a transformation from a REwLA of size $m$ to a deterministic finite automaton of $\mr{O}(2^{2^m})$ states. Next, we consider weighted regular expressions, which enable us to calculate submatch addressing. We propose a transformation from a weighted REwLA of size $m$ to a weighted nondeterministic finite automaton of $\mr{O}(2^{2^m})$ states.

We show that the tropical semiring over the natural numbers, when used as the grading space in graded coeffect typing, faithfully models the passage of time while simultaneously guaranteeing productivity of well-typed programs. A grade a, when assigned to a function parameter, indicates that the parameter is not necessarily available immediately, but will become available after a time steps. We investigate this idea through two formal systems. We first introduce a graded type system featuring recursive and polymorphic types, and show that, in this setting, a natural restriction on recursive types is sufficient to guarantee productivity, while still allowing the definition of streams and recursive programs on them. In particular, we prove that Nakano's later modality can be embedded directly into our system. We then show that tropical grading naturally suggests a novel form of intersection typing, in which the role traditionally played by sets or multisets of types is instead taken by "timed" sets, i.e., functions assigning to each type A the earliest time, represented as a grade, from which the underlying term is available with type A. For the resulting system, we prove not only that productivity is guaranteed, but that it is also characterized: the typable terms are exactly those with hereditarily head normal forms. Remarkably, the system is recursion-theoretically optimal, i.e., typability can be directly proved to be a $\Pi_0^2$ property in the arithmetical hierarchy.

Model checkpoints are growing in both number and size, which makes archival, transfer, and deployment increasingly costly. General-purpose compressors can reduce storage requirements but ignore tensor structure, whereas existing tensor-specific compressors rely on fixed and format-specific pipelines. We present Brevis, which formulates lossless tensor compression as program synthesis. We design a typed domain-specific language (DSL) that captures recurring tensor structures, such as repeated regions and floating-point fields, through a set of reversible operators. Given a tensor, Brevis synthesizes a self-contained DSL program that reconstructs it bit-exactly. A checkpoint-specific production prior, learned from a small representative sample of tensors, guides a bounded A* search to synthesize compact programs, which can later be executed directly for bit-exact decompression. On 10 public checkpoints spanning language, audio, and image generation models, Brevis reduces 2.13 TB of checkpoint data to 1.41 TB, a 33.93% storage reduction. It produces archives up to 30.87% smaller than those of four general-purpose compressors, including zstd and gzip, and smaller archives than the tensor-specific compressors ZipNN and DFloat11. Under a practical concurrency configuration, Brevis achieves 3.60 GB/s compression and 6.61 GB/s decompression while preserving every source byte.

This paper proposes an explainable digital twin for vulnerability detection in AArch64 machine code without access to source code. The digital twin reproduces the concrete execution of a program and preserves the state of registers, processor flags, memory, and live allocated blocks. Each instruction is transformed into a trace event containing the instruction name, operand values, and the post-instruction state. Vulnerabilities are represented as symbolic rules in Kleene algebra with tests: each rule specifies an event sequence and predicates over the machine state. This approach enables the detection of not only isolated unsafe instructions but also multi-step execution patterns. The rules are compiled into finite automata that scan the trace without using an SMT solver. The experimental evaluation covers three CWE classes: integer overflow (CWE-190), null pointer dereference (CWE-476), and heap buffer overflow (CWE-122). The system detected all three predefined vulnerabilities and produced no report on the safe trace. Each detection result includes the triggered rule, the trace position, and the concrete state values, thereby providing a reproducible explanation.

P4 is a domain-specific language for programming protocol-independent packet processors, where packet parsers describe how incoming bit-streams are structured into headers and fields. Building on work by Doenges et al. (2022), we present Octopus, a tool that translates P4 packet parsers into automata and then attempts to (symbolically) check their equivalence. Octopus produces evidence, either in the form of a bisimulation demonstrating equivalence, or a counterexample bit-stream witnessing a behavioral difference between the two parsers. In contrast with earlier work, our tool can check equivalence between non-trivial parsers within minutes, on consumer hardware. We report on the tool's implementation and evaluate its usability in networking contexts.

Training and deployed inference often cross export, conversion, and platform-specific runtime boundaries. Meganeura asks whether one compact native compiler can span both phases on consumer GPUs. Its typed static graph, automatic differentiation, optimizer, checkpoint, memory planner, and runtime lower specialized programs through Vulkan and Metal. We compare five matched workloads with PyTorch on NVIDIA and AMD discrete GPUs, an AMD APU, Apple silicon, and an Intel iGPU. The protocol separates strict f32 from validated fast paths and gates forward and backward independently. Forty-eight of 50 device-workload-mode cells pass both gates; the other two share one unresolved backward-reference disagreement on a newly supported APU. In strict f32, Meganeura wins 12 of 20 GPU-referenced minimal-latency cells and has a median valid training gap of 1.8x. On the discrete AMD GPU, four of five inference workloads are within 1.10x of compiled ROCm PyTorch and three training workloads are faster. Under accelerated contracts, the worst training gap is 4.6x. Compilation takes 0.1-2.4 seconds versus 6-96 seconds for torch.compile on supported GPU paths; the stripped binary is 13 MiB. Dispatch profiles localize the largest gaps to convolution derivatives and attention backward. A physical Android XR case study transfers a Meganeura-trained decoder into an Adreno/OpenXR application sharing the graphics queue. The results show that general consumer graphics APIs can support a compact shared train-to-deploy stack at useful, sometimes vendor-competitive performance. The measured gaps point to kernel coverage, scheduling, and arithmetic policy rather than an identified API limitation.