Towards Automated Verification of LLM-Synthesized C Programs

2024-10-18

arXiv.org

AI Summary

Automatically verifying C programs generated by large language models (LLMs) remains challenging due to their syntactic and semantic irregularities, which hinder formal verification. Method: This paper proposes SynVer—a novel framework that tightly integrates LLM-based program synthesis with formal verification. SynVer introduces verifiability-aware biasing mechanisms operating at both syntactic and semantic levels to guide LLMs toward generating verification-friendly code. It further incorporates separation logic (SL) specifications and the Verified Software Toolchain (VST) to enable end-to-end, fully automated verification—from specification to C implementation to machine-checked safety proofs. Results: Evaluated on diverse benchmarks covering basic coding tasks, SL assertions, and API specifications, SynVer significantly improves the automatic verification success rate of LLM-generated C programs. Empirical results demonstrate its scalability, robustness, and effectiveness in bridging the gap between neural code generation and rigorous formal assurance.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Large Vision Models

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

We present synver{}, a novel synthesis and verification framework for C programs, that deploys a Large Language Model (LLM) to search for a candidate program that satisfies the given specification. Our key idea is to impose syntactic and semantic biases on programs generated by LLMs, such that the synthesized program is more amenable to automated verification. Based on this idea, we propose a novel specification-verification tool, built on top of Verified Software Toolchain, that help automate the process. Our experiments on a diverse set of benchmarks drawn from the deductive program synthesis community, shows that this approach is scalable and extensible. The benchmarks constitute of specifications comprising of basic coding examples, Separation Logic based assertions, and API specifications.

Problem

• Automating verification of LLM-generated C programs
• Imposing syntactic and semantic biases for verification
• Developing scalable specification-verification tool for synthesis

Innovation

• LLM generates programs with syntactic biases
• Semantic biases make verification more efficient
• Tool built on Verified Software Toolchain automates process