SearcharxivSearch

arXiv · 2206.14606

Building a Secure Software Supply Chain with GNU Guix

Abstract

The software supply chain is becoming a widespread analogy to designate the series of steps taken to go from source code published by developers to executables running on the users? computers. A security vulnerability in any of these steps puts users at risk, and evidence shows that attacks on the supply chain are becoming more common. The consequences of an attack on the software supply chain can be tragic in a society that relies on many interconnected software systems, and this has led research interest as well as governmental incentives for supply chain security to rise. GNU Guix is a software deployment tool and software distribution that supports provenance tracking, reproducible builds, and reproducible software environments. Unlike many software distributions, it consists exclusively of source code: it provides a set of package definitions that describe how to build code from source. Together, these properties set it apart from many deployment tools that center on the distribution of binaries. This paper focuses on one research question: how can Guix and similar systems allow users to securely update their software? Guix source code is distributed using the Git version control system; updating Guix-installed software packages means, first, updating the local copy of the Guix source code. Prior work on secure software updates focuses on systems very different from Guix -- systems such as Debian, Fedora, or PyPI where updating consists in fetching metadata about the latest binary artifacts available -- and is largely inapplicable in the context of Guix. By contrast, the main threats for Guix are attacks on its source code repository, which could lead users to run inauthentic code or to downgrade their system. Deployment tools that more closely resemble Guix, from Nix to Portage, either lack secure update mechanisms or suffer from shortcomings. Our main contribution is a model and tool to authenticate new Git revisions. We further show how, building on Git semantics, we build protections against downgrade attacks and related threats. We explain implementation choices. This work has been deployed in production two years ago, giving us insight on its actual use at scale every day. The Git checkout authentication at its core is applicable beyond the specific use case of Guix, and we think it could benefit to developer teams that use Git. As attacks on the software supply chain appear, security research is now looking at every link of the supply chain. Secure updates are one important aspect of the supply chain, but this paper also looks at the broader context: how Guix models and implements the supply chain, from upstream source code to binaries running on computers. While much recent work focuses on attestation -- certifying each link of the supply chain -- Guix takes a more radical approach: enabling independent verification of each step, building on reproducible builds, "bootstrappable" builds, and provenance tracking. The big picture shows how Guix can be used as the foundation of secure software supply chains.

Explore related subjects

Keep this discovery

BibTeXRIS

Ludovic Courtès. 2022-06-28. Building a Secure Software Supply Chain with GNU Guix. https://doi.org/10.22152/programming-journal.org%2F2023%2F7%2F1

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

LLMVul: A Vulnerability-Labeled Dataset of LLM-Generated C/C++ Functions from Real Production Repositories

Large language models (LLMs) are increasingly used to generate and assist with software development, yet existing vulnerability datasets largely focus on human-written code or controlled prompting environments. This limits the ability to study security weaknesses in LLM-generated code as it appears in real-world software projects. We present LLMVul, a vulnerability-labeled dataset of LLM-generated C/C++ functions mined from real production repositories. We mine AI-assisted development activity from GitHub over a 4 year period, from November 13, 2022 to September 3, 2026, using provenance signals such as commit metadata and AI-related authorship evidence. After filtering and deduplication, LLMVul contains 21,430 unique C/C++ functions from 226 repositories, together with repository, commit, function, provenance, and AI-tool metadata. We establish vulnerability labels using an ensemble of complementary static-analysis and pattern-based techniques and assign Common Weakness Enumeration (CWE) categories to confirmed vulnerable functions. To assess labeling reliability, we additionally conduct independent manual annotation and measure inter-rater agreement using Cohen's kappa ($k=0.79$). LLMVul contains 1,540 ensemble-vulnerable functions spanning 17 unique CWE categories, providing substantially more real-world LLM-generated vulnerable C/C++ functions than existing vulnerability-oriented LLM code benchmarks. By preserving both code-level vulnerability labels and generation/provenance metadata, LLMVul enables reproducible research on vulnerability detection, security evaluation of LLM-generated code, and analysis of vulnerability patterns in AI-assisted software development. The LLMVul dataset is publicly available at https://doi.org/10.5281/zenodo.22668216.

cs.SE

What a Random Draw from the MCP Registry Contains, and What Tool-Use Benchmarks Contain Instead

Studies of the Model Context Protocol (MCP) server ecosystem draw their samples in ways that quietly select for servers that work: reference sets, popularity lists, hand-curated frames, or pipelines that repair a server until it starts. We report what an unrepaired probability sample actually contains. From a 24,135-server registry census we draw 400 npm/stdio servers with a published seed and probe each one over the wire. Only 48.8% complete an initialize handshake, against 66.7% for a hand-curated frame measured with the same instrument, and the dominant failure is not missing credentials (13.3%) but servers that never start at all (37.5%). Among the 195 that do run, hard conformance is total: zero fatal JSON Schema violations across 2,766 advertised tools. Optional safety annotations are the real variance, and the tool-level omission rate on a random draw is 58.8% against 41.5% on the curated frame, so curation flatters this figure too. We then compare the tool descriptions these servers advertise against two tool-use benchmark corpora using one method held constant. Real MCP tools show 2.8% near-duplication at cosine 0.70, and all of it lies within single servers: cross-author near-duplication is 0.0% at every threshold tested. BFCL v4 shows 16.7%, of which 16.4 points lie between independently presented tasks. UltraTool shows 0.3%, cleaner than real tools, so this is a property of BFCL and not of synthetic corpora as a class. Separately, 68.8% of raw BFCL rows and 85.6% of raw UltraTool rows are exact name-plus-description repeats, against 0.4% for real MCP, so any statistic computed over these releases without global deduplication measures repetition rather than tools. All figures regenerate from released scripts and a published seed.

cs.SE

Engineering Reliable Commit Gates for Agentic AI: Cost-Aware Verification Portfolios under Common-Mode Data Failures

Agentic systems commit state-changing actions, but additional verifiers can inherit the same upstream fault. We present VP-CONTROL, a runtime-assurance design and deterministic benchmark for cost-aware commit gates. Its 48 task templates yield 2,880 scenarios across six fault regimes. A fixed-call 2 x 2 experiment separates verifier-model diversity from evidence-source diversity. On frozen proposals from two local actor families, a cross-model vote over shared evidence approves 62.9% of unsafe proposals, versus 22.9% with an independent source. The source effect is 40.9 percentage points, compared with 11.3 for model diversity. A portfolio controller selects verification plans using only deployment-observable metadata. Approximate cluster-adjusted calibration at a nominal 5% per-task target yields 1.9% unsafe execution and 38.2% automated safe coverage on the locked test. Matched-budget portfolios also improve on fixed verification policies. Transfer remains conditional: unseen fault families yield 16-26% risk, and a FinQA check fails to reproduce the source effect with the tested small verifiers. A preregistered live HTTP/SQLite study tests concurrent writes and lost responses. After-check races defeat verifier-only gates; transactional partial guards prevent only covered failures, while a full atomic guard records no unsafe effects across 216 episodes. Idempotent request identifiers prevent duplicate effects after lost responses. The results motivate explicit evidence lineage, cost-aware selection, and commit-time enforcement, while exposing the limits of approximate calibration and local-tool generalization.

cs.SE