How CISA can improve OSS security

By Jim Miller

The US government recently issued a request for information (RFI) about open-source software (OSS) security. In this blog post, we will present a summary of our response and proposed solutions. Some of our solutions include rewriting widely used legacy code in memory safe languages such as Rust, funding OSS solutions to improve compliance, sponsoring research and development of vulnerability tracking and analysis tools, and educating developers on how to reduce attack surfaces and manage complex features.

Background details

The government entities responsible for the RFI were the Office of the National Cyber Director (ONCD), Cybersecurity Infrastructure Security Agency (CISA), National Science Foundation (NSF), Defense Advanced Research Projects Agency (DARPA), and Office of Management and Budget (OMB). The specific objective of this RFI was to gather public comments on future priorities and long-term areas of focus for OSS security. This RFI is a key part of the ongoing efforts by these organizations to identify systemic risks in OSS and foster the long-term sustainability of OSS communities.

The RFI includes five potential areas for long-term focus and prioritization. In response to this request, we are prioritizing the “Securing OSS Foundations” area and each of its four sub-areas: fostering the adoption of memory-safe programming languages, strengthening software supply chains, reducing entire classes of vulnerabilities at scale, and advancing developer education. We will provide suggested solutions for each of these four sub-areas below.

Fostering the adoption of memory-safe programming languages

Memory corruption vulnerabilities remain a grave threat to OSS security. This is demonstrated by the number and impact of several vulnerabilities such as the recent heap buffer overflow in libwebp, which was actively being exploited while we drafted our RFI response. Exploits such as these illustrate the need for solutions beyond runtime mitigations, and languages like Rust, which provide both memory and type safety, are the most promising.

In addition to dramatically reducing vulnerabilities, Rust also blends well with legacy codebases, offers high performance, and is relatively easy to use. Thus, our proposed solution centers on sponsoring strategic rewrites of important legacy codebases. Since rewrites are very costly, we specifically recommend undertaking a comprehensive and systematic analysis to identify the most suitable OSS candidates for transitioning to memory-safe languages like Rust. We propose a strong focus on software components that are widely used, poorly covered by tests, and prone to such memory safety vulnerabilities.

Strengthening software supply chains

Supply chain attacks, as demonstrated by the 2020 SolarWinds hack, represent another significant risk to OSS security. Supply chain security is a complex and multifaceted problem. Therefore, we propose improving protections across the entire software supply chain—from individual developers, to package indices, to downstream users.

Our suggested strategy includes establishing “strong link” guidelines that CISA could release. These would provide guidance for each of the critical OSS components: OSS developers, repository hosts, package indices, and consumers. In addition to this guidance, we also propose funding OSS solutions that better enable compliance, such as improving software bill of materials (SBOM) fidelity by integrating with build systems.

Reducing entire classes of vulnerabilities at scale

Another area of focus should be on large-scale reduction of vulnerabilities in the OSS ecosystem. Efforts such as OSS-Fuzz have successfully mitigated thousands of potential security issues, and we propose funding similar projects using this as a model. In addition, vulnerability tracking tools (like cargo-audit and pip-audit) have been successful at quickly remediating vulnerabilities that affect a wide number of users. A critical part of effectively maintaining these tools is properly maintaining the vulnerability database and not allowing over-reporting of insignificant security issues that could result in security fatigue, where developers ignore alerts because there are too many to process.

Therefore, our proposed solution is sponsoring the development and maintenance of tools for vulnerability tracking, analysis tools like Semgrep and CodeQL, and other novel techniques that could work at scale. We also recommend sponsoring research pertaining to novel tools and techniques to help solve specific high-value problems, such as secure HTTP parsing.

Advancing developer education

Lastly, we believe that improving developer education is an important long-term focus area for OSS security. In contrast to current educational efforts, which focus primarily on common vulnerabilities, we propose fostering an extension of developer education that covers areas like reducing attack surfaces, managing complex features, and “shifting left.” If done effectively, creating documentation and training materials specifically for these areas could have a substantially positive, long-term impact on OSS security.

Looking ahead

Addressing OSS security can be a complex challenge, but by making targeted interventions in these four areas, we can make significant improvements. We believe the US government can maximize impact through a combination of three strategies: provisioning comprehensive guidance, allocating funding through agencies like DARPA and ONR, and fostering collaboration with OSS foundations like OSTIF, OTF, and OpenSSF. This combined approach will enable the sponsorship and monetary support necessary to drive the research and engineering tasks outlined in our proposed solutions.

Together, these actions can build a safer future for open-source software. We welcome the initiative by ONCD, CISA, NSF, DARPA, and OMB for fostering such an open discussion and giving us the chance to contribute.

We welcome you to read our full response.

Leave a Reply