Linux kernel maintainers agreed at the 2025 Linux Kernel Maintainers Summit in Tokyo that Rust is no longer experimental and should be recognized as a core part of the kernel, according to The New Stack’s reporting. The agreement establishes its place alongside C, with the experimental label due to be removed.
Kernel developer Jonathan Corbet described the consensus among the assembled developers as recognizing that Rust is “here to stay,” adding that the experimental tag would come off. Maintainer Steven Rostedt told The New Stack there was “zero pushback” at the meeting.
The agreement followed comments by Linux creator Linus Torvalds at Open Source Summit Korea 2025. According to The New Stack, Torvalds said Rust was “truly becoming part of the kernel” and was no longer just an experiment.
C remains the dominant language in the codebase. The New Stack reported that, as of April 2025, the kernel contained approximately 34 million lines of C and 25,000 lines of Rust. The Rust effort has focused on enabling new kernel code rather than rewriting the existing kernel.
Memory safety and driver development
Memory safety was a central argument in early discussions about adopting Rust. At the 2019 Linux Security Summit, Alex Gaynor and Geoffrey Thomas cited research showing that roughly two-thirds of kernel vulnerabilities assigned CVEs in Android and Ubuntu stemmed from memory-safety issues, according to Nelson Elhage’s account for LWN.
Rust’s safety guarantees depend on the correctness of underlying unsafe code. In an account published by the Internet Security Research Group’s Prossimo project, Rust for Linux maintainer Miguel Ojeda explained that Rust prevents undefined behavior, including use-after-free errors, double frees and data races, provided that unsafe code is sound.
Drivers have long been a proposed starting point. In a 2021 interview with ZDNET, Torvalds described them as relatively small, independent components. He also noted that many drivers serve only a few architectures, making gaps in Rust’s architecture support less of an obstacle for those uses.
In the same report, Linux stable kernel maintainer Greg Kroah-Hartman described drivers as the “end leafs” of the kernel’s dependency tree: they depend on core functionality, but nothing depends on them. He said integrating Rust would require carefully mapping the kernel’s C structures and lifetime rules into Rust equivalents.
Development and maintenance challenges
Ojeda’s Prossimo account dates the creation of the Rust for Linux organization to summer 2019, with activity beginning the following summer. At the 2020 Linux Plumbers Conference, discussions addressed access to existing C APIs, architecture support and compatibility between the Rust and C application binary interfaces, according to Elhage’s LWN report.
In 2022, Kees Cook submitted a pull request for initial Rust support in Linux 6.1-rc1. He described the proposed code as the minimum needed to build Rust code in the kernel, with more interfaces and drivers to follow. The submission included build-system support, Rust crates and bindings, documentation and samples.
Learning and maintaining another language also raised concerns. During a 2021 discussion covered by LWN, Video4Linux maintainer Laurent Pinchart said he did not have time to stop other work and learn Rust. At Linux Plumbers 2024, Intel engineer Dan Williams described the difficulty of reviewing and debugging Rust code for maintainers unfamiliar with the language, according to The New Stack.
A dispute over proposed Rust abstractions for the Direct Memory Access API illustrated the disagreement over maintenance. LWN reported that Christoph Hellwig opposed the integration and said he had no interest in maintaining code in multiple languages. Rust developer Danilo Krummrich said contributors had written a shared abstraction around the C API for Rust drivers and were offering to maintain it themselves.