Rusix is taking an incremental path toward a Rust-native kernel: preserve Linux behavior where it matters, and change the implementation within.
The goal is not a line-by-line translation that keeps C-shaped internals forever. It is a Rust-native kernel with Linux-compatible external behavior.
Start with a small, defined boundary. Compare the existing implementation and its Rust replacement. Prove the replacement is in the real execution path. Then work toward coherent Rust subsystems.
Read architecture decision 0001Observable semantics, supported interfaces and data formats.
Ownership, lifetimes and Rust-native internal interfaces.
A migration policy, not a claim of full compatibility. Most of the kernel remains C.
A foundation to examine, not a kernel to install. Scope and limitations travel with every result.
int_sqrtRecorded kernel comparisons exercise the C implementation and Rust replacement against the same resolved configuration and tests. This establishes a narrow integer-square-root milestone—not a completed rewrite.
Read the verification recordA Rust-enabled Image build and QEMU virt KUnit boot have been verified on the local development branch. That work is not yet public evidence. It does not establish physical ARM hardware or Apple M1 support.
Module/export behavior, broader configurations, workloads and architecture coverage remain work. There is no production-readiness or performance-parity claim.
Follow the roadmapKernel work starts with understanding the contract. Documentation,
test coverage and review matter alongside implementation.
Read the architecture, workspace layout and toolchain notes before choosing work. This is a kernel checkout, not a standalone Cargo project.
Use isolated VMs. Keep build records, warnings and failures. Distinguish host tests, a kernel build and a real VM boot.
Identify callers, invariants and unchanged tests. Keep translation and redesign reviewable; document what remains C.