MRMCD2024 Rust in the Linux Kernel - A Status Check

Rust integration in the Linux kernel is still limited, with only a few drivers available in mainline (e.g., R null block driver, some file drivers)

Key differences from userspace Rust development:

• No standard library available, only core crate
• Custom allocators required
• No cargo build system
• Different printing functions (print_k instead of println)
• All C function calls must be wrapped in unsafe blocks

Benefits of Rust in kernel development:

• Eliminates entire classes of bugs (buffer overflows, memory leaks, race conditions)
• Clearer error handling through Result types
• Generally requires less code (30% reduction observed in some cases)
• Better type safety and ownership model

Current challenges:

• Longer compilation times compared to C
• Higher learning curve for kernel developers
• Limited platform support
• Social resistance from long-time kernel maintainers
• Need to create custom bindings for C functions

Development process requires:

• Enabling Rust support in kernel config
• Creating bindings for C functions using bindgen
• Implementing proper FFI (Foreign Function Interface) wrappers
• Following kernel module conventions
• Testing through QEMU or actual hardware

Integration with existing kernel code:

• Uses bindings_helper.h for C function calls
• Requires explicit handling of mutable/immutable parameters
• Must conform to C calling conventions
• Drivers can be written in either C or Rust, but not both simultaneously

Best use cases for Rust in kernel:

• New drivers without existing C implementation
• Code handling complex logic and user input
• Security-critical components
• Areas where memory safety is crucial