BeamFS filesystem with Reed-Solomon error correction proposed for Linux kernel

Developer Aurelien Desbrieres has proposed BeamFS for inclusion in the Linux kernel. The filesystem targets radiation-heavy environments like satellites and medical equipment, where it can repair corrupted data without requiring redundant hardware. It organizes 4 KB blocks into 16 interleaved Reed-Solomon RS(255,239) codewords, using about 7% extra storage. The submission is explicitly experimental and warns of known bugs causing data loss, security issues, and kernel crashes.

How BeamFS differs from existing filesystems

• Uses Reed-Solomon RS(255,239) forward error correction instead of checksums for both detection and repair
• Each 4 KB block contains 16 interleaved codewords, correcting up to 8 corrupted bytes per codeword
• Requires approximately 7% storage overhead for error-correction data
• Protects metadata including superblocks, inodes, directories, and allocation bitmaps
• Returns I/O error when recovery fails rather than silent data corruption
• Tested on x86-64 and ARM64 with Linux 7.3-rc5; recovered nearly all simulated single-bit corruptions
• No journaling or copy-on-write: unexpected shutdowns require full fsck.beamfs check

Why it matters

If eventually merged, BeamFS would fill a niche for single-device storage integrity in environments where redundant hardware is impractical. Its explicit failure mode (returning errors rather than corrupt data) aligns with safety-critical system needs, though the experimental status and lack of journaling mean it remains far from production use.