Linus submits first LLM-assisted kernel patch set

On October 15, reports said Linus Torvalds submitted the first set of kernel patches written with large language model assistance for review. The 10 patches defer freeing hot folios so data access can also happen inside RCU-protected regions; when user data copies do not fault, reference counting on the fast path is avoided. The report says the approach delivered an order-of-magnitude speedup in read()-heavy benchmarks. The patches are still under review, and review has found race conditions that need further changes.

Written by AI from the coverage · updated 2 hours ago

Not enough continuous observations to draw a trend yet.