tag: #kernel

bcachefs is, on technical merit, the most interesting copy-on-write filesystem to arrive on Linux in a decade. It is also, as of Linux 6.18, no longer in the kernel, ejected after a governance dispute between its author and Linus Torvalds and now shipping as a DKMS module like ZFS. This is the story of a filesystem whose engineering and whose politics point in opposite directions, and what that means for anyone deciding whether to run it.

The Linux kernel has had a written rule about commit hygiene since before the platforms most engineers use today existed. "Separate each logical change into a separate patch." The rule is currently being violated at an industrial scale by everyone using LLM assistants to write code, because the assistants produce sprawling multi-concern diffs by default. The kernel's discipline is the cleanest answer to the resulting mess, and it scales down to any project size.

Copy Fail, Dirty Frag, Fragnesia, ssh-keysign-pwn - four Linux kernel root exploits in three weeks. The kernel is fine. The admins are not.

eBPF is the most interesting thing to happen to the Linux kernel in the last decade and one of the least understood.

Demystifying address space translation for Linux and OS geeks

Exploring base and bound memory protection

A race in mm/gup.c lets any local user write to read-only memory and rewrite setuid binaries. It sat in the kernel for nine years. Here is how the race actually works, and why "it's only local" was always a lie, especially now that everyone runs containers.