The Bcachefs file system will be removed from the main Linux kernel tree. Despite its technical merits and active development, the project's fate has been complicated by an ongoing conflict between its author, Kent Overstreet, and the leading maintainers of the Virtual Memory (VM) and File Systems (FS) subsystems.
Bcachefs is one of the Linux file systems competing with Btrfs and ZFS, using the Copy-on-Write (COW) mechanism, where changes do not overwrite existing data — the new state is written to a new location, after which the pointer to the current state is changed. The exclusion of Bcachefs from the main kernel tree will complicate things for users who will have to rely on third-party builds or wait for its official return to the main development branch.
The conflict between Kent and several key maintainers has been ongoing for several years. According to participants, it's not so much about technical aspects or breaches of kernel development procedures, but rather a loss of trust. A recent dispute over whether the journal_rewind feature was a bug fix or new functionality was the last straw, and Linus Torvalds has stated that a long demonstration of constructive cooperation with other maintainers is necessary to restore trust.
Linus Torvalds believes that the problem is systemic and affects not only him personally but also the entire group of file system maintainers and the Virtual Memory subsystem. Any promises to 'behave better' are insufficient — real evidence is needed to show that interactions with other developers are conflict-free. According to Linus, this should happen while Bcachefs is supported outside the main kernel tree.
Kent Overstreet believes that Bcachefs is already stable and has a broad user base. He assures that he adheres to the rules for merging changes and makes fixes only in critical cases. At the same time, Kent advocates for a functional release preparation process that will allow for quick delivery of fixes to users.
The future of Bcachefs appears to be tied to its development outside the main kernel codebase—through Linux distributions and third-party repositories. Regaining trust may take years and will require Kent's active involvement in other kernel projects and collaboration with respected maintainers.
The material was written as a result of personal correspondence with Linus and Kent, which unfortunately does not allow for disclosing more detailed information.
Source: opennet.ru
