Ori Bernstein updated the experimental OpenBSD port of GEFS on October 31, 2026, reporting that users can now create a filesystem with mkfs on OpenBSD. But the developer's update identifies the consistency protocol as the biggest missing piece. The work makes the port easier to try without establishing that it can safely hold your files.
GEFS, the Good Enough File System, originated on 9front. Bernstein's September preview describes a snapshotting, copy-on-write filesystem being moved into the OpenBSD kernel. That announcement warned that the port was not ready for production and that data loss was expected, especially on errors. It also said inclusion in OpenBSD's source tree was some way off.
Formatting comes before snapshot controls
The October update reports some completed userspace tools, while snapshot control remains unfinished. Bernstein has also reviewed error handling and brought over bug fixes from 9front, though he cautions that errors may still be handled incorrectly.
The earlier preview had described booting 9front in a virtual machine to create or check a filesystem as an unsatisfactory arrangement. Native formatting removes part of that testing friction. It doesn't establish that repair tools or snapshot management are complete. We haven't built or tested the port.
Why write order still matters
In the original technical outline, Bernstein explains that the superblock write must follow all writes in the snapshot. In plain terms, the filesystem must finish storing the blocks before committing the metadata that points to them. That requirement explains why a working formatter alone cannot establish crash safety.
His latest proposal uses ordering barriers during commits while allowing new block I/O. It remains a design under discussion. For developers following GEFS, the next useful evidence is an implemented consistency protocol and tests of interrupted writes and error paths. Until then, the preview's explicit data-loss warning is the relevant starting point for any experiment.