A few months ago I started an ext4fs inspector. It can compute Linux's wrong implementation of crc32c that does not do the final xor, leaving us with incorrect results as per the RFC standard https://datatracker.ietf.org/doc/html/rfc3385.
Nevermind, I ported it to OpenBSD VFS using Anthropic's claude-code for checked inspection and searching of a massive amount of code. You can tag the partition as ext4fs in disklabel and it mounts and unmounts at reboot and reports correct information in df. For now it is a success for me as I'm completely new to Kernel side development.
I must say for this whole exercise I used a clean-room environment and never looked or cared for the linux implementation, only reverse engineering and documentation reading. And munching through claude-code.
https://github.com/thodg/OpenBSD-src/tree/current-ext4fs
Why ? I think hard disk drives over 2Tb with OpenBSD are a hassle : you have to wait for hours for fsck to find almost no problem most of the time. I usually patch rc.d to run fsck -y as I don't know how to repair a b0rked filesystem and pray for the best. I care for a journalized filesystem for large partitions and for compatibility with other Linux distro's drives. (or I forgot to mention that OpenBSD is not a Linux distro, ^^;)
For those who don't know, OpenBSD is a security oriented original UNIX and can provide shell and network services to a selection of stock hardware.