If you have been staying up-to-date on the happenings of Linux, then you most likely know about NTFSPlus, the fresh NTFS driver implementation for Linux, built by kernel developer Namjae Jeon. He is known for his work on the exFAT driver and ksmbd SMB server.
The driver has been steadily improving, with each patch iteration bringing new features and performance gains.
Now there's a new development worth talking about.
Same Function, New Name
Yesterday, Namjae submitted the v3 patch series for NTFSPlus to the Linux kernel mailing list, renaming NTFSPlus to just "NTFS" in the kernel. This has been done in a bid to restructure everything, using the old read-only NTFS driver as a base and applying changes on top of it.
According to Phoronix, this makes code review easier, and maintainers can focus on what's new instead of treating everything as fresh code.
Further explaining how this new NTFS implementation would work on Linux, Namjae mentions that the remade NTFS supports write operations plus essential features like iomap, no buffer-head usage, utilities, and xfstests.
Beyond the rename, this patchset brings many important improvements. New generic helpers handle cluster-to-folio conversions and byte-to-sector operations. The driver now allows readahead for the $MFT file, and the 2TB filesystem limitation on 32-bit systems has been removed.
Performance-wise, this NTFS adaptation is 3-5% faster than NTFS3 in single-threaded writes, but multi-threaded scenarios show 35-110% improvements. File listing is 12-14% quicker, and mount times are dramatically better: under a second for 4TB partitions versus 4+ seconds with NTFS3.
Do keep in mind that there are some features missing like full journaling support. Only journal replay exists right now, and even that doesn't work correctly according to Namjae. He plans to add proper journaling after the driver gets upstreamed.
The driver is currently marked as experimental, which makes sense given its development stage.
