Linux 6.6 NFS Enables NFSv4.2 READ_PLUS Option By Default

In addition to NFSD bringing a thrilling feature for Linux 6.6 in the form of NFSv4 write delegation support, the NFS client code for the in-development Linux 6.6 kernel also has a notable feature change.

The NFS client code for Linux 6.6 is enabling the READ_PLUS mode by default. Merged back in 2020 for Linux 5.10 was the NFSv4.2 READ_PLUS option. As explained back then, READ_PLUS is a variant of READ that supports efficiently transferring holes. In cases where READ_PLUS is supported by both the NFS client and server, this operation should always be used rather than READ. READ_PLUS should allow for better read performance when dealing with sparse files.

Now with Linux 6.6, the NFSv4.2 READ_PLUS feature is being enabled by default to provide for better performance.

The NFSv4.2 READ_PLUS operation by default landed with the nfs-for-6.6-1 merge. That batch of NFS updates also provides a number of bug fixes and code clean-ups. The main feature though is just that READ_PLUS by default for this set of Network File-System updates.