In addition to the surprising impact of /proc/filesystems read optimizations for Linux 7.2, another one of the VFS pull requests for this next kernel version is delivering some nice improvements for EXT4 and XFS around IOmap, the framework that maps file data offsets in memory to their physical locations on storage.

Among the IOmap changes now merged for the Linux kernel is an optimization showing nice benefit for EXT4 and XFS when running with NVMe SSD storage and using IO_uring:

"Skip the memset of the iomap in iomap_iter() once the iteration is done. In high-IOPS scenarios (4k randread NVMe polling via io_uring) the pointless memset wasted memory write bandwidth; this improves IOPS by about 5% on ext4 and xfs."

A 5% uplift to IOPS isn't bad for such a straight forward optimization. That optimization comes from Bytedance engineer Fengnan Chang after discovering the wasted memory write bandwidth due to the caller just discarding the iterator when it's done. The 5% IOPS gain comes at moving around two lines of code.

Separately, that same pull also adds the VFS infrastructure needed for implementing FS-VERITY support for XFS with a post-EOF Merkle tree, as another notable change.

Linux 7.2 continues to prove quite exciting as do most Linux kernel versions N+1.