LINUX STORAGE
The block subsystem and IO_uring changes were merged this week for Linux 7.1 in continuing to enhance Linux storage capabilities.

With the merged block changes for Linux 7.1 is introducing shared memory zero-copy I/O support for ublk, the user-space block driver. This zero-copy I/O support bypasses per-I/O copies between kernel and user-space for a nice performance and efficiency win.

The block changes also include work to support file-system initiated integrity operations, adding SED-OPAL ioctls for single-user mode management and its STACK_RESET command, IO_uring pass-through support to the BSG layer, and cleaning up the DRBD driver.

For Linux 7.1 NVMe support there is enhanced block queue limits support, a new write zeroes devicde quirk, and updates to NVMe Fabrics authentication handling.

The Linux MD “Soft RAID” code for Linux 7.1 fixes a RAID5 soft lock-up issue, a possible RAID10 deadlobk fix, fixing RAID1 overlapping writes on write-mostly drives, and other fixes.

This merge outlines all the prominent block changes for Linux 7.1.

Separately is also the IO_uring updates for Linux 7.1 that now allow implementing custom event loop logic, expanded IOPOLL capabilities, timeout improvements, zero-copy receive (ZCRX) updates, and various optimizations.