On Tue, Jul 29, 2025 at 12:13:42PM -0400, Tony Battersby wrote: > Improve writeback performance to RAID-4/5/6 by aligning writes to stripe > boundaries. This relies on io_opt being set to the stripe size (or > a multiple) when BLK_FEAT_RAID_PARTIAL_STRIPES_EXPENSIVE is set. You're not aligning anything. You are splitting I/O, which is exactly what we've been trying to avoid by moving to the immutable bio_vec model that moves the splitting to the place that needs it. > Benchmark of sequential writing to a large file on XFS using > io_uring with 8-disk md-raid6: > Before: 601.0 MB/s > After: 614.5 MB/s > Improvement: +2.3% Looks like you need to do some work on the bio splitting in RAID. It would help to Cc the maintainers as the driver is actually pretty actively worked on these days.
