On Sep 03, 2025 / 14:49, John Garry wrote: > When trying to find the value for max atomic write size, a user of > run_xfs_io_xstat calls something like the following: > > statx_atomic_min=$(run_xfs_io_xstat /dev/"$md_dev" "stat.atomic_write_unit_max") > > And run_xfs_io_xstat will grep for "stat.atomic_write_unit_max" in the raw > statx output. However, since new field "stat.atomic_write_unit_max_opt" was > added for statx, the grep returns 2x values (which is wrong). > > Fix by changing the grep to cover the full field name. > > This fixes md/002, nvme/059, and scsi/009. > > Signed-off-by: John Garry <john.g.garry@xxxxxxxxxx> I applied it. Thanks! FYI, Yi Zhang found the failure and reported with a GitHub issue [1]. [1] https://github.com/linux-blktests/blktests/issues/201
