guys, mkfs should consume "I/O hints" from block device (see ioctl BLKIOOPT, BLKPBSZGET etc) to layout ondisk data structure and data areas relative to the underlying storage's 'optimal_io_size'.
So the easiest solution is to eliminate block and sector size params from mkfs and let mkfs figure out optimal parameters.
look at mkfs.xfs source code - https://git.kernel.org/cgit/fs/xfs/xfsprogs-dev.git/tree/mkfs/xfs_mkfs.c#n1783 looks like it already uses optimal sector size parameters which we override for some reason.
guys, mkfs should consume "I/O hints" from block device (see ioctl BLKIOOPT, BLKPBSZGET etc) to layout ondisk data structure and data areas relative to the underlying storage's 'optimal_io_size'. /git.kernel. org/cgit/ fs/xfs/ xfsprogs- dev.git/ tree/mkfs/ xfs_mkfs. c#n1783 looks like it already uses optimal sector size parameters which we override for some reason.
So the easiest solution is to eliminate block and sector size params from mkfs and let mkfs figure out optimal parameters.
look at mkfs.xfs source code - https:/