On Mon, Feb 03, 2025 at 02:38:25PM +0100, Mikulas Patocka wrote: > > Not really. I mean it's always smart to not do tiny unaligned I/O > > unless you have to. So we're not just going to cap an exported value > > to a magic number because of something. > > The purpose of this patch is to avoid doing I/O not aligned on 4k > boundary. > > The 512-byte value that some SSDs report is just lie. That's very strong words. NVMe until fairly recently had no way to report this value at all, so defaulting to the LBA is not a lie. Similarly we ignore the device characteristics page is supposed to be skipped for usb attached SCSI device as it is buggy much more often than not. You still haven't mentioned what consumer of the value is affect that you care about, but it probably needs to be taught that opt_io == logical_block_size means that no opt_io size is provided at all. > Some USB-SATA bridges report optimal I/O size 33553920 bytes (that is > 512*65535). Well, that's clearly bogus and we'll need a tweak. That being said I was pretty sure we wouldn't even read that value for USB attachments.
