On 6/6/24 21:34, Christoph Hellwig wrote: > On Thu, Jun 06, 2024 at 09:14:46PM +0900, Damien Le Moal wrote: >> "also" ? your previous point was not about this function ? > > No, about all the attribute boilerplate code. Yeah. That boilerplate for the 2 drivers differ only with the internal data structure used to store the cdl_enable boolean. So I guess we could make these attributes more generic. Ideally though, we should do something similar to CDL and have scsi layer deal with that automatically to avoid SAS drivers to have to do that themselves. And libata also has the same attributes. I would like to get this fix in ASAP as I am getting reports back from the field of NCQ priority not working with mpi3mr. I can send a cleanup series for the attributes on top of this fix later if that is OK with you. -- Damien Le Moal Western Digital Research
