On 9/27/22 18:47, John Garry wrote:
> On 27/09/2022 10:28, Damien Le Moal wrote:
>>> Sure, we can use sas_to_ata_dev() to get the ata_device.
>>>
>>> I am just suggesting my way such that we can have a consistent method to
>>> get the ata_device between all libata users and we don't need to change
>>> the ata_change_queue_depth() interface. It would be something like:
>>>
>>> struct ata_device *ata_scsi_find_dev(struct ata_port *ap, const struct
>>> scsi_device *scsidev)
>>> {
>>> struct ata_link *link;
>>> struct ata_device *dev;
>>>
>>> ata_for_each_link(link, ap, EDGE) {
>>> ata_for_each_dev(dev, link, ENABLED) {
>>> if (scsidev == dev->sdev)
>>> return dev;
>>> }
>>> }
>>> // todo: check pmp
>>> return NULL;
>>> }
>> I see. Need to think about this one... This may also unify the pmp case.
>> Are you OK with the patch as is though ?
>
> I'm ok with your patchset, but let me test it and get back to you later
> today.
>
> We can improve with something
>> like the above on top later. Really need to fix that qd setting as it is
>> causing problems for testing devices with/without ncq commands.
>
> Out of curiosity, are you considering your patchset for 6.0?
Yes. But I can send it for 6.1 with cc: stable too.
>
>>
>
> Thanks,
> John
--
Damien Le Moal
Western Digital Research
