On 23.08.21 13:23, Mathias Homann wrote:
Am Montag, 23. August 2021, 13:12:50 CEST schrieb david allan finch:
On 08/23/21 11:11 AM, Andrei Borzenkov wrote:
Before this thread wanders too far in fairy land - there is no "default disk type". "sd" stays for "SCSI disk". Each block device driver enumerates supported devices and assigns devices names that are unique to this driver and all other different types of block devices.
Intresting I kind of remember it meaning 'standard disk' (block) and they used to be /dev/sd1 .. and partitions where sd1a ... and raw devices where rsd1a etc.
That's on netbsd - not on linux.
on linux we used to have /dev/hd[a..z] für old IDE/ATA drives, and /dev/ sd[a..z][a..z] for SCSI. Later, the SATA driver adopted the SCSI naming scheme.
Nitpicking: the SATA driver ceased to exist and was replaced by a SCSI-to-(S)ATA translation layer which, as the name implies, takes a SCSI CDB (Command Descriptor Block) and translates it into an (S)ATA command.
Today, we have /dev/nvme#n#p# for NVME storage, where the name signifies the controller, the next bit signifies the storage device, and the last bit the partition on it:
/dev/nvme0n1p1 would read as "first NVME controller, first storage device, first partition".
And seeing how there can be multiple controllers, each with multiple devices, I'd say it is unlikely that we'll ever see a "/dev/sd[a..za..z]' scheme for nvme storage.
Hm, we already can have multiple SCSI host adapters and/or (S)ATA controllers each with multiple devices. Please correct me if I'm wrong, but I see no obvious reason for the device name to describe the path. This can be (and is) done by udev and results in /dev/disk/by-path. So, IMHO, at one point we may actually see NVME devices being handled by the SCSI subsystem and then have /dev/disk/by-path/nvme#n#p# -> /dev/sdi Josef -- SUSE Software Solutions Germany GmbH Maxfeldstr. 5 90409 Nürnberg Germany (HRB 36809, AG Nürnberg) Geschäftsführer: Felix Imendörffer