Re: SCSI disk devices: even crazier naming scheme

David Hinds (dhinds@hyper.stanford.edu)
Wed, 7 May 1997 09:57:03 -0700

Andras Kadinger <bandit@freeside.elte.hu> wrote:
>
> To exercise this to an extreme level, Eric (or some brave soul) could
> enhance scsidev to provide rather longish and practically untypable
> names providing all (by some definition of all) available
> identification data of a scsi entity, and then everybody could make up
> simple scripts (which we have to do anyway in order to customize our
> systems) to use the magic of filename expansion to enumerate, retrieve
> and sort all the devices to their heart's content.

...

> I read somewhere that one big advantage of unices is the extensive
> recycling of ideas and concepts.

Don't you know it!

For the PCMCIA driver collection, I had the need for something like
scsidev, but didn't know that it existed at that time. So I wrote
something somewhat similar, but also along the lines of what you have
in mind. My "scsi_info" command produces information like:

# scsi_info /dev/sda1
SCSI_ID="0,4,0"
MODEL="Syquest EZ135S"
FW_REV="1_12"

I also have a similar tool for IDE drives:

# ide_info /dev/hda
MODEL="TOSHIBA MK1824FCV"
FW_REV="P0.09 B"
SERIAL_NO="Z4J60510"

Then, the scripts for configuring SCSI devices do things like:

eval `scsi_info /dev/$DEVICE`
ADDRESS="$SCHEME,$TYPE,$SOCKET,$SCSI_ID"

case $ADDRESS in
*,sr,*,*,*,*)
# stuff for CD-ROM's
*,*,*,0,4,0)
# stuff for my Syquest drive
;;
esac

-- Dave Hinds
dhinds@hyper.stanford.edu