Re: /proc/scsi/scsi

Matto Marjanovic (maddog@mir.com)
Fri, 10 Apr 1998 11:57:16 -0400 (EDT)

Mmmm... another problem with just looking at /proc/scsi/scsi is that
not all scanners provide useful vendor/model strings.
I don't know about others, but the Microtek scanners provide widely
varying strings which often do not mention "MICROTEK", nor even the
model of the scanner. The ID code in the backend looks at a number
of different clues from the device inquiry to prove to itself that
a device is indeed a Microtek scanner.

Perhaps a another suggestion would be to just scan /proc/scsi/scsi for
device *types* --- I don't know of any scanners that don't properly claim
to be Device Type 0x06 (scanner). In 99% of the cases, there will be only
one such device on the bus, and that will be the user's scanner. The list
of such devices could be passed to any and all backends that want a hint.

Then, perhaps the *.conf files could be made so that a device of "/dev/sgN"
means look at that device explicitly, but "PROC_SCSI_HINTS" means call the
/proc/scsi/scsi code and get a list of devices to check from there.

Perhaps also, we could borrow the syntax used in Samba smb.conf files (as
well as Win<gag> config files), and merge all the .conf files into one,
with a [defaults] section, as well as sections for individual backends,
as needed:

[defaults]
device = PROC_SCSI_HINTS

[dll]
umax
pnm

[microtek]
disable-mode-sense-1

and so on....

-matt m.

--
Source code, list archive, and docs: http://www.mostang.com/sane/
To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com