[sane-devel] sane-find-scanner finds multiple scanners

[Carsten Neumann]

On Tue, 07 May 2002, Douglas Gilbert wrote:
> Carsten Neumann wrote:
> > On Mon, 06 May 2002, Douglas Gilbert wrote:
> > 
> >>Johan,
> >>In your kernel config (assuming you have built that kernel)
> >>turn off CONFIG_SCSI_MULTI_LUN . Some SCSI scanners don't
> >>react properly to being probed for logical unit numbers (luns).
> >>They should only respond to lun==0 but they respond to all
> >>eight. That should leave you with:
> >>   /dev/scanner
> >>   /dev/sg1
> >>   /dev/sgb
> >>
> >>Doug Gilbert
> >>
> > 
> > Usually /dev/sga .. /dev/sgd are symlinks to /dev/sg0 .. /dev/sg3, respectively.
> 
> True, this is for backward compatibility; the /dev/sga,b,c
> sequence is deprecated.
> 
> > Also most harddisks will respond at every LUN.
> 
> I hope not. Could be a problem in lk 2.5 where we support
> 64 bit luns :-)

Which SCSI standard supports 64 bit LUNs??? 8-/

> 
> > Once again, I can't see any reason why probing all LUNs will prevent the scanner from working!
> 
> It is when applications (such as SANE) send other SCSI commands
> to the luns > 0 that don't exist there could be problems.

If a device ignores (doesn't decode) the LUN, why should it behave different on different LUNs?
But I would think that the 1st device found - the one with LUN == 0 - is used.
BTW: what does `other SCSI commands' mean?

Anyway, if that yields a kernel freeze it should be subject of further investigation. :-)

> 
> <snip/>
> 
> Doug Gilbert

	Carsten