On Thu, Jun 15, 2000 at 01:28:22PM +0200, Bernd Sieker wrote:
> 
> Hi,
> 
> I noticed that since recently the "siop" scsi driver is no longer
> marked "experimental", but instead the "ncr" driver is now marked "old
> driver".
> 
> What exactly would be the (dis)advantages of both?
> 
> What I noticed is that both drivers incorrectly detect my TekRam
> DC390U (Ultra-SCSI, narrow) as "ultra-wide".
> 
> At the moment all my scsi devices are narrow, but these days narrow
> disks are hard to get, almost all of them are wide.
> 
> So if I connect a wide drive to my narrow adapter via a narrow cable,
> will the driver and the disk negotiate wide transfers by default?
> 
> This would be a Bad Thing, since abviously I cannot do wide transfers
> via a narrow cable, or will the driver check if wide transfers are in
> fact possible and fall back to narrow transfers, if that fails?
> 
> Here's what dmesg reports:
> 
>   siop0 at pci0 dev 11 function 0: Symbios Logic 53c875 (ultra-wide scsi)
>   siop0: using on-board RAM
>   siop0: interrupting at irq 10

Hum, ok so this board is based on a wide chip but only the narrow connector
is used. Sure, with a wide disk the driver will negotiate wide transfers
and try to use them; there's no fall-back mechanism (hum depending on how
things are setup I'm not sure transfers will fail, they may just transfer
incorrect data with highter bits all 0 or all 1). I'm not sure how to handle
this; this is a quite unusual setup.

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
--