On Mon, Aug 08, 2005 at 12:15:48AM +0200, Hubert Feyrer wrote:
> On Sun, 7 Aug 2005, Jason Thorpe wrote:
> >I see the word "jumper" down there.  I don't think software is setting it, 
> >just that some backup software was able to deal with a mis-configured ATA 
> >bus.
> >
> >Also, SATA doesn't have "master" and "slave".  SATA is a point-to-point 
> >serial interface.  There is only one drive per channel in SATA, and thus 
> >no notion of "master" and "slave".
> 
> I may have mixed that up and it may have been old-school ATA.
> The point was to detect (in software) if there's wrong in hardware.

I think it really depends on how drives reacts. In general, I think there
would be errors reported by the kernel, such as commands timeout or DMA
errors. But I can also see (especially with PIO commands) how both drives
could accept and process the command without errors. If it's a write you write
the same data to both drives, if it's a read you get a logical or of both
data.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--