In article <20030929014021.GA12398@rezrov.net>
wileyc@rezrov.net wrote:

> Specific to the sgimips port, the odds of someone dropping another ahc into
> the sole PCI slot are IMHO pretty low.

But it still could happen and it _should_ work. "Low odds" would
never be an excuse (otherwise my news68k port never happened ;-p)
I've also tried various PCI devices on macppc, arc, sparc64 etc.
and fixed bugs in those drivers. Our goal's page
http://www.netbsd.org/Goals/system.html
also says "it doesn't work unless it's right."

> ahc0            at pci0 dev 1 function ? driveropts 0x80000004
 :
> Would that be acceptable?

I don't think so. It is not so differnt from #ifdef AHC_FOO.
Is there no way to detect whether the probed device is
(broken) O2 on-board ahc or normal one at run time?
---
Izumi Tsutsui
tsutsui@ceres.dti.ne.jp