On Mon, May 19, 2003 at 10:20:49PM +0200, Manuel Bouyer wrote:

=> > Does anyone know of an
=> > easy fix to get this working?  Will simply moving it to another slot
=> 
=> This is worth a try, it may work. Note that your siop0 is on pci1, where
=> all your other devices are on pci0. Try moving it to a slot on pci0, it's
=> likely a problem with pci1 itself

Well, I finally got some time to power down and open up the
AlphaServer 1000A.  The non-working SCSI card was, indeed, plugged
into a slot on PCI bus 2 (as it is designated on the motherboard).
Because PCI bus 1 was full, I ended up having to swap the SCSI card
with one of the two network cards.

So, now I have two working siop devices (hooray!:), but now only one
working network device (boo:), and so lose the ability to NAT I was
providing for an ailing DEC 3000/500S.  Anyway, I can live with that.

One odd side-effect of switching the cards is that tlp0 and tlp1
switched around.  I guess anything populating PCI bus 2 gets probed
before anything on PCI bus 1.  However, it seems that anything plugged
into PCI bus 2 won't work (at least in this 1000A), because of the
"unable to map interrupt" problem. :-(

Does anyone have an AlphaServer 1000A in which cards plugged into PCI
bus 2 work under NetBSD?  The devices in the one here show up and test
okay under SRM, BTW.

Cheers,

Paul.

e-mail: paul@gromit.dlib.vt.edu

"Without music to decorate it, time is just a bunch of boring production
 deadlines or dates by which bills must be paid."
        --- Frank Vincent Zappa