Subject: Re: kernel panic on 1000...
To: Ross Harvey <firstname.lastname@example.org>
From: mel kravitz <email@example.com>
Date: 04/24/2001 23:32:18
Ross Harvey wrote:
> Err, `hose1', on a 1000a? Now I'm confused. As a cia box it couldn't have
> more than one real hose, right? Perhaps a ppb and more than one bus, sure.
I am new to this machine there are two 'hose' pci-esia and pci-pci bus
structures as reported by the SRM console. Cards plug into the pci group marked
pci2 on the mother board, definetly are not initialized. My machine has 2
ESIA/3pci card slots grouped together, any card plugged into these slots works
as long as the kernel is configured to accept them(cyclades CY0 ISA card, vga
pci card etc..). the remaining 4 pci card slots marked pci2 and grouped
together, will not be initialized by the kernel on boot. If i move the vga card
to a pci2 slot, the machine will not boot and i get a kernel fault.
> I've made a set of install tools that will print the slot code, which will
> probably tell us how to fix it.
> It's true that the install can be made to work by either unplugging the
> keyboard and plugging in a serial console, or by moving the vga, but before
> or after doing that it would be really neat if the test kernel could be
> booted with the card in its original location.
Install of 1.5 with GENERIC kernel worked for me, currently this machine is
running xfree86-4.0.2 using a Matrox Mystique vga video card, running wmaker X
window at http://www.caseydoodle.com/netbsd.png
dmesg is shown there. Getting the remaining 4 pci slots to work would be cool!
> Please mail me the output it prints just before crashing. If you do get it
> going some other way, at least send me or `firstname.lastname@example.org'
> the dmesg or /kern/msgbuf. Thanks.
> // ross, who really likes visiting .it IRL, and who especially likes the
> Toscana wines
> Test installation binaries are in: