Re: MIATA variation breaking tulip found?
On Thu, May 28, 1998 at 03:48:39PM +0200, Kristoffer.Rose@ens-lyon.fr wrote:
> Mark:
>
> Bus 1, device 9, function 0:
> SCSI storage controller: Q Logic ISP1020 (rev 5).
> Medium devsel. IRQ 40. Master Capable. Latency=32.
> I/O at 0x9000.
> Non-prefetchable 32 bit memory at 0xb100000.
>
> Me:
>
> Bus 1, device 8, function 0:
> SCSI storage controller: Q Logic ISP1020 (rev 5).
> Medium devsel. IRQ 36. Master Capable. Latency=32.
> I/O at 0x9000.
> Non-prefetchable 32 bit memory at 0xb100000.
The PCI device number is determined based on exactly how a PCI device is
wired into the Host-to-PCI bridge. If the device number were incorrect, the
PCI configuration requests would fail and no device would appear at the
given device ID.
The device number is approximately the same as the "slot" number you plug a
device into on a "plug-in" PCI bus, except that even on-board PCI devices
get a device number.
Anyway, the point of all that is: the PCI device number must be correct on
both of your boxes. It looks like your Miata != Mark's Miata, and there may
be other physical hardware differences as well.
The IRQs mismatch also, but this may not be a problem. For "plug-in" PCI
buses each slot is normally assigned an IRQ or set of IRQs, so (sometimes) a
board is designed to have its onboard devices put together the same way.
If you want, you could try forcing the device driver to use Mark's IRQ
number -- the fact that the device ID is different doesn't necessarily mean
that the IRQ is wired to match.
In any case, I don't know why your SCSI controller would have an effect on
your ethernet card :) However, the fact that there's a difference with the
SCSI implies that there might be other, more subtle, differences elsewhere.
Another note -- if you look at an x86 box (ie. mine :)) the bus number is 0,
not 1. Are there some devices on a Miata with bus 0, and some with bus 1?
Normally the first PCI bus in a system is number 0. The existence of both 0
and 1 implies that there is a PCI-to-PCI bridge in the system. Making
things work with PCI-to-PCI bridges can be just wonderful fun.
Good luck!
Avery
--
To UNSUBSCRIBE, email to debian-alpha-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: