Izumi Tsutsui wrote:
> F.Lorenzen@gmx.de wrote:
[...]
>> There is one issue left i noticed during Testing:
>> If I enable only the second Card in Bios, which I don't do normally, re
>> says: 'can't map i/o space' and doesn't configure the Card.
> 
> I guess it's a BIOS problem (it doesn't initialize PCI I/O space
> properly but only does PCI MEM space in that case).
> Maybe you could remove "#define RE_USEIOSPACE" in if_re_pci.c
> to make it work, but could you please try the attached patch too?
> (to see if it's actually worth on such motherboard)

This morning i found the Time to test your Patch and found out the Issue
is gone away; but, this is the magical part, it's gone away without
doing anything!
After your last patch succeedet with -current i used it to build a
netbsd-4 installation-cdrom and installed the System freshly.
Now i took your "i/o space patch" and applied it to the syssrc.tgz
sources from the cd and built a new GENERIC.
Accidently i didn't reboot in the freshly build kernel but in the old
one and it recognised the (alone activated) second Card right.
I tried a bit around and even booted the original -current kernel from
the original Installation which _did_ produce the i/o space error in the
past, but it did not do it anymore.

Maybe there is some selfhealing integrated in the Card...
I am really wondering.


> Izumi Tsutsui


greetings
f