-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

On Apr 18, 2007, at 15:09, Nathan J. Williams wrote:

> Michael Lorenz <macallan@NetBSD.org> writes:
>
>> Still bogus but at least we know it's not the puc driver mucking 
>> things
>> up. I'm beginning to wonder if OF correctly handles devices which
>> hardwire the upper 16 bit in a used BAR to 0.
>
> Quite possibly not. DEC Alpha firmware didn't handle this abomination
> either (a loophole in the PCI spec that was unfortunately validated,
> instead of closed).
>
> Note also:
>
>      I/O space accesses: off
>      Memory space accesses: off
>
> ... so this device hasn't even been enabled.

Yeah, I noticed that. We can just enable and configure the device by 
hand but that's a funky hack at best. All the interrupt info we need is 
in the bridge's interrupt-mask property so what we'd need to do is 
enable it and program the BARs to something sa{n|f}e - not trivial 
since macppc relies on the firmware for that but probably not too hard 
either since not many devices use IO addresses there anyway.

have fun
Michael
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (Darwin)

iQEVAwUBRiaWaMpnzkX8Yg2nAQJ4qAgAg9ZUyAcSL/VqEepIazwoxwA+0eeuE3fD
uoFdS19K2PyzZAaATrdzV1M71P16uVCdL3qNvsda7MoBPUkQDAKEswGMhJfZ11Hk
xisjMirLGDjMvZqCK/pzFUuw8U2UZXxpmp9Dg+/nIprHOKKBpx+oZWEA70ZbdeiD
LgKd4OBPCFBtHyJbPxBvvMRTPW5+QZamSP0dFwt0FcFdBbOI3OcCtiH2Fe1gJL8X
cw3F0SG7kEoEmWvXimseYvVu3kiQ0O6mju9SjpbK3BJeNIP2RMorEOYuJ4wcL8yv
zUFMAMLxBvq7eygOGnemZCm1v3uIARLLgjOp/yIPjh3B1cLyzUjfWg==
=hqxT
-----END PGP SIGNATURE-----