>Here is an updated patch to the "Compaq PCI configuration problem".
>It's close to what I'd call a final solution. There are only minor
>unresolved questions:
>-Does it work at all?
>-Do we need this hack for "Triflex 2" (device ID 0x2000)? Nobody
>  with such a machine spoke up so far.
>-Are there more chipsets which need special treatment?
>-Can non-PCI machines be messed up by the probes?


1.3 will be around for a while. I think the patch would be even better
if it used a PCI_QUIRKS table instead of a hardcoded switch, and if
the PCI_QUIRKS table had at least one spare, patchable entry....