In message <199910180555.WAA05267@Greyhawk.DSG.Stanford.EDU>,
Jonathan Stone writes:

[snip]


>In message <199910180508.BAA02119@ginger.cmf.nrl.navy.mil>Ken Hornstein writes
>>Does setting "Plug and Play OS" to "No" in the BIOS help this at all?
>
>Not with the Cardbus vs. legacy ISA PCIC addresses, AFAICT.
>It does help get the the usb controller properly set up.
>
>Adding printfs to pccbb.c shows tat the `disable' of the legacy ISA
>addresses doesn't happen until the pccb is atytached -- which is ater
>the old PCMCIA code has found and attached a pcic at those same legacy
>addreses. I think that's a big part of the problem.
>
>I'll double-check tomorrow at work.



I've been using CardBus very happily with the PCIC entries removed
from my config file. I havent ever gotten the new Ricoh bridge to work
in CardBus mode if I leave the pcic entries in my config file -- not
with any combination of "PnPOS = no" and/or the PCIbios patches.

The PCIbios patches work fine. I think we should import them as soon
as the author thinks they're stable enough, to help out owners of
laptops where setting "PnP OS = no" isnt enough.

But what about the bridge-compatibility issues? the pcic at isa gets
probed before the CardBus at PCI. So if PCI cardbus hardware disables
PCI access after it sees legacy-ISA activity, we're sunk.

How about a `quirk table' or a check for balky bridges like this Ricoh
to the PCIBios, to disable the `legacy pcic compatiblity' on
known-screwy bridges? (in an #ifdef on the pci-to-cardbus bridge
attachment). 

Or should the 'pcic at isa' probe upcall into a PCI `scan-and-disable
PCI-cardbus legacy mode' hook?