In message <199910180655.PAA22309@hamachi.net.arch.sony.co.jp>,
Hayakawa Koichi writes:

>From: Jonathan Stone <jonathan@DSG.Stanford.EDU>

>Could you show me CardBus bridges your machine use?
>Sometimes Ricoh chips behave not very good.

Oops! Ricoh 5c478.  I will post complete dmesg when I get the machine
back in the same place as a working Ethernet card.  I added debugging
printfs, and the `reset legacy m ode code' for the Ricoh chips is
getting executed when the pccbb is attached. But by then, a pcic0 at
the legacy ISA addresses has already been attached.

Commenting out the 'pcic0 at isa' line seems to help: scanpci -v shows
sane data for the pccbb when the laptop goes multi-user.

>The code which sets intr to IRQ 10 should not be used.  I
>think we should use UCHIYAMA-san's patch unless BIOS sets
>intrline and bus number up correctly.

On this Vaio, Uchiyama-san's patch puts all devices not configured by
the BIOS at IRQ 9.  I ont know if that really works; it seems to mess
up the suspend/resume processing.