On Tue, Apr 22, 2003 at 02:39:26AM +0900, Izumi Tsutsui wrote:
> I wrote:
> 
> > In article <20030421162512.GA10989@bseis.eis.cs.tu-bs.de>
> > hannken@eis.cs.tu-bs.de wrote:
> > 
> > > But I still see no real need to attach the pcn to an artificial pci bus.
> > > There is no real PCI bridge/host controller. Not inside the CPU
> > > and not on the board.
> > > It is a PCI capable device attached to the plain processor bus.
> > 
> > At least, pcn supports full features of 79c971, including MII.
> > le (am79900) uses compatible mode with old LANCE.
> 
> I guess dreamcast also has similar (faked) PCI bridge for RTL8139
> and it has PCI attachment (gapspci_pci.c etc.) only for if_rtk_pci.c.
> 
> Furthermore, if you rewrite fb_elb.c (S3 driver) for generic
> PCI attachment, it could be shared with other ports.

I'm currently updating this one. I will replace rasops by S3-accelerator.
I suppose it should be split into an dev/ic/s3fb.c and bus attachment glue
like fb_elb.c.
Even with a fake pci attachment X11 will not work as the pci config
registers are not mapped into memory/io space.
-- 
Juergen Hannken-Illjes - hannken@eis.cs.tu-bs.de - TU Braunschweig (Germany)