Subject: Re: RISCServer 4200 Pseudo-patch
To: None <port-arc@netbsd.org>
From: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
List: port-arc
Date: 02/03/2004 21:33:51
In article <0F2AD2F1-55DD-11D8-A3AB-0003939563FA@perkins-tripp.com>
zach@perkins-tripp.com wrote:

> Here's the work I've done so far on getting the 4200 to boot. I still 
> get nothing on serial, and no video whatsoever, so I'm pretty sure I 
> have a hardware problem with the way I'm setting up the console and 
> maybe it's not the kernel, but who knows. This patch goes against 
> 1.6.1, not -current. I hope someone out there who can spot my mistake 
> can help out!

> *** ./arc/c_nec_pci.c	Wed Jun 13 10:23:22 2001

Maybe your video should be attached to EISA (not PCI),
so you can't use c_nec_pci_consinit() and maybe you
should have an R98 specific consinit fucntion.

But now I notice your R98 has a completely different
chipset from other NEC machines which are currently supported.
All jazzio bus devices on R94/RD94/J96/JC94 etc. has
physical address from 0x80000000 and mapped at virtual address
from 0xe0000000 (see sys/arch/arc/jazz/rd94.h).
But according to your arcdiag output, most devices on R98
are from PA 0x18x00000 and it seems PC-like chipset is used
for com/lpt/fdc etc:

>                  id: 6:[SONIC\x00]
>                    start: 0x 0 18600000, length: 0x1000

>                  id: 7:[I82077\x00]
>                     start: 0x 0 18c103f0, length: 0x1000

>                  id: 6:[I8742\x00]
>                    start: 0x 0 18c20060, length: 0x1000

>                  id: 6:[I8742\x00]
>                    start: 0x 0 18c20060, length: 0x1000

>                  id: 5:[COM1\x00] 
>                    start: 0x 0 18c103f8, length: 0x1000

>                  id: 5:[COM2\x00]
>                    start: 0x 0 18c102f8, length: 0x1000

>                  id: 5:[LPT1\x00]
>                    start: 0x 0 18c10278, length: 0x1000

>          id: 8:[NCRC710\x00]
>            start: 0x 0 18500000, length: 0x1000

>          id: 8:[NCRC710\x00]
>            start: 0x 0 18510000, length: 0x1000

I heard that R10000 variant of Express5800/230 had the
similar chipset, but I'm not sure how much work is needed
to support these new chipsets...

(Soda, how do you think about this?)
---
Izumi Tsutsui
tsutsuI@ceres.dti.ne.jp