Re: A3000 problems

[RadosÅaw Kujawa <radoslaw.kujawa%c0ff33.net@localhost>]

On Jan 2, 2011, at 4:38 PM, Frank Wille wrote:
> Because it cannot be compiled under NetBSD. And it is just a temporary
> solution. When amigappc ever reaches a usable state we should use/enhance
> the port-amiga bootblock code.

Certainly, but without it we can't even boot now. Besides, it's useful to be 
able to boot from AmigaOS (think about installation procedure). 

Can loadbsd (which resides in sys/arch/amiga/stand...) be compiled under NetBSD 
now? I think not ;).

> Ok, so it would theoretically work. But I think that it is practically
> useless to run PPC code in memory which is so slow as Z3 RAM.

That's true. It might be useful for swap space (provided someone will write 
something similar to Linux's z2ram block device driver). 

>> amigappc_batinit(0x00000000, BAT_BL_16M, BAT_I|BAT_G,
>>                    (startkernel & 0xf8000000), BAT_BL_128M, 0,
>>                    0xfff00000, BAT_BL_512K, 0,
>>                    0x40000000, BAT_BL_256M, BAT_I|BAT_G, 0,
>>                    0x50000000, BAT_BL_256M, BAT_I|BAT_G, 0,
>>                    0x60000000, BAT_BL_256M, BAT_I|BAT_G, 0,
>>                    0x70000000, BAT_BL_256M, BAT_I|BAT_G,
>>                    ~0);
>> 
> This should have worked. I have no more spontaneous ideas. You have to debug
> it.
> 
> Did you try to swap position with the Deneb, so the CV64/3D appears at
> 0x40000000?

I removed it, and tried different positions of the cards. I tried examining 
addresses from 0x4000 0000 - 0x5000 0000 range in DDB and it works ok. But if I 
try to read from 0x5000 0000 - 0x8000 0000 I'll always get DSI read trap. If I 
move the same cards below 0x5000 0000 then examining their space works fine. I 
think that's the source of all problems I've been experiencing here. If I 
understand correctly, DSI read trap happen when kernel is trying to access 
unavailable address. 

On Jan 2, 2011, at 5:14 PM, Ignatios Souvatzis wrote:

> On Sun, Jan 02, 2011 at 04:44:36PM +0100, Frank Wille wrote:
> 
>> AFAIK the rom.manid for all is 8512, the rom.prodid is 100 for CSPPC, 110
>> for BPPC (?) and probably 25 for PowerUp (because it is mounted on a
>> CSMk2)?
> 
> The former two: yes, the latter one I don't know.

Ok, so this is on my todo-list, right after busclock stuff.

--
Best regards,
RadosÅaw Kujawa