Re: better late than never: amigappc progress

[Jukka Andberg <jandberg%welho.com@localhost>]

On Sat, 13 Feb 2010 14:50:07 +0100 Frank Wille wrote:
> That's bad. Seems there is no memory mapped to 0xfff00000 (or it is
> protected). A CSPPC will hardware-map the last 512k of its Fast-RAM
> there. The PowerPC needs this region for exception vectors, so it should
> be the same for BPPC...?

Curious. Half a meg fast memory did go somewhere in my case, though. See below.

> In this example you inserted two 32MB SIMMs?

I had two SIMMs, 32M and 4M. I inserted each separately in different memory 
slots on BPPC. So there were four different test cases in total.

Case 1: 32MB SIMM alone in first slot:
  Memory Address: $76000020  Size: $01F7FFE0 (32255K)     Type=$00 CACHEABLE

Case 2: 32MB SIMM alone in second slot:
  Memory Address: $78000020  Size: $01F7FFE0 (32255K)     Type=$00 CACHEABLE

Case 3: 4MB SIMM alone in first slot:
  Memory Address: $57C00020  Size: $0037FFE0 (3583K)      Type=$00 CACHEABLE

Case 4: 4MB SIMM alone in second slot:
  Memory Address: $68000020  Size: $0037FFE0 (3583K)      Type=$00 CACHEABLE

As can be seen from these figures, half a meg is missing from the end in each 
case.

> Ignatios Souvatzis suggested to automatically relocate the kernel when
> loaded. I'm thinking about it. The rest would be easy and the kernel
> could autodetect everything when running.

Can't see why not.

--jukka