[9fans] forgot to mention ts7200 is working

[9fans] forgot to mention ts7200 is working

[Ronald G. Minnich]

this is the ep93xx core. 

it all seems to be working with cache etc. Due to the fact that phsyical 
memory is 4 segemnts at present only 16 MB of 32MB is usable. 

I'll drop the code on sources in a bit, when I'm sure it's reasonably 
reliable. Sorry folks, but the eth driver is derived from linux code, 
hence GPL. ah! wail! gnash teeth! 

you call can rewrite it if you wish :-)

I also include a modified xalloc.c that lets you allocate "flavors" of 
memory, e.g. uncached and cached regions. This may be handy at some point, 
esp. when we get to 64-bit machines where at minimum you needed "low 4G" 
and "above4G" flavors of memory.

At least I think it will, who knows.


ron

Re: [9fans] forgot to mention ts7200 is working

[Chris Collins]

Ronald G. Minnich was once rumoured to have said:
> this is the ep93xx core. 
> 
> it all seems to be working with cache etc. Due to the fact that phsyical 
> memory is 4 segemnts at present only 16 MB of 32MB is usable. 

This is the same problem that my current sparc32 code is plauged with
-- each bank of physical ram on a SS10 appears on a 64MB boundary, so
if a bank has less than 64MBs of RAM...

> I also include a modified xalloc.c that lets you allocate "flavors" of 
> memory, e.g. uncached and cached regions. This may be handy at some point, 
> esp. when we get to 64-bit machines where at minimum you needed "low 4G" 
> and "above4G" flavors of memory.

Actually, xalloc needs fixing to handle an arbitary number of physical
ranges (rather than just the 2 it currently supports).  I've been
saying this on #plan9 for the past... umm...  3-4 weeks?

I'll take a look at your code, try to remember how exactly I intended
on patching xalloc before I forgot (I put it aside for one week too
many) and see if I can fix it so we can use all the ram in both the
SS10 and in the ts7200 as well...

C.