Re: constraints on rbus_min_start?

[David Laight <david%l8s.co.uk@localhost>]

On Sat, Jan 05, 2008 at 07:18:55PM +0000, Steven M. Bellovin wrote:
> I'd like to port the code from arch/i386 that dynamically calculates the
> value for rbus_min_start from the memory size.  Are there any
> particular constraints I should be aware of for that value?
> 
> The i386 code in machdep.c is
> 
> #if NCARDBUS > 0
>         /* Tell RBUS how much RAM we have, so it can use heuristics.
>       rbus_min_start_hint(ptoa(physmem));
> #endif
> 
> and rbus_min_start_hint in rbus_machdep.c picks an appropriate value
> based on that number.  I'm assuming that any amd64 has a fair amount of
> memory and address lines, so simply rounding up to the next-highest 1G
> boundary will work.  But -- do I need to leave any gaps between real
> memory and rbus_min_start?  What boundary does it need to be at?

For amd64 (and i386 with PAE) the address allocated almost certainly
has to be inside the 'address hole' below 4GB.

        David

-- 
David Laight: david%l8s.co.uk@localhost