> In article <01030914514500.00460@pinky.paradox.demon.co.uk> you write:
> >On Thursday 08 March 2001 11:15 am, Richard Earnshaw wrote:
> >> Well the SA TRM definitely says that two banks aren't necessary iff the
> >> memory is unused for any other purpose.  (maybe this was a hack to work
> >> around not draining the write buffers properly :-)  I've been using this
> >> code for ~6 months in my own kernel and not seen any ill effects from it.
> >
> >just out of curiosity (and cos it's a mad idea :), would it be pheasible to 
> >use any constant block of 16k?  eg the first 16k of the kernel
> 
> Have you looked at the code?  It goes:
> 
> _C_LABEL(sa110_cache_clean_addr):
>         .word   0xf0000000
>         .global _C_LABEL(sa110_cache_clean_size)
> _C_LABEL(sa110_cache_clean_size):
>         .word   0x00008000
> 
> That looks like the first 32k (since this is the old version) of the kernel
> to me.  It might be a little better if it used a preprocessor constant
> there, though, since it won't behave quite as expected on (e.g.) hpcarm.
> 

That address is only used very early on.  Part of the bootstrap process 
allocates a new block of clean ram.  EG for the RPC:

rpc_sa110_cc_setup(void)
{
        int loop;
        paddr_t kaddr;
        pt_entry_t *pte;

        (void) pmap_extract(kernel_pmap, KERNEL_TEXT_BASE, &kaddr);
        for (loop = 0; loop < CPU_SA110_CACHE_CLEAN_SIZE; loop += NBPG) {
                pte = pmap_pte(kernel_pmap, (sa110_cc_base + loop));
                *pte = L2_PTE(kaddr, AP_KR);
        }
        sa110_cache_clean_addr = sa110_cc_base;
        sa110_cache_clean_size = CPU_SA110_CACHE_CLEAN_SIZE / 2;