Re: Weird calculation for kernel L2 segment table

[Ignatios Souvatzis <is%netbsd.org@localhost>]

On Sat, Jan 03, 2009 at 11:47:56AM +0900, Izumi Tsutsui wrote:

> The traditional hp300 derived pmap (designed for 68851 and HP MMU)
> uses two level table lookups for address translation.
> The first one is the segment table and the second is the page table.
> (there is a simple doc in arch/hp300/DOC/HPMMU.notes)
> 
> On 4kbytes/page systems (hp300 etc.), both have 1024 entries.
> On 8kbytes/page systems (i.e. on amiga and atari),
> the segment table has 256 entries and the page table has 2048
> (== NPTEPG == PAGE_SIZE / sizeof(pt_entry_t)) entries.
> (I don't know why these two traditional ports use different PAGE_SIZE)

I guess, because we can.

The 68030 MMU has only 22 TLB entries - although I never benchmarked
the effect, I felt that covering twice the VM with them would reduce
the pmap code traps. (I didn't *change* to 8k - it was that way when
I first looked tat the code).

(also from memory: 68851 has 64 TLB entries. Don't remember what 
68040 and 68060 have.)

HP300 has, I think, a motorola mmu like hardware built by HP, so it 
can only do 4k.

Somebody please correct me if I'm wrong.

        -is
-- 
seal your e-mail: http://www.gnupg.org/