On Thu, Jan 28, 2016 at 03:33:18PM +1100, Benjamin Herrenschmidt wrote:
> On Wed, 2016-01-27 at 21:13 +1100, David Gibson wrote:
> > When HPTEs are removed or modified by hypercalls on spapr, we need to
> > invalidate the relevant pages in the qemu TLB.
> > 
> > Currently we do that by doing some complicated calculations to work out the
> > right encoding for the tlbie instruction, then passing that to
> > ppc_tlb_invalidate_one()... which totally ignores the argument and flushes
> > the whole tlb.
> > 
> > Avoid that by adding a new flush-by-hpte helper in mmu-hash64.c.
> 
> Should we find a better "in between" so long run we implement tlbie
> properly ? IE, tlbie will give us the page size using the same encoding
> as the HPTE iirc when L=1 ? To be honest the encoding of tlbie in arch
> 2.07 is so completely insane I have a hard time figuring it out myself
> ... :-)

I'm not entirely sure what the better in-between would be.  Having the
pagesize in tlbie isn't enough on its own - the bigger problem is that
we need a way of invalidating a whole congruence class of entries in
the qemu TLB, which it doesn't currently provide a means to do.

> Otherwise,
> 
> Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson