Re: [Qemu-devel] [Qemu-ppc] [PATCH] pseries: Correct vmx/dfp handling in both KVM and TCG cases

[David Gibson]

On Mon, Oct 24, 2011 at 10:25:26AM -0700, Alexander Graf wrote:
> On 23.10.2011, at 22:29, David Gibson wrote:
> > On Thu, Oct 20, 2011 at 11:49:40PM -0700, Alexander Graf wrote:
[snip]
> >>> This gets further complicated in the case of the w-i-p patch I have to
> >>> properly advertise page sizes, where it's not just presence or absence
> >>> of a feature, but the specific SLB and HPTE encodings must be
> >>> advertised to the guest.
> >> 
> >> Yup, so we'd read out the host dt to find the host possible
> >> encodings (probably a bad idea, but that's a different story)
> > 
> > Um, a different story perhaps, but one I kind of need an answer to in
> > the near future...  I can query the host cpu's page sizes easily
> > enough, but I'm really not sure where this should be stashed before
> > filtering as suggested below.
> 
> Page sizes are usually powers of 2, so we should be ok with just
> having a bitmap there with each bit meaning 1 << (n + 12).

Not sufficient.  Again, it's not just the presence/absence of page
sizes I need, but the SLB and HPTE bit encodings.  And even if it
weren't for that, we need which base page size versus actual page
sizes combinations are supported, not just whether a given page size
is supported somehow.

I did have a draft patch adding more generalized multiple page size
support to TCG, which would have provided a solution except that a) I
don't really want to finish tha off - still a fair bit of work - just
in order to pass through host page sizes and b) I lost the draft in an
unfortunate encfs+git corruption incident :(.

-- 
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