Re: [Qemu-ppc] [PATCH 12/12] pseries: Generate unique LIOBNs for PCI host bridges

[Michael S. Tsirkin]

On Wed, Nov 21, 2012 at 11:09:50AM +0100, Alexander Graf wrote:
> 
> 
> On 21.11.2012, at 06:00, David Gibson <address@hidden> wrote:
> 
> > On Tue, Nov 20, 2012 at 10:27:11AM +0100, Alexander Graf wrote:
> >> 
> >> On 19.11.2012, at 23:51, David Gibson wrote:
> >> 
> >>> On Mon, Nov 19, 2012 at 05:34:12PM +0100, Alexander Graf wrote:
> >>>> 
> >>>> On 13.11.2012, at 03:47, David Gibson wrote:
> >>>> 
> >>>>> From: Alexey Kardashevskiy <address@hidden>
> >>>>> 
> >>>>> In future (with VFIO) we will have multiple PCI host bridges on
> >>>>> pseries.  Each one needs a unique LIOBN (IOMMU id).  At the moment we
> >>>>> derive these from the pci domain number, but the whole notion of
> >>>>> domain numbers on the qemu side is bogus and in any case they're not
> >>>>> actually uniquely allocated at this point.
> >>>>> 
> >>>>> This patch, therefore uses a simple sequence counter to generate
> >>>>> unique LIOBNs for PCI host bridges.
> >>>>> 
> >>>>> Signed-off-by: Alexey Kardashevskiy <address@hidden>
> >>>>> Signed-off-by: David Gibson <address@hidden>
> >>>> 
> >>>> I don't really like the idea of having a global variable just
> >>>> because our domain ID generation seems to not work as
> >>>> expected. Michael, any comments here?
> >>> 
> >>> Well, the patch I sent which changed domain id generation was
> >>> ignored.  In any case, as I said, the whole concept of domain numbers
> >> 
> >> Michael?
> >> 
> >>> makes no sense on the qemu side, so I don't think increasing reliance
> >>> on them by using them here is a good idea.
> >>> 
> >>> It would be conceptually nicer to derive the liobn from the buid, but
> >>> that would rely on the buid's being unique in the low 32-bits, which
> >>> is true in practice, but seems risky to rely on.
> >> 
> >> Well, there has to be some uniqueness from the guest's POV already,
> >> no?
> > 
> > Yes, the BUIDs are unique, but they are 64-bit, whereas the LIOBN is
> > only 32-bit.
> 
> Tricky. Michael, any ideas?
> 
> Alex

Whatever scheme we come up with, becomes part of guest ABI
that we have to maintain.
So rather than maintain it, I think it's easiest to require full
specification from users, and verify uniqueness.

> > 
> > -- 
> > 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