Re: [Qemu-devel] [PATCH] ppc/pnv: Add model for Power8 PHB3 PCIe Host bridge

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

On Wed, 2018-06-27 at 03:35 +0300, Michael S. Tsirkin wrote:
> 
> > +
> > +/* Extract field fname from val */
> > +#define GETFIELD(fname, val)                    \
> > +        (((val) & fname##_MASK) >> fname##_LSH)
> > +
> > +/* Set field fname of oval to fval
> > + * NOTE: oval isn't modified, the combined result is returned
> > + */
> > +#define SETFIELD(fname, oval, fval)                     \
> > +        (((oval) & ~fname##_MASK) | \
> > +         ((((typeof(oval))(fval)) << fname##_LSH) & fname##_MASK))
> > +
> 
> Pls don't make up macros like these. We can't have each device do it.

So what ? We move the macros in a generic place ? These are MUCH better
than open-coding the masks & shifts and much less error prone.

> > @@ -1031,6 +1110,7 @@ static Property pnv_chip_properties[] = {
> >      DEFINE_PROP_UINT64("ram-size", PnvChip, ram_size, 0),
> >      DEFINE_PROP_UINT32("nr-cores", PnvChip, nr_cores, 1),
> >      DEFINE_PROP_UINT64("cores-mask", PnvChip, cores_mask, 0x0),
> > +    DEFINE_PROP_UINT32("num-phbs", PnvChip, num_phbs, 1),
> >      DEFINE_PROP_END_OF_LIST(),
> >  };
> 
> How about instanciating each extra phb using -device?
> Seems better than teaching everyone about platform-specific
> options.

It's about which PHBs are enabled, not which are instanciated, if I
understand Cedric changes ...

This aims are implementing the POWER8 chip correctly, it has a fixed
number of PHBs per-chip at very specific XSCOM addresses, that the
firwmare knows about.

Cheers,
Ben.