On Sat, Aug 06, 2016 at 08:43:21AM +1000, Benjamin Herrenschmidt wrote: > On Fri, 2016-08-05 at 18:48 +0200, Cédric Le Goater wrote: > > The core initialization should be ok but building the device > > tree might be a bit of a burden if we have to 'cast' in the chip > > type we need. We will see. > > We don't cast, we use a method.
Right. You can have a look at the sPAPR VIO stuff for an example
where we use a method in the individual devices to do the device
specific fdt construction.
>
> > So what would be the big differences with what we have today ?
>
> The XSCOM controller has a different address decoding scheme, so
> we'll have two variants and a base class there. The number and
> location of functional units changes, so we'll probably need to make
> them properties or something, the core XSCOM addressing is completely
> different, we use PHB4 rather than PHB3, etc...
>
> (Note that for PHB we should probably rename PHB3 to PnvPhb with a
> subclass as well as there is a lot in common between the two, though
> the PBCQ bit is quite different).
>
--
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
signature.asc
Description: PGP signature
