On Wed, May 14, 2014 at 03:23:58PM +1000, Benjamin Herrenschmidt wrote: > On Wed, 2014-05-14 at 14:18 +1000, David Gibson wrote: > > I supposed it's arguable that a PCI ranges property really should have > > windows for both prefetchable and non-prefetchable areas, but since > > that's not done in practice, it's pretty much moot. > > They don't have to, only if there's a relevant difference, ie it's > perfectly legit for a firmware to assign prefetchable BARs in a > non-prefetchable region.
In what sense is the window region non-prefetchable? Isn't that a
property only of the final destination of the bus cycle?
> Now granted, those are somewhat special cases so maybe we should
> do just that ... special case PCI and try to not continue that mess
> with new bindings.
Yeah, I think that's the way to go.
--
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
pgpHs20oD_pI4.pgp
Description: PGP signature
