[Cbe-oss-dev] Spider DMA wrongness

[Benjamin Herrenschmidt]

On Tue, 2006-11-07 at 16:53 +1100, Anton Blanchard wrote:
>  Hi,
> 
> >  - First of all, not our fault, but alloc_etherdev doesn't (yet) have an
> > alloc_etherdev_node() version thus we end up with a data structure on
> > the wrong node, pretty bloody annoying. I'm not even talking about skb's
> > on the wrong node here but purely the netdevice and priv datas
> 
> Since the stuff that calls the pci ->probe routines does a:
> 
> 	if (node >= 0 && node_online(node))
> 		set_cpus_allowed(current, node_to_cpumask(node));
> 
> Wont we get memory on the right node?

Ah yes, I forgot about this PCI gross hack from hell... it will probably
get us memory on the right node indeed... for now. (And provided we get
the right properties in the device-tree for the associativity of that
PCI bus but I think we do).

Still, the code sucks... and the other proble remains of mixing up the
HW fields with the driver fields in the descriptor chain.

Cheers,
Ben.