[PATCH v2] gianfar: Add I/O barriers when touching buffer descriptor ownership.

[Scott Wood]

Kumar Gala wrote:
> I'd rather see a wmb() instead of eieio() to keep this code non-ppc  
> specific.  (also, we implement wmb as eieio, so I don't keep the  
> comment about it being too heavy, unless you mean generically).

wmb() is a sync, smp_wmb() is an eieio.  Andy told me he would not 
accept a sync in those spots.

And the driver is already ppc-specific; it uses in/out_be32.

-Scott