Memory allocation modifications in ibm_newemac driver

[Jonathan Haws]

Okay, I think I have all the issues worked out and can now send and receive any size packet without a hiccup.  I have tested this in our system setup as well with data being sent out to disk and did not see any problems there either (since it only ever allocates a single page, never more).

Is this something that may be wanted in the mainline?  I have not run full benchmarks, but I anticipate that my modified driver is slightly slower than the mainline driver because we keep track of an SKB ring, as well as a ring of pages and allocate both on each packet received.

Thanks,

Jonathan