From: Rafał Miłecki <zajec5@gmail.com>
Date: Wed, 30 Oct 2013 08:00:00 +0100

Copying whole packets with skb_copy_from_linear_data_offset is a pretty
bad idea. CPU was spending time in __copy_user_common and network
performance was lower. With the new solution iperf-measured speed
increased from 116Mb/s to 134Mb/s.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
---
Changes since [RFC TRY#2]:
1) Fixed arguments alignment
2) Dropped code fixing old slot in case of bgmac_dma_rx_skb_for_slot
failure. Thanks to Nathan patch bgmac_dma_rx_skb_for_slot doesn't
change anything in slot in case it failed somewhere.

Looks good, applied to net-next, thanks.