The patch uses netif_rx_ni() over netif_rx() to post buffers to
upper network code because it is always scheduled in a workqueue.
The problem was first observed from a dynamic ticks warning:

    "NOHZ: local_softirq_pending ..."
Signed-off-by: NZhu Yi <yi.zhu@intel.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>