Tested-by: Marvin Liu <yong.liu at intel.com> > -----Original Message----- > From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Reshma Pattan > Sent: Thursday, July 14, 2016 11:49 PM > To: dev at dpdk.org > Cc: Pattan, Reshma > Subject: [dpdk-dev] [PATCH] examples/distributor: fix Rx thread logic for > zero packets > > From: Reshma Pattan <reshma.pattan at intel.com> > > Zero packets can be returned by rte_eth_rx_burst() and > rte_distributor_returned_pkts() inside lcore_rx(), so > for zero packet scenario instead of proceeding to > next operations we should continue to the next iteration of the > loop to avoid unnecessary processing overhead which is causing > rx packets to be dropped and hence distributor failing to forward the > packets. > > Fixes: 07db4a97 ("examples/distributor: new sample app") > > Signed-off-by: Reshma Pattan <reshma.pattan at intel.com> > --- > examples/distributor/main.c | 10 +++++++++- > 1 file changed, 9 insertions(+), 1 deletion(-) > > diff --git a/examples/distributor/main.c b/examples/distributor/main.c > index 24857f2..537cee1 100644 > --- a/examples/distributor/main.c > +++ b/examples/distributor/main.c > @@ -221,14 +221,22 @@ lcore_rx(struct lcore_params *p) > struct rte_mbuf *bufs[BURST_SIZE*2]; > const uint16_t nb_rx = rte_eth_rx_burst(port, 0, bufs, > BURST_SIZE); > + if (unlikely(nb_rx == 0)) { > + if (++port == nb_ports) > + port = 0; > + continue; > + } > app_stats.rx.rx_pkts += nb_rx; > > rte_distributor_process(d, bufs, nb_rx); > const uint16_t nb_ret = rte_distributor_returned_pkts(d, > bufs, BURST_SIZE*2); > app_stats.rx.returned_pkts += nb_ret; > - if (unlikely(nb_ret == 0)) > + if (unlikely(nb_ret == 0)) { > + if (++port == nb_ports) > + port = 0; > continue; > + } > > uint16_t sent = rte_ring_enqueue_burst(r, (void *)bufs, > nb_ret); > app_stats.rx.enqueued_pkts += sent; > -- > 2.5.0
