On 18/08/2016 06:50, "Stokes, Ian" <ian.sto...@intel.com> wrote:
>> This patch introduces function 'netdev_dpdk_filter_packet_len()' which is
>> intended to find and remove all packets with 'pkt_len > max_packet_len'
>> from the Tx batch.
>>
>> It fixes inaccurate counting of 'tx_bytes' in vHost case if there was
>> dropped packets and allows to simplify send function.
>>
>> Fixes: 0072e931b207 ("netdev-dpdk: add support for jumbo frames")
>> Signed-off-by: Ilya Maximets <i.maxim...@samsung.com>
>> ---
>> lib/netdev-dpdk.c | 52 +++++++++++++++++++++++++++++---------------------
>> --
>> 1 file changed, 29 insertions(+), 23 deletions(-)
>>
>
>LGTM,
>
>Acked-by: Ian Stokes <ian.sto...@intel.com>
Thanks
I've rolled in the following incremental, which appears to improve the
throughput slightly on my system
index 59ae32e..d63e176 100644
--- a/lib/netdev-dpdk.c
+++ b/lib/netdev-dpdk.c
@@ -1449,9 +1449,13 @@ netdev_dpdk_filter_packet_len(struct netdev_dpdk *dev,
struct rte_mbuf **pkts,
VLOG_WARN_RL(&rl, "%s: Too big size %" PRIu32 " max_packet_len %d",
dev->up.name, pkt->pkt_len, dev->max_packet_len);
rte_pktmbuf_free(pkt);
- } else if (i != cnt++) {
- pkts[cnt - 1] = pkt;
+ continue;
+ }
+
+ if (OVS_UNLIKELY(i != cnt)) {
+ pkts[cnt] = pkt;
}
+ cnt++;
}
return cnt;
and applied this to master and branch-2.6
Thanks again,
Daniele
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
