[dpdk-dev] [PATCH v4 16/29] node: add ethdev Rx node

[Nithin Dabilpuram]

On Fri, Apr 10, 2020 at 01:05:08AM +0200, Andrzej Ostruszka wrote:
> On 4/5/20 10:56 AM, jerinj at marvell.com wrote:
> > From: Nithin Dabilpuram <ndabilpuram at marvell.com>
> > 
> > Add source rte_node ethdev_rx process function and register
> > it. This node is a source node that will be called periodically
> > and when called, performs rte_eth_rx_burst() on a specific
> > (port, queue) pair and enqueue them as stream of objects to
> > next node.
> > 
> > Signed-off-by: Nithin Dabilpuram <ndabilpuram at marvell.com>
> > Signed-off-by: Pavan Nikhilesh <pbhagavatula at marvell.com>
> > Signed-off-by: Kiran Kumar K <kirankumark at marvell.com>
> [...]
> > +/* Callback for soft ptype parsing */
> > +static uint16_t
> > +eth_pkt_parse_cb(uint16_t port, uint16_t queue, struct rte_mbuf **mbufs,
> > +		 uint16_t nb_pkts, uint16_t max_pkts, void *user_param)
> > +{
> > +	struct rte_mbuf *mbuf0, *mbuf1, *mbuf2, *mbuf3;
> > +	struct rte_ether_hdr *eth_hdr;
> > +	uint16_t etype, n_left;
> > +	struct rte_mbuf **pkts;
> > +
> > +	RTE_SET_USED(port);
> > +	RTE_SET_USED(queue);
> > +	RTE_SET_USED(max_pkts);
> > +	RTE_SET_USED(user_param);
> > +
> > +	pkts = mbufs;
> > +	n_left = nb_pkts;
> > +	while (n_left >= 12) {
> > +
> > +		/* Prefetch next-next mbufs */
> > +		rte_prefetch0(pkts[8]);
> > +		rte_prefetch0(pkts[9]);
> > +		rte_prefetch0(pkts[10]);
> > +		rte_prefetch0(pkts[11]);
> > +
> > +		/* Prefetch next mbuf data */
> > +		rte_prefetch0(
> > +			rte_pktmbuf_mtod(pkts[4], struct rte_ether_hdr *));
> > +		rte_prefetch0(
> > +			rte_pktmbuf_mtod(pkts[5], struct rte_ether_hdr *));
> > +		rte_prefetch0(
> > +			rte_pktmbuf_mtod(pkts[6], struct rte_ether_hdr *));
> > +		rte_prefetch0(
> > +			rte_pktmbuf_mtod(pkts[7], struct rte_ether_hdr *));
> 
> I know this is software fallback only (and not likely to be used) but is
> this aggressive prefetching always beneficial?  I guess you tested this
> on octeon and it works, but if this is supposed to be standard RX node
> then maybe this is not always good?
> 
> On the other hand if other platforms find that detrimental they can
> submit some improvements later :)

I tested it now in octeon and there is 6% increasing when using prefetch based while loop
instead of non-prefetch based while loop.

Yes, it is not intended to be used normally, but I just followed ideology of prefetch ahead 
before use. It could be changed later if needed for other platforms or split into platform 
dependent implementation like lookup node.

> 
> > +
> > +		mbuf0 = pkts[0];
> > +		mbuf1 = pkts[1];
> > +		mbuf2 = pkts[2];
> > +		mbuf3 = pkts[3];
> > +		pkts += 4;
> > +		n_left -= 4;
> > +
> > +		/* Extract ptype of mbuf0 */
> > +		eth_hdr = rte_pktmbuf_mtod(mbuf0, struct rte_ether_hdr *);
> > +		etype = eth_hdr->ether_type;
> > +		mbuf0->packet_type = l3_ptype(etype, 0);
> > +
> > +		/* Extract ptype of mbuf1 */
> > +		eth_hdr = rte_pktmbuf_mtod(mbuf1, struct rte_ether_hdr *);
> > +		etype = eth_hdr->ether_type;
> > +		mbuf1->packet_type = l3_ptype(etype, 0);
> > +
> > +		/* Extract ptype of mbuf2 */
> > +		eth_hdr = rte_pktmbuf_mtod(mbuf2, struct rte_ether_hdr *);
> > +		etype = eth_hdr->ether_type;
> > +		mbuf2->packet_type = l3_ptype(etype, 0);
> > +
> > +		/* Extract ptype of mbuf3 */
> > +		eth_hdr = rte_pktmbuf_mtod(mbuf3, struct rte_ether_hdr *);
> > +		etype = eth_hdr->ether_type;
> > +		mbuf3->packet_type = l3_ptype(etype, 0);
> > +	}
> > +
> > +	while (n_left > 0) {
> > +		mbuf0 = pkts[0];
> > +
> > +		pkts += 1;
> > +		n_left -= 1;
> > +
> > +		/* Extract ptype of mbuf0 */
> > +		eth_hdr = rte_pktmbuf_mtod(mbuf0, struct rte_ether_hdr *);
> > +		etype = eth_hdr->ether_type;
> > +		mbuf0->packet_type = l3_ptype(etype, 0);
> > +	}
> > +
> > +	return nb_pkts;
> > +}
> [...]
> 
> With regards
> Andrzej Ostruszka