[PATCH v5 12/12] net/nfp: add flower PF rxtx logic

[Chaoyong He <chaoyong.he@corigine.com>]

This commit implements the flower Rx logic. Fallback packets are
multiplexed to the correct representor port based on the prepended
metadata. The Rx poll is set to run on the existing service
infrastructure.

For Tx the existing NFP Tx logic is duplicated to keep the Tx two paths
distinct. Flower fallback also adds 8 bytes of metadata to the start of
the packet that has to be adjusted for in the Tx descriptor.

Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@corigine.com>
---
 drivers/net/nfp/flower/nfp_flower.c | 428 ++++++++++++++++++++++++++++++++++++
 drivers/net/nfp/flower/nfp_flower.h |   1 +
 2 files changed, 429 insertions(+)

diff --git a/drivers/net/nfp/flower/nfp_flower.c b/drivers/net/nfp/flower/nfp_flower.c
index d7772d6..bc0acff 100644
--- a/drivers/net/nfp/flower/nfp_flower.c
+++ b/drivers/net/nfp/flower/nfp_flower.c
@@ -23,6 +23,7 @@
 #include "nfp_flower_ovs_compat.h"
 #include "nfp_flower_ctrl.h"
 #include "nfp_flower_representor.h"
+#include "nfp_flower_cmsg.h"
 
 #define MAX_PKT_BURST 32
 #define MEMPOOL_CACHE_SIZE 512
@@ -218,6 +219,44 @@
 	.link_update            = nfp_flower_pf_link_update,
 };
 
+static void
+nfp_flower_pf_vnic_poll(struct nfp_app_flower *app_flower)
+{
+	uint16_t i;
+	uint16_t count;
+	uint16_t n_rxq;
+	uint16_t port_id;
+	struct nfp_net_hw *pf_hw;
+	struct rte_mbuf *pkts_burst[MAX_PKT_BURST];
+
+	pf_hw = app_flower->pf_hw;
+
+	n_rxq = pf_hw->eth_dev->data->nb_rx_queues;
+	port_id = pf_hw->eth_dev->data->port_id;
+
+	/* Add ability to run Rx queues on multiple service cores? */
+	for (i = 0; i < n_rxq; i++) {
+		count = rte_eth_rx_burst(port_id, i, pkts_burst, MAX_PKT_BURST);
+
+		/*
+		 * Don't expect packets here but free them in case they have
+		 * not been multiplexed to a representor
+		 */
+		if (unlikely(count > 0))
+			rte_pktmbuf_free_bulk(pkts_burst, count);
+	}
+}
+
+static int
+nfp_flower_pf_vnic_service(void *arg)
+{
+	struct nfp_app_flower *app_flower = arg;
+
+	nfp_flower_pf_vnic_poll(app_flower);
+
+	return 0;
+}
+
 static int
 nfp_flower_ctrl_vnic_service(void *arg)
 {
@@ -229,6 +268,10 @@
 }
 
 static struct rte_service_spec flower_services[NFP_FLOWER_SERVICE_MAX] = {
+	[NFP_FLOWER_SERVICE_PF] = {
+		.name         = "flower_pf_vnic_service",
+		.callback     = nfp_flower_pf_vnic_service,
+	},
 	[NFP_FLOWER_SERVICE_CTRL] = {
 		.name         = "flower_ctrl_vnic_service",
 		.callback     = nfp_flower_ctrl_vnic_service,
@@ -265,6 +308,387 @@
 	return ret;
 }
 
+static inline void
+nfp_flower_parse_metadata(struct nfp_net_rxq *rxq,
+		struct nfp_net_rx_desc *rxd,
+		struct rte_mbuf *mbuf,
+		uint32_t *portid)
+{
+	uint32_t meta_info;
+	uint8_t *meta_offset;
+	struct nfp_net_hw *hw;
+
+	hw = rxq->hw;
+	if (!((hw->ctrl & NFP_NET_CFG_CTRL_RSS) ||
+			(hw->ctrl & NFP_NET_CFG_CTRL_RSS2)))
+		return;
+
+	meta_offset = rte_pktmbuf_mtod(mbuf, uint8_t *);
+	meta_offset -= NFP_DESC_META_LEN(rxd);
+	meta_info = rte_be_to_cpu_32(*(uint32_t *)meta_offset);
+	meta_offset += 4;
+
+	while (meta_info) {
+		switch (meta_info & NFP_NET_META_FIELD_MASK) {
+		/* Expect flower firmware to only send packets with META_PORTID */
+		case NFP_NET_META_PORTID:
+			*portid = rte_be_to_cpu_32(*(uint32_t *)meta_offset);
+			meta_offset += 4;
+			meta_info >>= NFP_NET_META_FIELD_SIZE;
+			break;
+		default:
+			/* Unsupported metadata can be a performance issue */
+			return;
+		}
+	}
+}
+
+static inline struct nfp_flower_representor *
+nfp_flower_get_repr(struct nfp_net_hw *hw,
+		uint32_t port_id)
+{
+	uint8_t port;
+	struct nfp_app_flower *app_flower;
+
+	/* Obtain handle to app_flower here */
+	app_flower = NFP_APP_PRIV_TO_APP_FLOWER(hw->pf_dev->app_priv);
+
+	switch (NFP_FLOWER_CMSG_PORT_TYPE(port_id)) {
+	case NFP_FLOWER_CMSG_PORT_TYPE_PHYS_PORT:
+		port =  NFP_FLOWER_CMSG_PORT_PHYS_PORT_NUM(port_id);
+		return app_flower->phy_reprs[port];
+	case NFP_FLOWER_CMSG_PORT_TYPE_PCIE_PORT:
+		port = NFP_FLOWER_CMSG_PORT_VNIC(port_id);
+		return app_flower->vf_reprs[port];
+	default:
+		break;
+	}
+
+	return NULL;
+}
+
+static uint16_t
+nfp_flower_pf_recv_pkts(void *rx_queue,
+		struct rte_mbuf **rx_pkts,
+		uint16_t nb_pkts)
+{
+	/*
+	 * we need different counters for packets given to the caller
+	 * and packets sent to representors
+	 */
+	int avail = 0;
+	int avail_multiplexed = 0;
+	uint64_t dma_addr;
+	uint32_t meta_portid;
+	uint16_t nb_hold = 0;
+	struct rte_mbuf *mb;
+	struct nfp_net_hw *hw;
+	struct rte_mbuf *new_mb;
+	struct nfp_net_rxq *rxq;
+	struct nfp_net_rx_buff *rxb;
+	struct nfp_net_rx_desc *rxds;
+	struct nfp_flower_representor *repr;
+
+	rxq = rx_queue;
+	if (unlikely(rxq == NULL)) {
+		/*
+		 * DPDK just checks the queue is lower than max queues
+		 * enabled. But the queue needs to be configured
+		 */
+		RTE_LOG_DP(ERR, PMD, "RX Bad queue\n");
+		return 0;
+	}
+
+	hw = rxq->hw;
+
+	/*
+	 * This is tunable as we could allow to receive more packets than
+	 * requested if most are multiplexed.
+	 */
+	while (avail + avail_multiplexed < nb_pkts) {
+		rxb = &rxq->rxbufs[rxq->rd_p];
+		if (unlikely(rxb == NULL)) {
+			RTE_LOG_DP(ERR, PMD, "rxb does not exist!\n");
+			break;
+		}
+
+		rxds = &rxq->rxds[rxq->rd_p];
+		if ((rxds->rxd.meta_len_dd & PCIE_DESC_RX_DD) == 0)
+			break;
+
+		/*
+		 * Memory barrier to ensure that we won't do other
+		 * reads before the DD bit.
+		 */
+		rte_rmb();
+
+		/*
+		 * We got a packet. Let's alloc a new mbuf for refilling the
+		 * free descriptor ring as soon as possible
+		 */
+		new_mb = rte_pktmbuf_alloc(rxq->mem_pool);
+		if (unlikely(new_mb == NULL)) {
+			RTE_LOG_DP(DEBUG, PMD,
+			"RX mbuf alloc failed port_id=%u queue_id=%d\n",
+				rxq->port_id, rxq->qidx);
+			nfp_net_mbuf_alloc_failed(rxq);
+			break;
+		}
+
+		nb_hold++;
+
+		/*
+		 * Grab the mbuf and refill the descriptor with the
+		 * previously allocated mbuf
+		 */
+		mb = rxb->mbuf;
+		rxb->mbuf = new_mb;
+
+		PMD_RX_LOG(DEBUG, "Packet len: %u, mbuf_size: %u",
+			   rxds->rxd.data_len, rxq->mbuf_size);
+
+		/* Size of this segment */
+		mb->data_len = rxds->rxd.data_len - NFP_DESC_META_LEN(rxds);
+		/* Size of the whole packet. We just support 1 segment */
+		mb->pkt_len = rxds->rxd.data_len - NFP_DESC_META_LEN(rxds);
+
+		if (unlikely((mb->data_len + hw->rx_offset) > rxq->mbuf_size)) {
+			/*
+			 * This should not happen and the user has the
+			 * responsibility of avoiding it. But we have
+			 * to give some info about the error
+			 */
+			RTE_LOG_DP(ERR, PMD,
+				"mbuf overflow likely due to the RX offset.\n"
+				"\t\tYour mbuf size should have extra space for"
+				" RX offset=%u bytes.\n"
+				"\t\tCurrently you just have %u bytes available"
+				" but the received packet is %u bytes long",
+				hw->rx_offset,
+				rxq->mbuf_size - hw->rx_offset,
+				mb->data_len);
+			return 0;
+		}
+
+		/* Filling the received mbuf with packet info */
+		if (hw->rx_offset)
+			mb->data_off = RTE_PKTMBUF_HEADROOM + hw->rx_offset;
+		else
+			mb->data_off = RTE_PKTMBUF_HEADROOM +
+					NFP_DESC_META_LEN(rxds);
+
+		/* No scatter mode supported */
+		mb->nb_segs = 1;
+		mb->next = NULL;
+
+		mb->port = rxq->port_id;
+		meta_portid = 0;
+
+		/* Checking the RSS flag */
+		nfp_flower_parse_metadata(rxq, rxds, mb, &meta_portid);
+		PMD_RX_LOG(DEBUG, "Received from port %u type %u",
+				NFP_FLOWER_CMSG_PORT_VNIC(meta_portid),
+				NFP_FLOWER_CMSG_PORT_VNIC_TYPE(meta_portid));
+
+		/* Checking the checksum flag */
+		nfp_net_rx_cksum(rxq, rxds, mb);
+
+		if ((rxds->rxd.flags & PCIE_DESC_RX_VLAN) &&
+				(hw->ctrl & NFP_NET_CFG_CTRL_RXVLAN)) {
+			mb->vlan_tci = rte_cpu_to_le_32(rxds->rxd.vlan);
+			mb->ol_flags |= RTE_MBUF_F_RX_VLAN |
+					RTE_MBUF_F_RX_VLAN_STRIPPED;
+		}
+
+		repr = nfp_flower_get_repr(hw, meta_portid);
+		if (repr && repr->ring) {
+			PMD_RX_LOG(DEBUG, "Using representor %s", repr->name);
+			rte_ring_enqueue(repr->ring, (void *)mb);
+			avail_multiplexed++;
+		} else if (repr) {
+			PMD_RX_LOG(ERR, "[%u] No ring available for repr_port %s\n",
+					hw->idx, repr->name);
+			PMD_RX_LOG(DEBUG, "Adding the mbuf to the mbuf array passed by the app");
+			rx_pkts[avail++] = mb;
+		} else {
+			PMD_RX_LOG(DEBUG, "Adding the mbuf to the mbuf array passed by the app");
+			rx_pkts[avail++] = mb;
+		}
+
+		/* Now resetting and updating the descriptor */
+		rxds->vals[0] = 0;
+		rxds->vals[1] = 0;
+		dma_addr = rte_cpu_to_le_64(RTE_MBUF_DMA_ADDR_DEFAULT(new_mb));
+		rxds->fld.dd = 0;
+		rxds->fld.dma_addr_hi = (dma_addr >> 32) & 0xff;
+		rxds->fld.dma_addr_lo = dma_addr & 0xffffffff;
+
+		rxq->rd_p++;
+		if (unlikely(rxq->rd_p == rxq->rx_count)) /* wrapping?*/
+			rxq->rd_p = 0;
+	}
+
+	if (nb_hold == 0)
+		return nb_hold;
+
+	PMD_RX_LOG(DEBUG, "RX port_id=%u queue_id=%d, %d packets received",
+		   rxq->port_id, rxq->qidx, nb_hold);
+
+	nb_hold += rxq->nb_rx_hold;
+
+	/*
+	 * FL descriptors needs to be written before incrementing the
+	 * FL queue WR pointer
+	 */
+	rte_wmb();
+	if (nb_hold > rxq->rx_free_thresh) {
+		PMD_RX_LOG(DEBUG, "port=%u queue=%d nb_hold=%u avail=%d",
+			   rxq->port_id, rxq->qidx, nb_hold, avail);
+		nfp_qcp_ptr_add(rxq->qcp_fl, NFP_QCP_WRITE_PTR, nb_hold);
+		nb_hold = 0;
+	}
+
+	rxq->nb_rx_hold = nb_hold;
+
+	return avail;
+}
+
+static uint16_t
+nfp_flower_pf_xmit_pkts(void *tx_queue,
+		struct rte_mbuf **tx_pkts,
+		uint16_t nb_pkts)
+{
+	int i;
+	int pkt_size;
+	int dma_size;
+	uint64_t dma_addr;
+	uint16_t free_descs;
+	uint16_t issued_descs;
+	struct rte_mbuf *pkt;
+	struct nfp_net_hw *hw;
+	struct rte_mbuf **lmbuf;
+	struct nfp_net_txq *txq;
+	struct nfp_net_nfd3_tx_desc txd;
+	struct nfp_net_nfd3_tx_desc *txds;
+
+	txq = tx_queue;
+	hw = txq->hw;
+	txds = &txq->txds[txq->wr_p];
+
+	PMD_TX_LOG(DEBUG, "working for queue %d at pos %u and %u packets",
+			txq->qidx, txq->wr_p, nb_pkts);
+
+	if ((nfp_net_nfd3_free_tx_desc(txq) < nb_pkts) || (nfp_net_nfd3_txq_full(txq)))
+		nfp_net_tx_free_bufs(txq);
+
+	free_descs = (uint16_t)nfp_net_nfd3_free_tx_desc(txq);
+	if (unlikely(free_descs == 0))
+		return 0;
+
+	pkt = *tx_pkts;
+
+	i = 0;
+	issued_descs = 0;
+
+	/* Sending packets */
+	while ((i < nb_pkts) && free_descs) {
+		/* Grabbing the mbuf linked to the current descriptor */
+		lmbuf = &txq->txbufs[txq->wr_p].mbuf;
+		/* Warming the cache for releasing the mbuf later on */
+		RTE_MBUF_PREFETCH_TO_FREE(*lmbuf);
+
+		pkt = *(tx_pkts + i);
+
+		if (unlikely(pkt->nb_segs > 1 &&
+				!(hw->cap & NFP_NET_CFG_CTRL_GATHER))) {
+			PMD_INIT_LOG(INFO, "NFP_NET_CFG_CTRL_GATHER not set");
+			PMD_INIT_LOG(INFO, "Multisegment packet unsupported");
+			goto xmit_end;
+		}
+
+		/* Checking if we have enough descriptors */
+		if (unlikely(pkt->nb_segs > free_descs))
+			goto xmit_end;
+
+		/*
+		 * Checksum and VLAN flags just in the first descriptor for a
+		 * multisegment packet, but TSO info needs to be in all of them.
+		 */
+		txd.data_len = pkt->pkt_len;
+		nfp_net_nfd3_tx_tso(txq, &txd, pkt);
+		nfp_net_nfd3_tx_cksum(txq, &txd, pkt);
+
+		if ((pkt->ol_flags & RTE_MBUF_F_TX_VLAN) &&
+				(hw->cap & NFP_NET_CFG_CTRL_TXVLAN)) {
+			txd.flags |= PCIE_DESC_TX_VLAN;
+			txd.vlan = pkt->vlan_tci;
+		}
+
+		/*
+		 * mbuf data_len is the data in one segment and pkt_len data
+		 * in the whole packet. When the packet is just one segment,
+		 * then data_len = pkt_len
+		 */
+		pkt_size = pkt->pkt_len;
+
+		while (pkt) {
+			/* Copying TSO, VLAN and cksum info */
+			*txds = txd;
+
+			/* Releasing mbuf used by this descriptor previously*/
+			if (*lmbuf)
+				rte_pktmbuf_free_seg(*lmbuf);
+
+			/*
+			 * Linking mbuf with descriptor for being released
+			 * next time descriptor is used
+			 */
+			*lmbuf = pkt;
+
+			dma_size = pkt->data_len;
+			dma_addr = rte_mbuf_data_iova(pkt);
+
+			/* Filling descriptors fields */
+			txds->dma_len = dma_size;
+			txds->data_len = txd.data_len;
+			txds->dma_addr_hi = (dma_addr >> 32) & 0xff;
+			txds->dma_addr_lo = (dma_addr & 0xffffffff);
+			ASSERT(free_descs > 0);
+			free_descs--;
+
+			txq->wr_p++;
+			if (unlikely(txq->wr_p == txq->tx_count)) /* wrapping?*/
+				txq->wr_p = 0;
+
+			pkt_size -= dma_size;
+
+			/*
+			 * Making the EOP, packets with just one segment
+			 * the priority
+			 */
+			if (likely(!pkt_size))
+				txds->offset_eop = PCIE_DESC_TX_EOP |
+						FLOWER_PKT_DATA_OFFSET;
+			else
+				txds->offset_eop = 0;
+
+			pkt = pkt->next;
+			/* Referencing next free TX descriptor */
+			txds = &txq->txds[txq->wr_p];
+			lmbuf = &txq->txbufs[txq->wr_p].mbuf;
+			issued_descs++;
+		}
+		i++;
+	}
+
+xmit_end:
+	/* Increment write pointers. Force memory write before we let HW know */
+	rte_wmb();
+	nfp_qcp_ptr_add(txq->qcp_q, NFP_QCP_WRITE_PTR, issued_descs);
+
+	return i;
+}
+
 static void
 nfp_flower_pf_mp_init(__rte_unused struct rte_mempool *mp,
 		__rte_unused void *opaque_arg,
@@ -498,6 +922,8 @@
 
 	/* Add Rx/Tx functions */
 	eth_dev->dev_ops = &nfp_flower_pf_dev_ops;
+	eth_dev->rx_pkt_burst = nfp_flower_pf_recv_pkts;
+	eth_dev->tx_pkt_burst = nfp_flower_pf_xmit_pkts;
 
 	/* PF vNIC gets a random MAC */
 	eth_dev->data->mac_addrs = rte_zmalloc("mac_addr",
@@ -1117,6 +1543,8 @@
 
 	eth_dev->process_private = cpp;
 	eth_dev->dev_ops = &nfp_flower_pf_dev_ops;
+	eth_dev->rx_pkt_burst = nfp_flower_pf_recv_pkts;
+	eth_dev->tx_pkt_burst = nfp_flower_pf_xmit_pkts;
 	rte_eth_dev_probing_finish(eth_dev);
 
 	return 0;
diff --git a/drivers/net/nfp/flower/nfp_flower.h b/drivers/net/nfp/flower/nfp_flower.h
index 24fced3..4ac89d1 100644
--- a/drivers/net/nfp/flower/nfp_flower.h
+++ b/drivers/net/nfp/flower/nfp_flower.h
@@ -7,6 +7,7 @@
 #define _NFP_FLOWER_H_
 
 enum nfp_flower_service {
+	NFP_FLOWER_SERVICE_PF,
 	NFP_FLOWER_SERVICE_CTRL,
 	NFP_FLOWER_SERVICE_MAX
 };
-- 
1.8.3.1