[dpdk-dev] [PATCH v1 50/58] net/octeontx2: add Rx multi segment version

[<jerinj@marvell.com>]

From: Nithin Dabilpuram <ndabilpuram@marvell.com>

Add multi segment version of packet Receive function.

Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
---
 doc/guides/nics/features/octeontx2.ini     |  2 +
 doc/guides/nics/features/octeontx2_vec.ini |  1 +
 doc/guides/nics/features/octeontx2_vf.ini  |  2 +
 drivers/net/octeontx2/otx2_rx.c            | 25 ++++++++++
 drivers/net/octeontx2/otx2_rx.h            | 55 +++++++++++++++++++++-
 5 files changed, 84 insertions(+), 1 deletion(-)

diff --git a/doc/guides/nics/features/octeontx2.ini b/doc/guides/nics/features/octeontx2.ini
index 6117e1edf..18bcf81cf 100644
--- a/doc/guides/nics/features/octeontx2.ini
+++ b/doc/guides/nics/features/octeontx2.ini
@@ -24,6 +24,8 @@ Inner RSS            = Y
 VLAN filter          = Y
 Flow control         = Y
 Flow API             = Y
+Jumbo frame          = Y
+Scattered Rx         = Y
 VLAN offload         = Y
 QinQ offload         = Y
 Packet type parsing  = Y
diff --git a/doc/guides/nics/features/octeontx2_vec.ini b/doc/guides/nics/features/octeontx2_vec.ini
index 66c327cfc..97a24671e 100644
--- a/doc/guides/nics/features/octeontx2_vec.ini
+++ b/doc/guides/nics/features/octeontx2_vec.ini
@@ -24,6 +24,7 @@ Inner RSS            = Y
 VLAN filter          = Y
 Flow control         = Y
 Flow API             = Y
+Jumbo frame          = Y
 VLAN offload         = Y
 QinQ offload         = Y
 Packet type parsing  = Y
diff --git a/doc/guides/nics/features/octeontx2_vf.ini b/doc/guides/nics/features/octeontx2_vf.ini
index 3aa0491e1..916a6d7b0 100644
--- a/doc/guides/nics/features/octeontx2_vf.ini
+++ b/doc/guides/nics/features/octeontx2_vf.ini
@@ -19,6 +19,8 @@ RSS reta update      = Y
 Inner RSS            = Y
 VLAN filter          = Y
 Flow API             = Y
+Jumbo frame          = Y
+Scattered Rx         = Y
 VLAN offload         = Y
 QinQ offload         = Y
 Packet type parsing  = Y
diff --git a/drivers/net/octeontx2/otx2_rx.c b/drivers/net/octeontx2/otx2_rx.c
index b4a3e9d55..0f0919338 100644
--- a/drivers/net/octeontx2/otx2_rx.c
+++ b/drivers/net/octeontx2/otx2_rx.c
@@ -91,6 +91,14 @@ otx2_nix_recv_pkts_ ## name(void *rx_queue,				       \
 {									       \
 	return nix_recv_pkts(rx_queue, rx_pkts, pkts, (flags));		       \
 }									       \
+									       \
+static uint16_t __rte_noinline	__hot					       \
+otx2_nix_recv_pkts_mseg_ ## name(void *rx_queue,			       \
+			struct rte_mbuf **rx_pkts, uint16_t pkts)	       \
+{									       \
+	return nix_recv_pkts(rx_queue, rx_pkts, pkts,			       \
+			     (flags) | NIX_RX_MULTI_SEG_F);		       \
+}									       \
 
 NIX_RX_FASTPATH_MODES
 #undef R
@@ -114,15 +122,32 @@ pick_rx_func(struct rte_eth_dev *eth_dev,
 void
 otx2_eth_set_rx_function(struct rte_eth_dev *eth_dev)
 {
+	struct otx2_eth_dev *dev = otx2_eth_pmd_priv(eth_dev);
+
 	const eth_rx_burst_t nix_eth_rx_burst[2][2][2][2][2][2] = {
 #define R(name, f5, f4, f3, f2, f1, f0, flags)				\
 	[f5][f4][f3][f2][f1][f0] =  otx2_nix_recv_pkts_ ## name,
 
+NIX_RX_FASTPATH_MODES
+#undef R
+	};
+
+	const eth_rx_burst_t nix_eth_rx_burst_mseg[2][2][2][2][2][2] = {
+#define R(name, f5, f4, f3, f2, f1, f0, flags)				\
+	[f5][f4][f3][f2][f1][f0] =  otx2_nix_recv_pkts_mseg_ ## name,
+
 NIX_RX_FASTPATH_MODES
 #undef R
 	};
 
 	pick_rx_func(eth_dev, nix_eth_rx_burst);
 
+	if (dev->rx_offloads & DEV_RX_OFFLOAD_SCATTER)
+		pick_rx_func(eth_dev, nix_eth_rx_burst_mseg);
+
+	/* Copy multi seg version with no offload for tear down sequence */
+	if (rte_eal_process_type() == RTE_PROC_PRIMARY)
+		dev->rx_pkt_burst_no_offload =
+			nix_eth_rx_burst_mseg[0][0][0][0][0][0];
 	rte_mb();
 }
diff --git a/drivers/net/octeontx2/otx2_rx.h b/drivers/net/octeontx2/otx2_rx.h
index fc0e87d14..1d1150786 100644
--- a/drivers/net/octeontx2/otx2_rx.h
+++ b/drivers/net/octeontx2/otx2_rx.h
@@ -23,6 +23,11 @@
 #define NIX_RX_OFFLOAD_MARK_UPDATE_F   BIT(4)
 #define NIX_RX_OFFLOAD_TSTAMP_F        BIT(5)
 
+/* Flags to control cqe_to_mbuf conversion function.
+ * Defining it from backwards to denote its been
+ * not used as offload flags to pick function
+ */
+#define NIX_RX_MULTI_SEG_F            BIT(15)
 #define NIX_TIMESYNC_RX_OFFSET		8
 
 struct otx2_timesync_info {
@@ -133,6 +138,51 @@ nix_update_match_id(const uint16_t match_id, uint64_t ol_flags,
 	return ol_flags;
 }
 
+static __rte_always_inline void
+nix_cqe_xtract_mseg(const struct nix_rx_parse_s *rx,
+		    struct rte_mbuf *mbuf, uint64_t rearm)
+{
+	const rte_iova_t *iova_list;
+	struct rte_mbuf *head;
+	const rte_iova_t *eol;
+	uint8_t nb_segs;
+	uint64_t sg;
+
+	sg = *(const uint64_t *)(rx + 1);
+	nb_segs = (sg >> 48) & 0x3;
+	mbuf->nb_segs = nb_segs;
+	mbuf->data_len = sg & 0xFFFF;
+	sg = sg >> 16;
+
+	eol = ((const rte_iova_t *)(rx + 1) + ((rx->desc_sizem1 + 1) << 1));
+	/* Skip SG_S and first IOVA*/
+	iova_list = ((const rte_iova_t *)(rx + 1)) + 2;
+	nb_segs--;
+
+	rearm = rearm & ~0xFFFF;
+
+	head = mbuf;
+	while (nb_segs) {
+		mbuf->next = ((struct rte_mbuf *)*iova_list) - 1;
+		mbuf = mbuf->next;
+
+		__mempool_check_cookies(mbuf->pool, (void **)&mbuf, 1, 1);
+
+		mbuf->data_len = sg & 0xFFFF;
+		sg = sg >> 16;
+		*(uint64_t *)(&mbuf->rearm_data) = rearm;
+		nb_segs--;
+		iova_list++;
+
+		if (!nb_segs && (iova_list + 1 < eol)) {
+			sg = *(const uint64_t *)(iova_list);
+			nb_segs = (sg >> 48) & 0x3;
+			head->nb_segs += nb_segs;
+			iova_list = (const rte_iova_t *)(iova_list + 1);
+		}
+	}
+}
+
 static __rte_always_inline void
 otx2_nix_cqe_to_mbuf(const struct nix_cqe_hdr_s *cq, struct rte_mbuf *mbuf,
 		     const void *lookup_mem, const uint64_t val,
@@ -178,7 +228,10 @@ otx2_nix_cqe_to_mbuf(const struct nix_cqe_hdr_s *cq, struct rte_mbuf *mbuf,
 	*(uint64_t *)(&mbuf->rearm_data) = val;
 	mbuf->pkt_len = len;
 
-	mbuf->data_len = len;
+	if (flag & NIX_RX_MULTI_SEG_F)
+		nix_cqe_xtract_mseg(rx, mbuf, val);
+	else
+		mbuf->data_len = len;
 }
 
 #define CKSUM_F NIX_RX_OFFLOAD_CHECKSUM_F
-- 
2.21.0