[dpdk-stable] [PATCH 18.11 3/3] net/avf: fix vector Rx

[Jeff Guo <jia.guo@intel.com>]

[ upstream commit 851b22ff688e759a961ed969ea620372b20581d9 ]

The limitation of burst size in vector rx was removed, since it should
retrieve as much received packets as possible. And also the scattered
receive path should use a wrapper function to achieve the goal of
burst maximizing.

Bugzilla ID: 516
Fixes: 319c421f3890 ("net/avf: enable SSE Rx Tx")

Signed-off-by: Jeff Guo <jia.guo@intel.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Tested-by: Wei Ling <weix.ling@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
---
 drivers/net/avf/avf_rxtx_vec_sse.c | 49 ++++++++++++++++++++++--------
 1 file changed, 37 insertions(+), 12 deletions(-)

diff --git a/drivers/net/avf/avf_rxtx_vec_sse.c b/drivers/net/avf/avf_rxtx_vec_sse.c
index 13e94cebc0..4aa209f702 100644
--- a/drivers/net/avf/avf_rxtx_vec_sse.c
+++ b/drivers/net/avf/avf_rxtx_vec_sse.c
@@ -227,10 +227,12 @@ desc_to_ptype_v(__m128i descs[4], struct rte_mbuf **rx_pkts)
 	rx_pkts[3]->packet_type = type_table[_mm_extract_epi8(ptype1, 8)];
 }
 
-/* Notice:
+/**
+ * vPMD raw receive routine, only accept(nb_pkts >= AVF_VPMD_DESCS_PER_LOOP)
+ *
+ * Notice:
  * - nb_pkts < AVF_VPMD_DESCS_PER_LOOP, just return no packet
- * - nb_pkts > AVF_VPMD_RX_MAX_BURST, only scan AVF_VPMD_RX_MAX_BURST
- *   numbers of DD bits
+ * - floor align nb_pkts to a AVF_VPMD_DESCS_PER_LOOP power-of-two
  */
 static inline uint16_t
 _recv_raw_pkts_vec(struct avf_rx_queue *rxq, struct rte_mbuf **rx_pkts,
@@ -260,9 +262,6 @@ _recv_raw_pkts_vec(struct avf_rx_queue *rxq, struct rte_mbuf **rx_pkts,
 			offsetof(struct rte_mbuf, rx_descriptor_fields1) + 8);
 	__m128i dd_check, eop_check;
 
-	/* nb_pkts shall be less equal than AVF_VPMD_RX_MAX_BURST */
-	nb_pkts = RTE_MIN(nb_pkts, AVF_VPMD_RX_MAX_BURST);
-
 	/* nb_pkts has to be floor-aligned to AVF_VPMD_DESCS_PER_LOOP */
 	nb_pkts = RTE_ALIGN_FLOOR(nb_pkts, AVF_VPMD_DESCS_PER_LOOP);
 
@@ -486,15 +485,15 @@ avf_recv_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
 	return _recv_raw_pkts_vec(rx_queue, rx_pkts, nb_pkts, NULL);
 }
 
-/* vPMD receive routine that reassembles scattered packets
+/**
+ * vPMD receive routine that reassembles single burst of 32 scattered packets
+ *
  * Notice:
  * - nb_pkts < AVF_VPMD_DESCS_PER_LOOP, just return no packet
- * - nb_pkts > VPMD_RX_MAX_BURST, only scan AVF_VPMD_RX_MAX_BURST
- *   numbers of DD bits
  */
-uint16_t
-avf_recv_scattered_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
-			    uint16_t nb_pkts)
+static uint16_t
+avf_recv_scattered_burst_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
+			     uint16_t nb_pkts)
 {
 	struct avf_rx_queue *rxq = rx_queue;
 	uint8_t split_flags[AVF_VPMD_RX_MAX_BURST] = {0};
@@ -527,6 +526,32 @@ avf_recv_scattered_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
 		&split_flags[i]);
 }
 
+/**
+ * vPMD receive routine that reassembles scattered packets.
+ */
+uint16_t
+avf_recv_scattered_pkts_vec(void *rx_queue, struct rte_mbuf **rx_pkts,
+			    uint16_t nb_pkts)
+{
+	uint16_t retval = 0;
+
+	while (nb_pkts > AVF_VPMD_RX_MAX_BURST) {
+		uint16_t burst;
+
+		burst = avf_recv_scattered_burst_vec(rx_queue,
+						     rx_pkts + retval,
+						     AVF_VPMD_RX_MAX_BURST);
+		retval += burst;
+		nb_pkts -= burst;
+		if (burst < AVF_VPMD_RX_MAX_BURST)
+			return retval;
+	}
+
+	return retval + avf_recv_scattered_burst_vec(rx_queue,
+						     rx_pkts + retval,
+						     nb_pkts);
+}
+
 static inline void
 vtx1(volatile struct avf_tx_desc *txdp, struct rte_mbuf *pkt, uint64_t flags)
 {
-- 
2.20.1