From: Dekel Peled <dekelp@mellanox.com>
Date: Wed, 30 Jan 2019 06:43:29 +0000 (+0200)
Subject: net/mlx5: fix Tx metadata for multi-segment packet
X-Git-Url: http://git.droids-corp.org/?a=commitdiff_plain;h=7f4019d370f6edbd9d6ef26a9bc62442b0fd80a6;p=dpdk.git

net/mlx5: fix Tx metadata for multi-segment packet

Original patch implemented the use of match_metadata offload in the
different burst functions.
The concurrent use of match_metadata and multi_segs offloads was
not handled.

This patch updates function txq_scatter_v(), to pass metadata value
from mbuf to wqe, when indicated by offload flags.

Fixes: 6bd7fbd03c62 ("net/mlx5: support metadata as flow rule criteria")
Cc: stable@dpdk.org

Signed-off-by: Dekel Peled <dekelp@mellanox.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>
---

diff --git a/drivers/net/mlx5/mlx5_rxtx_vec_neon.h b/drivers/net/mlx5/mlx5_rxtx_vec_neon.h
index 883fe1bf91..38e915c5c9 100644
--- a/drivers/net/mlx5/mlx5_rxtx_vec_neon.h
+++ b/drivers/net/mlx5/mlx5_rxtx_vec_neon.h
@@ -104,6 +104,8 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
 		sizeof(struct mlx5_wqe) / MLX5_WQE_DWORD_SIZE;
 	unsigned int n;
 	volatile struct mlx5_wqe *wqe = NULL;
+	bool metadata_ol =
+		txq->offloads & DEV_TX_OFFLOAD_MATCH_METADATA ? true : false;
 
 	assert(elts_n > pkts_n);
 	mlx5_tx_complete(txq);
@@ -127,6 +129,9 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
 		uint8x16_t *t_wqe;
 		uint8_t *dseg;
 		uint8x16_t ctrl;
+		rte_be32_t metadata =
+			metadata_ol && (buf->ol_flags & PKT_TX_METADATA) ?
+			buf->tx_metadata : 0;
 
 		assert(segs_n);
 		max_elts = elts_n - (elts_head - txq->elts_tail);
@@ -164,9 +169,10 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
 		ctrl = vqtbl1q_u8(ctrl, ctrl_shuf_m);
 		vst1q_u8((void *)t_wqe, ctrl);
 		/* Fill ESEG in the header. */
-		vst1q_u16((void *)(t_wqe + 1),
-			  ((uint16x8_t) { 0, 0, cs_flags, rte_cpu_to_be_16(len),
-					  0, 0, 0, 0 }));
+		vst1q_u32((void *)(t_wqe + 1),
+			  ((uint32x4_t){ 0,
+					 cs_flags << 16 | rte_cpu_to_be_16(len),
+					 metadata, 0 }));
 		txq->wqe_ci = wqe_ci;
 	}
 	if (!n)
diff --git a/drivers/net/mlx5/mlx5_rxtx_vec_sse.h b/drivers/net/mlx5/mlx5_rxtx_vec_sse.h
index 14117c4bb4..fb384efde7 100644
--- a/drivers/net/mlx5/mlx5_rxtx_vec_sse.h
+++ b/drivers/net/mlx5/mlx5_rxtx_vec_sse.h
@@ -104,6 +104,8 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
 		sizeof(struct mlx5_wqe) / MLX5_WQE_DWORD_SIZE;
 	unsigned int n;
 	volatile struct mlx5_wqe *wqe = NULL;
+	bool metadata_ol =
+		txq->offloads & DEV_TX_OFFLOAD_MATCH_METADATA ? true : false;
 
 	assert(elts_n > pkts_n);
 	mlx5_tx_complete(txq);
@@ -125,6 +127,9 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
 		uint16_t max_wqe;
 		__m128i *t_wqe, *dseg;
 		__m128i ctrl;
+		rte_be32_t metadata =
+			metadata_ol && (buf->ol_flags & PKT_TX_METADATA) ?
+			buf->tx_metadata : 0;
 
 		assert(segs_n);
 		max_elts = elts_n - (elts_head - txq->elts_tail);
@@ -165,9 +170,9 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
 		_mm_store_si128(t_wqe, ctrl);
 		/* Fill ESEG in the header. */
 		_mm_store_si128(t_wqe + 1,
-				_mm_set_epi16(0, 0, 0, 0,
-					      rte_cpu_to_be_16(len), cs_flags,
-					      0, 0));
+				_mm_set_epi32(0, metadata,
+					      (rte_cpu_to_be_16(len) << 16) |
+					      cs_flags, 0));
 		txq->wqe_ci = wqe_ci;
 	}
 	if (!n)