[PATCH 1/3] net/mlx5: replace network to host macros

[PATCH 1/3] net/mlx5: replace network to host macros

[Shachar Beiser]

Fixes: 8bb5119634b7 ("net/mlx5: replace network byte order macro")
Cc: shacharbe@mellanox.com

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
 drivers/net/mlx5/mlx5_mac.c          |   8 ++-
 drivers/net/mlx5/mlx5_mr.c           |   2 +-
 drivers/net/mlx5/mlx5_rxmode.c       |   8 ++-
 drivers/net/mlx5/mlx5_rxq.c          |   9 +--
 drivers/net/mlx5/mlx5_rxtx.c         | 131 +++++++++++++++++++----------------
 drivers/net/mlx5/mlx5_rxtx.h         |  12 ++--
 drivers/net/mlx5/mlx5_rxtx_vec_sse.c |  12 ++--
 7 files changed, 102 insertions(+), 80 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_mac.c b/drivers/net/mlx5/mlx5_mac.c
index 45d23e4..b3c3fa2 100644
--- a/drivers/net/mlx5/mlx5_mac.c
+++ b/drivers/net/mlx5/mlx5_mac.c
@@ -263,11 +263,15 @@
 				(*mac)[0], (*mac)[1], (*mac)[2],
 				(*mac)[3], (*mac)[4], (*mac)[5]
 			},
-			.vlan_tag = (vlan_enabled ? htons(vlan_id) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(vlan_id)
+				     : 0),
 		},
 		.mask = {
 			.dst_mac = "\xff\xff\xff\xff\xff\xff",
-			.vlan_tag = (vlan_enabled ? htons(0xfff) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(0xfff) :
+				     0),
 		},
 	};
 	DEBUG("%p: adding MAC address %02x:%02x:%02x:%02x:%02x:%02x index %u"
diff --git a/drivers/net/mlx5/mlx5_mr.c b/drivers/net/mlx5/mlx5_mr.c
index 9593830..9a9f73a 100644
--- a/drivers/net/mlx5/mlx5_mr.c
+++ b/drivers/net/mlx5/mlx5_mr.c
@@ -203,7 +203,7 @@ struct ibv_mr *
 	txq_ctrl->txq.mp2mr[idx].start = (uintptr_t)mr->addr;
 	txq_ctrl->txq.mp2mr[idx].end = (uintptr_t)mr->addr + mr->length;
 	txq_ctrl->txq.mp2mr[idx].mr = mr;
-	txq_ctrl->txq.mp2mr[idx].lkey = htonl(mr->lkey);
+	txq_ctrl->txq.mp2mr[idx].lkey = rte_cpu_to_be_32(mr->lkey);
 	DEBUG("%p: new MR lkey for MP \"%s\" (%p): 0x%08" PRIu32,
 	      (void *)txq_ctrl, mp->name, (void *)mp,
 	      txq_ctrl->txq.mp2mr[idx].lkey);
diff --git a/drivers/net/mlx5/mlx5_rxmode.c b/drivers/net/mlx5/mlx5_rxmode.c
index 4a51e47..db2e05b 100644
--- a/drivers/net/mlx5/mlx5_rxmode.c
+++ b/drivers/net/mlx5/mlx5_rxmode.c
@@ -159,14 +159,18 @@
 				mac[0], mac[1], mac[2],
 				mac[3], mac[4], mac[5],
 			},
-			.vlan_tag = (vlan_enabled ? htons(vlan_id) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(vlan_id) :
+				     0),
 		},
 		.mask = {
 			.dst_mac = {
 				mask[0], mask[1], mask[2],
 				mask[3], mask[4], mask[5],
 			},
-			.vlan_tag = (vlan_enabled ? htons(0xfff) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(0xfff) :
+				     0),
 		},
 	};
 
diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index 35c5cb4..437dc02 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -672,9 +672,10 @@
 		/* scat->addr must be able to store a pointer. */
 		assert(sizeof(scat->addr) >= sizeof(uintptr_t));
 		*scat = (struct mlx5_wqe_data_seg){
-			.addr = htonll(rte_pktmbuf_mtod(buf, uintptr_t)),
-			.byte_count = htonl(DATA_LEN(buf)),
-			.lkey = htonl(rxq_ctrl->mr->lkey),
+			.addr =
+			    rte_cpu_to_be_64(rte_pktmbuf_mtod(buf, uintptr_t)),
+			.byte_count = rte_cpu_to_be_32(DATA_LEN(buf)),
+			.lkey = rte_cpu_to_be_32(rxq_ctrl->mr->lkey),
 		};
 		(*rxq_ctrl->rxq.elts)[i] = buf;
 	}
@@ -1077,7 +1078,7 @@
 	/* Update doorbell counter. */
 	rxq_ctrl->rxq.rq_ci = desc >> rxq_ctrl->rxq.sges_n;
 	rte_wmb();
-	*rxq_ctrl->rxq.rq_db = htonl(rxq_ctrl->rxq.rq_ci);
+	*rxq_ctrl->rxq.rq_db = rte_cpu_to_be_32(rxq_ctrl->rxq.rq_ci);
 	DEBUG("%p: rxq updated with %p", (void *)rxq_ctrl, (void *)&tmpl);
 	assert(ret == 0);
 	return 0;
diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index fe9e7ea..e1a35a3 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -306,7 +306,7 @@
 
 		op_own = cqe->op_own;
 		if (MLX5_CQE_FORMAT(op_own) == MLX5_COMPRESSED)
-			n = ntohl(cqe->byte_cnt);
+			n = rte_be_to_cpu_32(cqe->byte_cnt);
 		else
 			n = 1;
 		cq_ci += n;
@@ -434,7 +434,8 @@
 		raw = ((uint8_t *)(uintptr_t)wqe) + 2 * MLX5_WQE_DWORD_SIZE;
 		/* Replace the Ethernet type by the VLAN if necessary. */
 		if (buf->ol_flags & PKT_TX_VLAN_PKT) {
-			uint32_t vlan = htonl(0x81000000 | buf->vlan_tci);
+			uint32_t vlan = rte_cpu_to_be_32(0x81000000 |
+							 buf->vlan_tci);
 			unsigned int len = 2 * ETHER_ADDR_LEN - 2;
 
 			addr += 2;
@@ -510,8 +511,10 @@
 				} else {
 					/* NOP WQE. */
 					wqe->ctrl = (rte_v128u32_t){
-						     htonl(txq->wqe_ci << 8),
-						     htonl(txq->qp_num_8s | 1),
+						     rte_cpu_to_be_32(
+							txq->wqe_ci << 8),
+						     rte_cpu_to_be_32(
+							txq->qp_num_8s | 1),
 						     0,
 						     0,
 					};
@@ -550,7 +553,8 @@
 				max_wqe -= n;
 				if (tso) {
 					uint32_t inl =
-						htonl(copy_b | MLX5_INLINE_SEG);
+					rte_cpu_to_be_32(copy_b |
+							 MLX5_INLINE_SEG);
 
 					pkt_inline_sz =
 						MLX5_WQE_DS(tso_header_sz) *
@@ -603,9 +607,9 @@
 			ds = 3;
 use_dseg:
 			/* Add the remaining packet as a simple ds. */
-			naddr = htonll(addr);
+			naddr = rte_cpu_to_be_64(addr);
 			*dseg = (rte_v128u32_t){
-				htonl(length),
+				rte_cpu_to_be_32(length),
 				mlx5_tx_mb2mr(txq, buf),
 				naddr,
 				naddr >> 32,
@@ -642,9 +646,9 @@
 		total_length += length;
 #endif
 		/* Store segment information. */
-		naddr = htonll(rte_pktmbuf_mtod(buf, uintptr_t));
+		naddr = rte_cpu_to_be_64(rte_pktmbuf_mtod(buf, uintptr_t));
 		*dseg = (rte_v128u32_t){
-			htonl(length),
+			rte_cpu_to_be_32(length),
 			mlx5_tx_mb2mr(txq, buf),
 			naddr,
 			naddr >> 32,
@@ -663,21 +667,23 @@
 		/* Initialize known and common part of the WQE structure. */
 		if (tso) {
 			wqe->ctrl = (rte_v128u32_t){
-				htonl((txq->wqe_ci << 8) | MLX5_OPCODE_TSO),
-				htonl(txq->qp_num_8s | ds),
+				rte_cpu_to_be_32((txq->wqe_ci << 8) |
+						 MLX5_OPCODE_TSO),
+				rte_cpu_to_be_32(txq->qp_num_8s | ds),
 				0,
 				0,
 			};
 			wqe->eseg = (rte_v128u32_t){
 				0,
-				cs_flags | (htons(tso_segsz) << 16),
+				cs_flags | (rte_cpu_to_be_16(tso_segsz) << 16),
 				0,
-				(ehdr << 16) | htons(tso_header_sz),
+				(ehdr << 16) | rte_cpu_to_be_16(tso_header_sz),
 			};
 		} else {
 			wqe->ctrl = (rte_v128u32_t){
-				htonl((txq->wqe_ci << 8) | MLX5_OPCODE_SEND),
-				htonl(txq->qp_num_8s | ds),
+				rte_cpu_to_be_32((txq->wqe_ci << 8) |
+						 MLX5_OPCODE_SEND),
+				rte_cpu_to_be_32(txq->qp_num_8s | ds),
 				0,
 				0,
 			};
@@ -685,7 +691,7 @@
 				0,
 				cs_flags,
 				0,
-				(ehdr << 16) | htons(pkt_inline_sz),
+				(ehdr << 16) | rte_cpu_to_be_16(pkt_inline_sz),
 			};
 		}
 next_wqe:
@@ -705,7 +711,7 @@
 	comp = txq->elts_comp + i + j + k;
 	if (comp >= MLX5_TX_COMP_THRESH) {
 		/* Request completion on last WQE. */
-		last_wqe->ctrl2 = htonl(8);
+		last_wqe->ctrl2 = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		last_wqe->ctrl3 = txq->elts_head;
 		txq->elts_comp = 0;
@@ -744,13 +750,14 @@
 	mpw->len = length;
 	mpw->total_len = 0;
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->eseg.mss = htons(length);
+	mpw->wqe->eseg.mss = rte_cpu_to_be_16(length);
 	mpw->wqe->eseg.inline_hdr_sz = 0;
 	mpw->wqe->eseg.rsvd0 = 0;
 	mpw->wqe->eseg.rsvd1 = 0;
 	mpw->wqe->eseg.rsvd2 = 0;
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_MPW << 24) |
-				  (txq->wqe_ci << 8) | MLX5_OPCODE_TSO);
+	mpw->wqe->ctrl[0] = rte_cpu_to_be_32((MLX5_OPC_MOD_MPW << 24) |
+					     (txq->wqe_ci << 8) |
+					     MLX5_OPCODE_TSO);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
 	mpw->data.dseg[0] = (volatile struct mlx5_wqe_data_seg *)
@@ -779,7 +786,7 @@
 	 * Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | (2 + num));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s | (2 + num));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
 	if (num < 3)
 		++txq->wqe_ci;
@@ -886,9 +893,9 @@
 			dseg = mpw.data.dseg[mpw.pkts_n];
 			addr = rte_pktmbuf_mtod(buf, uintptr_t);
 			*dseg = (struct mlx5_wqe_data_seg){
-				.byte_count = htonl(DATA_LEN(buf)),
+				.byte_count = rte_cpu_to_be_32(DATA_LEN(buf)),
 				.lkey = mlx5_tx_mb2mr(txq, buf),
-				.addr = htonll(addr),
+				.addr = rte_cpu_to_be_64(addr),
 			};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 			length += DATA_LEN(buf);
@@ -916,7 +923,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -956,12 +963,12 @@
 	mpw->len = length;
 	mpw->total_len = 0;
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_MPW << 24) |
-				  (txq->wqe_ci << 8) |
-				  MLX5_OPCODE_TSO);
+	mpw->wqe->ctrl[0] = rte_cpu_to_be_32((MLX5_OPC_MOD_MPW << 24) |
+					     (txq->wqe_ci << 8) |
+					     MLX5_OPCODE_TSO);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
-	mpw->wqe->eseg.mss = htons(length);
+	mpw->wqe->eseg.mss = rte_cpu_to_be_16(length);
 	mpw->wqe->eseg.inline_hdr_sz = 0;
 	mpw->wqe->eseg.cs_flags = 0;
 	mpw->wqe->eseg.rsvd0 = 0;
@@ -992,9 +999,10 @@
 	 * Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | MLX5_WQE_DS(size));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s |
+					     MLX5_WQE_DS(size));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
-	inl->byte_cnt = htonl(mpw->total_len | MLX5_INLINE_SEG);
+	inl->byte_cnt = rte_cpu_to_be_32(mpw->total_len | MLX5_INLINE_SEG);
 	txq->wqe_ci += (size + (MLX5_WQE_SIZE - 1)) / MLX5_WQE_SIZE;
 }
 
@@ -1132,9 +1140,10 @@
 				dseg = mpw.data.dseg[mpw.pkts_n];
 				addr = rte_pktmbuf_mtod(buf, uintptr_t);
 				*dseg = (struct mlx5_wqe_data_seg){
-					.byte_count = htonl(DATA_LEN(buf)),
+					.byte_count =
+					       rte_cpu_to_be_32(DATA_LEN(buf)),
 					.lkey = mlx5_tx_mb2mr(txq, buf),
-					.addr = htonll(addr),
+					.addr = rte_cpu_to_be_64(addr),
 				};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 				length += DATA_LEN(buf);
@@ -1206,7 +1215,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -1246,9 +1255,10 @@
 	mpw->pkts_n = 0;
 	mpw->total_len = sizeof(struct mlx5_wqe);
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_ENHANCED_MPSW << 24) |
-				  (txq->wqe_ci << 8) |
-				  MLX5_OPCODE_ENHANCED_MPSW);
+	mpw->wqe->ctrl[0] =
+		rte_cpu_to_be_32((MLX5_OPC_MOD_ENHANCED_MPSW << 24) |
+				 (txq->wqe_ci << 8) |
+				 MLX5_OPCODE_ENHANCED_MPSW);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
 	memset((void *)(uintptr_t)&mpw->wqe->eseg, 0, MLX5_WQE_DWORD_SIZE);
@@ -1256,9 +1266,9 @@
 		uintptr_t addr = (uintptr_t)(mpw->wqe + 1);
 
 		/* Pad the first 2 DWORDs with zero-length inline header. */
-		*(volatile uint32_t *)addr = htonl(MLX5_INLINE_SEG);
+		*(volatile uint32_t *)addr = rte_cpu_to_be_32(MLX5_INLINE_SEG);
 		*(volatile uint32_t *)(addr + MLX5_WQE_DWORD_SIZE) =
-			htonl(MLX5_INLINE_SEG);
+			rte_cpu_to_be_32(MLX5_INLINE_SEG);
 		mpw->total_len += 2 * MLX5_WQE_DWORD_SIZE;
 		/* Start from the next WQEBB. */
 		mpw->data.raw = (volatile void *)(tx_mlx5_wqe(txq, idx + 1));
@@ -1286,7 +1296,8 @@
 	/* Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | MLX5_WQE_DS(mpw->total_len));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s |
+					     MLX5_WQE_DS(mpw->total_len));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
 	ret = (mpw->total_len + (MLX5_WQE_SIZE - 1)) / MLX5_WQE_SIZE;
 	txq->wqe_ci += ret;
@@ -1439,9 +1450,10 @@
 				dseg = mpw.data.dseg[mpw.pkts_n];
 				addr = rte_pktmbuf_mtod(buf, uintptr_t);
 				*dseg = (struct mlx5_wqe_data_seg){
-					.byte_count = htonl(DATA_LEN(buf)),
+					.byte_count = rte_cpu_to_be_32(
+								DATA_LEN(buf)),
 					.lkey = mlx5_tx_mb2mr(txq, buf),
-					.addr = htonll(addr),
+					.addr = rte_cpu_to_be_64(addr),
 				};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 				length += DATA_LEN(buf);
@@ -1464,7 +1476,7 @@
 
 			assert(mpw.state == MLX5_MPW_ENHANCED_STATE_OPENED);
 			assert(length == DATA_LEN(buf));
-			inl_hdr = htonl(length | MLX5_INLINE_SEG);
+			inl_hdr = rte_cpu_to_be_32(length | MLX5_INLINE_SEG);
 			addr = rte_pktmbuf_mtod(buf, uintptr_t);
 			mpw.data.raw = (volatile void *)
 				((uintptr_t)mpw.data.raw + inl_pad);
@@ -1520,9 +1532,9 @@
 			for (n = 0; n * RTE_CACHE_LINE_SIZE < length; n++)
 				rte_prefetch2((void *)(addr +
 						n * RTE_CACHE_LINE_SIZE));
-			naddr = htonll(addr);
+			naddr = rte_cpu_to_be_64(addr);
 			*dseg = (rte_v128u32_t) {
-				htonl(length),
+				rte_cpu_to_be_32(length),
 				mlx5_tx_mb2mr(txq, buf),
 				naddr,
 				naddr >> 32,
@@ -1550,7 +1562,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -1634,8 +1646,8 @@
 			(volatile struct mlx5_mini_cqe8 (*)[8])
 			(uintptr_t)(&(*rxq->cqes)[zip->ca & cqe_cnt].pkt_info);
 
-		len = ntohl((*mc)[zip->ai & 7].byte_cnt);
-		*rss_hash = ntohl((*mc)[zip->ai & 7].rx_hash_result);
+		len = rte_be_to_cpu_32((*mc)[zip->ai & 7].byte_cnt);
+		*rss_hash = rte_be_to_cpu_32((*mc)[zip->ai & 7].rx_hash_result);
 		if ((++zip->ai & 7) == 0) {
 			/* Invalidate consumed CQEs */
 			idx = zip->ca;
@@ -1683,7 +1695,7 @@
 							  cqe_cnt].pkt_info);
 
 			/* Fix endianness. */
-			zip->cqe_cnt = ntohl(cqe->byte_cnt);
+			zip->cqe_cnt = rte_be_to_cpu_32(cqe->byte_cnt);
 			/*
 			 * Current mini array position is the one returned by
 			 * check_cqe64().
@@ -1698,8 +1710,8 @@
 			--rxq->cq_ci;
 			zip->cq_ci = rxq->cq_ci + zip->cqe_cnt;
 			/* Get packet size to return. */
-			len = ntohl((*mc)[0].byte_cnt);
-			*rss_hash = ntohl((*mc)[0].rx_hash_result);
+			len = rte_be_to_cpu_32((*mc)[0].byte_cnt);
+			*rss_hash = rte_be_to_cpu_32((*mc)[0].rx_hash_result);
 			zip->ai = 1;
 			/* Prefetch all the entries to be invalidated */
 			idx = zip->ca;
@@ -1709,8 +1721,8 @@
 				++idx;
 			}
 		} else {
-			len = ntohl(cqe->byte_cnt);
-			*rss_hash = ntohl(cqe->rx_hash_res);
+			len = rte_be_to_cpu_32(cqe->byte_cnt);
+			*rss_hash = rte_be_to_cpu_32(cqe->rx_hash_res);
 		}
 		/* Error while receiving packet. */
 		if (unlikely(MLX5_CQE_OPCODE(op_own) == MLX5_CQE_RESP_ERR))
@@ -1734,7 +1746,7 @@
 rxq_cq_to_ol_flags(struct rxq *rxq, volatile struct mlx5_cqe *cqe)
 {
 	uint32_t ol_flags = 0;
-	uint16_t flags = ntohs(cqe->hdr_type_etc);
+	uint16_t flags = rte_be_to_cpu_16(cqe->hdr_type_etc);
 
 	ol_flags =
 		TRANSPOSE(flags,
@@ -1841,7 +1853,7 @@
 			    MLX5_FLOW_MARK_IS_VALID(cqe->sop_drop_qpn)) {
 				pkt->ol_flags |= PKT_RX_FDIR;
 				if (cqe->sop_drop_qpn !=
-				    htonl(MLX5_FLOW_MARK_DEFAULT)) {
+				    rte_cpu_to_be_32(MLX5_FLOW_MARK_DEFAULT)) {
 					uint32_t mark = cqe->sop_drop_qpn;
 
 					pkt->ol_flags |= PKT_RX_FDIR_ID;
@@ -1853,10 +1865,11 @@
 				pkt->ol_flags |= rxq_cq_to_ol_flags(rxq, cqe);
 			if (rxq->vlan_strip &&
 			    (cqe->hdr_type_etc &
-			     htons(MLX5_CQE_VLAN_STRIPPED))) {
+			     rte_cpu_to_be_16(MLX5_CQE_VLAN_STRIPPED))) {
 				pkt->ol_flags |= PKT_RX_VLAN_PKT |
 					PKT_RX_VLAN_STRIPPED;
-				pkt->vlan_tci = ntohs(cqe->vlan_info);
+				pkt->vlan_tci =
+					rte_be_to_cpu_16(cqe->vlan_info);
 			}
 			if (rxq->crc_present)
 				len -= ETHER_CRC_LEN;
@@ -1872,7 +1885,7 @@
 		 * of the buffers are already known, only the buffer address
 		 * changes.
 		 */
-		wqe->addr = htonll(rte_pktmbuf_mtod(rep, uintptr_t));
+		wqe->addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(rep, uintptr_t));
 		if (len > DATA_LEN(seg)) {
 			len -= DATA_LEN(seg);
 			++NB_SEGS(pkt);
@@ -1900,9 +1913,9 @@
 	/* Update the consumer index. */
 	rxq->rq_ci = rq_ci >> sges_n;
 	rte_wmb();
-	*rxq->cq_db = htonl(rxq->cq_ci);
+	*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
 	rte_wmb();
-	*rxq->rq_db = htonl(rxq->rq_ci);
+	*rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci);
 #ifdef MLX5_PMD_SOFT_COUNTERS
 	/* Increment packets counter. */
 	rxq->stats.ipackets += i;
diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h
index 033e70f..73a4ce8 100644
--- a/drivers/net/mlx5/mlx5_rxtx.h
+++ b/drivers/net/mlx5/mlx5_rxtx.h
@@ -269,7 +269,7 @@ struct txq {
 		uintptr_t start; /* Start address of MR */
 		uintptr_t end; /* End address of MR */
 		struct ibv_mr *mr; /* Memory Region (for mp). */
-		uint32_t lkey; /* htonl(mr->lkey) */
+		uint32_t lkey; /* rte_cpu_to_be_32(mr->lkey) */
 	} mp2mr[MLX5_PMD_TX_MP_CACHE]; /* MP to MR translation table. */
 	uint16_t mr_cache_idx; /* Index of last hit entry. */
 	struct rte_mbuf *(*elts)[]; /* TX elements. */
@@ -492,7 +492,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	}
 #endif /* NDEBUG */
 	++cq_ci;
-	txq->wqe_pi = ntohs(cqe->wqe_counter);
+	txq->wqe_pi = rte_be_to_cpu_16(cqe->wqe_counter);
 	ctrl = (volatile struct mlx5_wqe_ctrl *)
 		tx_mlx5_wqe(txq, txq->wqe_pi);
 	elts_tail = ctrl->ctrl3;
@@ -530,7 +530,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	txq->elts_tail = elts_tail;
 	/* Update the consumer index. */
 	rte_wmb();
-	*txq->cq_db = htonl(cq_ci);
+	*txq->cq_db = rte_cpu_to_be_32(cq_ci);
 }
 
 /**
@@ -581,7 +581,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 		if (txq->mp2mr[i].start <= addr &&
 		    txq->mp2mr[i].end >= addr) {
 			assert(txq->mp2mr[i].lkey != (uint32_t)-1);
-			assert(htonl(txq->mp2mr[i].mr->lkey) ==
+			assert(rte_cpu_to_be_32(txq->mp2mr[i].mr->lkey) ==
 			       txq->mp2mr[i].lkey);
 			txq->mr_cache_idx = i;
 			return txq->mp2mr[i].lkey;
@@ -605,8 +605,8 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	uint64_t *dst = (uint64_t *)((uintptr_t)txq->bf_reg);
 	volatile uint64_t *src = ((volatile uint64_t *)wqe);
 
-	rte_io_wmb();
-	*txq->qp_db = htonl(txq->wqe_ci);
+	rte_wmb();
+	*txq->qp_db = rte_cpu_to_be_32(txq->wqe_ci);
 	/* Ensure ordering between DB record and BF copy. */
 	rte_wmb();
 	*dst = *src;
diff --git a/drivers/net/mlx5/mlx5_rxtx_vec_sse.c b/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
index 37854a7..0a5d025 100644
--- a/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
+++ b/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
@@ -291,7 +291,7 @@
 		/* Fill ESEG in the header. */
 		_mm_store_si128(t_wqe + 1,
 				_mm_set_epi16(0, 0, 0, 0,
-					      htons(len), cs_flags,
+					      rte_cpu_to_be_16(len), cs_flags,
 					      0, 0));
 		txq->wqe_ci = wqe_ci;
 	}
@@ -300,7 +300,7 @@
 	txq->elts_comp += (uint16_t)(elts_head - txq->elts_head);
 	txq->elts_head = elts_head;
 	if (txq->elts_comp >= MLX5_TX_COMP_THRESH) {
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		wqe->ctrl[3] = txq->elts_head;
 		txq->elts_comp = 0;
 		++txq->cq_pi;
@@ -561,11 +561,11 @@
 		return;
 	}
 	for (i = 0; i < n; ++i)
-		wq[i].addr = htonll((uintptr_t)elts[i]->buf_addr +
-				    RTE_PKTMBUF_HEADROOM);
+		wq[i].addr = rte_cpu_to_be_64((uintptr_t)elts[i]->buf_addr +
+					      RTE_PKTMBUF_HEADROOM);
 	rxq->rq_ci += n;
 	rte_wmb();
-	*rxq->rq_db = htonl(rxq->rq_ci);
+	*rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci);
 }
 
 /**
@@ -1248,7 +1248,7 @@
 		}
 	}
 	rte_wmb();
-	*rxq->cq_db = htonl(rxq->cq_ci);
+	*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
 	return rcvd_pkt;
 }
 
-- 
1.8.3.1

[PATCH 2/3] net/mlx5: fix TSO MLNX OFED 3.3 verification

[Shachar Beiser]

Fixes: 3cf87e68d97b ("net/mlx5: remove old MLNX OFED 3.3 verification")
Cc: stable@dpdk.org

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
 drivers/net/mlx5/mlx5_prm.h | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_prm.h b/drivers/net/mlx5/mlx5_prm.h
index 608072f..8b82b5e 100644
--- a/drivers/net/mlx5/mlx5_prm.h
+++ b/drivers/net/mlx5/mlx5_prm.h
@@ -89,9 +89,6 @@
 /* Default max packet length to be inlined. */
 #define MLX5_EMPW_MAX_INLINE_LEN (4U * MLX5_WQE_SIZE)
 
-#ifndef HAVE_VERBS_MLX5_OPCODE_TSO
-#define MLX5_OPCODE_TSO MLX5_OPCODE_LSO_MPW /* Compat with OFED 3.3. */
-#endif
 
 #define MLX5_OPC_MOD_ENHANCED_MPSW 0
 #define MLX5_OPCODE_ENHANCED_MPSW 0x29
-- 
1.8.3.1

[PATCH 3/3] net/mlx5: fix interrupt enable return value

[Shachar Beiser]

Fixes: 3c7d44af252a ("net/mlx5: support user space Rx interrupt event")
Cc: stable@dpdk.org

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
 drivers/net/mlx5/mlx5_rxq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index 437dc02..24887fb 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -1330,7 +1330,7 @@
 	struct priv *priv = mlx5_get_priv(dev);
 	struct rxq *rxq = (*priv->rxqs)[rx_queue_id];
 	struct rxq_ctrl *rxq_ctrl = container_of(rxq, struct rxq_ctrl, rxq);
-	int ret;
+	int ret = 0;
 
 	if (!rxq || !rxq_ctrl->channel) {
 		ret = EINVAL;
-- 
1.8.3.1

Re: [PATCH 1/3] net/mlx5: replace network to host macros

[Nélio Laranjeiro]

The title is a little wrong, it also replace the host to network.

On Mon, Sep 04, 2017 at 11:48:45AM +0000, Shachar Beiser wrote:
> Fixes: 8bb5119634b7 ("net/mlx5: replace network byte order macro")

This commit does not exists, are you sure of it?

> Cc: shacharbe@mellanox.com
> 
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>

-- 
Nélio Laranjeiro
6WIND

Re: [PATCH 2/3] net/mlx5: fix TSO MLNX OFED 3.3 verification

[Nélio Laranjeiro]

On Mon, Sep 04, 2017 at 11:48:46AM +0000, Shachar Beiser wrote:
> Fixes: 3cf87e68d97b ("net/mlx5: remove old MLNX OFED 3.3 verification")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> ---
>  drivers/net/mlx5/mlx5_prm.h | 3 ---
>  1 file changed, 3 deletions(-)
> 
> diff --git a/drivers/net/mlx5/mlx5_prm.h b/drivers/net/mlx5/mlx5_prm.h
> index 608072f..8b82b5e 100644
> --- a/drivers/net/mlx5/mlx5_prm.h
> +++ b/drivers/net/mlx5/mlx5_prm.h
> @@ -89,9 +89,6 @@
>  /* Default max packet length to be inlined. */
>  #define MLX5_EMPW_MAX_INLINE_LEN (4U * MLX5_WQE_SIZE)
>  
> -#ifndef HAVE_VERBS_MLX5_OPCODE_TSO
> -#define MLX5_OPCODE_TSO MLX5_OPCODE_LSO_MPW /* Compat with OFED 3.3. */
> -#endif
>  
>  #define MLX5_OPC_MOD_ENHANCED_MPSW 0
>  #define MLX5_OPCODE_ENHANCED_MPSW 0x29
> -- 
> 1.8.3.1
> 

Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

-- 
Nélio Laranjeiro
6WIND

Re: [PATCH 3/3] net/mlx5: fix interrupt enable return value

[Nélio Laranjeiro]

On Mon, Sep 04, 2017 at 11:48:47AM +0000, Shachar Beiser wrote:
> Fixes: 3c7d44af252a ("net/mlx5: support user space Rx interrupt event")

It should fix commit e1016cb733 ("net/mlx5: fix Rx interrupts management")

> Cc: stable@dpdk.org
> 
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> ---
>  drivers/net/mlx5/mlx5_rxq.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
> index 437dc02..24887fb 100644
> --- a/drivers/net/mlx5/mlx5_rxq.c
> +++ b/drivers/net/mlx5/mlx5_rxq.c
> @@ -1330,7 +1330,7 @@
>  	struct priv *priv = mlx5_get_priv(dev);
>  	struct rxq *rxq = (*priv->rxqs)[rx_queue_id];
>  	struct rxq_ctrl *rxq_ctrl = container_of(rxq, struct rxq_ctrl, rxq);
> -	int ret;
> +	int ret = 0;
>  
>  	if (!rxq || !rxq_ctrl->channel) {
>  		ret = EINVAL;
> -- 
> 1.8.3.1
> 

-- 
Nélio Laranjeiro
6WIND

Re: [PATCH 3/3] net/mlx5: fix interrupt enable return value

[Shachar Beiser]

OK, 
I will fix to " commit e1016cb733 ("net/mlx5: fix Rx interrupts management") "


> -----Original Message-----
> From: Nélio Laranjeiro [mailto:nelio.laranjeiro@6wind.com]
> Sent: Monday, September 4, 2017 6:24 PM
> To: Shachar Beiser <shacharbe@mellanox.com>
> Cc: dev@dpdk.org; Adrien Mazarguil <adrien.mazarguil@6wind.com>;
> stable@dpdk.org
> Subject: Re: [PATCH 3/3] net/mlx5: fix interrupt enable return value
> 
> On Mon, Sep 04, 2017 at 11:48:47AM +0000, Shachar Beiser wrote:
> > Fixes: 3c7d44af252a ("net/mlx5: support user space Rx interrupt
> > event")
> 
> It should fix commit e1016cb733 ("net/mlx5: fix Rx interrupts management")
> 
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> > ---
> >  drivers/net/mlx5/mlx5_rxq.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
> > index 437dc02..24887fb 100644
> > --- a/drivers/net/mlx5/mlx5_rxq.c
> > +++ b/drivers/net/mlx5/mlx5_rxq.c
> > @@ -1330,7 +1330,7 @@
> >  	struct priv *priv = mlx5_get_priv(dev);
> >  	struct rxq *rxq = (*priv->rxqs)[rx_queue_id];
> >  	struct rxq_ctrl *rxq_ctrl = container_of(rxq, struct rxq_ctrl, rxq);
> > -	int ret;
> > +	int ret = 0;
> >
> >  	if (!rxq || !rxq_ctrl->channel) {
> >  		ret = EINVAL;
> > --
> > 1.8.3.1
> >
> 
> --
> Nélio Laranjeiro
> 6WIND

Re: [PATCH 1/3] net/mlx5: replace network to host macros

[Shachar Beiser]

Does it OK to change it to :
" Fixes: 1be17b6a5539 ("eal: introduce big and little endian types")"
?

> -----Original Message-----
> From: Nélio Laranjeiro [mailto:nelio.laranjeiro@6wind.com]
> Sent: Monday, September 4, 2017 6:15 PM
> To: Shachar Beiser <shacharbe@mellanox.com>
> Cc: dev@dpdk.org; Adrien Mazarguil <adrien.mazarguil@6wind.com>
> Subject: Re: [PATCH 1/3] net/mlx5: replace network to host macros
> 
> The title is a little wrong, it also replace the host to network.
> 
> On Mon, Sep 04, 2017 at 11:48:45AM +0000, Shachar Beiser wrote:
> > Fixes: 8bb5119634b7 ("net/mlx5: replace network byte order macro")
> 
> This commit does not exists, are you sure of it?
> 
> > Cc: shacharbe@mellanox.com
> >
> > Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> 
> --
> Nélio Laranjeiro
> 6WIND

[PATCH v2 1/3] net/mlx5: replace network to host macros

[Shachar Beiser]

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
 drivers/net/mlx5/mlx5_mac.c          |   8 ++-
 drivers/net/mlx5/mlx5_mr.c           |   2 +-
 drivers/net/mlx5/mlx5_rxmode.c       |   8 ++-
 drivers/net/mlx5/mlx5_rxq.c          |   9 +--
 drivers/net/mlx5/mlx5_rxtx.c         | 131 +++++++++++++++++++----------------
 drivers/net/mlx5/mlx5_rxtx.h         |  12 ++--
 drivers/net/mlx5/mlx5_rxtx_vec_sse.c |  12 ++--
 7 files changed, 102 insertions(+), 80 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_mac.c b/drivers/net/mlx5/mlx5_mac.c
index 45d23e4..b3c3fa2 100644
--- a/drivers/net/mlx5/mlx5_mac.c
+++ b/drivers/net/mlx5/mlx5_mac.c
@@ -263,11 +263,15 @@
 				(*mac)[0], (*mac)[1], (*mac)[2],
 				(*mac)[3], (*mac)[4], (*mac)[5]
 			},
-			.vlan_tag = (vlan_enabled ? htons(vlan_id) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(vlan_id)
+				     : 0),
 		},
 		.mask = {
 			.dst_mac = "\xff\xff\xff\xff\xff\xff",
-			.vlan_tag = (vlan_enabled ? htons(0xfff) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(0xfff) :
+				     0),
 		},
 	};
 	DEBUG("%p: adding MAC address %02x:%02x:%02x:%02x:%02x:%02x index %u"
diff --git a/drivers/net/mlx5/mlx5_mr.c b/drivers/net/mlx5/mlx5_mr.c
index 9593830..9a9f73a 100644
--- a/drivers/net/mlx5/mlx5_mr.c
+++ b/drivers/net/mlx5/mlx5_mr.c
@@ -203,7 +203,7 @@ struct ibv_mr *
 	txq_ctrl->txq.mp2mr[idx].start = (uintptr_t)mr->addr;
 	txq_ctrl->txq.mp2mr[idx].end = (uintptr_t)mr->addr + mr->length;
 	txq_ctrl->txq.mp2mr[idx].mr = mr;
-	txq_ctrl->txq.mp2mr[idx].lkey = htonl(mr->lkey);
+	txq_ctrl->txq.mp2mr[idx].lkey = rte_cpu_to_be_32(mr->lkey);
 	DEBUG("%p: new MR lkey for MP \"%s\" (%p): 0x%08" PRIu32,
 	      (void *)txq_ctrl, mp->name, (void *)mp,
 	      txq_ctrl->txq.mp2mr[idx].lkey);
diff --git a/drivers/net/mlx5/mlx5_rxmode.c b/drivers/net/mlx5/mlx5_rxmode.c
index 4a51e47..db2e05b 100644
--- a/drivers/net/mlx5/mlx5_rxmode.c
+++ b/drivers/net/mlx5/mlx5_rxmode.c
@@ -159,14 +159,18 @@
 				mac[0], mac[1], mac[2],
 				mac[3], mac[4], mac[5],
 			},
-			.vlan_tag = (vlan_enabled ? htons(vlan_id) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(vlan_id) :
+				     0),
 		},
 		.mask = {
 			.dst_mac = {
 				mask[0], mask[1], mask[2],
 				mask[3], mask[4], mask[5],
 			},
-			.vlan_tag = (vlan_enabled ? htons(0xfff) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(0xfff) :
+				     0),
 		},
 	};
 
diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index 35c5cb4..437dc02 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -672,9 +672,10 @@
 		/* scat->addr must be able to store a pointer. */
 		assert(sizeof(scat->addr) >= sizeof(uintptr_t));
 		*scat = (struct mlx5_wqe_data_seg){
-			.addr = htonll(rte_pktmbuf_mtod(buf, uintptr_t)),
-			.byte_count = htonl(DATA_LEN(buf)),
-			.lkey = htonl(rxq_ctrl->mr->lkey),
+			.addr =
+			    rte_cpu_to_be_64(rte_pktmbuf_mtod(buf, uintptr_t)),
+			.byte_count = rte_cpu_to_be_32(DATA_LEN(buf)),
+			.lkey = rte_cpu_to_be_32(rxq_ctrl->mr->lkey),
 		};
 		(*rxq_ctrl->rxq.elts)[i] = buf;
 	}
@@ -1077,7 +1078,7 @@
 	/* Update doorbell counter. */
 	rxq_ctrl->rxq.rq_ci = desc >> rxq_ctrl->rxq.sges_n;
 	rte_wmb();
-	*rxq_ctrl->rxq.rq_db = htonl(rxq_ctrl->rxq.rq_ci);
+	*rxq_ctrl->rxq.rq_db = rte_cpu_to_be_32(rxq_ctrl->rxq.rq_ci);
 	DEBUG("%p: rxq updated with %p", (void *)rxq_ctrl, (void *)&tmpl);
 	assert(ret == 0);
 	return 0;
diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index fe9e7ea..e1a35a3 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -306,7 +306,7 @@
 
 		op_own = cqe->op_own;
 		if (MLX5_CQE_FORMAT(op_own) == MLX5_COMPRESSED)
-			n = ntohl(cqe->byte_cnt);
+			n = rte_be_to_cpu_32(cqe->byte_cnt);
 		else
 			n = 1;
 		cq_ci += n;
@@ -434,7 +434,8 @@
 		raw = ((uint8_t *)(uintptr_t)wqe) + 2 * MLX5_WQE_DWORD_SIZE;
 		/* Replace the Ethernet type by the VLAN if necessary. */
 		if (buf->ol_flags & PKT_TX_VLAN_PKT) {
-			uint32_t vlan = htonl(0x81000000 | buf->vlan_tci);
+			uint32_t vlan = rte_cpu_to_be_32(0x81000000 |
+							 buf->vlan_tci);
 			unsigned int len = 2 * ETHER_ADDR_LEN - 2;
 
 			addr += 2;
@@ -510,8 +511,10 @@
 				} else {
 					/* NOP WQE. */
 					wqe->ctrl = (rte_v128u32_t){
-						     htonl(txq->wqe_ci << 8),
-						     htonl(txq->qp_num_8s | 1),
+						     rte_cpu_to_be_32(
+							txq->wqe_ci << 8),
+						     rte_cpu_to_be_32(
+							txq->qp_num_8s | 1),
 						     0,
 						     0,
 					};
@@ -550,7 +553,8 @@
 				max_wqe -= n;
 				if (tso) {
 					uint32_t inl =
-						htonl(copy_b | MLX5_INLINE_SEG);
+					rte_cpu_to_be_32(copy_b |
+							 MLX5_INLINE_SEG);
 
 					pkt_inline_sz =
 						MLX5_WQE_DS(tso_header_sz) *
@@ -603,9 +607,9 @@
 			ds = 3;
 use_dseg:
 			/* Add the remaining packet as a simple ds. */
-			naddr = htonll(addr);
+			naddr = rte_cpu_to_be_64(addr);
 			*dseg = (rte_v128u32_t){
-				htonl(length),
+				rte_cpu_to_be_32(length),
 				mlx5_tx_mb2mr(txq, buf),
 				naddr,
 				naddr >> 32,
@@ -642,9 +646,9 @@
 		total_length += length;
 #endif
 		/* Store segment information. */
-		naddr = htonll(rte_pktmbuf_mtod(buf, uintptr_t));
+		naddr = rte_cpu_to_be_64(rte_pktmbuf_mtod(buf, uintptr_t));
 		*dseg = (rte_v128u32_t){
-			htonl(length),
+			rte_cpu_to_be_32(length),
 			mlx5_tx_mb2mr(txq, buf),
 			naddr,
 			naddr >> 32,
@@ -663,21 +667,23 @@
 		/* Initialize known and common part of the WQE structure. */
 		if (tso) {
 			wqe->ctrl = (rte_v128u32_t){
-				htonl((txq->wqe_ci << 8) | MLX5_OPCODE_TSO),
-				htonl(txq->qp_num_8s | ds),
+				rte_cpu_to_be_32((txq->wqe_ci << 8) |
+						 MLX5_OPCODE_TSO),
+				rte_cpu_to_be_32(txq->qp_num_8s | ds),
 				0,
 				0,
 			};
 			wqe->eseg = (rte_v128u32_t){
 				0,
-				cs_flags | (htons(tso_segsz) << 16),
+				cs_flags | (rte_cpu_to_be_16(tso_segsz) << 16),
 				0,
-				(ehdr << 16) | htons(tso_header_sz),
+				(ehdr << 16) | rte_cpu_to_be_16(tso_header_sz),
 			};
 		} else {
 			wqe->ctrl = (rte_v128u32_t){
-				htonl((txq->wqe_ci << 8) | MLX5_OPCODE_SEND),
-				htonl(txq->qp_num_8s | ds),
+				rte_cpu_to_be_32((txq->wqe_ci << 8) |
+						 MLX5_OPCODE_SEND),
+				rte_cpu_to_be_32(txq->qp_num_8s | ds),
 				0,
 				0,
 			};
@@ -685,7 +691,7 @@
 				0,
 				cs_flags,
 				0,
-				(ehdr << 16) | htons(pkt_inline_sz),
+				(ehdr << 16) | rte_cpu_to_be_16(pkt_inline_sz),
 			};
 		}
 next_wqe:
@@ -705,7 +711,7 @@
 	comp = txq->elts_comp + i + j + k;
 	if (comp >= MLX5_TX_COMP_THRESH) {
 		/* Request completion on last WQE. */
-		last_wqe->ctrl2 = htonl(8);
+		last_wqe->ctrl2 = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		last_wqe->ctrl3 = txq->elts_head;
 		txq->elts_comp = 0;
@@ -744,13 +750,14 @@
 	mpw->len = length;
 	mpw->total_len = 0;
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->eseg.mss = htons(length);
+	mpw->wqe->eseg.mss = rte_cpu_to_be_16(length);
 	mpw->wqe->eseg.inline_hdr_sz = 0;
 	mpw->wqe->eseg.rsvd0 = 0;
 	mpw->wqe->eseg.rsvd1 = 0;
 	mpw->wqe->eseg.rsvd2 = 0;
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_MPW << 24) |
-				  (txq->wqe_ci << 8) | MLX5_OPCODE_TSO);
+	mpw->wqe->ctrl[0] = rte_cpu_to_be_32((MLX5_OPC_MOD_MPW << 24) |
+					     (txq->wqe_ci << 8) |
+					     MLX5_OPCODE_TSO);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
 	mpw->data.dseg[0] = (volatile struct mlx5_wqe_data_seg *)
@@ -779,7 +786,7 @@
 	 * Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | (2 + num));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s | (2 + num));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
 	if (num < 3)
 		++txq->wqe_ci;
@@ -886,9 +893,9 @@
 			dseg = mpw.data.dseg[mpw.pkts_n];
 			addr = rte_pktmbuf_mtod(buf, uintptr_t);
 			*dseg = (struct mlx5_wqe_data_seg){
-				.byte_count = htonl(DATA_LEN(buf)),
+				.byte_count = rte_cpu_to_be_32(DATA_LEN(buf)),
 				.lkey = mlx5_tx_mb2mr(txq, buf),
-				.addr = htonll(addr),
+				.addr = rte_cpu_to_be_64(addr),
 			};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 			length += DATA_LEN(buf);
@@ -916,7 +923,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -956,12 +963,12 @@
 	mpw->len = length;
 	mpw->total_len = 0;
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_MPW << 24) |
-				  (txq->wqe_ci << 8) |
-				  MLX5_OPCODE_TSO);
+	mpw->wqe->ctrl[0] = rte_cpu_to_be_32((MLX5_OPC_MOD_MPW << 24) |
+					     (txq->wqe_ci << 8) |
+					     MLX5_OPCODE_TSO);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
-	mpw->wqe->eseg.mss = htons(length);
+	mpw->wqe->eseg.mss = rte_cpu_to_be_16(length);
 	mpw->wqe->eseg.inline_hdr_sz = 0;
 	mpw->wqe->eseg.cs_flags = 0;
 	mpw->wqe->eseg.rsvd0 = 0;
@@ -992,9 +999,10 @@
 	 * Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | MLX5_WQE_DS(size));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s |
+					     MLX5_WQE_DS(size));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
-	inl->byte_cnt = htonl(mpw->total_len | MLX5_INLINE_SEG);
+	inl->byte_cnt = rte_cpu_to_be_32(mpw->total_len | MLX5_INLINE_SEG);
 	txq->wqe_ci += (size + (MLX5_WQE_SIZE - 1)) / MLX5_WQE_SIZE;
 }
 
@@ -1132,9 +1140,10 @@
 				dseg = mpw.data.dseg[mpw.pkts_n];
 				addr = rte_pktmbuf_mtod(buf, uintptr_t);
 				*dseg = (struct mlx5_wqe_data_seg){
-					.byte_count = htonl(DATA_LEN(buf)),
+					.byte_count =
+					       rte_cpu_to_be_32(DATA_LEN(buf)),
 					.lkey = mlx5_tx_mb2mr(txq, buf),
-					.addr = htonll(addr),
+					.addr = rte_cpu_to_be_64(addr),
 				};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 				length += DATA_LEN(buf);
@@ -1206,7 +1215,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -1246,9 +1255,10 @@
 	mpw->pkts_n = 0;
 	mpw->total_len = sizeof(struct mlx5_wqe);
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_ENHANCED_MPSW << 24) |
-				  (txq->wqe_ci << 8) |
-				  MLX5_OPCODE_ENHANCED_MPSW);
+	mpw->wqe->ctrl[0] =
+		rte_cpu_to_be_32((MLX5_OPC_MOD_ENHANCED_MPSW << 24) |
+				 (txq->wqe_ci << 8) |
+				 MLX5_OPCODE_ENHANCED_MPSW);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
 	memset((void *)(uintptr_t)&mpw->wqe->eseg, 0, MLX5_WQE_DWORD_SIZE);
@@ -1256,9 +1266,9 @@
 		uintptr_t addr = (uintptr_t)(mpw->wqe + 1);
 
 		/* Pad the first 2 DWORDs with zero-length inline header. */
-		*(volatile uint32_t *)addr = htonl(MLX5_INLINE_SEG);
+		*(volatile uint32_t *)addr = rte_cpu_to_be_32(MLX5_INLINE_SEG);
 		*(volatile uint32_t *)(addr + MLX5_WQE_DWORD_SIZE) =
-			htonl(MLX5_INLINE_SEG);
+			rte_cpu_to_be_32(MLX5_INLINE_SEG);
 		mpw->total_len += 2 * MLX5_WQE_DWORD_SIZE;
 		/* Start from the next WQEBB. */
 		mpw->data.raw = (volatile void *)(tx_mlx5_wqe(txq, idx + 1));
@@ -1286,7 +1296,8 @@
 	/* Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | MLX5_WQE_DS(mpw->total_len));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s |
+					     MLX5_WQE_DS(mpw->total_len));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
 	ret = (mpw->total_len + (MLX5_WQE_SIZE - 1)) / MLX5_WQE_SIZE;
 	txq->wqe_ci += ret;
@@ -1439,9 +1450,10 @@
 				dseg = mpw.data.dseg[mpw.pkts_n];
 				addr = rte_pktmbuf_mtod(buf, uintptr_t);
 				*dseg = (struct mlx5_wqe_data_seg){
-					.byte_count = htonl(DATA_LEN(buf)),
+					.byte_count = rte_cpu_to_be_32(
+								DATA_LEN(buf)),
 					.lkey = mlx5_tx_mb2mr(txq, buf),
-					.addr = htonll(addr),
+					.addr = rte_cpu_to_be_64(addr),
 				};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 				length += DATA_LEN(buf);
@@ -1464,7 +1476,7 @@
 
 			assert(mpw.state == MLX5_MPW_ENHANCED_STATE_OPENED);
 			assert(length == DATA_LEN(buf));
-			inl_hdr = htonl(length | MLX5_INLINE_SEG);
+			inl_hdr = rte_cpu_to_be_32(length | MLX5_INLINE_SEG);
 			addr = rte_pktmbuf_mtod(buf, uintptr_t);
 			mpw.data.raw = (volatile void *)
 				((uintptr_t)mpw.data.raw + inl_pad);
@@ -1520,9 +1532,9 @@
 			for (n = 0; n * RTE_CACHE_LINE_SIZE < length; n++)
 				rte_prefetch2((void *)(addr +
 						n * RTE_CACHE_LINE_SIZE));
-			naddr = htonll(addr);
+			naddr = rte_cpu_to_be_64(addr);
 			*dseg = (rte_v128u32_t) {
-				htonl(length),
+				rte_cpu_to_be_32(length),
 				mlx5_tx_mb2mr(txq, buf),
 				naddr,
 				naddr >> 32,
@@ -1550,7 +1562,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -1634,8 +1646,8 @@
 			(volatile struct mlx5_mini_cqe8 (*)[8])
 			(uintptr_t)(&(*rxq->cqes)[zip->ca & cqe_cnt].pkt_info);
 
-		len = ntohl((*mc)[zip->ai & 7].byte_cnt);
-		*rss_hash = ntohl((*mc)[zip->ai & 7].rx_hash_result);
+		len = rte_be_to_cpu_32((*mc)[zip->ai & 7].byte_cnt);
+		*rss_hash = rte_be_to_cpu_32((*mc)[zip->ai & 7].rx_hash_result);
 		if ((++zip->ai & 7) == 0) {
 			/* Invalidate consumed CQEs */
 			idx = zip->ca;
@@ -1683,7 +1695,7 @@
 							  cqe_cnt].pkt_info);
 
 			/* Fix endianness. */
-			zip->cqe_cnt = ntohl(cqe->byte_cnt);
+			zip->cqe_cnt = rte_be_to_cpu_32(cqe->byte_cnt);
 			/*
 			 * Current mini array position is the one returned by
 			 * check_cqe64().
@@ -1698,8 +1710,8 @@
 			--rxq->cq_ci;
 			zip->cq_ci = rxq->cq_ci + zip->cqe_cnt;
 			/* Get packet size to return. */
-			len = ntohl((*mc)[0].byte_cnt);
-			*rss_hash = ntohl((*mc)[0].rx_hash_result);
+			len = rte_be_to_cpu_32((*mc)[0].byte_cnt);
+			*rss_hash = rte_be_to_cpu_32((*mc)[0].rx_hash_result);
 			zip->ai = 1;
 			/* Prefetch all the entries to be invalidated */
 			idx = zip->ca;
@@ -1709,8 +1721,8 @@
 				++idx;
 			}
 		} else {
-			len = ntohl(cqe->byte_cnt);
-			*rss_hash = ntohl(cqe->rx_hash_res);
+			len = rte_be_to_cpu_32(cqe->byte_cnt);
+			*rss_hash = rte_be_to_cpu_32(cqe->rx_hash_res);
 		}
 		/* Error while receiving packet. */
 		if (unlikely(MLX5_CQE_OPCODE(op_own) == MLX5_CQE_RESP_ERR))
@@ -1734,7 +1746,7 @@
 rxq_cq_to_ol_flags(struct rxq *rxq, volatile struct mlx5_cqe *cqe)
 {
 	uint32_t ol_flags = 0;
-	uint16_t flags = ntohs(cqe->hdr_type_etc);
+	uint16_t flags = rte_be_to_cpu_16(cqe->hdr_type_etc);
 
 	ol_flags =
 		TRANSPOSE(flags,
@@ -1841,7 +1853,7 @@
 			    MLX5_FLOW_MARK_IS_VALID(cqe->sop_drop_qpn)) {
 				pkt->ol_flags |= PKT_RX_FDIR;
 				if (cqe->sop_drop_qpn !=
-				    htonl(MLX5_FLOW_MARK_DEFAULT)) {
+				    rte_cpu_to_be_32(MLX5_FLOW_MARK_DEFAULT)) {
 					uint32_t mark = cqe->sop_drop_qpn;
 
 					pkt->ol_flags |= PKT_RX_FDIR_ID;
@@ -1853,10 +1865,11 @@
 				pkt->ol_flags |= rxq_cq_to_ol_flags(rxq, cqe);
 			if (rxq->vlan_strip &&
 			    (cqe->hdr_type_etc &
-			     htons(MLX5_CQE_VLAN_STRIPPED))) {
+			     rte_cpu_to_be_16(MLX5_CQE_VLAN_STRIPPED))) {
 				pkt->ol_flags |= PKT_RX_VLAN_PKT |
 					PKT_RX_VLAN_STRIPPED;
-				pkt->vlan_tci = ntohs(cqe->vlan_info);
+				pkt->vlan_tci =
+					rte_be_to_cpu_16(cqe->vlan_info);
 			}
 			if (rxq->crc_present)
 				len -= ETHER_CRC_LEN;
@@ -1872,7 +1885,7 @@
 		 * of the buffers are already known, only the buffer address
 		 * changes.
 		 */
-		wqe->addr = htonll(rte_pktmbuf_mtod(rep, uintptr_t));
+		wqe->addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(rep, uintptr_t));
 		if (len > DATA_LEN(seg)) {
 			len -= DATA_LEN(seg);
 			++NB_SEGS(pkt);
@@ -1900,9 +1913,9 @@
 	/* Update the consumer index. */
 	rxq->rq_ci = rq_ci >> sges_n;
 	rte_wmb();
-	*rxq->cq_db = htonl(rxq->cq_ci);
+	*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
 	rte_wmb();
-	*rxq->rq_db = htonl(rxq->rq_ci);
+	*rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci);
 #ifdef MLX5_PMD_SOFT_COUNTERS
 	/* Increment packets counter. */
 	rxq->stats.ipackets += i;
diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h
index 033e70f..73a4ce8 100644
--- a/drivers/net/mlx5/mlx5_rxtx.h
+++ b/drivers/net/mlx5/mlx5_rxtx.h
@@ -269,7 +269,7 @@ struct txq {
 		uintptr_t start; /* Start address of MR */
 		uintptr_t end; /* End address of MR */
 		struct ibv_mr *mr; /* Memory Region (for mp). */
-		uint32_t lkey; /* htonl(mr->lkey) */
+		uint32_t lkey; /* rte_cpu_to_be_32(mr->lkey) */
 	} mp2mr[MLX5_PMD_TX_MP_CACHE]; /* MP to MR translation table. */
 	uint16_t mr_cache_idx; /* Index of last hit entry. */
 	struct rte_mbuf *(*elts)[]; /* TX elements. */
@@ -492,7 +492,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	}
 #endif /* NDEBUG */
 	++cq_ci;
-	txq->wqe_pi = ntohs(cqe->wqe_counter);
+	txq->wqe_pi = rte_be_to_cpu_16(cqe->wqe_counter);
 	ctrl = (volatile struct mlx5_wqe_ctrl *)
 		tx_mlx5_wqe(txq, txq->wqe_pi);
 	elts_tail = ctrl->ctrl3;
@@ -530,7 +530,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	txq->elts_tail = elts_tail;
 	/* Update the consumer index. */
 	rte_wmb();
-	*txq->cq_db = htonl(cq_ci);
+	*txq->cq_db = rte_cpu_to_be_32(cq_ci);
 }
 
 /**
@@ -581,7 +581,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 		if (txq->mp2mr[i].start <= addr &&
 		    txq->mp2mr[i].end >= addr) {
 			assert(txq->mp2mr[i].lkey != (uint32_t)-1);
-			assert(htonl(txq->mp2mr[i].mr->lkey) ==
+			assert(rte_cpu_to_be_32(txq->mp2mr[i].mr->lkey) ==
 			       txq->mp2mr[i].lkey);
 			txq->mr_cache_idx = i;
 			return txq->mp2mr[i].lkey;
@@ -605,8 +605,8 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	uint64_t *dst = (uint64_t *)((uintptr_t)txq->bf_reg);
 	volatile uint64_t *src = ((volatile uint64_t *)wqe);
 
-	rte_io_wmb();
-	*txq->qp_db = htonl(txq->wqe_ci);
+	rte_wmb();
+	*txq->qp_db = rte_cpu_to_be_32(txq->wqe_ci);
 	/* Ensure ordering between DB record and BF copy. */
 	rte_wmb();
 	*dst = *src;
diff --git a/drivers/net/mlx5/mlx5_rxtx_vec_sse.c b/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
index 37854a7..0a5d025 100644
--- a/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
+++ b/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
@@ -291,7 +291,7 @@
 		/* Fill ESEG in the header. */
 		_mm_store_si128(t_wqe + 1,
 				_mm_set_epi16(0, 0, 0, 0,
-					      htons(len), cs_flags,
+					      rte_cpu_to_be_16(len), cs_flags,
 					      0, 0));
 		txq->wqe_ci = wqe_ci;
 	}
@@ -300,7 +300,7 @@
 	txq->elts_comp += (uint16_t)(elts_head - txq->elts_head);
 	txq->elts_head = elts_head;
 	if (txq->elts_comp >= MLX5_TX_COMP_THRESH) {
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		wqe->ctrl[3] = txq->elts_head;
 		txq->elts_comp = 0;
 		++txq->cq_pi;
@@ -561,11 +561,11 @@
 		return;
 	}
 	for (i = 0; i < n; ++i)
-		wq[i].addr = htonll((uintptr_t)elts[i]->buf_addr +
-				    RTE_PKTMBUF_HEADROOM);
+		wq[i].addr = rte_cpu_to_be_64((uintptr_t)elts[i]->buf_addr +
+					      RTE_PKTMBUF_HEADROOM);
 	rxq->rq_ci += n;
 	rte_wmb();
-	*rxq->rq_db = htonl(rxq->rq_ci);
+	*rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci);
 }
 
 /**
@@ -1248,7 +1248,7 @@
 		}
 	}
 	rte_wmb();
-	*rxq->cq_db = htonl(rxq->cq_ci);
+	*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
 	return rcvd_pkt;
 }
 
-- 
1.8.3.1

[PATCH v2 2/3] net/mlx5: fix TSO MLNX OFED 3.3 verification

[Shachar Beiser]

Fixes: 3cf87e68d97b ("net/mlx5: remove old MLNX OFED 3.3 verification")
Cc: stable@dpdk.org

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
 drivers/net/mlx5/mlx5_prm.h | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_prm.h b/drivers/net/mlx5/mlx5_prm.h
index 608072f..8b82b5e 100644
--- a/drivers/net/mlx5/mlx5_prm.h
+++ b/drivers/net/mlx5/mlx5_prm.h
@@ -89,9 +89,6 @@
 /* Default max packet length to be inlined. */
 #define MLX5_EMPW_MAX_INLINE_LEN (4U * MLX5_WQE_SIZE)
 
-#ifndef HAVE_VERBS_MLX5_OPCODE_TSO
-#define MLX5_OPCODE_TSO MLX5_OPCODE_LSO_MPW /* Compat with OFED 3.3. */
-#endif
 
 #define MLX5_OPC_MOD_ENHANCED_MPSW 0
 #define MLX5_OPCODE_ENHANCED_MPSW 0x29
-- 
1.8.3.1

[PATCH v2 3/3] net/mlx5: fix interrupt enable return value

[Shachar Beiser]

return value is sometimes returned uninitialized

Fixes: e1016cb73383 ("net/mlx5: fix Rx interrupts management")
Fixes: b18042fb8f49 ("net/mlx5: fix misplaced Rx interrupts functions")
Cc: adrien.mazarguil@6wind.com
Cc: stable@dpdk.org

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
 drivers/net/mlx5/mlx5_rxq.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index 437dc02..24887fb 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -1330,7 +1330,7 @@
 	struct priv *priv = mlx5_get_priv(dev);
 	struct rxq *rxq = (*priv->rxqs)[rx_queue_id];
 	struct rxq_ctrl *rxq_ctrl = container_of(rxq, struct rxq_ctrl, rxq);
-	int ret;
+	int ret = 0;
 
 	if (!rxq || !rxq_ctrl->channel) {
 		ret = EINVAL;
-- 
1.8.3.1

Re: [PATCH v2 3/3] net/mlx5: fix interrupt enable return value

[Adrien Mazarguil]

Hi Shachar,

On Tue, Sep 05, 2017 at 01:04:38PM +0000, Shachar Beiser wrote:
> return value is sometimes returned uninitialized
> 
> Fixes: e1016cb73383 ("net/mlx5: fix Rx interrupts management")
> Fixes: b18042fb8f49 ("net/mlx5: fix misplaced Rx interrupts functions")
> 
> Cc: adrien.mazarguil@6wind.com
> Cc: stable@dpdk.org
> 
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>

Looks like in both commits, ret is properly initialized so I'm wondering if
the fixes line is right? Did you even get a compilation error?

Otherwise, you should drop this patch from the series.

> ---
>  drivers/net/mlx5/mlx5_rxq.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
> index 437dc02..24887fb 100644
> --- a/drivers/net/mlx5/mlx5_rxq.c
> +++ b/drivers/net/mlx5/mlx5_rxq.c
> @@ -1330,7 +1330,7 @@
>  	struct priv *priv = mlx5_get_priv(dev);
>  	struct rxq *rxq = (*priv->rxqs)[rx_queue_id];
>  	struct rxq_ctrl *rxq_ctrl = container_of(rxq, struct rxq_ctrl, rxq);
> -	int ret;
> +	int ret = 0;
>  
>  	if (!rxq || !rxq_ctrl->channel) {
>  		ret = EINVAL;
> -- 
> 1.8.3.1
> 

-- 
Adrien Mazarguil
6WIND

Re: [PATCH v2 1/3] net/mlx5: replace network to host macros

[Nélio Laranjeiro]

On Tue, Sep 05, 2017 at 01:04:36PM +0000, Shachar Beiser wrote:
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

Thanks,

-- 
Nélio Laranjeiro
6WIND

Re: [PATCH v2 2/3] net/mlx5: fix TSO MLNX OFED 3.3 verification

[Nélio Laranjeiro]

On Tue, Sep 05, 2017 at 01:04:37PM +0000, Shachar Beiser wrote:
> Fixes: 3cf87e68d97b ("net/mlx5: remove old MLNX OFED 3.3 verification")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> ---
>  drivers/net/mlx5/mlx5_prm.h | 3 ---
>  1 file changed, 3 deletions(-)
> 
> diff --git a/drivers/net/mlx5/mlx5_prm.h b/drivers/net/mlx5/mlx5_prm.h
> index 608072f..8b82b5e 100644
> --- a/drivers/net/mlx5/mlx5_prm.h
> +++ b/drivers/net/mlx5/mlx5_prm.h
> @@ -89,9 +89,6 @@
>  /* Default max packet length to be inlined. */
>  #define MLX5_EMPW_MAX_INLINE_LEN (4U * MLX5_WQE_SIZE)
>  
> -#ifndef HAVE_VERBS_MLX5_OPCODE_TSO
> -#define MLX5_OPCODE_TSO MLX5_OPCODE_LSO_MPW /* Compat with OFED 3.3. */
> -#endif
>  
>  #define MLX5_OPC_MOD_ENHANCED_MPSW 0
>  #define MLX5_OPCODE_ENHANCED_MPSW 0x29
> -- 
> 1.8.3.1
> 


Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

-- 
Nélio Laranjeiro
6WIND

Re: [PATCH v2 3/3] net/mlx5: fix interrupt enable return value

[Shachar Beiser]

Hi ,

       The problem was that the value ret  was not set to 0.
        As a result we got a warning message:
	WARN("unable to arm interrupt on rx queue %d", rx_queue_id);
       The fix is ret =0;
        I was looking with Nelio for the right commits that this bug fixes I am 
pretty sure that this ret value was not initialized.

              -Shachar Beiser.

> -----Original Message-----
> From: Adrien Mazarguil [mailto:adrien.mazarguil@6wind.com]
> Sent: Tuesday, September 5, 2017 4:27 PM
> To: Shachar Beiser <shacharbe@mellanox.com>
> Cc: dev@dpdk.org; Nélio Laranjeiro <nelio.laranjeiro@6wind.com>;
> stable@dpdk.org
> Subject: Re: [PATCH v2 3/3] net/mlx5: fix interrupt enable return value
> 
> Hi Shachar,
> 
> On Tue, Sep 05, 2017 at 01:04:38PM +0000, Shachar Beiser wrote:
> > return value is sometimes returned uninitialized
> >
> > Fixes: e1016cb73383 ("net/mlx5: fix Rx interrupts management")
> > Fixes: b18042fb8f49 ("net/mlx5: fix misplaced Rx interrupts
> > functions")
> >
> > Cc: adrien.mazarguil@6wind.com
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> 
> Looks like in both commits, ret is properly initialized so I'm wondering if the
> fixes line is right? Did you even get a compilation error?
> 
> Otherwise, you should drop this patch from the series.
> 
> > ---
> >  drivers/net/mlx5/mlx5_rxq.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
> > index 437dc02..24887fb 100644
> > --- a/drivers/net/mlx5/mlx5_rxq.c
> > +++ b/drivers/net/mlx5/mlx5_rxq.c
> > @@ -1330,7 +1330,7 @@
> >  	struct priv *priv = mlx5_get_priv(dev);
> >  	struct rxq *rxq = (*priv->rxqs)[rx_queue_id];
> >  	struct rxq_ctrl *rxq_ctrl = container_of(rxq, struct rxq_ctrl, rxq);
> > -	int ret;
> > +	int ret = 0;
> >
> >  	if (!rxq || !rxq_ctrl->channel) {
> >  		ret = EINVAL;
> > --
> > 1.8.3.1
> >
> 
> --
> Adrien Mazarguil
> 6WIND

[PATCH v3 1/2] net/mlx5: replace network to host macros

[Shachar Beiser]

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
There are only 2 patches left since 
I have squashed :
[PATCH v2 3/3] “net/mlx5: fix interrupt enable return”
http://dpdk.org/dev/patchwork/patch/28380/
into [dpdk-dev,v5] net/mlx5: support upstream rdma-core
---
 drivers/net/mlx5/mlx5_mac.c          |   8 ++-
 drivers/net/mlx5/mlx5_mr.c           |   2 +-
 drivers/net/mlx5/mlx5_rxmode.c       |   8 ++-
 drivers/net/mlx5/mlx5_rxq.c          |   9 +--
 drivers/net/mlx5/mlx5_rxtx.c         | 131 +++++++++++++++++++----------------
 drivers/net/mlx5/mlx5_rxtx.h         |  12 ++--
 drivers/net/mlx5/mlx5_rxtx_vec_sse.c |  12 ++--
 7 files changed, 102 insertions(+), 80 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_mac.c b/drivers/net/mlx5/mlx5_mac.c
index 45d23e4..b3c3fa2 100644
--- a/drivers/net/mlx5/mlx5_mac.c
+++ b/drivers/net/mlx5/mlx5_mac.c
@@ -263,11 +263,15 @@
 				(*mac)[0], (*mac)[1], (*mac)[2],
 				(*mac)[3], (*mac)[4], (*mac)[5]
 			},
-			.vlan_tag = (vlan_enabled ? htons(vlan_id) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(vlan_id)
+				     : 0),
 		},
 		.mask = {
 			.dst_mac = "\xff\xff\xff\xff\xff\xff",
-			.vlan_tag = (vlan_enabled ? htons(0xfff) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(0xfff) :
+				     0),
 		},
 	};
 	DEBUG("%p: adding MAC address %02x:%02x:%02x:%02x:%02x:%02x index %u"
diff --git a/drivers/net/mlx5/mlx5_mr.c b/drivers/net/mlx5/mlx5_mr.c
index 9593830..9a9f73a 100644
--- a/drivers/net/mlx5/mlx5_mr.c
+++ b/drivers/net/mlx5/mlx5_mr.c
@@ -203,7 +203,7 @@ struct ibv_mr *
 	txq_ctrl->txq.mp2mr[idx].start = (uintptr_t)mr->addr;
 	txq_ctrl->txq.mp2mr[idx].end = (uintptr_t)mr->addr + mr->length;
 	txq_ctrl->txq.mp2mr[idx].mr = mr;
-	txq_ctrl->txq.mp2mr[idx].lkey = htonl(mr->lkey);
+	txq_ctrl->txq.mp2mr[idx].lkey = rte_cpu_to_be_32(mr->lkey);
 	DEBUG("%p: new MR lkey for MP \"%s\" (%p): 0x%08" PRIu32,
 	      (void *)txq_ctrl, mp->name, (void *)mp,
 	      txq_ctrl->txq.mp2mr[idx].lkey);
diff --git a/drivers/net/mlx5/mlx5_rxmode.c b/drivers/net/mlx5/mlx5_rxmode.c
index 4a51e47..db2e05b 100644
--- a/drivers/net/mlx5/mlx5_rxmode.c
+++ b/drivers/net/mlx5/mlx5_rxmode.c
@@ -159,14 +159,18 @@
 				mac[0], mac[1], mac[2],
 				mac[3], mac[4], mac[5],
 			},
-			.vlan_tag = (vlan_enabled ? htons(vlan_id) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(vlan_id) :
+				     0),
 		},
 		.mask = {
 			.dst_mac = {
 				mask[0], mask[1], mask[2],
 				mask[3], mask[4], mask[5],
 			},
-			.vlan_tag = (vlan_enabled ? htons(0xfff) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(0xfff) :
+				     0),
 		},
 	};
 
diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index 35c5cb4..437dc02 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -672,9 +672,10 @@
 		/* scat->addr must be able to store a pointer. */
 		assert(sizeof(scat->addr) >= sizeof(uintptr_t));
 		*scat = (struct mlx5_wqe_data_seg){
-			.addr = htonll(rte_pktmbuf_mtod(buf, uintptr_t)),
-			.byte_count = htonl(DATA_LEN(buf)),
-			.lkey = htonl(rxq_ctrl->mr->lkey),
+			.addr =
+			    rte_cpu_to_be_64(rte_pktmbuf_mtod(buf, uintptr_t)),
+			.byte_count = rte_cpu_to_be_32(DATA_LEN(buf)),
+			.lkey = rte_cpu_to_be_32(rxq_ctrl->mr->lkey),
 		};
 		(*rxq_ctrl->rxq.elts)[i] = buf;
 	}
@@ -1077,7 +1078,7 @@
 	/* Update doorbell counter. */
 	rxq_ctrl->rxq.rq_ci = desc >> rxq_ctrl->rxq.sges_n;
 	rte_wmb();
-	*rxq_ctrl->rxq.rq_db = htonl(rxq_ctrl->rxq.rq_ci);
+	*rxq_ctrl->rxq.rq_db = rte_cpu_to_be_32(rxq_ctrl->rxq.rq_ci);
 	DEBUG("%p: rxq updated with %p", (void *)rxq_ctrl, (void *)&tmpl);
 	assert(ret == 0);
 	return 0;
diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index fe9e7ea..e1a35a3 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -306,7 +306,7 @@
 
 		op_own = cqe->op_own;
 		if (MLX5_CQE_FORMAT(op_own) == MLX5_COMPRESSED)
-			n = ntohl(cqe->byte_cnt);
+			n = rte_be_to_cpu_32(cqe->byte_cnt);
 		else
 			n = 1;
 		cq_ci += n;
@@ -434,7 +434,8 @@
 		raw = ((uint8_t *)(uintptr_t)wqe) + 2 * MLX5_WQE_DWORD_SIZE;
 		/* Replace the Ethernet type by the VLAN if necessary. */
 		if (buf->ol_flags & PKT_TX_VLAN_PKT) {
-			uint32_t vlan = htonl(0x81000000 | buf->vlan_tci);
+			uint32_t vlan = rte_cpu_to_be_32(0x81000000 |
+							 buf->vlan_tci);
 			unsigned int len = 2 * ETHER_ADDR_LEN - 2;
 
 			addr += 2;
@@ -510,8 +511,10 @@
 				} else {
 					/* NOP WQE. */
 					wqe->ctrl = (rte_v128u32_t){
-						     htonl(txq->wqe_ci << 8),
-						     htonl(txq->qp_num_8s | 1),
+						     rte_cpu_to_be_32(
+							txq->wqe_ci << 8),
+						     rte_cpu_to_be_32(
+							txq->qp_num_8s | 1),
 						     0,
 						     0,
 					};
@@ -550,7 +553,8 @@
 				max_wqe -= n;
 				if (tso) {
 					uint32_t inl =
-						htonl(copy_b | MLX5_INLINE_SEG);
+					rte_cpu_to_be_32(copy_b |
+							 MLX5_INLINE_SEG);
 
 					pkt_inline_sz =
 						MLX5_WQE_DS(tso_header_sz) *
@@ -603,9 +607,9 @@
 			ds = 3;
 use_dseg:
 			/* Add the remaining packet as a simple ds. */
-			naddr = htonll(addr);
+			naddr = rte_cpu_to_be_64(addr);
 			*dseg = (rte_v128u32_t){
-				htonl(length),
+				rte_cpu_to_be_32(length),
 				mlx5_tx_mb2mr(txq, buf),
 				naddr,
 				naddr >> 32,
@@ -642,9 +646,9 @@
 		total_length += length;
 #endif
 		/* Store segment information. */
-		naddr = htonll(rte_pktmbuf_mtod(buf, uintptr_t));
+		naddr = rte_cpu_to_be_64(rte_pktmbuf_mtod(buf, uintptr_t));
 		*dseg = (rte_v128u32_t){
-			htonl(length),
+			rte_cpu_to_be_32(length),
 			mlx5_tx_mb2mr(txq, buf),
 			naddr,
 			naddr >> 32,
@@ -663,21 +667,23 @@
 		/* Initialize known and common part of the WQE structure. */
 		if (tso) {
 			wqe->ctrl = (rte_v128u32_t){
-				htonl((txq->wqe_ci << 8) | MLX5_OPCODE_TSO),
-				htonl(txq->qp_num_8s | ds),
+				rte_cpu_to_be_32((txq->wqe_ci << 8) |
+						 MLX5_OPCODE_TSO),
+				rte_cpu_to_be_32(txq->qp_num_8s | ds),
 				0,
 				0,
 			};
 			wqe->eseg = (rte_v128u32_t){
 				0,
-				cs_flags | (htons(tso_segsz) << 16),
+				cs_flags | (rte_cpu_to_be_16(tso_segsz) << 16),
 				0,
-				(ehdr << 16) | htons(tso_header_sz),
+				(ehdr << 16) | rte_cpu_to_be_16(tso_header_sz),
 			};
 		} else {
 			wqe->ctrl = (rte_v128u32_t){
-				htonl((txq->wqe_ci << 8) | MLX5_OPCODE_SEND),
-				htonl(txq->qp_num_8s | ds),
+				rte_cpu_to_be_32((txq->wqe_ci << 8) |
+						 MLX5_OPCODE_SEND),
+				rte_cpu_to_be_32(txq->qp_num_8s | ds),
 				0,
 				0,
 			};
@@ -685,7 +691,7 @@
 				0,
 				cs_flags,
 				0,
-				(ehdr << 16) | htons(pkt_inline_sz),
+				(ehdr << 16) | rte_cpu_to_be_16(pkt_inline_sz),
 			};
 		}
 next_wqe:
@@ -705,7 +711,7 @@
 	comp = txq->elts_comp + i + j + k;
 	if (comp >= MLX5_TX_COMP_THRESH) {
 		/* Request completion on last WQE. */
-		last_wqe->ctrl2 = htonl(8);
+		last_wqe->ctrl2 = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		last_wqe->ctrl3 = txq->elts_head;
 		txq->elts_comp = 0;
@@ -744,13 +750,14 @@
 	mpw->len = length;
 	mpw->total_len = 0;
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->eseg.mss = htons(length);
+	mpw->wqe->eseg.mss = rte_cpu_to_be_16(length);
 	mpw->wqe->eseg.inline_hdr_sz = 0;
 	mpw->wqe->eseg.rsvd0 = 0;
 	mpw->wqe->eseg.rsvd1 = 0;
 	mpw->wqe->eseg.rsvd2 = 0;
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_MPW << 24) |
-				  (txq->wqe_ci << 8) | MLX5_OPCODE_TSO);
+	mpw->wqe->ctrl[0] = rte_cpu_to_be_32((MLX5_OPC_MOD_MPW << 24) |
+					     (txq->wqe_ci << 8) |
+					     MLX5_OPCODE_TSO);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
 	mpw->data.dseg[0] = (volatile struct mlx5_wqe_data_seg *)
@@ -779,7 +786,7 @@
 	 * Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | (2 + num));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s | (2 + num));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
 	if (num < 3)
 		++txq->wqe_ci;
@@ -886,9 +893,9 @@
 			dseg = mpw.data.dseg[mpw.pkts_n];
 			addr = rte_pktmbuf_mtod(buf, uintptr_t);
 			*dseg = (struct mlx5_wqe_data_seg){
-				.byte_count = htonl(DATA_LEN(buf)),
+				.byte_count = rte_cpu_to_be_32(DATA_LEN(buf)),
 				.lkey = mlx5_tx_mb2mr(txq, buf),
-				.addr = htonll(addr),
+				.addr = rte_cpu_to_be_64(addr),
 			};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 			length += DATA_LEN(buf);
@@ -916,7 +923,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -956,12 +963,12 @@
 	mpw->len = length;
 	mpw->total_len = 0;
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_MPW << 24) |
-				  (txq->wqe_ci << 8) |
-				  MLX5_OPCODE_TSO);
+	mpw->wqe->ctrl[0] = rte_cpu_to_be_32((MLX5_OPC_MOD_MPW << 24) |
+					     (txq->wqe_ci << 8) |
+					     MLX5_OPCODE_TSO);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
-	mpw->wqe->eseg.mss = htons(length);
+	mpw->wqe->eseg.mss = rte_cpu_to_be_16(length);
 	mpw->wqe->eseg.inline_hdr_sz = 0;
 	mpw->wqe->eseg.cs_flags = 0;
 	mpw->wqe->eseg.rsvd0 = 0;
@@ -992,9 +999,10 @@
 	 * Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | MLX5_WQE_DS(size));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s |
+					     MLX5_WQE_DS(size));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
-	inl->byte_cnt = htonl(mpw->total_len | MLX5_INLINE_SEG);
+	inl->byte_cnt = rte_cpu_to_be_32(mpw->total_len | MLX5_INLINE_SEG);
 	txq->wqe_ci += (size + (MLX5_WQE_SIZE - 1)) / MLX5_WQE_SIZE;
 }
 
@@ -1132,9 +1140,10 @@
 				dseg = mpw.data.dseg[mpw.pkts_n];
 				addr = rte_pktmbuf_mtod(buf, uintptr_t);
 				*dseg = (struct mlx5_wqe_data_seg){
-					.byte_count = htonl(DATA_LEN(buf)),
+					.byte_count =
+					       rte_cpu_to_be_32(DATA_LEN(buf)),
 					.lkey = mlx5_tx_mb2mr(txq, buf),
-					.addr = htonll(addr),
+					.addr = rte_cpu_to_be_64(addr),
 				};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 				length += DATA_LEN(buf);
@@ -1206,7 +1215,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -1246,9 +1255,10 @@
 	mpw->pkts_n = 0;
 	mpw->total_len = sizeof(struct mlx5_wqe);
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_ENHANCED_MPSW << 24) |
-				  (txq->wqe_ci << 8) |
-				  MLX5_OPCODE_ENHANCED_MPSW);
+	mpw->wqe->ctrl[0] =
+		rte_cpu_to_be_32((MLX5_OPC_MOD_ENHANCED_MPSW << 24) |
+				 (txq->wqe_ci << 8) |
+				 MLX5_OPCODE_ENHANCED_MPSW);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
 	memset((void *)(uintptr_t)&mpw->wqe->eseg, 0, MLX5_WQE_DWORD_SIZE);
@@ -1256,9 +1266,9 @@
 		uintptr_t addr = (uintptr_t)(mpw->wqe + 1);
 
 		/* Pad the first 2 DWORDs with zero-length inline header. */
-		*(volatile uint32_t *)addr = htonl(MLX5_INLINE_SEG);
+		*(volatile uint32_t *)addr = rte_cpu_to_be_32(MLX5_INLINE_SEG);
 		*(volatile uint32_t *)(addr + MLX5_WQE_DWORD_SIZE) =
-			htonl(MLX5_INLINE_SEG);
+			rte_cpu_to_be_32(MLX5_INLINE_SEG);
 		mpw->total_len += 2 * MLX5_WQE_DWORD_SIZE;
 		/* Start from the next WQEBB. */
 		mpw->data.raw = (volatile void *)(tx_mlx5_wqe(txq, idx + 1));
@@ -1286,7 +1296,8 @@
 	/* Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | MLX5_WQE_DS(mpw->total_len));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s |
+					     MLX5_WQE_DS(mpw->total_len));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
 	ret = (mpw->total_len + (MLX5_WQE_SIZE - 1)) / MLX5_WQE_SIZE;
 	txq->wqe_ci += ret;
@@ -1439,9 +1450,10 @@
 				dseg = mpw.data.dseg[mpw.pkts_n];
 				addr = rte_pktmbuf_mtod(buf, uintptr_t);
 				*dseg = (struct mlx5_wqe_data_seg){
-					.byte_count = htonl(DATA_LEN(buf)),
+					.byte_count = rte_cpu_to_be_32(
+								DATA_LEN(buf)),
 					.lkey = mlx5_tx_mb2mr(txq, buf),
-					.addr = htonll(addr),
+					.addr = rte_cpu_to_be_64(addr),
 				};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 				length += DATA_LEN(buf);
@@ -1464,7 +1476,7 @@
 
 			assert(mpw.state == MLX5_MPW_ENHANCED_STATE_OPENED);
 			assert(length == DATA_LEN(buf));
-			inl_hdr = htonl(length | MLX5_INLINE_SEG);
+			inl_hdr = rte_cpu_to_be_32(length | MLX5_INLINE_SEG);
 			addr = rte_pktmbuf_mtod(buf, uintptr_t);
 			mpw.data.raw = (volatile void *)
 				((uintptr_t)mpw.data.raw + inl_pad);
@@ -1520,9 +1532,9 @@
 			for (n = 0; n * RTE_CACHE_LINE_SIZE < length; n++)
 				rte_prefetch2((void *)(addr +
 						n * RTE_CACHE_LINE_SIZE));
-			naddr = htonll(addr);
+			naddr = rte_cpu_to_be_64(addr);
 			*dseg = (rte_v128u32_t) {
-				htonl(length),
+				rte_cpu_to_be_32(length),
 				mlx5_tx_mb2mr(txq, buf),
 				naddr,
 				naddr >> 32,
@@ -1550,7 +1562,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -1634,8 +1646,8 @@
 			(volatile struct mlx5_mini_cqe8 (*)[8])
 			(uintptr_t)(&(*rxq->cqes)[zip->ca & cqe_cnt].pkt_info);
 
-		len = ntohl((*mc)[zip->ai & 7].byte_cnt);
-		*rss_hash = ntohl((*mc)[zip->ai & 7].rx_hash_result);
+		len = rte_be_to_cpu_32((*mc)[zip->ai & 7].byte_cnt);
+		*rss_hash = rte_be_to_cpu_32((*mc)[zip->ai & 7].rx_hash_result);
 		if ((++zip->ai & 7) == 0) {
 			/* Invalidate consumed CQEs */
 			idx = zip->ca;
@@ -1683,7 +1695,7 @@
 							  cqe_cnt].pkt_info);
 
 			/* Fix endianness. */
-			zip->cqe_cnt = ntohl(cqe->byte_cnt);
+			zip->cqe_cnt = rte_be_to_cpu_32(cqe->byte_cnt);
 			/*
 			 * Current mini array position is the one returned by
 			 * check_cqe64().
@@ -1698,8 +1710,8 @@
 			--rxq->cq_ci;
 			zip->cq_ci = rxq->cq_ci + zip->cqe_cnt;
 			/* Get packet size to return. */
-			len = ntohl((*mc)[0].byte_cnt);
-			*rss_hash = ntohl((*mc)[0].rx_hash_result);
+			len = rte_be_to_cpu_32((*mc)[0].byte_cnt);
+			*rss_hash = rte_be_to_cpu_32((*mc)[0].rx_hash_result);
 			zip->ai = 1;
 			/* Prefetch all the entries to be invalidated */
 			idx = zip->ca;
@@ -1709,8 +1721,8 @@
 				++idx;
 			}
 		} else {
-			len = ntohl(cqe->byte_cnt);
-			*rss_hash = ntohl(cqe->rx_hash_res);
+			len = rte_be_to_cpu_32(cqe->byte_cnt);
+			*rss_hash = rte_be_to_cpu_32(cqe->rx_hash_res);
 		}
 		/* Error while receiving packet. */
 		if (unlikely(MLX5_CQE_OPCODE(op_own) == MLX5_CQE_RESP_ERR))
@@ -1734,7 +1746,7 @@
 rxq_cq_to_ol_flags(struct rxq *rxq, volatile struct mlx5_cqe *cqe)
 {
 	uint32_t ol_flags = 0;
-	uint16_t flags = ntohs(cqe->hdr_type_etc);
+	uint16_t flags = rte_be_to_cpu_16(cqe->hdr_type_etc);
 
 	ol_flags =
 		TRANSPOSE(flags,
@@ -1841,7 +1853,7 @@
 			    MLX5_FLOW_MARK_IS_VALID(cqe->sop_drop_qpn)) {
 				pkt->ol_flags |= PKT_RX_FDIR;
 				if (cqe->sop_drop_qpn !=
-				    htonl(MLX5_FLOW_MARK_DEFAULT)) {
+				    rte_cpu_to_be_32(MLX5_FLOW_MARK_DEFAULT)) {
 					uint32_t mark = cqe->sop_drop_qpn;
 
 					pkt->ol_flags |= PKT_RX_FDIR_ID;
@@ -1853,10 +1865,11 @@
 				pkt->ol_flags |= rxq_cq_to_ol_flags(rxq, cqe);
 			if (rxq->vlan_strip &&
 			    (cqe->hdr_type_etc &
-			     htons(MLX5_CQE_VLAN_STRIPPED))) {
+			     rte_cpu_to_be_16(MLX5_CQE_VLAN_STRIPPED))) {
 				pkt->ol_flags |= PKT_RX_VLAN_PKT |
 					PKT_RX_VLAN_STRIPPED;
-				pkt->vlan_tci = ntohs(cqe->vlan_info);
+				pkt->vlan_tci =
+					rte_be_to_cpu_16(cqe->vlan_info);
 			}
 			if (rxq->crc_present)
 				len -= ETHER_CRC_LEN;
@@ -1872,7 +1885,7 @@
 		 * of the buffers are already known, only the buffer address
 		 * changes.
 		 */
-		wqe->addr = htonll(rte_pktmbuf_mtod(rep, uintptr_t));
+		wqe->addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(rep, uintptr_t));
 		if (len > DATA_LEN(seg)) {
 			len -= DATA_LEN(seg);
 			++NB_SEGS(pkt);
@@ -1900,9 +1913,9 @@
 	/* Update the consumer index. */
 	rxq->rq_ci = rq_ci >> sges_n;
 	rte_wmb();
-	*rxq->cq_db = htonl(rxq->cq_ci);
+	*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
 	rte_wmb();
-	*rxq->rq_db = htonl(rxq->rq_ci);
+	*rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci);
 #ifdef MLX5_PMD_SOFT_COUNTERS
 	/* Increment packets counter. */
 	rxq->stats.ipackets += i;
diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h
index 033e70f..73a4ce8 100644
--- a/drivers/net/mlx5/mlx5_rxtx.h
+++ b/drivers/net/mlx5/mlx5_rxtx.h
@@ -269,7 +269,7 @@ struct txq {
 		uintptr_t start; /* Start address of MR */
 		uintptr_t end; /* End address of MR */
 		struct ibv_mr *mr; /* Memory Region (for mp). */
-		uint32_t lkey; /* htonl(mr->lkey) */
+		uint32_t lkey; /* rte_cpu_to_be_32(mr->lkey) */
 	} mp2mr[MLX5_PMD_TX_MP_CACHE]; /* MP to MR translation table. */
 	uint16_t mr_cache_idx; /* Index of last hit entry. */
 	struct rte_mbuf *(*elts)[]; /* TX elements. */
@@ -492,7 +492,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	}
 #endif /* NDEBUG */
 	++cq_ci;
-	txq->wqe_pi = ntohs(cqe->wqe_counter);
+	txq->wqe_pi = rte_be_to_cpu_16(cqe->wqe_counter);
 	ctrl = (volatile struct mlx5_wqe_ctrl *)
 		tx_mlx5_wqe(txq, txq->wqe_pi);
 	elts_tail = ctrl->ctrl3;
@@ -530,7 +530,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	txq->elts_tail = elts_tail;
 	/* Update the consumer index. */
 	rte_wmb();
-	*txq->cq_db = htonl(cq_ci);
+	*txq->cq_db = rte_cpu_to_be_32(cq_ci);
 }
 
 /**
@@ -581,7 +581,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 		if (txq->mp2mr[i].start <= addr &&
 		    txq->mp2mr[i].end >= addr) {
 			assert(txq->mp2mr[i].lkey != (uint32_t)-1);
-			assert(htonl(txq->mp2mr[i].mr->lkey) ==
+			assert(rte_cpu_to_be_32(txq->mp2mr[i].mr->lkey) ==
 			       txq->mp2mr[i].lkey);
 			txq->mr_cache_idx = i;
 			return txq->mp2mr[i].lkey;
@@ -605,8 +605,8 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	uint64_t *dst = (uint64_t *)((uintptr_t)txq->bf_reg);
 	volatile uint64_t *src = ((volatile uint64_t *)wqe);
 
-	rte_io_wmb();
-	*txq->qp_db = htonl(txq->wqe_ci);
+	rte_wmb();
+	*txq->qp_db = rte_cpu_to_be_32(txq->wqe_ci);
 	/* Ensure ordering between DB record and BF copy. */
 	rte_wmb();
 	*dst = *src;
diff --git a/drivers/net/mlx5/mlx5_rxtx_vec_sse.c b/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
index 37854a7..0a5d025 100644
--- a/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
+++ b/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
@@ -291,7 +291,7 @@
 		/* Fill ESEG in the header. */
 		_mm_store_si128(t_wqe + 1,
 				_mm_set_epi16(0, 0, 0, 0,
-					      htons(len), cs_flags,
+					      rte_cpu_to_be_16(len), cs_flags,
 					      0, 0));
 		txq->wqe_ci = wqe_ci;
 	}
@@ -300,7 +300,7 @@
 	txq->elts_comp += (uint16_t)(elts_head - txq->elts_head);
 	txq->elts_head = elts_head;
 	if (txq->elts_comp >= MLX5_TX_COMP_THRESH) {
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		wqe->ctrl[3] = txq->elts_head;
 		txq->elts_comp = 0;
 		++txq->cq_pi;
@@ -561,11 +561,11 @@
 		return;
 	}
 	for (i = 0; i < n; ++i)
-		wq[i].addr = htonll((uintptr_t)elts[i]->buf_addr +
-				    RTE_PKTMBUF_HEADROOM);
+		wq[i].addr = rte_cpu_to_be_64((uintptr_t)elts[i]->buf_addr +
+					      RTE_PKTMBUF_HEADROOM);
 	rxq->rq_ci += n;
 	rte_wmb();
-	*rxq->rq_db = htonl(rxq->rq_ci);
+	*rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci);
 }
 
 /**
@@ -1248,7 +1248,7 @@
 		}
 	}
 	rte_wmb();
-	*rxq->cq_db = htonl(rxq->cq_ci);
+	*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
 	return rcvd_pkt;
 }
 
-- 
1.8.3.1

[PATCH v3 2/2] net/mlx5: fix TSO MLNX OFED 3.3 verification

[Shachar Beiser]

Fixes: 3cf87e68d97b ("net/mlx5: remove old MLNX OFED 3.3 verification")
Cc: stable@dpdk.org

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
There are only 2 patches left since
I have squashed :
[PATCH v2 3/3] “net/mlx5: fix interrupt enable return”
http://dpdk.org/dev/patchwork/patch/28380/
into [dpdk-dev,v5] net/mlx5: support upstream rdma-core
---
 drivers/net/mlx5/mlx5_prm.h | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_prm.h b/drivers/net/mlx5/mlx5_prm.h
index 608072f..8b82b5e 100644
--- a/drivers/net/mlx5/mlx5_prm.h
+++ b/drivers/net/mlx5/mlx5_prm.h
@@ -89,9 +89,6 @@
 /* Default max packet length to be inlined. */
 #define MLX5_EMPW_MAX_INLINE_LEN (4U * MLX5_WQE_SIZE)
 
-#ifndef HAVE_VERBS_MLX5_OPCODE_TSO
-#define MLX5_OPCODE_TSO MLX5_OPCODE_LSO_MPW /* Compat with OFED 3.3. */
-#endif
 
 #define MLX5_OPC_MOD_ENHANCED_MPSW 0
 #define MLX5_OPCODE_ENHANCED_MPSW 0x29
-- 
1.8.3.1

Re: [PATCH v3 1/2] net/mlx5: replace network to host macros

[Yongseok Koh]

> On Sep 14, 2017, at 6:43 AM, Shachar Beiser <shacharbe@mellanox.com> wrote:
> 
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> ---
[...]
> @@ -550,7 +553,8 @@
> 				max_wqe -= n;
> 				if (tso) {
> 					uint32_t inl =
> -						htonl(copy_b | MLX5_INLINE_SEG);
> +					rte_cpu_to_be_32(copy_b |
> +							 MLX5_INLINE_SEG);
Wrong indentation.

[...]
> @@ -605,8 +605,8 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
> 	uint64_t *dst = (uint64_t *)((uintptr_t)txq->bf_reg);
> 	volatile uint64_t *src = ((volatile uint64_t *)wqe);
> 
> -	rte_io_wmb();
> -	*txq->qp_db = htonl(txq->wqe_ci);
> +	rte_wmb();
Look like a mistake when rebasing. This should not be touched by this patch.

> +	*txq->qp_db = rte_cpu_to_be_32(txq->wqe_ci);
> 	/* Ensure ordering between DB record and BF copy. */
> 	rte_wmb();
> 	*dst = *src;

--
Thanks,
Yongseok

[PATCH v4 1/2] net/mlx5: replace network to host macros

[Shachar Beiser]

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
I have rebased [PATCH v3 1/2] net/mlx5: replace network to host macros
I have fixed a rebase conflict in mlx5_rxtx.h in line 609
-rte_wmb();
+rte_io_wmb();

---
 drivers/net/mlx5/mlx5_mac.c          |   8 +-
 drivers/net/mlx5/mlx5_mr.c           |   2 +-
 drivers/net/mlx5/mlx5_rxmode.c       |   8 +-
 drivers/net/mlx5/mlx5_rxq.c          |   9 ++-
 drivers/net/mlx5/mlx5_rxtx.c         | 137 ++++++++++++++++++++---------------
 drivers/net/mlx5/mlx5_rxtx.h         |  10 +--
 drivers/net/mlx5/mlx5_rxtx_vec_sse.c |  12 +--
 7 files changed, 107 insertions(+), 79 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_mac.c b/drivers/net/mlx5/mlx5_mac.c
index 45d23e4..b3c3fa2 100644
--- a/drivers/net/mlx5/mlx5_mac.c
+++ b/drivers/net/mlx5/mlx5_mac.c
@@ -263,11 +263,15 @@
 				(*mac)[0], (*mac)[1], (*mac)[2],
 				(*mac)[3], (*mac)[4], (*mac)[5]
 			},
-			.vlan_tag = (vlan_enabled ? htons(vlan_id) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(vlan_id)
+				     : 0),
 		},
 		.mask = {
 			.dst_mac = "\xff\xff\xff\xff\xff\xff",
-			.vlan_tag = (vlan_enabled ? htons(0xfff) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(0xfff) :
+				     0),
 		},
 	};
 	DEBUG("%p: adding MAC address %02x:%02x:%02x:%02x:%02x:%02x index %u"
diff --git a/drivers/net/mlx5/mlx5_mr.c b/drivers/net/mlx5/mlx5_mr.c
index 9593830..9a9f73a 100644
--- a/drivers/net/mlx5/mlx5_mr.c
+++ b/drivers/net/mlx5/mlx5_mr.c
@@ -203,7 +203,7 @@ struct ibv_mr *
 	txq_ctrl->txq.mp2mr[idx].start = (uintptr_t)mr->addr;
 	txq_ctrl->txq.mp2mr[idx].end = (uintptr_t)mr->addr + mr->length;
 	txq_ctrl->txq.mp2mr[idx].mr = mr;
-	txq_ctrl->txq.mp2mr[idx].lkey = htonl(mr->lkey);
+	txq_ctrl->txq.mp2mr[idx].lkey = rte_cpu_to_be_32(mr->lkey);
 	DEBUG("%p: new MR lkey for MP \"%s\" (%p): 0x%08" PRIu32,
 	      (void *)txq_ctrl, mp->name, (void *)mp,
 	      txq_ctrl->txq.mp2mr[idx].lkey);
diff --git a/drivers/net/mlx5/mlx5_rxmode.c b/drivers/net/mlx5/mlx5_rxmode.c
index 4a51e47..db2e05b 100644
--- a/drivers/net/mlx5/mlx5_rxmode.c
+++ b/drivers/net/mlx5/mlx5_rxmode.c
@@ -159,14 +159,18 @@
 				mac[0], mac[1], mac[2],
 				mac[3], mac[4], mac[5],
 			},
-			.vlan_tag = (vlan_enabled ? htons(vlan_id) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(vlan_id) :
+				     0),
 		},
 		.mask = {
 			.dst_mac = {
 				mask[0], mask[1], mask[2],
 				mask[3], mask[4], mask[5],
 			},
-			.vlan_tag = (vlan_enabled ? htons(0xfff) : 0),
+			.vlan_tag = (vlan_enabled ?
+				     rte_cpu_to_be_16(0xfff) :
+				     0),
 		},
 	};
 
diff --git a/drivers/net/mlx5/mlx5_rxq.c b/drivers/net/mlx5/mlx5_rxq.c
index 35c5cb4..437dc02 100644
--- a/drivers/net/mlx5/mlx5_rxq.c
+++ b/drivers/net/mlx5/mlx5_rxq.c
@@ -672,9 +672,10 @@
 		/* scat->addr must be able to store a pointer. */
 		assert(sizeof(scat->addr) >= sizeof(uintptr_t));
 		*scat = (struct mlx5_wqe_data_seg){
-			.addr = htonll(rte_pktmbuf_mtod(buf, uintptr_t)),
-			.byte_count = htonl(DATA_LEN(buf)),
-			.lkey = htonl(rxq_ctrl->mr->lkey),
+			.addr =
+			    rte_cpu_to_be_64(rte_pktmbuf_mtod(buf, uintptr_t)),
+			.byte_count = rte_cpu_to_be_32(DATA_LEN(buf)),
+			.lkey = rte_cpu_to_be_32(rxq_ctrl->mr->lkey),
 		};
 		(*rxq_ctrl->rxq.elts)[i] = buf;
 	}
@@ -1077,7 +1078,7 @@
 	/* Update doorbell counter. */
 	rxq_ctrl->rxq.rq_ci = desc >> rxq_ctrl->rxq.sges_n;
 	rte_wmb();
-	*rxq_ctrl->rxq.rq_db = htonl(rxq_ctrl->rxq.rq_ci);
+	*rxq_ctrl->rxq.rq_db = rte_cpu_to_be_32(rxq_ctrl->rxq.rq_ci);
 	DEBUG("%p: rxq updated with %p", (void *)rxq_ctrl, (void *)&tmpl);
 	assert(ret == 0);
 	return 0;
diff --git a/drivers/net/mlx5/mlx5_rxtx.c b/drivers/net/mlx5/mlx5_rxtx.c
index bc1f85c..3f1e2f4 100644
--- a/drivers/net/mlx5/mlx5_rxtx.c
+++ b/drivers/net/mlx5/mlx5_rxtx.c
@@ -306,7 +306,7 @@
 
 		op_own = cqe->op_own;
 		if (MLX5_CQE_FORMAT(op_own) == MLX5_COMPRESSED)
-			n = ntohl(cqe->byte_cnt);
+			n = rte_be_to_cpu_32(cqe->byte_cnt);
 		else
 			n = 1;
 		cq_ci += n;
@@ -436,7 +436,8 @@
 		raw = ((uint8_t *)(uintptr_t)wqe) + 2 * MLX5_WQE_DWORD_SIZE;
 		/* Replace the Ethernet type by the VLAN if necessary. */
 		if (buf->ol_flags & PKT_TX_VLAN_PKT) {
-			uint32_t vlan = htonl(0x81000000 | buf->vlan_tci);
+			uint32_t vlan = rte_cpu_to_be_32(0x81000000 |
+							 buf->vlan_tci);
 			unsigned int len = 2 * ETHER_ADDR_LEN - 2;
 
 			addr += 2;
@@ -511,8 +512,10 @@
 				} else {
 					/* NOP WQE. */
 					wqe->ctrl = (rte_v128u32_t){
-						     htonl(txq->wqe_ci << 8),
-						     htonl(txq->qp_num_8s | 1),
+						     rte_cpu_to_be_32(
+							txq->wqe_ci << 8),
+						     rte_cpu_to_be_32(
+							txq->qp_num_8s | 1),
 						     0,
 						     0,
 					};
@@ -551,7 +554,14 @@
 					break;
 				max_wqe -= n;
 				if (tso) {
-					inl = htonl(copy_b | MLX5_INLINE_SEG);
+					uint32_t inl =
+					rte_cpu_to_be_32(copy_b |
+							 MLX5_INLINE_SEG);
+
+					pkt_inline_sz =
+						MLX5_WQE_DS(tso_header_sz) *
+						MLX5_WQE_DWORD_SIZE;
+
 					rte_memcpy((void *)raw,
 						   (void *)&inl, sizeof(inl));
 					raw += sizeof(inl);
@@ -600,9 +610,9 @@
 			ds = 3;
 use_dseg:
 			/* Add the remaining packet as a simple ds. */
-			naddr = htonll(addr);
+			naddr = rte_cpu_to_be_64(addr);
 			*dseg = (rte_v128u32_t){
-				htonl(length),
+				rte_cpu_to_be_32(length),
 				mlx5_tx_mb2mr(txq, buf),
 				naddr,
 				naddr >> 32,
@@ -639,9 +649,9 @@
 		total_length += length;
 #endif
 		/* Store segment information. */
-		naddr = htonll(rte_pktmbuf_mtod(buf, uintptr_t));
+		naddr = rte_cpu_to_be_64(rte_pktmbuf_mtod(buf, uintptr_t));
 		*dseg = (rte_v128u32_t){
-			htonl(length),
+			rte_cpu_to_be_32(length),
 			mlx5_tx_mb2mr(txq, buf),
 			naddr,
 			naddr >> 32,
@@ -664,21 +674,23 @@
 		/* Initialize known and common part of the WQE structure. */
 		if (tso) {
 			wqe->ctrl = (rte_v128u32_t){
-				htonl((txq->wqe_ci << 8) | MLX5_OPCODE_TSO),
-				htonl(txq->qp_num_8s | ds),
+				rte_cpu_to_be_32((txq->wqe_ci << 8) |
+						 MLX5_OPCODE_TSO),
+				rte_cpu_to_be_32(txq->qp_num_8s | ds),
 				0,
 				0,
 			};
 			wqe->eseg = (rte_v128u32_t){
 				0,
-				cs_flags | (htons(tso_segsz) << 16),
+				cs_flags | (rte_cpu_to_be_16(tso_segsz) << 16),
 				0,
-				(ehdr << 16) | htons(tso_header_sz),
+				(ehdr << 16) | rte_cpu_to_be_16(tso_header_sz),
 			};
 		} else {
 			wqe->ctrl = (rte_v128u32_t){
-				htonl((txq->wqe_ci << 8) | MLX5_OPCODE_SEND),
-				htonl(txq->qp_num_8s | ds),
+				rte_cpu_to_be_32((txq->wqe_ci << 8) |
+						 MLX5_OPCODE_SEND),
+				rte_cpu_to_be_32(txq->qp_num_8s | ds),
 				0,
 				0,
 			};
@@ -686,7 +698,7 @@
 				0,
 				cs_flags,
 				0,
-				(ehdr << 16) | htons(pkt_inline_sz),
+				(ehdr << 16) | rte_cpu_to_be_16(pkt_inline_sz),
 			};
 		}
 next_wqe:
@@ -706,7 +718,7 @@
 	comp = txq->elts_comp + i + j + k;
 	if (comp >= MLX5_TX_COMP_THRESH) {
 		/* Request completion on last WQE. */
-		last_wqe->ctrl2 = htonl(8);
+		last_wqe->ctrl2 = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		last_wqe->ctrl3 = txq->elts_head;
 		txq->elts_comp = 0;
@@ -745,13 +757,14 @@
 	mpw->len = length;
 	mpw->total_len = 0;
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->eseg.mss = htons(length);
+	mpw->wqe->eseg.mss = rte_cpu_to_be_16(length);
 	mpw->wqe->eseg.inline_hdr_sz = 0;
 	mpw->wqe->eseg.rsvd0 = 0;
 	mpw->wqe->eseg.rsvd1 = 0;
 	mpw->wqe->eseg.rsvd2 = 0;
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_MPW << 24) |
-				  (txq->wqe_ci << 8) | MLX5_OPCODE_TSO);
+	mpw->wqe->ctrl[0] = rte_cpu_to_be_32((MLX5_OPC_MOD_MPW << 24) |
+					     (txq->wqe_ci << 8) |
+					     MLX5_OPCODE_TSO);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
 	mpw->data.dseg[0] = (volatile struct mlx5_wqe_data_seg *)
@@ -780,7 +793,7 @@
 	 * Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | (2 + num));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s | (2 + num));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
 	if (num < 3)
 		++txq->wqe_ci;
@@ -889,9 +902,9 @@
 			dseg = mpw.data.dseg[mpw.pkts_n];
 			addr = rte_pktmbuf_mtod(buf, uintptr_t);
 			*dseg = (struct mlx5_wqe_data_seg){
-				.byte_count = htonl(DATA_LEN(buf)),
+				.byte_count = rte_cpu_to_be_32(DATA_LEN(buf)),
 				.lkey = mlx5_tx_mb2mr(txq, buf),
-				.addr = htonll(addr),
+				.addr = rte_cpu_to_be_64(addr),
 			};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 			length += DATA_LEN(buf);
@@ -919,7 +932,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -959,12 +972,12 @@
 	mpw->len = length;
 	mpw->total_len = 0;
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_MPW << 24) |
-				  (txq->wqe_ci << 8) |
-				  MLX5_OPCODE_TSO);
+	mpw->wqe->ctrl[0] = rte_cpu_to_be_32((MLX5_OPC_MOD_MPW << 24) |
+					     (txq->wqe_ci << 8) |
+					     MLX5_OPCODE_TSO);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
-	mpw->wqe->eseg.mss = htons(length);
+	mpw->wqe->eseg.mss = rte_cpu_to_be_16(length);
 	mpw->wqe->eseg.inline_hdr_sz = 0;
 	mpw->wqe->eseg.cs_flags = 0;
 	mpw->wqe->eseg.rsvd0 = 0;
@@ -995,9 +1008,10 @@
 	 * Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | MLX5_WQE_DS(size));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s |
+					     MLX5_WQE_DS(size));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
-	inl->byte_cnt = htonl(mpw->total_len | MLX5_INLINE_SEG);
+	inl->byte_cnt = rte_cpu_to_be_32(mpw->total_len | MLX5_INLINE_SEG);
 	txq->wqe_ci += (size + (MLX5_WQE_SIZE - 1)) / MLX5_WQE_SIZE;
 }
 
@@ -1137,9 +1151,10 @@
 				dseg = mpw.data.dseg[mpw.pkts_n];
 				addr = rte_pktmbuf_mtod(buf, uintptr_t);
 				*dseg = (struct mlx5_wqe_data_seg){
-					.byte_count = htonl(DATA_LEN(buf)),
+					.byte_count =
+					       rte_cpu_to_be_32(DATA_LEN(buf)),
 					.lkey = mlx5_tx_mb2mr(txq, buf),
-					.addr = htonll(addr),
+					.addr = rte_cpu_to_be_64(addr),
 				};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 				length += DATA_LEN(buf);
@@ -1211,7 +1226,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -1251,9 +1266,10 @@
 	mpw->pkts_n = 0;
 	mpw->total_len = sizeof(struct mlx5_wqe);
 	mpw->wqe = (volatile struct mlx5_wqe *)tx_mlx5_wqe(txq, idx);
-	mpw->wqe->ctrl[0] = htonl((MLX5_OPC_MOD_ENHANCED_MPSW << 24) |
-				  (txq->wqe_ci << 8) |
-				  MLX5_OPCODE_ENHANCED_MPSW);
+	mpw->wqe->ctrl[0] =
+		rte_cpu_to_be_32((MLX5_OPC_MOD_ENHANCED_MPSW << 24) |
+				 (txq->wqe_ci << 8) |
+				 MLX5_OPCODE_ENHANCED_MPSW);
 	mpw->wqe->ctrl[2] = 0;
 	mpw->wqe->ctrl[3] = 0;
 	memset((void *)(uintptr_t)&mpw->wqe->eseg, 0, MLX5_WQE_DWORD_SIZE);
@@ -1261,9 +1277,9 @@
 		uintptr_t addr = (uintptr_t)(mpw->wqe + 1);
 
 		/* Pad the first 2 DWORDs with zero-length inline header. */
-		*(volatile uint32_t *)addr = htonl(MLX5_INLINE_SEG);
+		*(volatile uint32_t *)addr = rte_cpu_to_be_32(MLX5_INLINE_SEG);
 		*(volatile uint32_t *)(addr + MLX5_WQE_DWORD_SIZE) =
-			htonl(MLX5_INLINE_SEG);
+			rte_cpu_to_be_32(MLX5_INLINE_SEG);
 		mpw->total_len += 2 * MLX5_WQE_DWORD_SIZE;
 		/* Start from the next WQEBB. */
 		mpw->data.raw = (volatile void *)(tx_mlx5_wqe(txq, idx + 1));
@@ -1291,7 +1307,8 @@
 	/* Store size in multiple of 16 bytes. Control and Ethernet segments
 	 * count as 2.
 	 */
-	mpw->wqe->ctrl[1] = htonl(txq->qp_num_8s | MLX5_WQE_DS(mpw->total_len));
+	mpw->wqe->ctrl[1] = rte_cpu_to_be_32(txq->qp_num_8s |
+					     MLX5_WQE_DS(mpw->total_len));
 	mpw->state = MLX5_MPW_STATE_CLOSED;
 	ret = (mpw->total_len + (MLX5_WQE_SIZE - 1)) / MLX5_WQE_SIZE;
 	txq->wqe_ci += ret;
@@ -1446,9 +1463,10 @@
 				dseg = mpw.data.dseg[mpw.pkts_n];
 				addr = rte_pktmbuf_mtod(buf, uintptr_t);
 				*dseg = (struct mlx5_wqe_data_seg){
-					.byte_count = htonl(DATA_LEN(buf)),
+					.byte_count = rte_cpu_to_be_32(
+								DATA_LEN(buf)),
 					.lkey = mlx5_tx_mb2mr(txq, buf),
-					.addr = htonll(addr),
+					.addr = rte_cpu_to_be_64(addr),
 				};
 #if defined(MLX5_PMD_SOFT_COUNTERS) || !defined(NDEBUG)
 				length += DATA_LEN(buf);
@@ -1471,7 +1489,7 @@
 
 			assert(mpw.state == MLX5_MPW_ENHANCED_STATE_OPENED);
 			assert(length == DATA_LEN(buf));
-			inl_hdr = htonl(length | MLX5_INLINE_SEG);
+			inl_hdr = rte_cpu_to_be_32(length | MLX5_INLINE_SEG);
 			addr = rte_pktmbuf_mtod(buf, uintptr_t);
 			mpw.data.raw = (volatile void *)
 				((uintptr_t)mpw.data.raw + inl_pad);
@@ -1527,9 +1545,9 @@
 			for (n = 0; n * RTE_CACHE_LINE_SIZE < length; n++)
 				rte_prefetch2((void *)(addr +
 						n * RTE_CACHE_LINE_SIZE));
-			naddr = htonll(addr);
+			naddr = rte_cpu_to_be_64(addr);
 			*dseg = (rte_v128u32_t) {
-				htonl(length),
+				rte_cpu_to_be_32(length),
 				mlx5_tx_mb2mr(txq, buf),
 				naddr,
 				naddr >> 32,
@@ -1557,7 +1575,7 @@
 		volatile struct mlx5_wqe *wqe = mpw.wqe;
 
 		/* Request completion on last WQE. */
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		/* Save elts_head in unused "immediate" field of WQE. */
 		wqe->ctrl[3] = elts_head;
 		txq->elts_comp = 0;
@@ -1641,8 +1659,8 @@
 			(volatile struct mlx5_mini_cqe8 (*)[8])
 			(uintptr_t)(&(*rxq->cqes)[zip->ca & cqe_cnt].pkt_info);
 
-		len = ntohl((*mc)[zip->ai & 7].byte_cnt);
-		*rss_hash = ntohl((*mc)[zip->ai & 7].rx_hash_result);
+		len = rte_be_to_cpu_32((*mc)[zip->ai & 7].byte_cnt);
+		*rss_hash = rte_be_to_cpu_32((*mc)[zip->ai & 7].rx_hash_result);
 		if ((++zip->ai & 7) == 0) {
 			/* Invalidate consumed CQEs */
 			idx = zip->ca;
@@ -1690,7 +1708,7 @@
 							  cqe_cnt].pkt_info);
 
 			/* Fix endianness. */
-			zip->cqe_cnt = ntohl(cqe->byte_cnt);
+			zip->cqe_cnt = rte_be_to_cpu_32(cqe->byte_cnt);
 			/*
 			 * Current mini array position is the one returned by
 			 * check_cqe64().
@@ -1705,8 +1723,8 @@
 			--rxq->cq_ci;
 			zip->cq_ci = rxq->cq_ci + zip->cqe_cnt;
 			/* Get packet size to return. */
-			len = ntohl((*mc)[0].byte_cnt);
-			*rss_hash = ntohl((*mc)[0].rx_hash_result);
+			len = rte_be_to_cpu_32((*mc)[0].byte_cnt);
+			*rss_hash = rte_be_to_cpu_32((*mc)[0].rx_hash_result);
 			zip->ai = 1;
 			/* Prefetch all the entries to be invalidated */
 			idx = zip->ca;
@@ -1716,8 +1734,8 @@
 				++idx;
 			}
 		} else {
-			len = ntohl(cqe->byte_cnt);
-			*rss_hash = ntohl(cqe->rx_hash_res);
+			len = rte_be_to_cpu_32(cqe->byte_cnt);
+			*rss_hash = rte_be_to_cpu_32(cqe->rx_hash_res);
 		}
 		/* Error while receiving packet. */
 		if (unlikely(MLX5_CQE_OPCODE(op_own) == MLX5_CQE_RESP_ERR))
@@ -1741,7 +1759,7 @@
 rxq_cq_to_ol_flags(struct rxq *rxq, volatile struct mlx5_cqe *cqe)
 {
 	uint32_t ol_flags = 0;
-	uint16_t flags = ntohs(cqe->hdr_type_etc);
+	uint16_t flags = rte_be_to_cpu_16(cqe->hdr_type_etc);
 
 	ol_flags =
 		TRANSPOSE(flags,
@@ -1848,7 +1866,7 @@
 			    MLX5_FLOW_MARK_IS_VALID(cqe->sop_drop_qpn)) {
 				pkt->ol_flags |= PKT_RX_FDIR;
 				if (cqe->sop_drop_qpn !=
-				    htonl(MLX5_FLOW_MARK_DEFAULT)) {
+				    rte_cpu_to_be_32(MLX5_FLOW_MARK_DEFAULT)) {
 					uint32_t mark = cqe->sop_drop_qpn;
 
 					pkt->ol_flags |= PKT_RX_FDIR_ID;
@@ -1860,10 +1878,11 @@
 				pkt->ol_flags |= rxq_cq_to_ol_flags(rxq, cqe);
 			if (rxq->vlan_strip &&
 			    (cqe->hdr_type_etc &
-			     htons(MLX5_CQE_VLAN_STRIPPED))) {
+			     rte_cpu_to_be_16(MLX5_CQE_VLAN_STRIPPED))) {
 				pkt->ol_flags |= PKT_RX_VLAN_PKT |
 					PKT_RX_VLAN_STRIPPED;
-				pkt->vlan_tci = ntohs(cqe->vlan_info);
+				pkt->vlan_tci =
+					rte_be_to_cpu_16(cqe->vlan_info);
 			}
 			if (rxq->crc_present)
 				len -= ETHER_CRC_LEN;
@@ -1879,7 +1898,7 @@
 		 * of the buffers are already known, only the buffer address
 		 * changes.
 		 */
-		wqe->addr = htonll(rte_pktmbuf_mtod(rep, uintptr_t));
+		wqe->addr = rte_cpu_to_be_64(rte_pktmbuf_mtod(rep, uintptr_t));
 		if (len > DATA_LEN(seg)) {
 			len -= DATA_LEN(seg);
 			++NB_SEGS(pkt);
@@ -1907,9 +1926,9 @@
 	/* Update the consumer index. */
 	rxq->rq_ci = rq_ci >> sges_n;
 	rte_wmb();
-	*rxq->cq_db = htonl(rxq->cq_ci);
+	*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
 	rte_wmb();
-	*rxq->rq_db = htonl(rxq->rq_ci);
+	*rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci);
 #ifdef MLX5_PMD_SOFT_COUNTERS
 	/* Increment packets counter. */
 	rxq->stats.ipackets += i;
diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net/mlx5/mlx5_rxtx.h
index 107ada0..9375aa8 100644
--- a/drivers/net/mlx5/mlx5_rxtx.h
+++ b/drivers/net/mlx5/mlx5_rxtx.h
@@ -269,7 +269,7 @@ struct txq {
 		uintptr_t start; /* Start address of MR */
 		uintptr_t end; /* End address of MR */
 		struct ibv_mr *mr; /* Memory Region (for mp). */
-		uint32_t lkey; /* htonl(mr->lkey) */
+		uint32_t lkey; /* rte_cpu_to_be_32(mr->lkey) */
 	} mp2mr[MLX5_PMD_TX_MP_CACHE]; /* MP to MR translation table. */
 	uint16_t mr_cache_idx; /* Index of last hit entry. */
 	struct rte_mbuf *(*elts)[]; /* TX elements. */
@@ -492,7 +492,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	}
 #endif /* NDEBUG */
 	++cq_ci;
-	txq->wqe_pi = ntohs(cqe->wqe_counter);
+	txq->wqe_pi = rte_be_to_cpu_16(cqe->wqe_counter);
 	ctrl = (volatile struct mlx5_wqe_ctrl *)
 		tx_mlx5_wqe(txq, txq->wqe_pi);
 	elts_tail = ctrl->ctrl3;
@@ -530,7 +530,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	txq->elts_tail = elts_tail;
 	/* Update the consumer index. */
 	rte_wmb();
-	*txq->cq_db = htonl(cq_ci);
+	*txq->cq_db = rte_cpu_to_be_32(cq_ci);
 }
 
 /**
@@ -581,7 +581,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 		if (txq->mp2mr[i].start <= addr &&
 		    txq->mp2mr[i].end >= addr) {
 			assert(txq->mp2mr[i].lkey != (uint32_t)-1);
-			assert(htonl(txq->mp2mr[i].mr->lkey) ==
+			assert(rte_cpu_to_be_32(txq->mp2mr[i].mr->lkey) ==
 			       txq->mp2mr[i].lkey);
 			txq->mr_cache_idx = i;
 			return txq->mp2mr[i].lkey;
@@ -606,7 +606,7 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *, uint16_t, uint16_t, unsigned int,
 	volatile uint64_t *src = ((volatile uint64_t *)wqe);
 
 	rte_io_wmb();
-	*txq->qp_db = htonl(txq->wqe_ci);
+	*txq->qp_db = rte_cpu_to_be_32(txq->wqe_ci);
 	/* Ensure ordering between DB record and BF copy. */
 	rte_wmb();
 	*dst = *src;
diff --git a/drivers/net/mlx5/mlx5_rxtx_vec_sse.c b/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
index 39c7325..aff3359 100644
--- a/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
+++ b/drivers/net/mlx5/mlx5_rxtx_vec_sse.c
@@ -293,7 +293,7 @@
 		/* Fill ESEG in the header. */
 		_mm_store_si128(t_wqe + 1,
 				_mm_set_epi16(0, 0, 0, 0,
-					      htons(len), cs_flags,
+					      rte_cpu_to_be_16(len), cs_flags,
 					      0, 0));
 		txq->wqe_ci = wqe_ci;
 	}
@@ -302,7 +302,7 @@
 	txq->elts_comp += (uint16_t)(elts_head - txq->elts_head);
 	txq->elts_head = elts_head;
 	if (txq->elts_comp >= MLX5_TX_COMP_THRESH) {
-		wqe->ctrl[2] = htonl(8);
+		wqe->ctrl[2] = rte_cpu_to_be_32(8);
 		wqe->ctrl[3] = txq->elts_head;
 		txq->elts_comp = 0;
 		++txq->cq_pi;
@@ -564,11 +564,11 @@
 		return;
 	}
 	for (i = 0; i < n; ++i)
-		wq[i].addr = htonll((uintptr_t)elts[i]->buf_addr +
-				    RTE_PKTMBUF_HEADROOM);
+		wq[i].addr = rte_cpu_to_be_64((uintptr_t)elts[i]->buf_addr +
+					      RTE_PKTMBUF_HEADROOM);
 	rxq->rq_ci += n;
 	rte_wmb();
-	*rxq->rq_db = htonl(rxq->rq_ci);
+	*rxq->rq_db = rte_cpu_to_be_32(rxq->rq_ci);
 }
 
 /**
@@ -1251,7 +1251,7 @@
 		}
 	}
 	rte_wmb();
-	*rxq->cq_db = htonl(rxq->cq_ci);
+	*rxq->cq_db = rte_cpu_to_be_32(rxq->cq_ci);
 	return rcvd_pkt;
 }
 
-- 
1.8.3.1

[PATCH v4 2/2] net/mlx5: fix TSO MLNX OFED 3.3 verification

[Shachar Beiser]

Fixes: 3cf87e68d97b ("net/mlx5: remove old MLNX OFED 3.3 verification")
Cc: stable@dpdk.org

Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
---
I have rebased last patch [PATCH v3 2/2] net/mlx5: fix TSO MLNX OFED 3.3 verification
---
 drivers/net/mlx5/mlx5_prm.h | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/net/mlx5/mlx5_prm.h b/drivers/net/mlx5/mlx5_prm.h
index c61e4d8..e00be81 100644
--- a/drivers/net/mlx5/mlx5_prm.h
+++ b/drivers/net/mlx5/mlx5_prm.h
@@ -89,9 +89,6 @@
 /* Default max packet length to be inlined. */
 #define MLX5_EMPW_MAX_INLINE_LEN (4U * MLX5_WQE_SIZE)
 
-#ifndef HAVE_VERBS_MLX5_OPCODE_TSO
-#define MLX5_OPCODE_TSO MLX5_OPCODE_LSO_MPW /* Compat with OFED 3.3. */
-#endif
 
 #define MLX5_OPC_MOD_ENHANCED_MPSW 0
 #define MLX5_OPCODE_ENHANCED_MPSW 0x29
-- 
1.8.3.1

Re: [PATCH v3 1/2] net/mlx5: replace network to host macros

[Shachar Beiser]

> -----Original Message-----
> From: Yongseok Koh
> Sent: Friday, September 15, 2017 11:50 PM
> To: Shachar Beiser <shacharbe@mellanox.com>
> Cc: dev@dpdk.org; Adrien Mazarguil <adrien.mazarguil@6wind.com>; Nélio
> Laranjeiro <nelio.laranjeiro@6wind.com>
> Subject: Re: [dpdk-dev] [PATCH v3 1/2] net/mlx5: replace network to host
> macros
> 
> 
> > On Sep 14, 2017, at 6:43 AM, Shachar Beiser <shacharbe@mellanox.com>
> wrote:
> >
> > Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> > ---
> [...]
> > @@ -550,7 +553,8 @@
> > 				max_wqe -= n;
> > 				if (tso) {
> > 					uint32_t inl =
> > -						htonl(copy_b |
> MLX5_INLINE_SEG);
> > +					rte_cpu_to_be_32(copy_b |
> > +							 MLX5_INLINE_SEG);
> Wrong indentation.
> 
Since there is a constrain of 80 characters a line , this indentation was accepted by
Nélio . 
> [...]
> > @@ -605,8 +605,8 @@ int mlx5_tx_queue_setup(struct rte_eth_dev *,
> uint16_t, uint16_t, unsigned int,
> > 	uint64_t *dst = (uint64_t *)((uintptr_t)txq->bf_reg);
> > 	volatile uint64_t *src = ((volatile uint64_t *)wqe);
> >
> > -	rte_io_wmb();
> > -	*txq->qp_db = htonl(txq->wqe_ci);
> > +	rte_wmb();
> Look like a mistake when rebasing. This should not be touched by this patch.
> 
> > +	*txq->qp_db = rte_cpu_to_be_32(txq->wqe_ci);
> > 	/* Ensure ordering between DB record and BF copy. */
> > 	rte_wmb();
> > 	*dst = *src;
> 

Yes , this is true. There is a fix for this issue in the coming patch:
[dpdk-dev,v4,1/2] net/mlx5: replace network to host macros
http://dpdk.org/dev/patchwork/patch/28804/


> --
> Thanks,
> Yongseok
> 

Re: [PATCH v4 2/2] net/mlx5: fix TSO MLNX OFED 3.3 verification

[Yongseok Koh]

> On Sep 17, 2017, at 3:42 AM, Shachar Beiser <shacharbe@mellanox.com> wrote:
> 
> Fixes: 3cf87e68d97b ("net/mlx5: remove old MLNX OFED 3.3 verification")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> ---
> I have rebased last patch [PATCH v3 2/2] net/mlx5: fix TSO MLNX OFED 3.3 verification
> ---
I think this patch had been acked by Nelio since v2.  Then you can keep
'Acked-by' tag through updated versions unless there's any change in the patch.

Acked-by: Yongseok Koh <yskoh@mellanox.com>

Thanks
Yongseok

Re: [PATCH v4 1/2] net/mlx5: replace network to host macros

[Yongseok Koh]

> On Sep 17, 2017, at 3:42 AM, Shachar Beiser <shacharbe@mellanox.com> wrote:
> 
> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
> ---
> I have rebased [PATCH v3 1/2] net/mlx5: replace network to host macros
> I have fixed a rebase conflict in mlx5_rxtx.h in line 609
> -rte_wmb();
> +rte_io_wmb();
> 
> ---
Acked-by: Yongseok Koh <yskoh@mellanox.com>

Thanks
Yongseok

Re: [PATCH v4 1/2] net/mlx5: replace network to host macros

[Nélio Laranjeiro]

For the series
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

-- 
Nélio Laranjeiro
6WIND

Re: [dpdk-stable] [PATCH v4 2/2] net/mlx5: fix TSO MLNX OFED 3.3 verification

[Ferruh Yigit]

On 9/18/2017 6:52 PM, Yongseok Koh wrote:
> 
>> On Sep 17, 2017, at 3:42 AM, Shachar Beiser <shacharbe@mellanox.com> wrote:
>>
>> Fixes: 3cf87e68d97b ("net/mlx5: remove old MLNX OFED 3.3 verification")
>> Cc: stable@dpdk.org
>>
>> Signed-off-by: Shachar Beiser <shacharbe@mellanox.com>
>> ---
>> I have rebased last patch [PATCH v3 2/2] net/mlx5: fix TSO MLNX OFED 3.3 verification
>> ---
> I think this patch had been acked by Nelio since v2.  Then you can keep
> 'Acked-by' tag through updated versions unless there's any change in the patch.
> 
> Acked-by: Yongseok Koh <yskoh@mellanox.com>

This commit is fixing one that is within this release, so this commit
squashed into relevant one.

> 
> Thanks
> Yongseok
>