1 files changed, 233 insertions, 103 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx4/en_tx.c b/drivers/net/ethernet/mellanox/mlx4/en_tx.c
index 7c492382da0..5045bab5963 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_tx.c
@@ -39,51 +39,62 @@
 #include <linux/if_vlan.h>
 #include <linux/vmalloc.h>
 #include <linux/tcp.h>
+#include <linux/ip.h>
 #include <linux/moduleparam.h>
 
 #include "mlx4_en.h"
 
-enum {
-	MAX_INLINE = 104, /* 128 - 16 - 4 - 4 */
-	MAX_BF = 256,
-};
-
-static int inline_thold __read_mostly = MAX_INLINE;
-
-module_param_named(inline_thold, inline_thold, int, 0444);
-MODULE_PARM_DESC(inline_thold, "threshold for using inline data");
-
 int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
-			   struct mlx4_en_tx_ring *ring, int qpn, u32 size,
-			   u16 stride)
+			   struct mlx4_en_tx_ring **pring, int qpn, u32 size,
+			   u16 stride, int node, int queue_index)
 {
 	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_tx_ring *ring;
 	int tmp;
 	int err;
 
+	ring = kzalloc_node(sizeof(*ring), GFP_KERNEL, node);
+	if (!ring) {
+		ring = kzalloc(sizeof(*ring), GFP_KERNEL);
+		if (!ring) {
+			en_err(priv, "Failed allocating TX ring\n");
+			return -ENOMEM;
+		}
+	}
+
 	ring->size = size;
 	ring->size_mask = size - 1;
 	ring->stride = stride;
-
-	inline_thold = min(inline_thold, MAX_INLINE);
+	ring->inline_thold = priv->prof->inline_thold;
 
 	tmp = size * sizeof(struct mlx4_en_tx_info);
-	ring->tx_info = vmalloc(tmp);
-	if (!ring->tx_info)
-		return -ENOMEM;
+	ring->tx_info = vmalloc_node(tmp, node);
+	if (!ring->tx_info) {
+		ring->tx_info = vmalloc(tmp);
+		if (!ring->tx_info) {
+			err = -ENOMEM;
+			goto err_ring;
+		}
+	}
 
 	en_dbg(DRV, priv, "Allocated tx_info ring at addr:%p size:%d\n",
 		 ring->tx_info, tmp);
 
-	ring->bounce_buf = kmalloc(MAX_DESC_SIZE, GFP_KERNEL);
+	ring->bounce_buf = kmalloc_node(MAX_DESC_SIZE, GFP_KERNEL, node);
 	if (!ring->bounce_buf) {
-		err = -ENOMEM;
-		goto err_tx;
+		ring->bounce_buf = kmalloc(MAX_DESC_SIZE, GFP_KERNEL);
+		if (!ring->bounce_buf) {
+			err = -ENOMEM;
+			goto err_info;
+		}
 	}
 	ring->buf_size = ALIGN(size * ring->stride, MLX4_EN_PAGE_SIZE);
 
+	/* Allocate HW buffers on provided NUMA node */
+	set_dev_node(&mdev->dev->pdev->dev, node);
 	err = mlx4_alloc_hwq_res(mdev->dev, &ring->wqres, ring->buf_size,
 				 2 * PAGE_SIZE);
+	set_dev_node(&mdev->dev->pdev->dev, mdev->dev->numa_node);
 	if (err) {
 		en_err(priv, "Failed allocating hwq resources\n");
 		goto err_bounce;
@@ -97,21 +108,21 @@ int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
 
 	ring->buf = ring->wqres.buf.direct.buf;
 
-	en_dbg(DRV, priv, "Allocated TX ring (addr:%p) - buf:%p size:%d "
-	       "buf_size:%d dma:%llx\n", ring, ring->buf, ring->size,
-	       ring->buf_size, (unsigned long long) ring->wqres.buf.direct.map);
+	en_dbg(DRV, priv, "Allocated TX ring (addr:%p) - buf:%p size:%d buf_size:%d dma:%llx\n",
+	       ring, ring->buf, ring->size, ring->buf_size,
+	       (unsigned long long) ring->wqres.buf.direct.map);
 
 	ring->qpn = qpn;
-	err = mlx4_qp_alloc(mdev->dev, ring->qpn, &ring->qp);
+	err = mlx4_qp_alloc(mdev->dev, ring->qpn, &ring->qp, GFP_KERNEL);
 	if (err) {
 		en_err(priv, "Failed allocating qp %d\n", ring->qpn);
 		goto err_map;
 	}
 	ring->qp.event = mlx4_en_sqp_event;
 
-	err = mlx4_bf_alloc(mdev->dev, &ring->bf);
+	err = mlx4_bf_alloc(mdev->dev, &ring->bf, node);
 	if (err) {
-		en_dbg(DRV, priv, "working without blueflame (%d)", err);
+		en_dbg(DRV, priv, "working without blueflame (%d)\n", err);
 		ring->bf.uar = &mdev->priv_uar;
 		ring->bf.uar->map = mdev->uar_map;
 		ring->bf_enabled = false;
@@ -119,7 +130,12 @@ int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
 		ring->bf_enabled = true;
 
 	ring->hwtstamp_tx_type = priv->hwtstamp_config.tx_type;
+	ring->queue_index = queue_index;
+
+	if (queue_index < priv->num_tx_rings_p_up && cpu_online(queue_index))
+		cpumask_set_cpu(queue_index, &ring->affinity_mask);
 
+	*pring = ring;
 	return 0;
 
 err_map:
@@ -129,16 +145,20 @@ err_hwq_res:
 err_bounce:
 	kfree(ring->bounce_buf);
 	ring->bounce_buf = NULL;
-err_tx:
+err_info:
 	vfree(ring->tx_info);
 	ring->tx_info = NULL;
+err_ring:
+	kfree(ring);
+	*pring = NULL;
 	return err;
 }
 
 void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv,
-			     struct mlx4_en_tx_ring *ring)
+			     struct mlx4_en_tx_ring **pring)
 {
 	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_tx_ring *ring = *pring;
 	en_dbg(DRV, priv, "Destroying tx ring, qpn: %d\n", ring->qpn);
 
 	if (ring->bf_enabled)
@@ -151,6 +171,8 @@ void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv,
 	ring->bounce_buf = NULL;
 	vfree(ring->tx_info);
 	ring->tx_info = NULL;
+	kfree(ring);
+	*pring = NULL;
 }
 
 int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv,
@@ -178,6 +200,9 @@ int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv,
 
 	err = mlx4_qp_to_ready(mdev->dev, &ring->wqres.mtt, &ring->context,
 			       &ring->qp, &ring->qp_state);
+	if (!user_prio && cpu_online(ring->queue_index))
+		netif_set_xps_queue(priv->dev, &ring->affinity_mask,
+				    ring->queue_index);
 
 	return err;
 }
@@ -191,6 +216,39 @@ void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv *priv,
 		       MLX4_QP_STATE_RST, NULL, 0, 0, &ring->qp);
 }
 
+static void mlx4_en_stamp_wqe(struct mlx4_en_priv *priv,
+			      struct mlx4_en_tx_ring *ring, int index,
+			      u8 owner)
+{
+	__be32 stamp = cpu_to_be32(STAMP_VAL | (!!owner << STAMP_SHIFT));
+	struct mlx4_en_tx_desc *tx_desc = ring->buf + index * TXBB_SIZE;
+	struct mlx4_en_tx_info *tx_info = &ring->tx_info[index];
+	void *end = ring->buf + ring->buf_size;
+	__be32 *ptr = (__be32 *)tx_desc;
+	int i;
+
+	/* Optimize the common case when there are no wraparounds */
+	if (likely((void *)tx_desc + tx_info->nr_txbb * TXBB_SIZE <= end)) {
+		/* Stamp the freed descriptor */
+		for (i = 0; i < tx_info->nr_txbb * TXBB_SIZE;
+		     i += STAMP_STRIDE) {
+			*ptr = stamp;
+			ptr += STAMP_DWORDS;
+		}
+	} else {
+		/* Stamp the freed descriptor */
+		for (i = 0; i < tx_info->nr_txbb * TXBB_SIZE;
+		     i += STAMP_STRIDE) {
+			*ptr = stamp;
+			ptr += STAMP_DWORDS;
+			if ((void *)ptr >= end) {
+				ptr = ring->buf;
+				stamp ^= cpu_to_be32(0x80000000);
+			}
+		}
+	}
+}
+
 
 static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
 				struct mlx4_en_tx_ring *ring,
@@ -205,8 +263,6 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
 	void *end = ring->buf + ring->buf_size;
 	int frags = skb_shinfo(skb)->nr_frags;
 	int i;
-	__be32 *ptr = (__be32 *)tx_desc;
-	__be32 stamp = cpu_to_be32(STAMP_VAL | (!!owner << STAMP_SHIFT));
 	struct skb_shared_hwtstamps hwts;
 
 	if (timestamp) {
@@ -232,12 +288,6 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
 					skb_frag_size(frag), PCI_DMA_TODEVICE);
 			}
 		}
-		/* Stamp the freed descriptor */
-		for (i = 0; i < tx_info->nr_txbb * TXBB_SIZE; i += STAMP_STRIDE) {
-			*ptr = stamp;
-			ptr += STAMP_DWORDS;
-		}
-
 	} else {
 		if (!tx_info->inl) {
 			if ((void *) data >= end) {
@@ -263,16 +313,6 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
 				++data;
 			}
 		}
-		/* Stamp the freed descriptor */
-		for (i = 0; i < tx_info->nr_txbb * TXBB_SIZE; i += STAMP_STRIDE) {
-			*ptr = stamp;
-			ptr += STAMP_DWORDS;
-			if ((void *) ptr >= end) {
-				ptr = ring->buf;
-				stamp ^= cpu_to_be32(0x80000000);
-			}
-		}
-
 	}
 	dev_kfree_skb_any(skb);
 	return tx_info->nr_txbb;
@@ -311,15 +351,17 @@ int mlx4_en_free_tx_buf(struct net_device *dev, struct mlx4_en_tx_ring *ring)
 	return cnt;
 }
 
-static void mlx4_en_process_tx_cq(struct net_device *dev, struct mlx4_en_cq *cq)
+static bool mlx4_en_process_tx_cq(struct net_device *dev,
+				 struct mlx4_en_cq *cq)
 {
 	struct mlx4_en_priv *priv = netdev_priv(dev);
 	struct mlx4_cq *mcq = &cq->mcq;
-	struct mlx4_en_tx_ring *ring = &priv->tx_ring[cq->ring];
+	struct mlx4_en_tx_ring *ring = priv->tx_ring[cq->ring];
 	struct mlx4_cqe *cqe;
 	u16 index;
-	u16 new_index, ring_index;
+	u16 new_index, ring_index, stamp_index;
 	u32 txbbs_skipped = 0;
+	u32 txbbs_stamp = 0;
 	u32 cons_index = mcq->cons_index;
 	int size = cq->size;
 	u32 size_mask = ring->size_mask;
@@ -328,23 +370,35 @@ static void mlx4_en_process_tx_cq(struct net_device *dev, struct mlx4_en_cq *cq)
 	u32 bytes = 0;
 	int factor = priv->cqe_factor;
 	u64 timestamp = 0;
+	int done = 0;
+	int budget = priv->tx_work_limit;
 
 	if (!priv->port_up)
-		return;
+		return true;
 
 	index = cons_index & size_mask;
 	cqe = &buf[(index << factor) + factor];
 	ring_index = ring->cons & size_mask;
+	stamp_index = ring_index;
 
 	/* Process all completed CQEs */
 	while (XNOR(cqe->owner_sr_opcode & MLX4_CQE_OWNER_MASK,
-			cons_index & size)) {
+			cons_index & size) && (done < budget)) {
 		/*
 		 * make sure we read the CQE after we read the
 		 * ownership bit
 		 */
 		rmb();
 
+		if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) ==
+			     MLX4_CQE_OPCODE_ERROR)) {
+			struct mlx4_err_cqe *cqe_err = (struct mlx4_err_cqe *)cqe;
+
+			en_err(priv, "CQE error - vendor syndrome: 0x%x syndrome: 0x%x\n",
+			       cqe_err->vendor_err_syndrome,
+			       cqe_err->syndrome);
+		}
+
 		/* Skip over last polled CQE */
 		new_index = be16_to_cpu(cqe->wqe_index) & size_mask;
 
@@ -359,9 +413,15 @@ static void mlx4_en_process_tx_cq(struct net_device *dev, struct mlx4_en_cq *cq)
 					priv, ring, ring_index,
 					!!((ring->cons + txbbs_skipped) &
 					ring->size), timestamp);
+
+			mlx4_en_stamp_wqe(priv, ring, stamp_index,
+					  !!((ring->cons + txbbs_stamp) &
+						ring->size));
+			stamp_index = ring_index;
+			txbbs_stamp = txbbs_skipped;
 			packets++;
 			bytes += ring->tx_info[ring_index].nr_bytes;
-		} while (ring_index != new_index);
+		} while ((++done < budget) && (ring_index != new_index));
 
 		++cons_index;
 		index = cons_index & size_mask;
@@ -385,8 +445,9 @@ static void mlx4_en_process_tx_cq(struct net_device *dev, struct mlx4_en_cq *cq)
 	 */
 	if (netif_tx_queue_stopped(ring->tx_queue) && txbbs_skipped > 0) {
 		netif_tx_wake_queue(ring->tx_queue);
-		priv->port_stats.wake_queue++;
+		ring->wake_queue++;
 	}
+	return done < budget;
 }
 
 void mlx4_en_tx_irq(struct mlx4_cq *mcq)
@@ -394,10 +455,29 @@ void mlx4_en_tx_irq(struct mlx4_cq *mcq)
 	struct mlx4_en_cq *cq = container_of(mcq, struct mlx4_en_cq, mcq);
 	struct mlx4_en_priv *priv = netdev_priv(cq->dev);
 
-	mlx4_en_process_tx_cq(cq->dev, cq);
-	mlx4_en_arm_cq(priv, cq);
+	if (priv->port_up)
+		napi_schedule(&cq->napi);
+	else
+		mlx4_en_arm_cq(priv, cq);
 }
 
+/* TX CQ polling - called by NAPI */
+int mlx4_en_poll_tx_cq(struct napi_struct *napi, int budget)
+{
+	struct mlx4_en_cq *cq = container_of(napi, struct mlx4_en_cq, napi);
+	struct net_device *dev = cq->dev;
+	struct mlx4_en_priv *priv = netdev_priv(dev);
+	int clean_complete;
+
+	clean_complete = mlx4_en_process_tx_cq(dev, cq);
+	if (!clean_complete)
+		return budget;
+
+	napi_complete(napi);
+	mlx4_en_arm_cq(priv, cq);
+
+	return 0;
+}
 
 static struct mlx4_en_tx_desc *mlx4_en_bounce_to_desc(struct mlx4_en_priv *priv,
 						      struct mlx4_en_tx_ring *ring,
@@ -427,7 +507,7 @@ static struct mlx4_en_tx_desc *mlx4_en_bounce_to_desc(struct mlx4_en_priv *priv,
 	return ring->buf + index * TXBB_SIZE;
 }
 
-static int is_inline(struct sk_buff *skb, void **pfrag)
+static int is_inline(int inline_thold, struct sk_buff *skb, void **pfrag)
 {
 	void *ptr;
 
@@ -468,7 +548,10 @@ static int get_real_size(struct sk_buff *skb, struct net_device *dev,
 	int real_size;
 
 	if (skb_is_gso(skb)) {
-		*lso_header_size = skb_transport_offset(skb) + tcp_hdrlen(skb);
+		if (skb->encapsulation)
+			*lso_header_size = (skb_inner_transport_header(skb) - skb->data) + inner_tcp_hdrlen(skb);
+		else
+			*lso_header_size = skb_transport_offset(skb) + tcp_hdrlen(skb);
 		real_size = CTRL_SIZE + skb_shinfo(skb)->nr_frags * DS_SIZE +
 			ALIGN(*lso_header_size + 4, DS_SIZE);
 		if (unlikely(*lso_header_size != skb_headlen(skb))) {
@@ -484,7 +567,7 @@ static int get_real_size(struct sk_buff *skb, struct net_device *dev,
 		}
 	} else {
 		*lso_header_size = 0;
-		if (!is_inline(skb, NULL))
+		if (!is_inline(priv->prof->inline_thold, skb, NULL))
 			real_size = CTRL_SIZE + (skb_shinfo(skb)->nr_frags + 1) * DS_SIZE;
 		else
 			real_size = inline_size(skb);
@@ -500,7 +583,13 @@ static void build_inline_wqe(struct mlx4_en_tx_desc *tx_desc, struct sk_buff *sk
 	int spc = MLX4_INLINE_ALIGN - CTRL_SIZE - sizeof *inl;
 
 	if (skb->len <= spc) {
-		inl->byte_count = cpu_to_be32(1 << 31 | skb->len);
+		if (likely(skb->len >= MIN_PKT_LEN)) {
+			inl->byte_count = cpu_to_be32(1 << 31 | skb->len);
+		} else {
+			inl->byte_count = cpu_to_be32(1 << 31 | MIN_PKT_LEN);
+			memset(((void *)(inl + 1)) + skb->len, 0,
+			       MIN_PKT_LEN - skb->len);
+		}
 		skb_copy_from_linear_data(skb, inl + 1, skb_headlen(skb));
 		if (skb_shinfo(skb)->nr_frags)
 			memcpy(((void *)(inl + 1)) + skb_headlen(skb), fragptr,
@@ -532,7 +621,8 @@ static void build_inline_wqe(struct mlx4_en_tx_desc *tx_desc, struct sk_buff *sk
 	}
 }
 
-u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb)
+u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb,
+			 void *accel_priv, select_queue_fallback_t fallback)
 {
 	struct mlx4_en_priv *priv = netdev_priv(dev);
 	u16 rings_p_up = priv->num_tx_rings_p_up;
@@ -544,7 +634,7 @@ u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb)
 	if (vlan_tx_tag_present(skb))
 		up = vlan_tx_tag_get(skb) >> VLAN_PRIO_SHIFT;
 
-	return __netdev_pick_tx(dev, skb) % rings_p_up + up * rings_p_up;
+	return fallback(dev, skb) % rings_p_up + up * rings_p_up;
 }
 
 static void mlx4_bf_copy(void __iomem *dst, unsigned long *src, unsigned bytecnt)
@@ -556,17 +646,15 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 {
 	struct mlx4_en_priv *priv = netdev_priv(dev);
 	struct mlx4_en_dev *mdev = priv->mdev;
+	struct device *ddev = priv->ddev;
 	struct mlx4_en_tx_ring *ring;
 	struct mlx4_en_tx_desc *tx_desc;
 	struct mlx4_wqe_data_seg *data;
-	struct skb_frag_struct *frag;
 	struct mlx4_en_tx_info *tx_info;
-	struct ethhdr *ethh;
 	int tx_ind = 0;
 	int nr_txbb;
 	int desc_size;
 	int real_size;
-	dma_addr_t dma;
 	u32 index, bf_index;
 	__be32 op_own;
 	u16 vlan_tag = 0;
@@ -592,7 +680,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	tx_ind = skb->queue_mapping;
-	ring = &priv->tx_ring[tx_ind];
+	ring = priv->tx_ring[tx_ind];
 	if (vlan_tx_tag_present(skb))
 		vlan_tag = vlan_tx_tag_get(skb);
 
@@ -601,7 +689,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 		     ring->size - HEADROOM - MAX_DESC_TXBBS)) {
 		/* every full Tx ring stops queue */
 		netif_tx_stop_queue(ring->tx_queue);
-		priv->port_stats.queue_stopped++;
+		ring->queue_stopped++;
 
 		/* If queue was emptied after the if, and before the
 		 * stop_queue - need to wake the queue, or else it will remain
@@ -614,7 +702,7 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 		if (unlikely(((int)(ring->prod - ring->cons)) <=
 			     ring->size - HEADROOM - MAX_DESC_TXBBS)) {
 			netif_tx_wake_queue(ring->tx_queue);
-			priv->port_stats.wake_queue++;
+			ring->wake_queue++;
 		} else {
 			return NETDEV_TX_BUSY;
 		}
@@ -642,6 +730,61 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 	tx_info->skb = skb;
 	tx_info->nr_txbb = nr_txbb;
 
+	if (lso_header_size)
+		data = ((void *)&tx_desc->lso + ALIGN(lso_header_size + 4,
+						      DS_SIZE));
+	else
+		data = &tx_desc->data;
+
+	/* valid only for none inline segments */
+	tx_info->data_offset = (void *)data - (void *)tx_desc;
+
+	tx_info->linear = (lso_header_size < skb_headlen(skb) &&
+			   !is_inline(ring->inline_thold, skb, NULL)) ? 1 : 0;
+
+	data += skb_shinfo(skb)->nr_frags + tx_info->linear - 1;
+
+	if (is_inline(ring->inline_thold, skb, &fragptr)) {
+		tx_info->inl = 1;
+	} else {
+		/* Map fragments */
+		for (i = skb_shinfo(skb)->nr_frags - 1; i >= 0; i--) {
+			struct skb_frag_struct *frag;
+			dma_addr_t dma;
+
+			frag = &skb_shinfo(skb)->frags[i];
+			dma = skb_frag_dma_map(ddev, frag,
+					       0, skb_frag_size(frag),
+					       DMA_TO_DEVICE);
+			if (dma_mapping_error(ddev, dma))
+				goto tx_drop_unmap;
+
+			data->addr = cpu_to_be64(dma);
+			data->lkey = cpu_to_be32(mdev->mr.key);
+			wmb();
+			data->byte_count = cpu_to_be32(skb_frag_size(frag));
+			--data;
+		}
+
+		/* Map linear part */
+		if (tx_info->linear) {
+			u32 byte_count = skb_headlen(skb) - lso_header_size;
+			dma_addr_t dma;
+
+			dma = dma_map_single(ddev, skb->data +
+					     lso_header_size, byte_count,
+					     PCI_DMA_TODEVICE);
+			if (dma_mapping_error(ddev, dma))
+				goto tx_drop_unmap;
+
+			data->addr = cpu_to_be64(dma);
+			data->lkey = cpu_to_be32(mdev->mr.key);
+			wmb();
+			data->byte_count = cpu_to_be32(byte_count);
+		}
+		tx_info->inl = 0;
+	}
+
 	/*
 	 * For timestamping add flag to skb_shinfo and
 	 * set flag for further reference
@@ -666,6 +809,8 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	if (priv->flags & MLX4_EN_FLAG_ENABLE_HW_LOOPBACK) {
+		struct ethhdr *ethh;
+
 		/* Copy dst mac address to wqe. This allows loopback in eSwitch,
 		 * so that VFs and PF can communicate with each other
 		 */
@@ -688,8 +833,6 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 		/* Copy headers;
 		 * note that we already verified that it is linear */
 		memcpy(tx_desc->lso.header, skb->data, lso_header_size);
-		data = ((void *) &tx_desc->lso +
-			ALIGN(lso_header_size + 4, DS_SIZE));
 
 		priv->port_stats.tso_packets++;
 		i = ((skb->len - lso_header_size) / skb_shinfo(skb)->gso_size) +
@@ -701,7 +844,6 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 		op_own = cpu_to_be32(MLX4_OPCODE_SEND) |
 			((ring->prod & ring->size) ?
 			 cpu_to_be32(MLX4_EN_BIT_DESC_OWN) : 0);
-		data = &tx_desc->data;
 		tx_info->nr_bytes = max_t(unsigned int, skb->len, ETH_ZLEN);
 		ring->packets++;
 
@@ -710,42 +852,19 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 	netdev_tx_sent_queue(ring->tx_queue, tx_info->nr_bytes);
 	AVG_PERF_COUNTER(priv->pstats.tx_pktsz_avg, skb->len);
 
-
-	/* valid only for none inline segments */
-	tx_info->data_offset = (void *) data - (void *) tx_desc;
-
-	tx_info->linear = (lso_header_size < skb_headlen(skb) && !is_inline(skb, NULL)) ? 1 : 0;
-	data += skb_shinfo(skb)->nr_frags + tx_info->linear - 1;
-
-	if (!is_inline(skb, &fragptr)) {
-		/* Map fragments */
-		for (i = skb_shinfo(skb)->nr_frags - 1; i >= 0; i--) {
-			frag = &skb_shinfo(skb)->frags[i];
-			dma = skb_frag_dma_map(priv->ddev, frag,
-					       0, skb_frag_size(frag),
-					       DMA_TO_DEVICE);
-			data->addr = cpu_to_be64(dma);
-			data->lkey = cpu_to_be32(mdev->mr.key);
-			wmb();
-			data->byte_count = cpu_to_be32(skb_frag_size(frag));
-			--data;
-		}
-
-		/* Map linear part */
-		if (tx_info->linear) {
-			dma = dma_map_single(priv->ddev, skb->data + lso_header_size,
-					     skb_headlen(skb) - lso_header_size, PCI_DMA_TODEVICE);
-			data->addr = cpu_to_be64(dma);
-			data->lkey = cpu_to_be32(mdev->mr.key);
-			wmb();
-			data->byte_count = cpu_to_be32(skb_headlen(skb) - lso_header_size);
-		}
-		tx_info->inl = 0;
-	} else {
+	if (tx_info->inl) {
 		build_inline_wqe(tx_desc, skb, real_size, &vlan_tag, tx_ind, fragptr);
 		tx_info->inl = 1;
 	}
 
+	if (skb->encapsulation) {
+		struct iphdr *ipv4 = (struct iphdr *)skb_inner_network_header(skb);
+		if (ipv4->protocol == IPPROTO_TCP || ipv4->protocol == IPPROTO_UDP)
+			op_own |= cpu_to_be32(MLX4_WQE_CTRL_IIP | MLX4_WQE_CTRL_ILP);
+		else
+			op_own |= cpu_to_be32(MLX4_WQE_CTRL_IIP);
+	}
+
 	ring->prod += nr_txbb;
 
 	/* If we used a bounce buffer then copy descriptor back into place */
@@ -755,7 +874,8 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 	skb_tx_timestamp(skb);
 
 	if (ring->bf_enabled && desc_size <= MAX_BF && !bounce && !vlan_tx_tag_present(skb)) {
-		*(__be32 *) (&tx_desc->ctrl.vlan_tag) |= cpu_to_be32(ring->doorbell_qpn);
+		tx_desc->ctrl.bf_qpn |= cpu_to_be32(ring->doorbell_qpn);
+
 		op_own |= htonl((bf_index & 0xffff) << 8);
 		/* Ensure new descirptor hits memory
 		* before setting ownership of this descriptor to HW */
@@ -781,6 +901,16 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 
 	return NETDEV_TX_OK;
 
+tx_drop_unmap:
+	en_err(priv, "DMA mapping error\n");
+
+	for (i++; i < skb_shinfo(skb)->nr_frags; i++) {
+		data++;
+		dma_unmap_page(ddev, (dma_addr_t) be64_to_cpu(data->addr),
+			       be32_to_cpu(data->byte_count),
+			       PCI_DMA_TODEVICE);
+	}
+
 tx_drop:
 	dev_kfree_skb_any(skb);
 	priv->stats.tx_dropped++;