4 files changed, 43 insertions, 28 deletions

diff --git a/drivers/net/mlx4/en_netdev.c b/drivers/net/mlx4/en_netdev.c
index 15bb38d99304..9f6644a44030 100644
--- a/drivers/net/mlx4/en_netdev.c
+++ b/drivers/net/mlx4/en_netdev.c
@@ -952,6 +952,7 @@ static const struct net_device_ops mlx4_netdev_ops = {
 	.ndo_get_stats		= mlx4_en_get_stats,
 	.ndo_set_multicast_list	= mlx4_en_set_multicast,
 	.ndo_set_mac_address	= mlx4_en_set_mac,
+	.ndo_validate_addr	= eth_validate_addr,
 	.ndo_change_mtu		= mlx4_en_change_mtu,
 	.ndo_tx_timeout		= mlx4_en_tx_timeout,
 	.ndo_vlan_rx_register	= mlx4_en_vlan_rx_register,
diff --git a/drivers/net/mlx4/en_params.c b/drivers/net/mlx4/en_params.c
index cfeef0f1bacc..c1bd040b9e05 100644
--- a/drivers/net/mlx4/en_params.c
+++ b/drivers/net/mlx4/en_params.c
@@ -399,8 +399,10 @@ static int mlx4_en_set_ringparam(struct net_device *dev,
 
 	rx_size = roundup_pow_of_two(param->rx_pending);
 	rx_size = max_t(u32, rx_size, MLX4_EN_MIN_RX_SIZE);
+	rx_size = min_t(u32, rx_size, MLX4_EN_MAX_RX_SIZE);
 	tx_size = roundup_pow_of_two(param->tx_pending);
 	tx_size = max_t(u32, tx_size, MLX4_EN_MIN_TX_SIZE);
+	tx_size = min_t(u32, tx_size, MLX4_EN_MAX_TX_SIZE);
 
 	if (rx_size == priv->prof->rx_ring_size &&
 	    tx_size == priv->prof->tx_ring_size)
@@ -440,8 +442,8 @@ static void mlx4_en_get_ringparam(struct net_device *dev,
 	struct mlx4_en_dev *mdev = priv->mdev;
 
 	memset(param, 0, sizeof(*param));
-	param->rx_max_pending = mdev->dev->caps.max_rq_sg;
-	param->tx_max_pending = mdev->dev->caps.max_sq_sg;
+	param->rx_max_pending = MLX4_EN_MAX_RX_SIZE;
+	param->tx_max_pending = MLX4_EN_MAX_TX_SIZE;
 	param->rx_pending = mdev->profile.prof[priv->port].rx_ring_size;
 	param->tx_pending = mdev->profile.prof[priv->port].tx_ring_size;
 }
diff --git a/drivers/net/mlx4/en_tx.c b/drivers/net/mlx4/en_tx.c
index ff4d75205c25..4afd5993e31c 100644
--- a/drivers/net/mlx4/en_tx.c
+++ b/drivers/net/mlx4/en_tx.c
@@ -203,19 +203,21 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
 
 	/* Optimize the common case when there are no wraparounds */
 	if (likely((void *) tx_desc + tx_info->nr_txbb * TXBB_SIZE <= end)) {
-		if (tx_info->linear) {
-			pci_unmap_single(mdev->pdev,
-					 (dma_addr_t) be64_to_cpu(data->addr),
+		if (!tx_info->inl) {
+			if (tx_info->linear) {
+				pci_unmap_single(mdev->pdev,
+					(dma_addr_t) be64_to_cpu(data->addr),
 					 be32_to_cpu(data->byte_count),
 					 PCI_DMA_TODEVICE);
-			++data;
-		}
+				++data;
+			}
 
-		for (i = 0; i < frags; i++) {
-			frag = &skb_shinfo(skb)->frags[i];
-			pci_unmap_page(mdev->pdev,
-				       (dma_addr_t) be64_to_cpu(data[i].addr),
-				       frag->size, PCI_DMA_TODEVICE);
+			for (i = 0; i < frags; i++) {
+				frag = &skb_shinfo(skb)->frags[i];
+				pci_unmap_page(mdev->pdev,
+					(dma_addr_t) be64_to_cpu(data[i].addr),
+					frag->size, PCI_DMA_TODEVICE);
+			}
 		}
 		/* Stamp the freed descriptor */
 		for (i = 0; i < tx_info->nr_txbb * TXBB_SIZE; i += STAMP_STRIDE) {
@@ -224,27 +226,29 @@ static u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
 		}
 
 	} else {
-		if ((void *) data >= end) {
-			data = (struct mlx4_wqe_data_seg *)
-					(ring->buf + ((void *) data - end));
-		}
+		if (!tx_info->inl) {
+			if ((void *) data >= end) {
+				data = (struct mlx4_wqe_data_seg *)
+						(ring->buf + ((void *) data - end));
+			}
 
-		if (tx_info->linear) {
-			pci_unmap_single(mdev->pdev,
-					 (dma_addr_t) be64_to_cpu(data->addr),
+			if (tx_info->linear) {
+				pci_unmap_single(mdev->pdev,
+					(dma_addr_t) be64_to_cpu(data->addr),
 					 be32_to_cpu(data->byte_count),
 					 PCI_DMA_TODEVICE);
-			++data;
-		}
+				++data;
+			}
 
-		for (i = 0; i < frags; i++) {
-			/* Check for wraparound before unmapping */
-			if ((void *) data >= end)
-				data = (struct mlx4_wqe_data_seg *) ring->buf;
-			frag = &skb_shinfo(skb)->frags[i];
-			pci_unmap_page(mdev->pdev,
+			for (i = 0; i < frags; i++) {
+				/* Check for wraparound before unmapping */
+				if ((void *) data >= end)
+					data = (struct mlx4_wqe_data_seg *) ring->buf;
+				frag = &skb_shinfo(skb)->frags[i];
+				pci_unmap_page(mdev->pdev,
 					(dma_addr_t) be64_to_cpu(data->addr),
 					 frag->size, PCI_DMA_TODEVICE);
+			}
 		}
 		/* Stamp the freed descriptor */
 		for (i = 0; i < tx_info->nr_txbb * TXBB_SIZE; i += STAMP_STRIDE) {
@@ -790,8 +794,11 @@ int mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
 			wmb();
 			data->byte_count = cpu_to_be32(skb_headlen(skb) - lso_header_size);
 		}
-	} else
+		tx_info->inl = 0;
+	} else {
 		build_inline_wqe(tx_desc, skb, real_size, &vlan_tag, tx_ind, fragptr);
+		tx_info->inl = 1;
+	}
 
 	ring->prod += nr_txbb;
 
diff --git a/drivers/net/mlx4/mlx4_en.h b/drivers/net/mlx4/mlx4_en.h
index 2e96c7b2180a..e9af32d41ca4 100644
--- a/drivers/net/mlx4/mlx4_en.h
+++ b/drivers/net/mlx4/mlx4_en.h
@@ -115,6 +115,10 @@ enum {
 };
 #define MLX4_EN_MAX_RX_FRAGS	4
 
+/* Maximum ring sizes */
+#define MLX4_EN_MAX_TX_SIZE	8192
+#define MLX4_EN_MAX_RX_SIZE	8192
+
 /* Minimum ring size for our page-allocation sceme to work */
 #define MLX4_EN_MIN_RX_SIZE	(MLX4_EN_ALLOC_SIZE / SMP_CACHE_BYTES)
 #define MLX4_EN_MIN_TX_SIZE	(4096 / TXBB_SIZE)
@@ -202,6 +206,7 @@ struct mlx4_en_tx_info {
 	u32 nr_txbb;
 	u8 linear;
 	u8 data_offset;
+	u8 inl;
 };