net/mlx4_en: Optimize Rx fast path filter checks

Currently, RX path code that does RX filtering is not optimized
and does an expensive conversion. In order to use ether_addr_equal_64bits
which is optimized for such cases, we need the MAC address kept by the device
to be in the form of unsigned char array instead of u64. Store the MAC address
as unsigned char array and convert to/from u64 out of the fast path when needed.
Side effect of this is that we no longer need priv->mac, since it's the same
as dev->dev_addr.

This optimization was suggested by Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Yan Burman <yanb@mellanox.com>
Signed-off-by: Amir Vadai <amirv@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/mellanox/mlx4/en_ethtool.c

@@ -712,16 +712,13 @@ static int mlx4_en_ethtool_add_mac_rule_by_ipv4(struct mlx4_en_priv *priv,
 						__be32 ipv4_dst)
 {
 #ifdef CONFIG_INET
-	__be64 be_mac = 0;
 	unsigned char mac[ETH_ALEN];
 
 	if (!ipv4_is_multicast(ipv4_dst)) {
-		if (cmd->fs.flow_type & FLOW_MAC_EXT) {
+		if (cmd->fs.flow_type & FLOW_MAC_EXT)
 			memcpy(&mac, cmd->fs.h_ext.h_dest, ETH_ALEN);
-		} else {
-			be_mac = cpu_to_be64((priv->mac & MLX4_MAC_MASK) << 16);
-			memcpy(&mac, &be_mac, ETH_ALEN);
-		}
+		else
+			memcpy(&mac, priv->dev->dev_addr, ETH_ALEN);
 	} else {
 		ip_eth_mc_map(ipv4_dst, mac);
 	}

drivers/net/ethernet/mellanox/mlx4/en_netdev.c

@@ -132,17 +132,14 @@ static void mlx4_en_filter_work(struct work_struct *work)
 		.priority = MLX4_DOMAIN_RFS,
 	};
 	int rc;
-	__be64 mac;
 	__be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16);
 
 	list_add_tail(&spec_eth.list, &rule.list);
 	list_add_tail(&spec_ip.list, &rule.list);
 	list_add_tail(&spec_tcp.list, &rule.list);
 
-	mac = cpu_to_be64((priv->mac & MLX4_MAC_MASK) << 16);
-
 	rule.qpn = priv->rss_map.qps[filter->rxq_index].qpn;
-	memcpy(spec_eth.eth.dst_mac, &mac, ETH_ALEN);
+	memcpy(spec_eth.eth.dst_mac, priv->dev->dev_addr, ETH_ALEN);
 	memcpy(spec_eth.eth.dst_mac_msk, &mac_mask, ETH_ALEN);
 
 	filter->activated = 0;
@@ -413,6 +410,16 @@ static int mlx4_en_vlan_rx_kill_vid(struct net_device *dev, unsigned short vid)
 	return 0;
 }
 
+static void mlx4_en_u64_to_mac(unsigned char dst_mac[ETH_ALEN + 2], u64 src_mac)
+{
+	unsigned int i;
+	for (i = ETH_ALEN - 1; i; --i) {
+		dst_mac[i] = src_mac & 0xff;
+		src_mac >>= 8;
+	}
+	memset(&dst_mac[ETH_ALEN], 0, 2);
+}
+
 u64 mlx4_en_mac_to_u64(u8 *addr)
 {
 	u64 mac = 0;
@@ -435,7 +442,6 @@ static int mlx4_en_set_mac(struct net_device *dev, void *addr)
 		return -EADDRNOTAVAIL;
 
 	memcpy(dev->dev_addr, saddr->sa_data, ETH_ALEN);
-	priv->mac = mlx4_en_mac_to_u64(dev->dev_addr);
 	queue_work(mdev->workqueue, &priv->mac_task);
 	return 0;
 }
@@ -450,10 +456,13 @@ static void mlx4_en_do_set_mac(struct work_struct *work)
 	mutex_lock(&mdev->state_lock);
 	if (priv->port_up) {
 		/* Remove old MAC and insert the new one */
+		u64 mac = mlx4_en_mac_to_u64(priv->dev->dev_addr);
 		err = mlx4_replace_mac(mdev->dev, priv->port,
-				       priv->base_qpn, priv->mac);
+				       priv->base_qpn, mac);
 		if (err)
 			en_err(priv, "Failed changing HW MAC address\n");
+		memcpy(priv->prev_mac, priv->dev->dev_addr,
+		       sizeof(priv->prev_mac));
 	} else
 		en_dbg(HW, priv, "Port is down while "
 				 "registering mac, exiting...\n");
@@ -1031,6 +1040,7 @@ int mlx4_en_start_port(struct net_device *dev)
 	int i;
 	int j;
 	u8 mc_list[16] = {0};
+	u64 mac = mlx4_en_mac_to_u64(dev->dev_addr);
 
 	if (priv->port_up) {
 		en_dbg(DRV, priv, "start port called while port already up\n");
@@ -1078,7 +1088,7 @@ int mlx4_en_start_port(struct net_device *dev)
 	/* Set qp number */
 	en_dbg(DRV, priv, "Getting qp number for port %d\n", priv->port);
 	err = mlx4_get_eth_qp(mdev->dev, priv->port,
-				priv->mac, &priv->base_qpn);
+			      mac, &priv->base_qpn);
 	if (err) {
 		en_err(priv, "Failed getting eth qp\n");
 		goto cq_err;
@@ -1191,7 +1201,7 @@ int mlx4_en_start_port(struct net_device *dev)
 rss_err:
 	mlx4_en_release_rss_steer(priv);
 mac_err:
-	mlx4_put_eth_qp(mdev->dev, priv->port, priv->mac, priv->base_qpn);
+	mlx4_put_eth_qp(mdev->dev, priv->port, mac, priv->base_qpn);
 cq_err:
 	while (rx_index--)
 		mlx4_en_deactivate_cq(priv, &priv->rx_cq[rx_index]);
@@ -1210,6 +1220,7 @@ void mlx4_en_stop_port(struct net_device *dev, int detach)
 	struct ethtool_flow_id *flow, *tmp_flow;
 	int i;
 	u8 mc_list[16] = {0};
+	u64 mac = mlx4_en_mac_to_u64(dev->dev_addr);
 
 	if (!priv->port_up) {
 		en_dbg(DRV, priv, "stop port called while port already down\n");
@@ -1290,7 +1301,7 @@ void mlx4_en_stop_port(struct net_device *dev, int detach)
 	mlx4_en_release_rss_steer(priv);
 
 	/* Unregister Mac address for the port */
-	mlx4_put_eth_qp(mdev->dev, priv->port, priv->mac, priv->base_qpn);
+	mlx4_put_eth_qp(mdev->dev, priv->port, mac, priv->base_qpn);
 	if (!(mdev->dev->caps.flags2 & MLX4_DEV_CAP_FLAGS2_REASSIGN_MAC_EN))
 		mdev->mac_removed[priv->port] = 1;
 
@@ -1597,7 +1608,6 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
 {
 	struct net_device *dev;
 	struct mlx4_en_priv *priv;
-	int i;
 	int err;
 
 	dev = alloc_etherdev_mqs(sizeof(struct mlx4_en_priv),
@@ -1658,14 +1668,19 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
 
 	/* Query for default mac and max mtu */
 	priv->max_mtu = mdev->dev->caps.eth_mtu_cap[priv->port];
-	priv->mac = mdev->dev->caps.def_mac[priv->port];
-	if (ILLEGAL_MAC(priv->mac)) {
-		en_err(priv, "Port: %d, invalid mac burned: 0x%llx, quiting\n",
-			 priv->port, priv->mac);
+
+	/* Set default MAC */
+	dev->addr_len = ETH_ALEN;
+	mlx4_en_u64_to_mac(dev->dev_addr, mdev->dev->caps.def_mac[priv->port]);
+	if (!is_valid_ether_addr(dev->dev_addr)) {
+		en_err(priv, "Port: %d, invalid mac burned: %pM, quiting\n",
+		       priv->port, dev->dev_addr);
 		err = -EINVAL;
 		goto out;
 	}
 
+	memcpy(priv->prev_mac, dev->dev_addr, sizeof(priv->prev_mac));
+
 	priv->stride = roundup_pow_of_two(sizeof(struct mlx4_en_rx_desc) +
 					  DS_SIZE * MLX4_EN_MAX_RX_FRAGS);
 	err = mlx4_en_alloc_resources(priv);
@@ -1696,11 +1711,6 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
 
 	SET_ETHTOOL_OPS(dev, &mlx4_en_ethtool_ops);
 
-	/* Set defualt MAC */
-	dev->addr_len = ETH_ALEN;
-	for (i = 0; i < ETH_ALEN; i++)
-		dev->dev_addr[ETH_ALEN - 1 - i] = (u8) (priv->mac >> (8 * i));
-
 	/*
 	 * Set driver features
 	 */

drivers/net/ethernet/mellanox/mlx4/en_rx.c

@@ -606,7 +606,6 @@ int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int bud
 		if (priv->flags & MLX4_EN_FLAG_RX_FILTER_NEEDED) {
 			struct ethhdr *ethh;
 			dma_addr_t dma;
-			u64 s_mac;
 			/* Get pointer to first fragment since we haven't
 			 * skb yet and cast it to ethhdr struct
 			 */
@@ -617,8 +616,8 @@ int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int bud
 						 frags[0].offset);
 
 			/* Drop the packet, since HW loopback-ed it */
-			s_mac = mlx4_en_mac_to_u64(ethh->h_source);
-			if (s_mac == priv->mac)
+			if (ether_addr_equal_64bits(dev->dev_addr,
+						    ethh->h_source))
 				goto next;
 		}
 

drivers/net/ethernet/mellanox/mlx4/mlx4_en.h

@@ -198,7 +198,6 @@ enum cq_type {
  */
 #define ROUNDUP_LOG2(x)		ilog2(roundup_pow_of_two(x))
 #define XNOR(x, y)		(!(x) == !(y))
-#define ILLEGAL_MAC(addr)	(addr == 0xffffffffffffULL || addr == 0x0)
 
 
 struct mlx4_en_tx_info {
@@ -483,7 +482,7 @@ struct mlx4_en_priv {
 	int registered;
 	int allocated;
 	int stride;
-	u64 mac;
+	unsigned char prev_mac[ETH_ALEN + 2];
 	int mac_index;
 	unsigned max_mtu;
 	int base_qpn;