Merge branch 'mlx4'

Amir Vadai says:

====================
net/mlx4: Mellanox driver update 07-11-2013

This patchset contains some enhancements and bug fixes for the mlx4_* drivers.
Patchset was applied and tested against commit: "9bb8ca8 virtio-net: switch to
use XPS to choose txq"
====================

Signed-off-by: David S. Miller <[e-mail]>

drivers/infiniband/hw/mlx4/main.c

···
 	if (IS_ERR(mailbox))
 		return 0;
 
-	memset(mailbox->buf, 0, 256);
 	memcpy(mailbox->buf, props->node_desc, 64);
 	mlx4_cmd(to_mdev(ibdev)->dev, mailbox->dma, 1, 0,
 		 MLX4_CMD_SET_NODE, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
···
 	mailbox = mlx4_alloc_cmd_mailbox(dev->dev);
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
-
-	memset(mailbox->buf, 0, 256);
 
 	if (dev->dev->flags & MLX4_FLAG_OLD_PORT_CMDS) {
 		*(u8 *) mailbox->buf	     = !!reset_qkey_viols << 6;
···
 	struct mlx4_ib_dev *mdev = to_mdev(qp->device);
 	struct mlx4_cmd_mailbox *mailbox;
 	struct mlx4_net_trans_rule_hw_ctrl *ctrl;
-	size_t rule_size = sizeof(struct mlx4_net_trans_rule_hw_ctrl) +
-			   (sizeof(struct _rule_hw) * flow_attr->num_of_specs);
 
 	static const u16 __mlx4_domain[] = {
 		[IB_FLOW_DOMAIN_USER] = MLX4_DOMAIN_UVERBS,
···
 	mailbox = mlx4_alloc_cmd_mailbox(mdev->dev);
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
-	memset(mailbox->buf, 0, rule_size);
 	ctrl = mailbox->buf;
 
 	ctrl->prio = cpu_to_be16(__mlx4_domain[domain] |

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

···
 	return ret;
 }
 
-static int calculate_transition(u16 oper_vlan, u16 admin_vlan)
-{
-	return (2 * (oper_vlan == MLX4_VGT) + (admin_vlan == MLX4_VGT));
-}
-
 static int mlx4_master_immediate_activate_vlan_qos(struct mlx4_priv *priv,
 					    int slave, int port)
 {
···
 	struct mlx4_dev *dev = &(priv->dev);
 	int err;
 	int admin_vlan_ix = NO_INDX;
-	enum mlx4_vlan_transition vlan_trans;
 
 	vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
 	vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
···
 	    vp_oper->state.link_state == vp_admin->link_state)
 		return 0;
 
-	vlan_trans = calculate_transition(vp_oper->state.default_vlan,
-					  vp_admin->default_vlan);
-
 	if (!(priv->mfunc.master.slave_state[slave].active &&
-	      dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP &&
-	      vlan_trans == MLX4_VLAN_TRANSITION_VST_VST)) {
+	      dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_UPDATE_QP)) {
 		/* even if the UPDATE_QP command isn't supported, we still want
 		 * to set this VF link according to the admin directive
 		 */
···
 		return -ENOMEM;
 
 	if (vp_oper->state.default_vlan != vp_admin->default_vlan) {
-		err = __mlx4_register_vlan(&priv->dev, port,
-					   vp_admin->default_vlan,
-					   &admin_vlan_ix);
-		if (err) {
-			kfree(work);
-			mlx4_warn((&priv->dev),
-				  "No vlan resources slave %d, port %d\n",
-				  slave, port);
-			return err;
+		if (MLX4_VGT != vp_admin->default_vlan) {
+			err = __mlx4_register_vlan(&priv->dev, port,
+						   vp_admin->default_vlan,
+						   &admin_vlan_ix);
+			if (err) {
+				kfree(work);
+				mlx4_warn((&priv->dev),
+					  "No vlan resources slave %d, port %d\n",
+					  slave, port);
+				return err;
+			}
+		} else {
+			admin_vlan_ix = NO_INDX;
 		}
 		work->flags |= MLX4_VF_IMMED_VLAN_FLAG_VLAN;
 		mlx4_dbg((&(priv->dev)),
···
 		kfree(mailbox);
 		return ERR_PTR(-ENOMEM);
 	}
+
+	memset(mailbox->buf, 0, MLX4_MAILBOX_SIZE);
 
 	return mailbox;
 }

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

···
 		return PTR_ERR(mailbox);
 
 	cq_context = mailbox->buf;
-	memset(cq_context, 0, sizeof *cq_context);
-
 	cq_context->cq_max_count = cpu_to_be16(count);
 	cq_context->cq_period    = cpu_to_be16(period);
 
···
 		return PTR_ERR(mailbox);
 
 	cq_context = mailbox->buf;
-	memset(cq_context, 0, sizeof *cq_context);
-
 	cq_context->logsize_usrpage = cpu_to_be32(ilog2(entries) << 24);
 	cq_context->log_page_size   = mtt->page_shift - 12;
 	mtt_addr = mlx4_mtt_addr(dev, mtt);
···
 	}
 
 	cq_context = mailbox->buf;
-	memset(cq_context, 0, sizeof *cq_context);
-
 	cq_context->flags	    = cpu_to_be32(!!collapsed << 18);
 	if (timestamp_en)
 		cq_context->flags  |= cpu_to_be32(1 << 19);

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

···
 
 
 int mlx4_en_create_cq(struct mlx4_en_priv *priv,
-		      struct mlx4_en_cq *cq,
-		      int entries, int ring, enum cq_type mode)
+		      struct mlx4_en_cq **pcq,
+		      int entries, int ring, enum cq_type mode,
+		      int node)
 {
 	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_cq *cq;
 	int err;
+
+	cq = kzalloc_node(sizeof(*cq), GFP_KERNEL, node);
+	if (!cq) {
+		cq = kzalloc(sizeof(*cq), GFP_KERNEL);
+		if (!cq) {
+			en_err(priv, "Failed to allocate CQ structure\n");
+			return -ENOMEM;
+		}
+	}
 
 	cq->size = entries;
 	cq->buf_size = cq->size * mdev->dev->caps.cqe_size;
···
 	cq->is_tx = mode;
 	spin_lock_init(&cq->lock);
 
+	/* Allocate HW buffers on provided NUMA node.
+	 * dev->numa_node is used in mtt range allocation flow.
+	 */
+	set_dev_node(&mdev->dev->pdev->dev, node);
 	err = mlx4_alloc_hwq_res(mdev->dev, &cq->wqres,
 				cq->buf_size, 2 * PAGE_SIZE);
+	set_dev_node(&mdev->dev->pdev->dev, mdev->dev->numa_node);
 	if (err)
-		return err;
+		goto err_cq;
 
 	err = mlx4_en_map_buffer(&cq->wqres.buf);
 	if (err)
-		mlx4_free_hwq_res(mdev->dev, &cq->wqres, cq->buf_size);
-	else
-		cq->buf = (struct mlx4_cqe *) cq->wqres.buf.direct.buf;
+		goto err_res;
 
+	cq->buf = (struct mlx4_cqe *)cq->wqres.buf.direct.buf;
+	*pcq = cq;
+
+	return 0;
+
+err_res:
+	mlx4_free_hwq_res(mdev->dev, &cq->wqres, cq->buf_size);
+err_cq:
+	kfree(cq);
+	*pcq = NULL;
 	return err;
 }
 
···
 		struct mlx4_en_cq *rx_cq;
 
 		cq_idx = cq_idx % priv->rx_ring_num;
-		rx_cq = &priv->rx_cq[cq_idx];
+		rx_cq = priv->rx_cq[cq_idx];
 		cq->vector = rx_cq->vector;
 	}
 
 	if (!cq->is_tx)
-		cq->size = priv->rx_ring[cq->ring].actual_size;
+		cq->size = priv->rx_ring[cq->ring]->actual_size;
 
 	if ((cq->is_tx && priv->hwtstamp_config.tx_type) ||
 	    (!cq->is_tx && priv->hwtstamp_config.rx_filter))
···
 	return 0;
 }
 
-void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
+void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq)
 {
 	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_cq *cq = *pcq;
 
 	mlx4_en_unmap_buffer(&cq->wqres.buf);
 	mlx4_free_hwq_res(mdev->dev, &cq->wqres, cq->buf_size);
···
 	cq->vector = 0;
 	cq->buf_size = 0;
 	cq->buf = NULL;
+	kfree(cq);
+	*pcq = NULL;
 }
 
 void mlx4_en_deactivate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)

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

···
 	int err = 0;
 
 	for (i = 0; i < priv->tx_ring_num; i++) {
-		priv->tx_cq[i].moder_cnt = priv->tx_frames;
-		priv->tx_cq[i].moder_time = priv->tx_usecs;
+		priv->tx_cq[i]->moder_cnt = priv->tx_frames;
+		priv->tx_cq[i]->moder_time = priv->tx_usecs;
 		if (priv->port_up) {
-			err = mlx4_en_set_cq_moder(priv, &priv->tx_cq[i]);
+			err = mlx4_en_set_cq_moder(priv, priv->tx_cq[i]);
 			if (err)
 				return err;
 		}
···
 		return 0;
 
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		priv->rx_cq[i].moder_cnt = priv->rx_frames;
-		priv->rx_cq[i].moder_time = priv->rx_usecs;
+		priv->rx_cq[i]->moder_cnt = priv->rx_frames;
+		priv->rx_cq[i]->moder_time = priv->rx_usecs;
 		priv->last_moder_time[i] = MLX4_EN_AUTO_CONF;
 		if (priv->port_up) {
-			err = mlx4_en_set_cq_moder(priv, &priv->rx_cq[i]);
+			err = mlx4_en_set_cq_moder(priv, priv->rx_cq[i]);
 			if (err)
 				return err;
 		}
···
 		}
 	}
 	for (i = 0; i < priv->tx_ring_num; i++) {
-		data[index++] = priv->tx_ring[i].packets;
-		data[index++] = priv->tx_ring[i].bytes;
+		data[index++] = priv->tx_ring[i]->packets;
+		data[index++] = priv->tx_ring[i]->bytes;
 	}
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		data[index++] = priv->rx_ring[i].packets;
-		data[index++] = priv->rx_ring[i].bytes;
+		data[index++] = priv->rx_ring[i]->packets;
+		data[index++] = priv->rx_ring[i]->bytes;
 #ifdef CONFIG_NET_RX_BUSY_POLL
-		data[index++] = priv->rx_ring[i].yields;
-		data[index++] = priv->rx_ring[i].misses;
-		data[index++] = priv->rx_ring[i].cleaned;
+		data[index++] = priv->rx_ring[i]->yields;
+		data[index++] = priv->rx_ring[i]->misses;
+		data[index++] = priv->rx_ring[i]->cleaned;
 #endif
 	}
 	spin_unlock_bh(&priv->stats_lock);
···
 	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->port_up ? priv->rx_ring[0].actual_size :
-					priv->rx_ring[0].size) &&
-	    tx_size == priv->tx_ring[0].size)
+	if (rx_size == (priv->port_up ? priv->rx_ring[0]->actual_size :
+					priv->rx_ring[0]->size) &&
+	    tx_size == priv->tx_ring[0]->size)
 		return 0;
 
 	mutex_lock(&mdev->state_lock);
···
 	param->rx_max_pending = MLX4_EN_MAX_RX_SIZE;
 	param->tx_max_pending = MLX4_EN_MAX_TX_SIZE;
 	param->rx_pending = priv->port_up ?
-		priv->rx_ring[0].actual_size : priv->rx_ring[0].size;
-	param->tx_pending = priv->tx_ring[0].size;
+		priv->rx_ring[0]->actual_size : priv->rx_ring[0]->size;
+	param->tx_pending = priv->tx_ring[0]->size;
 }
 
 static u32 mlx4_en_get_rxfh_indir_size(struct net_device *dev)

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

···
 	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);
-	struct mlx4_en_rx_ring *rx_ring = &priv->rx_ring[cq->ring];
+	struct mlx4_en_rx_ring *rx_ring = priv->rx_ring[cq->ring];
 	int done;
 
 	if (!priv->port_up)
···
 	struct list_head next;
 	struct work_struct work;
 
+	u8     ip_proto;
 	__be32 src_ip;
 	__be32 dst_ip;
 	__be16 src_port;
···
 
 static void mlx4_en_filter_rfs_expire(struct mlx4_en_priv *priv);
 
+static enum mlx4_net_trans_rule_id mlx4_ip_proto_to_trans_rule_id(u8 ip_proto)
+{
+	switch (ip_proto) {
+	case IPPROTO_UDP:
+		return MLX4_NET_TRANS_RULE_ID_UDP;
+	case IPPROTO_TCP:
+		return MLX4_NET_TRANS_RULE_ID_TCP;
+	default:
+		return -EPROTONOSUPPORT;
+	}
+};
+
 static void mlx4_en_filter_work(struct work_struct *work)
 {
 	struct mlx4_en_filter *filter = container_of(work,
 						     struct mlx4_en_filter,
 						     work);
 	struct mlx4_en_priv *priv = filter->priv;
-	struct mlx4_spec_list spec_tcp = {
-		.id = MLX4_NET_TRANS_RULE_ID_TCP,
+	struct mlx4_spec_list spec_tcp_udp = {
+		.id = mlx4_ip_proto_to_trans_rule_id(filter->ip_proto),
 		{
 			.tcp_udp = {
 				.dst_port = filter->dst_port,
···
 	int rc;
 	__be64 mac_mask = cpu_to_be64(MLX4_MAC_MASK << 16);
 
+	if (spec_tcp_udp.id < 0) {
+		en_warn(priv, "RFS: ignoring unsupported ip protocol (%d)\n",
+			filter->ip_proto);
+		goto ignore;
+	}
 	list_add_tail(&spec_eth.list, &rule.list);
 	list_add_tail(&spec_ip.list, &rule.list);
-	list_add_tail(&spec_tcp.list, &rule.list);
+	list_add_tail(&spec_tcp_udp.list, &rule.list);
 
 	rule.qpn = priv->rss_map.qps[filter->rxq_index].qpn;
 	memcpy(spec_eth.eth.dst_mac, priv->dev->dev_addr, ETH_ALEN);
···
 	if (rc)
 		en_err(priv, "Error attaching flow. err = %d\n", rc);
 
+ignore:
 	mlx4_en_filter_rfs_expire(priv);
 
 	filter->activated = 1;
···
 
 static struct mlx4_en_filter *
 mlx4_en_filter_alloc(struct mlx4_en_priv *priv, int rxq_index, __be32 src_ip,
-		     __be32 dst_ip, __be16 src_port, __be16 dst_port,
-		     u32 flow_id)
+		     __be32 dst_ip, u8 ip_proto, __be16 src_port,
+		     __be16 dst_port, u32 flow_id)
 {
 	struct mlx4_en_filter *filter = NULL;
 
···
 
 	filter->src_ip = src_ip;
 	filter->dst_ip = dst_ip;
+	filter->ip_proto = ip_proto;
 	filter->src_port = src_port;
 	filter->dst_port = dst_port;
 
···
 
 static inline struct mlx4_en_filter *
 mlx4_en_filter_find(struct mlx4_en_priv *priv, __be32 src_ip, __be32 dst_ip,
-		    __be16 src_port, __be16 dst_port)
+		    u8 ip_proto, __be16 src_port, __be16 dst_port)
 {
 	struct mlx4_en_filter *filter;
 	struct mlx4_en_filter *ret = NULL;
···
 			     filter_chain) {
 		if (filter->src_ip == src_ip &&
 		    filter->dst_ip == dst_ip &&
+		    filter->ip_proto == ip_proto &&
 		    filter->src_port == src_port &&
 		    filter->dst_port == dst_port) {
 			ret = filter;
···
 	struct mlx4_en_filter *filter;
 	const struct iphdr *ip;
 	const __be16 *ports;
+	u8 ip_proto;
 	__be32 src_ip;
 	__be32 dst_ip;
 	__be16 src_port;
···
 	if (ip_is_fragment(ip))
 		return -EPROTONOSUPPORT;
 
+	if ((ip->protocol != IPPROTO_TCP) && (ip->protocol != IPPROTO_UDP))
+		return -EPROTONOSUPPORT;
 	ports = (const __be16 *)(skb->data + nhoff + 4 * ip->ihl);
 
+	ip_proto = ip->protocol;
 	src_ip = ip->saddr;
 	dst_ip = ip->daddr;
 	src_port = ports[0];
 	dst_port = ports[1];
 
-	if (ip->protocol != IPPROTO_TCP)
-		return -EPROTONOSUPPORT;
-
 	spin_lock_bh(&priv->filters_lock);
-	filter = mlx4_en_filter_find(priv, src_ip, dst_ip, src_port, dst_port);
+	filter = mlx4_en_filter_find(priv, src_ip, dst_ip, ip_proto,
+				     src_port, dst_port);
 	if (filter) {
 		if (filter->rxq_index == rxq_index)
 			goto out;
···
 		filter->rxq_index = rxq_index;
 	} else {
 		filter = mlx4_en_filter_alloc(priv, rxq_index,
-					      src_ip, dst_ip,
+					      src_ip, dst_ip, ip_proto,
 					      src_port, dst_port, flow_id);
 		if (!filter) {
 			ret = -ENOMEM;
···
 	return ret;
 }
 
-void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv,
-			     struct mlx4_en_rx_ring *rx_ring)
+void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv)
 {
 	struct mlx4_en_filter *filter, *tmp;
 	LIST_HEAD(del_list);
···
 	int i;
 
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		cq = &priv->rx_cq[i];
+		cq = priv->rx_cq[i];
 		spin_lock_irqsave(&cq->lock, flags);
 		napi_synchronize(&cq->napi);
 		mlx4_en_process_rx_cq(dev, cq, 0);
···
 		if (!netif_tx_queue_stopped(netdev_get_tx_queue(dev, i)))
 			continue;
 		en_warn(priv, "TX timeout on queue: %d, QP: 0x%x, CQ: 0x%x, Cons: 0x%x, Prod: 0x%x\n",
-			i, priv->tx_ring[i].qpn, priv->tx_ring[i].cqn,
-			priv->tx_ring[i].cons, priv->tx_ring[i].prod);
+			i, priv->tx_ring[i]->qpn, priv->tx_ring[i]->cqn,
+			priv->tx_ring[i]->cons, priv->tx_ring[i]->prod);
 	}
 
 	priv->port_stats.tx_timeout++;
···
 
 	/* Setup cq moderation params */
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		cq = &priv->rx_cq[i];
+		cq = priv->rx_cq[i];
 		cq->moder_cnt = priv->rx_frames;
 		cq->moder_time = priv->rx_usecs;
 		priv->last_moder_time[i] = MLX4_EN_AUTO_CONF;
···
 	}
 
 	for (i = 0; i < priv->tx_ring_num; i++) {
-		cq = &priv->tx_cq[i];
+		cq = priv->tx_cq[i];
 		cq->moder_cnt = priv->tx_frames;
 		cq->moder_time = priv->tx_usecs;
 	}
···
 
 	for (ring = 0; ring < priv->rx_ring_num; ring++) {
 		spin_lock_bh(&priv->stats_lock);
-		rx_packets = priv->rx_ring[ring].packets;
-		rx_bytes = priv->rx_ring[ring].bytes;
+		rx_packets = priv->rx_ring[ring]->packets;
+		rx_bytes = priv->rx_ring[ring]->bytes;
 		spin_unlock_bh(&priv->stats_lock);
 
 		rx_pkt_diff = ((unsigned long) (rx_packets -
···
 
 		if (moder_time != priv->last_moder_time[ring]) {
 			priv->last_moder_time[ring] = moder_time;
-			cq = &priv->rx_cq[ring];
+			cq = priv->rx_cq[ring];
 			cq->moder_time = moder_time;
 			cq->moder_cnt = priv->rx_frames;
 			err = mlx4_en_set_cq_moder(priv, cq);
···
 		return err;
 	}
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		cq = &priv->rx_cq[i];
+		cq = priv->rx_cq[i];
 
 		mlx4_en_cq_init_lock(cq);
 
···
 			goto cq_err;
 		}
 		mlx4_en_arm_cq(priv, cq);
-		priv->rx_ring[i].cqn = cq->mcq.cqn;
+		priv->rx_ring[i]->cqn = cq->mcq.cqn;
 		++rx_index;
 	}
 
···
 	/* Configure tx cq's and rings */
 	for (i = 0; i < priv->tx_ring_num; i++) {
 		/* Configure cq */
-		cq = &priv->tx_cq[i];
+		cq = priv->tx_cq[i];
 		err = mlx4_en_activate_cq(priv, cq, i);
 		if (err) {
 			en_err(priv, "Failed allocating Tx CQ\n");
···
 		cq->buf->wqe_index = cpu_to_be16(0xffff);
 
 		/* Configure ring */
-		tx_ring = &priv->tx_ring[i];
+		tx_ring = priv->tx_ring[i];
 		err = mlx4_en_activate_tx_ring(priv, tx_ring, cq->mcq.cqn,
 			i / priv->num_tx_rings_p_up);
 		if (err) {
···
 
 tx_err:
 	while (tx_index--) {
-		mlx4_en_deactivate_tx_ring(priv, &priv->tx_ring[tx_index]);
-		mlx4_en_deactivate_cq(priv, &priv->tx_cq[tx_index]);
+		mlx4_en_deactivate_tx_ring(priv, priv->tx_ring[tx_index]);
+		mlx4_en_deactivate_cq(priv, priv->tx_cq[tx_index]);
 	}
 	mlx4_en_destroy_drop_qp(priv);
 rss_err:
···
 	mlx4_en_put_qp(priv);
 cq_err:
 	while (rx_index--)
-		mlx4_en_deactivate_cq(priv, &priv->rx_cq[rx_index]);
+		mlx4_en_deactivate_cq(priv, priv->rx_cq[rx_index]);
 	for (i = 0; i < priv->rx_ring_num; i++)
-		mlx4_en_deactivate_rx_ring(priv, &priv->rx_ring[i]);
+		mlx4_en_deactivate_rx_ring(priv, priv->rx_ring[i]);
 
 	return err; /* need to close devices */
 }
···
 
 	/* Free TX Rings */
 	for (i = 0; i < priv->tx_ring_num; i++) {
-		mlx4_en_deactivate_tx_ring(priv, &priv->tx_ring[i]);
-		mlx4_en_deactivate_cq(priv, &priv->tx_cq[i]);
+		mlx4_en_deactivate_tx_ring(priv, priv->tx_ring[i]);
+		mlx4_en_deactivate_cq(priv, priv->tx_cq[i]);
 	}
 	msleep(10);
 
 	for (i = 0; i < priv->tx_ring_num; i++)
-		mlx4_en_free_tx_buf(dev, &priv->tx_ring[i]);
+		mlx4_en_free_tx_buf(dev, priv->tx_ring[i]);
 
 	/* Free RSS qps */
 	mlx4_en_release_rss_steer(priv);
···
 
 	/* Free RX Rings */
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		struct mlx4_en_cq *cq = &priv->rx_cq[i];
+		struct mlx4_en_cq *cq = priv->rx_cq[i];
 
 		local_bh_disable();
 		while (!mlx4_en_cq_lock_napi(cq)) {
···
 
 		while (test_bit(NAPI_STATE_SCHED, &cq->napi.state))
 			msleep(1);
-		mlx4_en_deactivate_rx_ring(priv, &priv->rx_ring[i]);
+		mlx4_en_deactivate_rx_ring(priv, priv->rx_ring[i]);
 		mlx4_en_deactivate_cq(priv, cq);
 	}
 }
···
 	memset(&priv->port_stats, 0, sizeof(priv->port_stats));
 
 	for (i = 0; i < priv->tx_ring_num; i++) {
-		priv->tx_ring[i].bytes = 0;
-		priv->tx_ring[i].packets = 0;
-		priv->tx_ring[i].tx_csum = 0;
+		priv->tx_ring[i]->bytes = 0;
+		priv->tx_ring[i]->packets = 0;
+		priv->tx_ring[i]->tx_csum = 0;
 	}
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		priv->rx_ring[i].bytes = 0;
-		priv->rx_ring[i].packets = 0;
-		priv->rx_ring[i].csum_ok = 0;
-		priv->rx_ring[i].csum_none = 0;
+		priv->rx_ring[i]->bytes = 0;
+		priv->rx_ring[i]->packets = 0;
+		priv->rx_ring[i]->csum_ok = 0;
+		priv->rx_ring[i]->csum_none = 0;
 	}
 }
 
···
 #endif
 
 	for (i = 0; i < priv->tx_ring_num; i++) {
-		if (priv->tx_ring[i].tx_info)
+		if (priv->tx_ring && priv->tx_ring[i])
 			mlx4_en_destroy_tx_ring(priv, &priv->tx_ring[i]);
-		if (priv->tx_cq[i].buf)
+		if (priv->tx_cq && priv->tx_cq[i])
 			mlx4_en_destroy_cq(priv, &priv->tx_cq[i]);
 	}
 
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		if (priv->rx_ring[i].rx_info)
+		if (priv->rx_ring[i])
 			mlx4_en_destroy_rx_ring(priv, &priv->rx_ring[i],
 				priv->prof->rx_ring_size, priv->stride);
-		if (priv->rx_cq[i].buf)
+		if (priv->rx_cq[i])
 			mlx4_en_destroy_cq(priv, &priv->rx_cq[i]);
 	}
 
···
 	struct mlx4_en_port_profile *prof = priv->prof;
 	int i;
 	int err;
+	int node;
 
 	err = mlx4_qp_reserve_range(priv->mdev->dev, priv->tx_ring_num, 256, &priv->base_tx_qpn);
 	if (err) {
···
 
 	/* Create tx Rings */
 	for (i = 0; i < priv->tx_ring_num; i++) {
+		node = cpu_to_node(i % num_online_cpus());
 		if (mlx4_en_create_cq(priv, &priv->tx_cq[i],
-				      prof->tx_ring_size, i, TX))
+				      prof->tx_ring_size, i, TX, node))
 			goto err;
 
 		if (mlx4_en_create_tx_ring(priv, &priv->tx_ring[i], priv->base_tx_qpn + i,
-					   prof->tx_ring_size, TXBB_SIZE))
+					   prof->tx_ring_size, TXBB_SIZE, node))
 			goto err;
 	}
 
 	/* Create rx Rings */
 	for (i = 0; i < priv->rx_ring_num; i++) {
+		node = cpu_to_node(i % num_online_cpus());
 		if (mlx4_en_create_cq(priv, &priv->rx_cq[i],
-				      prof->rx_ring_size, i, RX))
+				      prof->rx_ring_size, i, RX, node))
 			goto err;
 
 		if (mlx4_en_create_rx_ring(priv, &priv->rx_ring[i],
-					   prof->rx_ring_size, priv->stride))
+					   prof->rx_ring_size, priv->stride,
+					   node))
 			goto err;
 	}
 
···
 
 err:
 	en_err(priv, "Failed to allocate NIC resources\n");
+	for (i = 0; i < priv->rx_ring_num; i++) {
+		if (priv->rx_ring[i])
+			mlx4_en_destroy_rx_ring(priv, &priv->rx_ring[i],
+						prof->rx_ring_size,
+						priv->stride);
+		if (priv->rx_cq[i])
+			mlx4_en_destroy_cq(priv, &priv->rx_cq[i]);
+	}
+	for (i = 0; i < priv->tx_ring_num; i++) {
+		if (priv->tx_ring[i])
+			mlx4_en_destroy_tx_ring(priv, &priv->tx_ring[i]);
+		if (priv->tx_cq[i])
+			mlx4_en_destroy_cq(priv, &priv->tx_cq[i]);
+	}
 	return -ENOMEM;
 }
 
···
 	priv->num_tx_rings_p_up = mdev->profile.num_tx_rings_p_up;
 	priv->tx_ring_num = prof->tx_ring_num;
 
-	priv->tx_ring = kzalloc(sizeof(struct mlx4_en_tx_ring) * MAX_TX_RINGS,
+	priv->tx_ring = kzalloc(sizeof(struct mlx4_en_tx_ring *) * MAX_TX_RINGS,
 				GFP_KERNEL);
 	if (!priv->tx_ring) {
 		err = -ENOMEM;
 		goto out;
 	}
-	priv->tx_cq = kzalloc(sizeof(struct mlx4_en_cq) * MAX_TX_RINGS,
+	priv->tx_cq = kzalloc(sizeof(struct mlx4_en_cq *) * MAX_TX_RINGS,
 			      GFP_KERNEL);
 	if (!priv->tx_cq) {
 		err = -ENOMEM;

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

···
 		return PTR_ERR(mailbox);
 
 	filter = mailbox->buf;
-	memset(filter, 0, sizeof(*filter));
 	for (i = VLAN_FLTR_SIZE - 1; i >= 0; i--) {
 		entry = 0;
 		for (j = 0; j < 32; j++)
···
 	mailbox = mlx4_alloc_cmd_mailbox(mdev->dev);
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
-	memset(mailbox->buf, 0, sizeof(*qport_context));
 	err = mlx4_cmd_box(mdev->dev, 0, mailbox->dma, port, 0,
 			   MLX4_CMD_QUERY_PORT, MLX4_CMD_TIME_CLASS_B,
 			   MLX4_CMD_WRAPPED);
···
 	mailbox = mlx4_alloc_cmd_mailbox(mdev->dev);
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
-	memset(mailbox->buf, 0, sizeof(*mlx4_en_stats));
 	err = mlx4_cmd_box(mdev->dev, 0, mailbox->dma, in_mod, 0,
 			   MLX4_CMD_DUMP_ETH_STATS, MLX4_CMD_TIME_CLASS_B,
 			   MLX4_CMD_WRAPPED);
···
 	priv->port_stats.rx_chksum_good = 0;
 	priv->port_stats.rx_chksum_none = 0;
 	for (i = 0; i < priv->rx_ring_num; i++) {
-		stats->rx_packets += priv->rx_ring[i].packets;
-		stats->rx_bytes += priv->rx_ring[i].bytes;
-		priv->port_stats.rx_chksum_good += priv->rx_ring[i].csum_ok;
-		priv->port_stats.rx_chksum_none += priv->rx_ring[i].csum_none;
+		stats->rx_packets += priv->rx_ring[i]->packets;
+		stats->rx_bytes += priv->rx_ring[i]->bytes;
+		priv->port_stats.rx_chksum_good += priv->rx_ring[i]->csum_ok;
+		priv->port_stats.rx_chksum_none += priv->rx_ring[i]->csum_none;
 	}
 	stats->tx_packets = 0;
 	stats->tx_bytes = 0;
 	priv->port_stats.tx_chksum_offload = 0;
 	for (i = 0; i < priv->tx_ring_num; i++) {
-		stats->tx_packets += priv->tx_ring[i].packets;
-		stats->tx_bytes += priv->tx_ring[i].bytes;
-		priv->port_stats.tx_chksum_offload += priv->tx_ring[i].tx_csum;
+		stats->tx_packets += priv->tx_ring[i]->packets;
+		stats->tx_bytes += priv->tx_ring[i]->bytes;
+		priv->port_stats.tx_chksum_offload += priv->tx_ring[i]->tx_csum;
 	}
 
 	stats->rx_errors = be64_to_cpu(mlx4_en_stats->PCS) +

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

···
 
 	for (buf_ind = 0; buf_ind < priv->prof->rx_ring_size; buf_ind++) {
 		for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
-			ring = &priv->rx_ring[ring_ind];
+			ring = priv->rx_ring[ring_ind];
 
 			if (mlx4_en_prepare_rx_desc(priv, ring,
 						    ring->actual_size,
···
 
 reduce_rings:
 	for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
-		ring = &priv->rx_ring[ring_ind];
+		ring = priv->rx_ring[ring_ind];
 		while (ring->actual_size > new_size) {
 			ring->actual_size--;
 			ring->prod--;
···
 }
 
 int mlx4_en_create_rx_ring(struct mlx4_en_priv *priv,
-			   struct mlx4_en_rx_ring *ring, u32 size, u16 stride)
+			   struct mlx4_en_rx_ring **pring,
+			   u32 size, u16 stride, int node)
 {
 	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_rx_ring *ring;
 	int err = -ENOMEM;
 	int tmp;
+
+	ring = kzalloc_node(sizeof(*ring), GFP_KERNEL, node);
+	if (!ring) {
+		ring = kzalloc(sizeof(*ring), GFP_KERNEL);
+		if (!ring) {
+			en_err(priv, "Failed to allocate RX ring structure\n");
+			return -ENOMEM;
+		}
+	}
 
 	ring->prod = 0;
 	ring->cons = 0;
···
 
 	tmp = size * roundup_pow_of_two(MLX4_EN_MAX_RX_FRAGS *
 					sizeof(struct mlx4_en_rx_alloc));
-	ring->rx_info = vmalloc(tmp);
-	if (!ring->rx_info)
-		return -ENOMEM;
+	ring->rx_info = vmalloc_node(tmp, node);
+	if (!ring->rx_info) {
+		ring->rx_info = vmalloc(tmp);
+		if (!ring->rx_info) {
+			err = -ENOMEM;
+			goto err_ring;
+		}
+	}
 
 	en_dbg(DRV, priv, "Allocated rx_info ring at addr:%p size:%d\n",
 		 ring->rx_info, tmp);
 
+	/* 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)
-		goto err_ring;
+		goto err_info;
 
 	err = mlx4_en_map_buffer(&ring->wqres.buf);
 	if (err) {
···
 
 	ring->hwtstamp_rx_filter = priv->hwtstamp_config.rx_filter;
 
+	*pring = ring;
 	return 0;
 
 err_hwq:
 	mlx4_free_hwq_res(mdev->dev, &ring->wqres, ring->buf_size);
-err_ring:
+err_info:
 	vfree(ring->rx_info);
 	ring->rx_info = NULL;
+err_ring:
+	kfree(ring);
+	*pring = NULL;
+
 	return err;
 }
 
···
 					DS_SIZE * priv->num_frags);
 
 	for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
-		ring = &priv->rx_ring[ring_ind];
+		ring = priv->rx_ring[ring_ind];
 
 		ring->prod = 0;
 		ring->cons = 0;
 		ring->actual_size = 0;
-		ring->cqn = priv->rx_cq[ring_ind].mcq.cqn;
+		ring->cqn = priv->rx_cq[ring_ind]->mcq.cqn;
 
 		ring->stride = stride;
 		if (ring->stride <= TXBB_SIZE)
···
 		goto err_buffers;
 
 	for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++) {
-		ring = &priv->rx_ring[ring_ind];
+		ring = priv->rx_ring[ring_ind];
 
 		ring->size_mask = ring->actual_size - 1;
 		mlx4_en_update_rx_prod_db(ring);
···
 
 err_buffers:
 	for (ring_ind = 0; ring_ind < priv->rx_ring_num; ring_ind++)
-		mlx4_en_free_rx_buf(priv, &priv->rx_ring[ring_ind]);
+		mlx4_en_free_rx_buf(priv, priv->rx_ring[ring_ind]);
 
 	ring_ind = priv->rx_ring_num - 1;
 err_allocator:
 	while (ring_ind >= 0) {
-		if (priv->rx_ring[ring_ind].stride <= TXBB_SIZE)
-			priv->rx_ring[ring_ind].buf -= TXBB_SIZE;
-		mlx4_en_destroy_allocator(priv, &priv->rx_ring[ring_ind]);
+		if (priv->rx_ring[ring_ind]->stride <= TXBB_SIZE)
+			priv->rx_ring[ring_ind]->buf -= TXBB_SIZE;
+		mlx4_en_destroy_allocator(priv, priv->rx_ring[ring_ind]);
 		ring_ind--;
 	}
 	return err;
 }
 
 void mlx4_en_destroy_rx_ring(struct mlx4_en_priv *priv,
-			     struct mlx4_en_rx_ring *ring, u32 size, u16 stride)
+			     struct mlx4_en_rx_ring **pring,
+			     u32 size, u16 stride)
 {
 	struct mlx4_en_dev *mdev = priv->mdev;
+	struct mlx4_en_rx_ring *ring = *pring;
 
 	mlx4_en_unmap_buffer(&ring->wqres.buf);
 	mlx4_free_hwq_res(mdev->dev, &ring->wqres, size * stride + TXBB_SIZE);
 	vfree(ring->rx_info);
 	ring->rx_info = NULL;
+	kfree(ring);
+	*pring = NULL;
 #ifdef CONFIG_RFS_ACCEL
-	mlx4_en_cleanup_filters(priv, ring);
+	mlx4_en_cleanup_filters(priv);
 #endif
 }
 
···
 	struct mlx4_en_priv *priv = netdev_priv(dev);
 	struct mlx4_en_dev *mdev = priv->mdev;
 	struct mlx4_cqe *cqe;
-	struct mlx4_en_rx_ring *ring = &priv->rx_ring[cq->ring];
+	struct mlx4_en_rx_ring *ring = priv->rx_ring[cq->ring];
 	struct mlx4_en_rx_alloc *frags;
 	struct mlx4_en_rx_desc *rx_desc;
 	struct sk_buff *skb;
···
 
 	for (i = 0; i < priv->rx_ring_num; i++) {
 		qpn = rss_map->base_qpn + i;
-		err = mlx4_en_config_rss_qp(priv, qpn, &priv->rx_ring[i],
+		err = mlx4_en_config_rss_qp(priv, qpn, priv->rx_ring[i],
 					    &rss_map->state[i],
 					    &rss_map->qps[i]);
 		if (err)
···
 	}
 	rss_map->indir_qp.event = mlx4_en_sqp_event;
 	mlx4_en_fill_qp_context(priv, 0, 0, 0, 1, priv->base_qpn,
-				priv->rx_ring[0].cqn, -1, &context);
+				priv->rx_ring[0]->cqn, -1, &context);
 
 	if (!priv->prof->rss_rings || priv->prof->rss_rings > priv->rx_ring_num)
 		rss_rings = priv->rx_ring_num;

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

···
 		 * since we turned the carrier off */
 		msleep(200);
 		for (i = 0; i < priv->tx_ring_num && carrier_ok; i++) {
-			tx_ring = &priv->tx_ring[i];
+			tx_ring = priv->tx_ring[i];
 			if (tx_ring->prod != (tx_ring->cons + tx_ring->last_nr_txbb))
 				goto retry_tx;
 		}

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

···
 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)
 {
 	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;
···
 	inline_thold = min(inline_thold, MAX_INLINE);
 
 	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;
···
 	}
 	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);
 		ring->bf.uar = &mdev->priv_uar;
···
 
 	ring->hwtstamp_tx_type = priv->hwtstamp_config.tx_type;
 
+	*pring = ring;
 	return 0;
 
 err_map:
···
 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)
···
 	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,
···
 {
 	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, stamp_index;
···
 	}
 
 	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);
 

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

···
 	if (err)
 		goto err_out_free_mtt;
 
-	memset(eq_context, 0, sizeof *eq_context);
 	eq_context->flags	  = cpu_to_be32(MLX4_EQ_STATUS_OK   |
 						MLX4_EQ_STATE_ARMED);
 	eq_context->log_eq_size	  = ilog2(eq->nent);

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

···
 		return PTR_ERR(mailbox);
 	inbox = mailbox->buf;
 
-	memset(inbox, 0, MOD_STAT_CFG_IN_SIZE);
-
 	MLX4_PUT(inbox, cfg->log_pg_sz, MOD_STAT_CFG_PG_SZ_OFFSET);
 	MLX4_PUT(inbox, cfg->log_pg_sz_m, MOD_STAT_CFG_PG_SZ_M_OFFSET);
 
···
 	mailbox = mlx4_alloc_cmd_mailbox(dev);
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
-	memset(mailbox->buf, 0, MLX4_MAILBOX_SIZE);
 	pages = mailbox->buf;
 
 	for (mlx4_icm_first(icm, &iter);
···
 		return PTR_ERR(mailbox);
 	inbox = mailbox->buf;
 
-	memset(inbox, 0, INIT_HCA_IN_SIZE);
-
 	*((u8 *) mailbox->buf + INIT_HCA_VERSION_OFFSET) = INIT_HCA_VERSION;
 
 	*((u8 *) mailbox->buf + INIT_HCA_CACHELINE_SZ_OFFSET) =
···
 		if (IS_ERR(mailbox))
 			return PTR_ERR(mailbox);
 		inbox = mailbox->buf;
-
-		memset(inbox, 0, INIT_PORT_IN_SIZE);
 
 		flags = 0;
 		flags |= (dev->caps.vl_cap[port] & 0xf) << INIT_PORT_VL_SHIFT;

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

···
 	kfree(icm);
 }
 
-static int mlx4_alloc_icm_pages(struct scatterlist *mem, int order, gfp_t gfp_mask)
+static int mlx4_alloc_icm_pages(struct scatterlist *mem, int order,
+				gfp_t gfp_mask, int node)
 {
 	struct page *page;
 
-	page = alloc_pages(gfp_mask, order);
-	if (!page)
-		return -ENOMEM;
+	page = alloc_pages_node(node, gfp_mask, order);
+	if (!page) {
+		page = alloc_pages(gfp_mask, order);
+		if (!page)
+			return -ENOMEM;
+	}
 
 	sg_set_page(mem, page, PAGE_SIZE << order, 0);
 	return 0;
···
 	/* We use sg_set_buf for coherent allocs, which assumes low memory */
 	BUG_ON(coherent && (gfp_mask & __GFP_HIGHMEM));
 
-	icm = kmalloc(sizeof *icm, gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
-	if (!icm)
-		return NULL;
+	icm = kmalloc_node(sizeof(*icm),
+			   gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN),
+			   dev->numa_node);
+	if (!icm) {
+		icm = kmalloc(sizeof(*icm),
+			      gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
+		if (!icm)
+			return NULL;
+	}
 
 	icm->refcount = 0;
 	INIT_LIST_HEAD(&icm->chunk_list);
···
 
 	while (npages > 0) {
 		if (!chunk) {
-			chunk = kmalloc(sizeof *chunk,
-					gfp_mask & ~(__GFP_HIGHMEM | __GFP_NOWARN));
-			if (!chunk)
-				goto fail;
+			chunk = kmalloc_node(sizeof(*chunk),
+					     gfp_mask & ~(__GFP_HIGHMEM |
+							  __GFP_NOWARN),
+					     dev->numa_node);
+			if (!chunk) {
+				chunk = kmalloc(sizeof(*chunk),
+						gfp_mask & ~(__GFP_HIGHMEM |
+							     __GFP_NOWARN));
+				if (!chunk)
+					goto fail;
+			}
 
 			sg_init_table(chunk->mem, MLX4_ICM_CHUNK_LEN);
 			chunk->npages = 0;
···
 						      cur_order, gfp_mask);
 		else
 			ret = mlx4_alloc_icm_pages(&chunk->mem[chunk->npages],
-						   cur_order, gfp_mask);
+						   cur_order, gfp_mask,
+						   dev->numa_node);
 
 		if (ret) {
 			if (--cur_order < 0)

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

···
 	mutex_init(&priv->bf_mutex);
 
 	dev->rev_id = pdev->revision;
+	dev->numa_node = dev_to_node(&pdev->dev);
 	/* Detect if this device is a virtual function */
 	if (pci_dev_data & MLX4_PCI_DEV_IS_VF) {
 		/* When acting as pf, we normally skip vfs unless explicitly

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

···
 		goto out_list;
 	}
 	mgm = mailbox->buf;
-	memset(mgm, 0, sizeof *mgm);
 	members_count = 0;
 	list_for_each_entry(dqp, &s_steer->promisc_qps[steer], list)
 		mgm->qp[members_count++] = cpu_to_be32(dqp->qpn & MGM_QPN_MASK);
···
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
 
-	memset(mailbox->buf, 0, sizeof(struct mlx4_net_trans_rule_hw_ctrl));
 	trans_rule_ctrl_to_hw(rule, mailbox->buf);
 
 	size += sizeof(struct mlx4_net_trans_rule_hw_ctrl);

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

···
 	u16 num_frags;
 	u16 log_rx_info;
 
-	struct mlx4_en_tx_ring *tx_ring;
-	struct mlx4_en_rx_ring rx_ring[MAX_RX_RINGS];
-	struct mlx4_en_cq *tx_cq;
-	struct mlx4_en_cq rx_cq[MAX_RX_RINGS];
+	struct mlx4_en_tx_ring **tx_ring;
+	struct mlx4_en_rx_ring *rx_ring[MAX_RX_RINGS];
+	struct mlx4_en_cq **tx_cq;
+	struct mlx4_en_cq *rx_cq[MAX_RX_RINGS];
 	struct mlx4_qp drop_qp;
 	struct work_struct rx_mode_task;
 	struct work_struct watchdog_task;
···
 	if ((cq->state & MLX4_CQ_LOCKED)) {
 		struct net_device *dev = cq->dev;
 		struct mlx4_en_priv *priv = netdev_priv(dev);
-		struct mlx4_en_rx_ring *rx_ring = &priv->rx_ring[cq->ring];
+		struct mlx4_en_rx_ring *rx_ring = priv->rx_ring[cq->ring];
 
 		cq->state |= MLX4_EN_CQ_STATE_POLL_YIELD;
 		rc = false;
···
 void mlx4_en_free_resources(struct mlx4_en_priv *priv);
 int mlx4_en_alloc_resources(struct mlx4_en_priv *priv);
 
-int mlx4_en_create_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
-		      int entries, int ring, enum cq_type mode);
-void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
+int mlx4_en_create_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq,
+		      int entries, int ring, enum cq_type mode, int node);
+void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq);
 int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
 			int cq_idx);
 void mlx4_en_deactivate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
···
 u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb);
 netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev);
 
-int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring,
-			   int qpn, u32 size, u16 stride);
-void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring);
+int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
+			   struct mlx4_en_tx_ring **pring,
+			   int qpn, u32 size, u16 stride, int node);
+void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv,
+			     struct mlx4_en_tx_ring **pring);
 int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv,
 			     struct mlx4_en_tx_ring *ring,
 			     int cq, int user_prio);
···
 				struct mlx4_en_tx_ring *ring);
 
 int mlx4_en_create_rx_ring(struct mlx4_en_priv *priv,
-			   struct mlx4_en_rx_ring *ring,
-			   u32 size, u16 stride);
+			   struct mlx4_en_rx_ring **pring,
+			   u32 size, u16 stride, int node);
 void mlx4_en_destroy_rx_ring(struct mlx4_en_priv *priv,
-			     struct mlx4_en_rx_ring *ring,
+			     struct mlx4_en_rx_ring **pring,
 			     u32 size, u16 stride);
 int mlx4_en_activate_rx_rings(struct mlx4_en_priv *priv);
 void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv *priv,
···
 int mlx4_en_setup_tc(struct net_device *dev, u8 up);
 
 #ifdef CONFIG_RFS_ACCEL
-void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv,
-			     struct mlx4_en_rx_ring *rx_ring);
+void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv);
 #endif
 
 #define MLX4_EN_NUM_SELF_TEST	5

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

···
 		goto err_table;
 	}
 	mpt_entry = mailbox->buf;
-
-	memset(mpt_entry, 0, sizeof *mpt_entry);
-
 	mpt_entry->flags = cpu_to_be32(MLX4_MPT_FLAG_MIO	 |
 				       MLX4_MPT_FLAG_REGION	 |
 				       mr->access);
···
 		goto err_table;
 	}
 	mpt_entry = mailbox->buf;
-
-	memset(mpt_entry, 0, sizeof(*mpt_entry));
 
 	/* Note that the MLX4_MPT_FLAG_REGION bit in mpt_entry->flags is turned
 	 * off, thus creating a memory window and not a memory region.

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

···
 }
 EXPORT_SYMBOL_GPL(mlx4_uar_free);
 
-int mlx4_bf_alloc(struct mlx4_dev *dev, struct mlx4_bf *bf)
+int mlx4_bf_alloc(struct mlx4_dev *dev, struct mlx4_bf *bf, int node)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
 	struct mlx4_uar *uar;
···
 			err = -ENOMEM;
 			goto out;
 		}
-		uar = kmalloc(sizeof *uar, GFP_KERNEL);
+		uar = kmalloc_node(sizeof(*uar), GFP_KERNEL, node);
 		if (!uar) {
-			err = -ENOMEM;
-			goto out;
+			uar = kmalloc(sizeof(*uar), GFP_KERNEL);
+			if (!uar) {
+				err = -ENOMEM;
+				goto out;
+			}
 		}
 		err = mlx4_uar_alloc(dev, uar);
 		if (err)

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

···
 
 	inbuf = inmailbox->buf;
 	outbuf = outmailbox->buf;
-	memset(inbuf, 0, 256);
-	memset(outbuf, 0, 256);
 	inbuf[0] = 1;
 	inbuf[1] = 1;
 	inbuf[2] = 1;
···
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
 
-	memset(mailbox->buf, 0, 256);
-
 	((__be32 *) mailbox->buf)[1] = dev->caps.ib_port_def_cap[port];
 
 	if (pkey_tbl_sz >= 0 && mlx4_is_master(dev)) {
···
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
 	context = mailbox->buf;
-	memset(context, 0, sizeof *context);
-
 	context->flags = SET_PORT_GEN_ALL_VALID;
 	context->mtu = cpu_to_be16(mtu);
 	context->pptx = (pptx * (!pfctx)) << 7;
···
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
 	context = mailbox->buf;
-	memset(context, 0, sizeof *context);
-
 	context->base_qpn = cpu_to_be32(base_qpn);
 	context->n_mac = dev->caps.log_num_macs;
 	context->promisc = cpu_to_be32(promisc << SET_PORT_PROMISC_SHIFT |
···
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
 	context = mailbox->buf;
-	memset(context, 0, sizeof *context);
-
 	for (i = 0; i < MLX4_NUM_UP; i += 2)
 		context->prio2tc[i >> 1] = prio2tc[i] << 4 | prio2tc[i + 1];
 
···
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
 	context = mailbox->buf;
-	memset(context, 0, sizeof *context);
 
 	for (i = 0; i < MLX4_NUM_TC; i++) {
 		struct mlx4_port_scheduler_tc_cfg_be *tc = &context->tc[i];

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

···
 	int			local_qpn;
 	atomic_t		ref_count;
 	u32			qpc_flags;
+	/* saved qp params before VST enforcement in order to restore on VGT */
 	u8			sched_queue;
+	__be32			param3;
+	u8			vlan_control;
+	u8			fvl_rx;
+	u8			pri_path_fl;
+	u8			vlan_index;
+	u8			feup;
 };
 
 enum res_mtt_states {
···
 		return err;
 	qp->local_qpn = local_qpn;
 	qp->sched_queue = 0;
+	qp->param3 = 0;
+	qp->vlan_control = 0;
+	qp->fvl_rx = 0;
+	qp->pri_path_fl = 0;
+	qp->vlan_index = 0;
+	qp->feup = 0;
 	qp->qpc_flags = be32_to_cpu(qpc->flags);
 
 	err = get_res(dev, slave, mtt_base, RES_MTT, &mtt);
···
 	int qpn = vhcr->in_modifier & 0x7fffff;
 	struct res_qp *qp;
 	u8 orig_sched_queue;
+	__be32	orig_param3 = qpc->param3;
+	u8 orig_vlan_control = qpc->pri_path.vlan_control;
+	u8 orig_fvl_rx = qpc->pri_path.fvl_rx;
+	u8 orig_pri_path_fl = qpc->pri_path.fl;
+	u8 orig_vlan_index = qpc->pri_path.vlan_index;
+	u8 orig_feup = qpc->pri_path.feup;
 
 	err = verify_qp_parameters(dev, inbox, QP_TRANS_INIT2RTR, slave);
 	if (err)
···
 	 * essentially the QOS value provided by the VF. This will be useful
 	 * if we allow dynamic changes from VST back to VGT
 	 */
-	if (!err)
+	if (!err) {
 		qp->sched_queue = orig_sched_queue;
-
+		qp->param3	= orig_param3;
+		qp->vlan_control = orig_vlan_control;
+		qp->fvl_rx	=  orig_fvl_rx;
+		qp->pri_path_fl = orig_pri_path_fl;
+		qp->vlan_index  = orig_vlan_index;
+		qp->feup	= orig_feup;
+	}
 	put_res(dev, slave, qpn, RES_QP);
 	return err;
 }
···
 		&tracker->slave_list[work->slave].res_list[RES_QP];
 	struct res_qp *qp;
 	struct res_qp *tmp;
-	u64 qp_mask = ((1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_UNTAGGED) |
+	u64 qp_path_mask_vlan_ctrl =
+		       ((1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_UNTAGGED) |
 		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_1P) |
 		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_TX_BLOCK_TAGGED) |
 		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_UNTAGGED) |
 		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_1P) |
-		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_TAGGED) |
-		       (1ULL << MLX4_UPD_QP_PATH_MASK_VLAN_INDEX) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_RX_BLOCK_TAGGED));
+
+	u64 qp_path_mask = ((1ULL << MLX4_UPD_QP_PATH_MASK_VLAN_INDEX) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_FVL) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_CV) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_ETH_HIDE_CQE_VLAN) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_FEUP) |
+		       (1ULL << MLX4_UPD_QP_PATH_MASK_FVL_RX) |
 		       (1ULL << MLX4_UPD_QP_PATH_MASK_SCHED_QUEUE));
 
 	int err;
···
 			MLX4_VLAN_CTRL_ETH_RX_BLOCK_UNTAGGED;
 
 	upd_context = mailbox->buf;
-	upd_context->primary_addr_path_mask = cpu_to_be64(qp_mask);
-	upd_context->qp_context.pri_path.vlan_control = vlan_control;
-	upd_context->qp_context.pri_path.vlan_index = work->vlan_ix;
+	upd_context->qp_mask = cpu_to_be64(MLX4_UPD_QP_MASK_VSD);
 
 	spin_lock_irq(mlx4_tlock(dev));
 	list_for_each_entry_safe(qp, tmp, qp_list, com.list) {
···
 				spin_lock_irq(mlx4_tlock(dev));
 				continue;
 			}
-			upd_context->qp_context.pri_path.sched_queue =
-				qp->sched_queue & 0xC7;
-			upd_context->qp_context.pri_path.sched_queue |=
-				((work->qos & 0x7) << 3);
+			if (MLX4_QP_ST_RC == ((qp->qpc_flags >> 16) & 0xff))
+				upd_context->primary_addr_path_mask = cpu_to_be64(qp_path_mask);
+			else
+				upd_context->primary_addr_path_mask =
+					cpu_to_be64(qp_path_mask | qp_path_mask_vlan_ctrl);
+			if (work->vlan_id == MLX4_VGT) {
+				upd_context->qp_context.param3 = qp->param3;
+				upd_context->qp_context.pri_path.vlan_control = qp->vlan_control;
+				upd_context->qp_context.pri_path.fvl_rx = qp->fvl_rx;
+				upd_context->qp_context.pri_path.vlan_index = qp->vlan_index;
+				upd_context->qp_context.pri_path.fl = qp->pri_path_fl;
+				upd_context->qp_context.pri_path.feup = qp->feup;
+				upd_context->qp_context.pri_path.sched_queue =
+					qp->sched_queue;
+			} else {
+				upd_context->qp_context.param3 = qp->param3 & ~cpu_to_be32(MLX4_STRIP_VLAN);
+				upd_context->qp_context.pri_path.vlan_control = vlan_control;
+				upd_context->qp_context.pri_path.vlan_index = work->vlan_ix;
+				upd_context->qp_context.pri_path.fvl_rx =
+					qp->fvl_rx | MLX4_FVL_RX_FORCE_ETH_VLAN;
+				upd_context->qp_context.pri_path.fl =
+					qp->pri_path_fl | MLX4_FL_CV | MLX4_FL_ETH_HIDE_CQE_VLAN;
+				upd_context->qp_context.pri_path.feup =
+					qp->feup | MLX4_FEUP_FORCE_ETH_UP | MLX4_FVL_FORCE_ETH_VLAN;
+				upd_context->qp_context.pri_path.sched_queue =
+					qp->sched_queue & 0xC7;
+				upd_context->qp_context.pri_path.sched_queue |=
+					((work->qos & 0x7) << 3);
+			}
 
 			err = mlx4_cmd(dev, mailbox->dma,
 				       qp->local_qpn & 0xffffff,

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

···
 	}
 
 	srq_context = mailbox->buf;
-	memset(srq_context, 0, sizeof *srq_context);
-
 	srq_context->state_logsize_srqn = cpu_to_be32((ilog2(srq->max) << 24) |
 						      srq->srqn);
 	srq_context->logstride          = srq->wqe_shift - 4;

include/linux/mlx4/device.h

···
 	u8			rev_id;
 	char			board_id[MLX4_BOARD_ID_LEN];
 	int			num_vfs;
+	int			numa_node;
 	int			oper_log_mgm_entry_size;
 	u64			regid_promisc_array[MLX4_MAX_PORTS + 1];
 	u64			regid_allmulti_array[MLX4_MAX_PORTS + 1];
···
 
 int mlx4_uar_alloc(struct mlx4_dev *dev, struct mlx4_uar *uar);
 void mlx4_uar_free(struct mlx4_dev *dev, struct mlx4_uar *uar);
-int mlx4_bf_alloc(struct mlx4_dev *dev, struct mlx4_bf *bf);
+int mlx4_bf_alloc(struct mlx4_dev *dev, struct mlx4_bf *bf, int node);
 void mlx4_bf_free(struct mlx4_dev *dev, struct mlx4_bf *bf);
 
 int mlx4_mtt_init(struct mlx4_dev *dev, int npages, int page_shift,