Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next

Pull networking updates from David Miller:
 "Here is my initial pull request for the networking subsystem during
  this merge window:

   1) Support for ESN in AH (RFC 4302) from Fan Du.

   2) Add full kernel doc for ethtool command structures, from Ben
      Hutchings.

   3) Add BCM7xxx PHY driver, from Florian Fainelli.

   4) Export computed TCP rate information in netlink socket dumps, from
      Eric Dumazet.

   5) Allow IPSEC SA to be dumped partially using a filter, from Nicolas
      Dichtel.

   6) Convert many drivers to pci_enable_msix_range(), from Alexander
      Gordeev.

   7) Record SKB timestamps more efficiently, from Eric Dumazet.

   8) Switch to microsecond resolution for TCP round trip times, also
      from Eric Dumazet.

   9) Clean up and fix 6lowpan fragmentation handling by making use of
      the existing inet_frag api for it's implementation.

  10) Add TX grant mapping to xen-netback driver, from Zoltan Kiss.

  11) Auto size SKB lengths when composing netlink messages based upon
      past message sizes used, from Eric Dumazet.

  12) qdisc dumps can take a long time, add a cond_resched(), From Eric
      Dumazet.

  13) Sanitize netpoll core and drivers wrt.  SKB handling semantics.
      Get rid of never-used-in-tree netpoll RX handling.  From Eric W
      Biederman.

  14) Support inter-address-family and namespace changing in VTI tunnel
      driver(s).  From Steffen Klassert.

  15) Add Altera TSE driver, from Vince Bridgers.

  16) Optimizing csum_replace2() so that it doesn't adjust the checksum
      by checksumming the entire header, from Eric Dumazet.

  17) Expand BPF internal implementation for faster interpreting, more
      direct translations into JIT'd code, and much cleaner uses of BPF
      filtering in non-socket ocntexts.  From Daniel Borkmann and Alexei
      Starovoitov"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1976 commits)
  netpoll: Use skb_irq_freeable to make zap_completion_queue safe.
  net: Add a test to see if a skb is freeable in irq context
  qlcnic: Fix build failure due to undefined reference to `vxlan_get_rx_port'
  net: ptp: move PTP classifier in its own file
  net: sxgbe: make "core_ops" static
  net: sxgbe: fix logical vs bitwise operation
  net: sxgbe: sxgbe_mdio_register() frees the bus
  Call efx_set_channels() before efx->type->dimension_resources()
  xen-netback: disable rogue vif in kthread context
  net/mlx4: Set proper build dependancy with vxlan
  be2net: fix build dependency on VxLAN
  mac802154: make csma/cca parameters per-wpan
  mac802154: allow only one WPAN to be up at any given time
  net: filter: minor: fix kdoc in __sk_run_filter
  netlink: don't compare the nul-termination in nla_strcmp
  can: c_can: Avoid led toggling for every packet.
  can: c_can: Simplify TX interrupt cleanup
  can: c_can: Store dlc private
  can: c_can: Reduce register access
  can: c_can: Make the code readable
  ...

9 files changed, 514 insertions, 116 deletions

diff --git a/drivers/infiniband/hw/mlx4/alias_GUID.c b/drivers/infiniband/hw/mlx4/alias_GUID.c
index 2f215b93db6b..0eb141c41416 100644
--- a/drivers/infiniband/hw/mlx4/alias_GUID.c
+++ b/drivers/infiniband/hw/mlx4/alias_GUID.c
@@ -154,7 +154,7 @@ void mlx4_ib_notify_slaves_on_guid_change(struct mlx4_ib_dev *dev,
 			continue;
 
 		slave_id = (block_num * NUM_ALIAS_GUID_IN_REC) + i ;
-		if (slave_id >= dev->dev->num_slaves)
+		if (slave_id >= dev->dev->num_vfs + 1)
 			return;
 		tmp_cur_ag = *(__be64 *)&p_data[i * GUID_REC_SIZE];
 		form_cache_ag = get_cached_alias_guid(dev, port_num,
diff --git a/drivers/infiniband/hw/mlx4/cm.c b/drivers/infiniband/hw/mlx4/cm.c
index d1f5f1dd77b0..56a593e0ae5d 100644
--- a/drivers/infiniband/hw/mlx4/cm.c
+++ b/drivers/infiniband/hw/mlx4/cm.c
@@ -61,6 +61,11 @@ struct cm_generic_msg {
 	__be32 remote_comm_id;
 };
 
+struct cm_sidr_generic_msg {
+	struct ib_mad_hdr hdr;
+	__be32 request_id;
+};
+
 struct cm_req_msg {
 	unsigned char unused[0x60];
 	union ib_gid primary_path_sgid;
@@ -69,28 +74,62 @@ struct cm_req_msg {
 
 static void set_local_comm_id(struct ib_mad *mad, u32 cm_id)
 {
-	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
-	msg->local_comm_id = cpu_to_be32(cm_id);
+	if (mad->mad_hdr.attr_id == CM_SIDR_REQ_ATTR_ID) {
+		struct cm_sidr_generic_msg *msg =
+			(struct cm_sidr_generic_msg *)mad;
+		msg->request_id = cpu_to_be32(cm_id);
+	} else if (mad->mad_hdr.attr_id == CM_SIDR_REP_ATTR_ID) {
+		pr_err("trying to set local_comm_id in SIDR_REP\n");
+		return;
+	} else {
+		struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+		msg->local_comm_id = cpu_to_be32(cm_id);
+	}
 }
 
 static u32 get_local_comm_id(struct ib_mad *mad)
 {
-	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
-
-	return be32_to_cpu(msg->local_comm_id);
+	if (mad->mad_hdr.attr_id == CM_SIDR_REQ_ATTR_ID) {
+		struct cm_sidr_generic_msg *msg =
+			(struct cm_sidr_generic_msg *)mad;
+		return be32_to_cpu(msg->request_id);
+	} else if (mad->mad_hdr.attr_id == CM_SIDR_REP_ATTR_ID) {
+		pr_err("trying to set local_comm_id in SIDR_REP\n");
+		return -1;
+	} else {
+		struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+		return be32_to_cpu(msg->local_comm_id);
+	}
 }
 
 static void set_remote_comm_id(struct ib_mad *mad, u32 cm_id)
 {
-	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
-	msg->remote_comm_id = cpu_to_be32(cm_id);
+	if (mad->mad_hdr.attr_id == CM_SIDR_REP_ATTR_ID) {
+		struct cm_sidr_generic_msg *msg =
+			(struct cm_sidr_generic_msg *)mad;
+		msg->request_id = cpu_to_be32(cm_id);
+	} else if (mad->mad_hdr.attr_id == CM_SIDR_REQ_ATTR_ID) {
+		pr_err("trying to set remote_comm_id in SIDR_REQ\n");
+		return;
+	} else {
+		struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+		msg->remote_comm_id = cpu_to_be32(cm_id);
+	}
 }
 
 static u32 get_remote_comm_id(struct ib_mad *mad)
 {
-	struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
-
-	return be32_to_cpu(msg->remote_comm_id);
+	if (mad->mad_hdr.attr_id == CM_SIDR_REP_ATTR_ID) {
+		struct cm_sidr_generic_msg *msg =
+			(struct cm_sidr_generic_msg *)mad;
+		return be32_to_cpu(msg->request_id);
+	} else if (mad->mad_hdr.attr_id == CM_SIDR_REQ_ATTR_ID) {
+		pr_err("trying to set remote_comm_id in SIDR_REQ\n");
+		return -1;
+	} else {
+		struct cm_generic_msg *msg = (struct cm_generic_msg *)mad;
+		return be32_to_cpu(msg->remote_comm_id);
+	}
 }
 
 static union ib_gid gid_from_req_msg(struct ib_device *ibdev, struct ib_mad *mad)
@@ -282,19 +321,21 @@ int mlx4_ib_multiplex_cm_handler(struct ib_device *ibdev, int port, int slave_id
 	u32 sl_cm_id;
 	int pv_cm_id = -1;
 
-	sl_cm_id = get_local_comm_id(mad);
-
 	if (mad->mad_hdr.attr_id == CM_REQ_ATTR_ID ||
-			mad->mad_hdr.attr_id == CM_REP_ATTR_ID) {
+			mad->mad_hdr.attr_id == CM_REP_ATTR_ID ||
+			mad->mad_hdr.attr_id == CM_SIDR_REQ_ATTR_ID) {
+		sl_cm_id = get_local_comm_id(mad);
 		id = id_map_alloc(ibdev, slave_id, sl_cm_id);
 		if (IS_ERR(id)) {
 			mlx4_ib_warn(ibdev, "%s: id{slave: %d, sl_cm_id: 0x%x} Failed to id_map_alloc\n",
 				__func__, slave_id, sl_cm_id);
 			return PTR_ERR(id);
 		}
-	} else if (mad->mad_hdr.attr_id == CM_REJ_ATTR_ID) {
+	} else if (mad->mad_hdr.attr_id == CM_REJ_ATTR_ID ||
+		   mad->mad_hdr.attr_id == CM_SIDR_REP_ATTR_ID) {
 		return 0;
 	} else {
+		sl_cm_id = get_local_comm_id(mad);
 		id = id_map_get(ibdev, &pv_cm_id, slave_id, sl_cm_id);
 	}
 
@@ -315,14 +356,18 @@ int mlx4_ib_multiplex_cm_handler(struct ib_device *ibdev, int port, int slave_id
 }
 
 int mlx4_ib_demux_cm_handler(struct ib_device *ibdev, int port, int *slave,
-							     struct ib_mad *mad)
+			     struct ib_mad *mad)
 {
 	u32 pv_cm_id;
 	struct id_map_entry *id;
 
-	if (mad->mad_hdr.attr_id == CM_REQ_ATTR_ID) {
+	if (mad->mad_hdr.attr_id == CM_REQ_ATTR_ID ||
+	    mad->mad_hdr.attr_id == CM_SIDR_REQ_ATTR_ID) {
 		union ib_gid gid;
 
+		if (!slave)
+			return 0;
+
 		gid = gid_from_req_msg(ibdev, mad);
 		*slave = mlx4_ib_find_real_gid(ibdev, port, gid.global.interface_id);
 		if (*slave < 0) {
@@ -341,7 +386,8 @@ int mlx4_ib_demux_cm_handler(struct ib_device *ibdev, int port, int *slave,
 		return -ENOENT;
 	}
 
-	*slave = id->slave_id;
+	if (slave)
+		*slave = id->slave_id;
 	set_remote_comm_id(mad, id->sl_cm_id);
 
 	if (mad->mad_hdr.attr_id == CM_DREQ_ATTR_ID)
diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index cc40f08ca8f1..5f640814cc81 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -564,7 +564,7 @@ static int mlx4_ib_ipoib_csum_ok(__be16 status, __be16 checksum)
 }
 
 static int use_tunnel_data(struct mlx4_ib_qp *qp, struct mlx4_ib_cq *cq, struct ib_wc *wc,
-			   unsigned tail, struct mlx4_cqe *cqe)
+			   unsigned tail, struct mlx4_cqe *cqe, int is_eth)
 {
 	struct mlx4_ib_proxy_sqp_hdr *hdr;
 
@@ -574,12 +574,20 @@ static int use_tunnel_data(struct mlx4_ib_qp *qp, struct mlx4_ib_cq *cq, struct
 				   DMA_FROM_DEVICE);
 	hdr = (struct mlx4_ib_proxy_sqp_hdr *) (qp->sqp_proxy_rcv[tail].addr);
 	wc->pkey_index	= be16_to_cpu(hdr->tun.pkey_index);
-	wc->slid	= be16_to_cpu(hdr->tun.slid_mac_47_32);
-	wc->sl		= (u8) (be16_to_cpu(hdr->tun.sl_vid) >> 12);
 	wc->src_qp	= be32_to_cpu(hdr->tun.flags_src_qp) & 0xFFFFFF;
 	wc->wc_flags   |= (hdr->tun.g_ml_path & 0x80) ? (IB_WC_GRH) : 0;
 	wc->dlid_path_bits = 0;
 
+	if (is_eth) {
+		wc->vlan_id = be16_to_cpu(hdr->tun.sl_vid);
+		memcpy(&(wc->smac[0]), (char *)&hdr->tun.mac_31_0, 4);
+		memcpy(&(wc->smac[4]), (char *)&hdr->tun.slid_mac_47_32, 2);
+		wc->wc_flags |= (IB_WC_WITH_VLAN | IB_WC_WITH_SMAC);
+	} else {
+		wc->slid        = be16_to_cpu(hdr->tun.slid_mac_47_32);
+		wc->sl          = (u8) (be16_to_cpu(hdr->tun.sl_vid) >> 12);
+	}
+
 	return 0;
 }
 
@@ -594,6 +602,7 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 	struct mlx4_srq *msrq = NULL;
 	int is_send;
 	int is_error;
+	int is_eth;
 	u32 g_mlpath_rqpn;
 	u16 wqe_ctr;
 	unsigned tail = 0;
@@ -778,11 +787,15 @@ repoll:
 			break;
 		}
 
+		is_eth = (rdma_port_get_link_layer(wc->qp->device,
+						  (*cur_qp)->port) ==
+			  IB_LINK_LAYER_ETHERNET);
 		if (mlx4_is_mfunc(to_mdev(cq->ibcq.device)->dev)) {
 			if ((*cur_qp)->mlx4_ib_qp_type &
 			    (MLX4_IB_QPT_PROXY_SMI_OWNER |
 			     MLX4_IB_QPT_PROXY_SMI | MLX4_IB_QPT_PROXY_GSI))
-				return use_tunnel_data(*cur_qp, cq, wc, tail, cqe);
+				return use_tunnel_data(*cur_qp, cq, wc, tail,
+						       cqe, is_eth);
 		}
 
 		wc->slid	   = be16_to_cpu(cqe->rlid);
@@ -793,20 +806,21 @@ repoll:
 		wc->pkey_index     = be32_to_cpu(cqe->immed_rss_invalid) & 0x7f;
 		wc->wc_flags	  |= mlx4_ib_ipoib_csum_ok(cqe->status,
 					cqe->checksum) ? IB_WC_IP_CSUM_OK : 0;
-		if (rdma_port_get_link_layer(wc->qp->device,
-				(*cur_qp)->port) == IB_LINK_LAYER_ETHERNET)
+		if (is_eth) {
 			wc->sl  = be16_to_cpu(cqe->sl_vid) >> 13;
-		else
-			wc->sl  = be16_to_cpu(cqe->sl_vid) >> 12;
-		if (be32_to_cpu(cqe->vlan_my_qpn) & MLX4_CQE_VLAN_PRESENT_MASK) {
-			wc->vlan_id = be16_to_cpu(cqe->sl_vid) &
-				MLX4_CQE_VID_MASK;
+			if (be32_to_cpu(cqe->vlan_my_qpn) &
+					MLX4_CQE_VLAN_PRESENT_MASK) {
+				wc->vlan_id = be16_to_cpu(cqe->sl_vid) &
+					MLX4_CQE_VID_MASK;
+			} else {
+				wc->vlan_id = 0xffff;
+			}
+			memcpy(wc->smac, cqe->smac, ETH_ALEN);
+			wc->wc_flags |= (IB_WC_WITH_VLAN | IB_WC_WITH_SMAC);
 		} else {
+			wc->sl  = be16_to_cpu(cqe->sl_vid) >> 12;
 			wc->vlan_id = 0xffff;
 		}
-		wc->wc_flags |= IB_WC_WITH_VLAN;
-		memcpy(wc->smac, cqe->smac, ETH_ALEN);
-		wc->wc_flags |= IB_WC_WITH_SMAC;
 	}
 
 	return 0;
diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c
index f2a3f48107e7..fd36ec672632 100644
--- a/drivers/infiniband/hw/mlx4/mad.c
+++ b/drivers/infiniband/hw/mlx4/mad.c
@@ -467,6 +467,7 @@ int mlx4_ib_send_to_slave(struct mlx4_ib_dev *dev, int slave, u8 port,
 	int ret = 0;
 	u16 tun_pkey_ix;
 	u16 cached_pkey;
+	u8 is_eth = dev->dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH;
 
 	if (dest_qpt > IB_QPT_GSI)
 		return -EINVAL;
@@ -509,6 +510,10 @@ int mlx4_ib_send_to_slave(struct mlx4_ib_dev *dev, int slave, u8 port,
 	 * The driver will set the force loopback bit in post_send */
 	memset(&attr, 0, sizeof attr);
 	attr.port_num = port;
+	if (is_eth) {
+		memcpy(&attr.grh.dgid.raw[0], &grh->dgid.raw[0], 16);
+		attr.ah_flags = IB_AH_GRH;
+	}
 	ah = ib_create_ah(tun_ctx->pd, &attr);
 	if (IS_ERR(ah))
 		return -ENOMEM;
@@ -540,11 +545,36 @@ int mlx4_ib_send_to_slave(struct mlx4_ib_dev *dev, int slave, u8 port,
 
 	/* adjust tunnel data */
 	tun_mad->hdr.pkey_index = cpu_to_be16(tun_pkey_ix);
-	tun_mad->hdr.sl_vid = cpu_to_be16(((u16)(wc->sl)) << 12);
-	tun_mad->hdr.slid_mac_47_32 = cpu_to_be16(wc->slid);
 	tun_mad->hdr.flags_src_qp = cpu_to_be32(wc->src_qp & 0xFFFFFF);
 	tun_mad->hdr.g_ml_path = (grh && (wc->wc_flags & IB_WC_GRH)) ? 0x80 : 0;
 
+	if (is_eth) {
+		u16 vlan = 0;
+		if (mlx4_get_slave_default_vlan(dev->dev, port, slave, &vlan,
+						NULL)) {
+			/* VST mode */
+			if (vlan != wc->vlan_id)
+				/* Packet vlan is not the VST-assigned vlan.
+				 * Drop the packet.
+				 */
+				goto out;
+			 else
+				/* Remove the vlan tag before forwarding
+				 * the packet to the VF.
+				 */
+				vlan = 0xffff;
+		} else {
+			vlan = wc->vlan_id;
+		}
+
+		tun_mad->hdr.sl_vid = cpu_to_be16(vlan);
+		memcpy((char *)&tun_mad->hdr.mac_31_0, &(wc->smac[0]), 4);
+		memcpy((char *)&tun_mad->hdr.slid_mac_47_32, &(wc->smac[4]), 2);
+	} else {
+		tun_mad->hdr.sl_vid = cpu_to_be16(((u16)(wc->sl)) << 12);
+		tun_mad->hdr.slid_mac_47_32 = cpu_to_be16(wc->slid);
+	}
+
 	ib_dma_sync_single_for_device(&dev->ib_dev,
 				      tun_qp->tx_ring[tun_tx_ix].buf.map,
 				      sizeof (struct mlx4_rcv_tunnel_mad),
@@ -580,6 +610,41 @@ static int mlx4_ib_demux_mad(struct ib_device *ibdev, u8 port,
 	int err;
 	int slave;
 	u8 *slave_id;
+	int is_eth = 0;
+
+	if (rdma_port_get_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND)
+		is_eth = 0;
+	else
+		is_eth = 1;
+
+	if (is_eth) {
+		if (!(wc->wc_flags & IB_WC_GRH)) {
+			mlx4_ib_warn(ibdev, "RoCE grh not present.\n");
+			return -EINVAL;
+		}
+		if (mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_CM) {
+			mlx4_ib_warn(ibdev, "RoCE mgmt class is not CM\n");
+			return -EINVAL;
+		}
+		if (mlx4_get_slave_from_roce_gid(dev->dev, port, grh->dgid.raw, &slave)) {
+			mlx4_ib_warn(ibdev, "failed matching grh\n");
+			return -ENOENT;
+		}
+		if (slave >= dev->dev->caps.sqp_demux) {
+			mlx4_ib_warn(ibdev, "slave id: %d is bigger than allowed:%d\n",
+				     slave, dev->dev->caps.sqp_demux);
+			return -ENOENT;
+		}
+
+		if (mlx4_ib_demux_cm_handler(ibdev, port, NULL, mad))
+			return 0;
+
+		err = mlx4_ib_send_to_slave(dev, slave, port, wc->qp->qp_type, wc, grh, mad);
+		if (err)
+			pr_debug("failed sending to slave %d via tunnel qp (%d)\n",
+				 slave, err);
+		return 0;
+	}
 
 	/* Initially assume that this mad is for us */
 	slave = mlx4_master_func_num(dev->dev);
@@ -1076,8 +1141,9 @@ static int is_proxy_qp0(struct mlx4_ib_dev *dev, int qpn, int slave)
 
 
 int mlx4_ib_send_to_wire(struct mlx4_ib_dev *dev, int slave, u8 port,
-			 enum ib_qp_type dest_qpt, u16 pkey_index, u32 remote_qpn,
-			 u32 qkey, struct ib_ah_attr *attr, struct ib_mad *mad)
+			 enum ib_qp_type dest_qpt, u16 pkey_index,
+			 u32 remote_qpn, u32 qkey, struct ib_ah_attr *attr,
+			 u8 *s_mac, struct ib_mad *mad)
 {
 	struct ib_sge list;
 	struct ib_send_wr wr, *bad_wr;
@@ -1166,6 +1232,9 @@ int mlx4_ib_send_to_wire(struct mlx4_ib_dev *dev, int slave, u8 port,
 	wr.num_sge = 1;
 	wr.opcode = IB_WR_SEND;
 	wr.send_flags = IB_SEND_SIGNALED;
+	if (s_mac)
+		memcpy(to_mah(ah)->av.eth.s_mac, s_mac, 6);
+
 
 	ret = ib_post_send(send_qp, &wr, &bad_wr);
 out:
@@ -1174,6 +1243,22 @@ out:
 	return ret;
 }
 
+static int get_slave_base_gid_ix(struct mlx4_ib_dev *dev, int slave, int port)
+{
+	if (rdma_port_get_link_layer(&dev->ib_dev, port) == IB_LINK_LAYER_INFINIBAND)
+		return slave;
+	return mlx4_get_base_gid_ix(dev->dev, slave, port);
+}
+
+static void fill_in_real_sgid_index(struct mlx4_ib_dev *dev, int slave, int port,
+				    struct ib_ah_attr *ah_attr)
+{
+	if (rdma_port_get_link_layer(&dev->ib_dev, port) == IB_LINK_LAYER_INFINIBAND)
+		ah_attr->grh.sgid_index = slave;
+	else
+		ah_attr->grh.sgid_index += get_slave_base_gid_ix(dev, slave, port);
+}
+
 static void mlx4_ib_multiplex_mad(struct mlx4_ib_demux_pv_ctx *ctx, struct ib_wc *wc)
 {
 	struct mlx4_ib_dev *dev = to_mdev(ctx->ib_dev);
@@ -1184,6 +1269,7 @@ static void mlx4_ib_multiplex_mad(struct mlx4_ib_demux_pv_ctx *ctx, struct ib_wc
 	struct ib_ah_attr ah_attr;
 	u8 *slave_id;
 	int slave;
+	int port;
 
 	/* Get slave that sent this packet */
 	if (wc->src_qp < dev->dev->phys_caps.base_proxy_sqpn ||
@@ -1260,12 +1346,18 @@ static void mlx4_ib_multiplex_mad(struct mlx4_ib_demux_pv_ctx *ctx, struct ib_wc
 	memcpy(&ah.av, &tunnel->hdr.av, sizeof (struct mlx4_av));
 	ah.ibah.device = ctx->ib_dev;
 	mlx4_ib_query_ah(&ah.ibah, &ah_attr);
-	if ((ah_attr.ah_flags & IB_AH_GRH) &&
-	    (ah_attr.grh.sgid_index != slave)) {
-		mlx4_ib_warn(ctx->ib_dev, "slave:%d accessed invalid sgid_index:%d\n",
-			     slave, ah_attr.grh.sgid_index);
+	if (ah_attr.ah_flags & IB_AH_GRH)
+		fill_in_real_sgid_index(dev, slave, ctx->port, &ah_attr);
+
+	port = mlx4_slave_convert_port(dev->dev, slave, ah_attr.port_num);
+	if (port < 0)
 		return;
-	}
+	ah_attr.port_num = port;
+	memcpy(ah_attr.dmac, tunnel->hdr.mac, 6);
+	ah_attr.vlan_id = be16_to_cpu(tunnel->hdr.vlan);
+	/* if slave have default vlan use it */
+	mlx4_get_slave_default_vlan(dev->dev, ctx->port, slave,
+				    &ah_attr.vlan_id, &ah_attr.sl);
 
 	mlx4_ib_send_to_wire(dev, slave, ctx->port,
 			     is_proxy_qp0(dev, wc->src_qp, slave) ?
@@ -1273,7 +1365,7 @@ static void mlx4_ib_multiplex_mad(struct mlx4_ib_demux_pv_ctx *ctx, struct ib_wc
 			     be16_to_cpu(tunnel->hdr.pkey_index),
 			     be32_to_cpu(tunnel->hdr.remote_qpn),
 			     be32_to_cpu(tunnel->hdr.qkey),
-			     &ah_attr, &tunnel->mad);
+			     &ah_attr, wc->smac, &tunnel->mad);
 }
 
 static int mlx4_ib_alloc_pv_bufs(struct mlx4_ib_demux_pv_ctx *ctx,
@@ -1850,7 +1942,15 @@ static int mlx4_ib_alloc_demux_ctx(struct mlx4_ib_dev *dev,
 	ctx->port = port;
 	ctx->ib_dev = &dev->ib_dev;
 
-	for (i = 0; i < dev->dev->caps.sqp_demux; i++) {
+	for (i = 0;
+	     i < min(dev->dev->caps.sqp_demux, (u16)(dev->dev->num_vfs + 1));
+	     i++) {
+		struct mlx4_active_ports actv_ports =
+			mlx4_get_active_ports(dev->dev, i);
+
+		if (!test_bit(port - 1, actv_ports.ports))
+			continue;
+
 		ret = alloc_pv_object(dev, i, port, &ctx->tun[i]);
 		if (ret) {
 			ret = -ENOMEM;
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index f9c12e92fdd6..6cb85467dde7 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -1546,7 +1546,7 @@ static int mlx4_ib_addr_event(int event, struct net_device *event_netdev,
 	iboe = &ibdev->iboe;
 	spin_lock(&iboe->lock);
 
-	for (port = 1; port <= MLX4_MAX_PORTS; ++port)
+	for (port = 1; port <= ibdev->dev->caps.num_ports; ++port)
 		if ((netif_is_bond_master(real_dev) &&
 		     (real_dev == iboe->masters[port - 1])) ||
 		     (!netif_is_bond_master(real_dev) &&
@@ -1569,14 +1569,14 @@ static u8 mlx4_ib_get_dev_port(struct net_device *dev,
 
 	iboe = &ibdev->iboe;
 
-	for (port = 1; port <= MLX4_MAX_PORTS; ++port)
+	for (port = 1; port <= ibdev->dev->caps.num_ports; ++port)
 		if ((netif_is_bond_master(real_dev) &&
 		     (real_dev == iboe->masters[port - 1])) ||
 		     (!netif_is_bond_master(real_dev) &&
 		     (real_dev == iboe->netdevs[port - 1])))
 			break;
 
-	if ((port == 0) || (port > MLX4_MAX_PORTS))
+	if ((port == 0) || (port > ibdev->dev->caps.num_ports))
 		return 0;
 	else
 		return port;
@@ -1626,7 +1626,7 @@ static void mlx4_ib_get_dev_addr(struct net_device *dev,
 	union ib_gid gid;
 
 
-	if ((port == 0) || (port > MLX4_MAX_PORTS))
+	if ((port == 0) || (port > ibdev->dev->caps.num_ports))
 		return;
 
 	/* IPv4 gids */
@@ -1888,14 +1888,6 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 
 	pr_info_once("%s", mlx4_ib_version);
 
-	mlx4_foreach_non_ib_transport_port(i, dev)
-		num_ports++;
-
-	if (mlx4_is_mfunc(dev) && num_ports) {
-		dev_err(&dev->pdev->dev, "RoCE is not supported over SRIOV as yet\n");
-		return NULL;
-	}
-
 	num_ports = 0;
 	mlx4_foreach_ib_transport_port(i, dev)
 		num_ports++;
@@ -2331,17 +2323,24 @@ static void do_slave_init(struct mlx4_ib_dev *ibdev, int slave, int do_init)
 	struct mlx4_dev *dev = ibdev->dev;
 	int i;
 	unsigned long flags;
+	struct mlx4_active_ports actv_ports;
+	unsigned int ports;
+	unsigned int first_port;
 
 	if (!mlx4_is_master(dev))
 		return;
 
-	dm = kcalloc(dev->caps.num_ports, sizeof *dm, GFP_ATOMIC);
+	actv_ports = mlx4_get_active_ports(dev, slave);
+	ports = bitmap_weight(actv_ports.ports, dev->caps.num_ports);
+	first_port = find_first_bit(actv_ports.ports, dev->caps.num_ports);
+
+	dm = kcalloc(ports, sizeof(*dm), GFP_ATOMIC);
 	if (!dm) {
 		pr_err("failed to allocate memory for tunneling qp update\n");
 		goto out;
 	}
 
-	for (i = 0; i < dev->caps.num_ports; i++) {
+	for (i = 0; i < ports; i++) {
 		dm[i] = kmalloc(sizeof (struct mlx4_ib_demux_work), GFP_ATOMIC);
 		if (!dm[i]) {
 			pr_err("failed to allocate memory for tunneling qp update work struct\n");
@@ -2353,9 +2352,9 @@ static void do_slave_init(struct mlx4_ib_dev *ibdev, int slave, int do_init)
 		}
 	}
 	/* initialize or tear down tunnel QPs for the slave */
-	for (i = 0; i < dev->caps.num_ports; i++) {
+	for (i = 0; i < ports; i++) {
 		INIT_WORK(&dm[i]->work, mlx4_ib_tunnels_update_work);
-		dm[i]->port = i + 1;
+		dm[i]->port = first_port + i + 1;
 		dm[i]->slave = slave;
 		dm[i]->do_init = do_init;
 		dm[i]->dev = ibdev;
diff --git a/drivers/infiniband/hw/mlx4/mcg.c b/drivers/infiniband/hw/mlx4/mcg.c
index 25b2cdff00f8..ed327e6c8fdc 100644
--- a/drivers/infiniband/hw/mlx4/mcg.c
+++ b/drivers/infiniband/hw/mlx4/mcg.c
@@ -215,8 +215,9 @@ static int send_mad_to_wire(struct mlx4_ib_demux_ctx *ctx, struct ib_mad *mad)
 	}
 	mlx4_ib_query_ah(dev->sm_ah[ctx->port - 1], &ah_attr);
 	spin_unlock(&dev->sm_lock);
-	return mlx4_ib_send_to_wire(dev, mlx4_master_func_num(dev->dev), ctx->port,
-				    IB_QPT_GSI, 0, 1, IB_QP1_QKEY, &ah_attr, mad);
+	return mlx4_ib_send_to_wire(dev, mlx4_master_func_num(dev->dev),
+				    ctx->port, IB_QPT_GSI, 0, 1, IB_QP1_QKEY,
+				    &ah_attr, NULL, mad);
 }
 
 static int send_mad_to_slave(int slave, struct mlx4_ib_demux_ctx *ctx,
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index a230683af940..f589522fddfd 100644
--- a/drivers/infiniband/hw/mlx4/mlx4_ib.h
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -241,6 +241,22 @@ struct mlx4_ib_proxy_sqp_hdr {
 	struct mlx4_rcv_tunnel_hdr tun;
 }  __packed;
 
+struct mlx4_roce_smac_vlan_info {
+	u64 smac;
+	int smac_index;
+	int smac_port;
+	u64 candidate_smac;
+	int candidate_smac_index;
+	int candidate_smac_port;
+	u16 vid;
+	int vlan_index;
+	int vlan_port;
+	u16 candidate_vid;
+	int candidate_vlan_index;
+	int candidate_vlan_port;
+	int update_vid;
+};
+
 struct mlx4_ib_qp {
 	struct ib_qp		ibqp;
 	struct mlx4_qp		mqp;
@@ -273,8 +289,9 @@ struct mlx4_ib_qp {
 	struct list_head	gid_list;
 	struct list_head	steering_rules;
 	struct mlx4_ib_buf	*sqp_proxy_rcv;
+	struct mlx4_roce_smac_vlan_info pri;
+	struct mlx4_roce_smac_vlan_info alt;
 	u64			reg_id;
-
 };
 
 struct mlx4_ib_srq {
@@ -720,9 +737,12 @@ void mlx4_ib_tunnels_update_work(struct work_struct *work);
 int mlx4_ib_send_to_slave(struct mlx4_ib_dev *dev, int slave, u8 port,
 			  enum ib_qp_type qpt, struct ib_wc *wc,
 			  struct ib_grh *grh, struct ib_mad *mad);
+
 int mlx4_ib_send_to_wire(struct mlx4_ib_dev *dev, int slave, u8 port,
 			 enum ib_qp_type dest_qpt, u16 pkey_index, u32 remote_qpn,
-			 u32 qkey, struct ib_ah_attr *attr, struct ib_mad *mad);
+			 u32 qkey, struct ib_ah_attr *attr, u8 *s_mac,
+			 struct ib_mad *mad);
+
 __be64 mlx4_ib_get_new_demux_tid(struct mlx4_ib_demux_ctx *ctx);
 
 int mlx4_ib_demux_cm_handler(struct ib_device *ibdev, int port, int *slave,
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index d8f4d1fe8494..aadf7f82e1f3 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -662,10 +662,14 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd,
 			if (!sqp)
 				return -ENOMEM;
 			qp = &sqp->qp;
+			qp->pri.vid = 0xFFFF;
+			qp->alt.vid = 0xFFFF;
 		} else {
 			qp = kzalloc(sizeof (struct mlx4_ib_qp), GFP_KERNEL);
 			if (!qp)
 				return -ENOMEM;
+			qp->pri.vid = 0xFFFF;
+			qp->alt.vid = 0xFFFF;
 		}
 	} else
 		qp = *caller_qp;
@@ -940,11 +944,32 @@ static void destroy_qp_common(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp,
 {
 	struct mlx4_ib_cq *send_cq, *recv_cq;
 
-	if (qp->state != IB_QPS_RESET)
+	if (qp->state != IB_QPS_RESET) {
 		if (mlx4_qp_modify(dev->dev, NULL, to_mlx4_state(qp->state),
 				   MLX4_QP_STATE_RST, NULL, 0, 0, &qp->mqp))
 			pr_warn("modify QP %06x to RESET failed.\n",
 			       qp->mqp.qpn);
+		if (qp->pri.smac) {
+			mlx4_unregister_mac(dev->dev, qp->pri.smac_port, qp->pri.smac);
+			qp->pri.smac = 0;
+		}
+		if (qp->alt.smac) {
+			mlx4_unregister_mac(dev->dev, qp->alt.smac_port, qp->alt.smac);
+			qp->alt.smac = 0;
+		}
+		if (qp->pri.vid < 0x1000) {
+			mlx4_unregister_vlan(dev->dev, qp->pri.vlan_port, qp->pri.vid);
+			qp->pri.vid = 0xFFFF;
+			qp->pri.candidate_vid = 0xFFFF;
+			qp->pri.update_vid = 0;
+		}
+		if (qp->alt.vid < 0x1000) {
+			mlx4_unregister_vlan(dev->dev, qp->alt.vlan_port, qp->alt.vid);
+			qp->alt.vid = 0xFFFF;
+			qp->alt.candidate_vid = 0xFFFF;
+			qp->alt.update_vid = 0;
+		}
+	}
 
 	get_cqs(qp, &send_cq, &recv_cq);
 
@@ -1057,6 +1082,8 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
 		qp = kzalloc(sizeof *qp, GFP_KERNEL);
 		if (!qp)
 			return ERR_PTR(-ENOMEM);
+		qp->pri.vid = 0xFFFF;
+		qp->alt.vid = 0xFFFF;
 		/* fall through */
 	case IB_QPT_UD:
 	{
@@ -1188,12 +1215,13 @@ static void mlx4_set_sched(struct mlx4_qp_path *path, u8 port)
 
 static int _mlx4_set_path(struct mlx4_ib_dev *dev, const struct ib_ah_attr *ah,
 			  u64 smac, u16 vlan_tag, struct mlx4_qp_path *path,
-			  u8 port)
+			  struct mlx4_roce_smac_vlan_info *smac_info, u8 port)
 {
 	int is_eth = rdma_port_get_link_layer(&dev->ib_dev, port) ==
 		IB_LINK_LAYER_ETHERNET;
 	int vidx;
 	int smac_index;
+	int err;
 
 
 	path->grh_mylmc     = ah->src_path_bits & 0x7f;
@@ -1223,61 +1251,103 @@ static int _mlx4_set_path(struct mlx4_ib_dev *dev, const struct ib_ah_attr *ah,
 	}
 
 	if (is_eth) {
-		path->sched_queue = MLX4_IB_DEFAULT_SCHED_QUEUE |
-			((port - 1) << 6) | ((ah->sl & 7) << 3);
-
 		if (!(ah->ah_flags & IB_AH_GRH))
 			return -1;
 
-		memcpy(path->dmac, ah->dmac, ETH_ALEN);
-		path->ackto = MLX4_IB_LINK_TYPE_ETH;
-		/* find the index  into MAC table for IBoE */
-		if (!is_zero_ether_addr((const u8 *)&smac)) {
-			if (mlx4_find_cached_mac(dev->dev, port, smac,
-						 &smac_index))
-				return -ENOENT;
-		} else {
-			smac_index = 0;
-		}
-
-		path->grh_mylmc &= 0x80 | smac_index;
+		path->sched_queue = MLX4_IB_DEFAULT_SCHED_QUEUE |
+			((port - 1) << 6) | ((ah->sl & 7) << 3);
 
 		path->feup |= MLX4_FEUP_FORCE_ETH_UP;
 		if (vlan_tag < 0x1000) {
-			if (mlx4_find_cached_vlan(dev->dev, port, vlan_tag, &vidx))
-				return -ENOENT;
-
-			path->vlan_index = vidx;
-			path->fl = 1 << 6;
+			if (smac_info->vid < 0x1000) {
+				/* both valid vlan ids */
+				if (smac_info->vid != vlan_tag) {
+					/* different VIDs.  unreg old and reg new */
+					err = mlx4_register_vlan(dev->dev, port, vlan_tag, &vidx);
+					if (err)
+						return err;
+					smac_info->candidate_vid = vlan_tag;
+					smac_info->candidate_vlan_index = vidx;
+					smac_info->candidate_vlan_port = port;
+					smac_info->update_vid = 1;
+					path->vlan_index = vidx;
+				} else {
+					path->vlan_index = smac_info->vlan_index;
+				}
+			} else {
+				/* no current vlan tag in qp */
+				err = mlx4_register_vlan(dev->dev, port, vlan_tag, &vidx);
+				if (err)
+					return err;
+				smac_info->candidate_vid = vlan_tag;
+				smac_info->candidate_vlan_index = vidx;
+				smac_info->candidate_vlan_port = port;
+				smac_info->update_vid = 1;
+				path->vlan_index = vidx;
+			}
 			path->feup |= MLX4_FVL_FORCE_ETH_VLAN;
+			path->fl = 1 << 6;
+		} else {
+			/* have current vlan tag. unregister it at modify-qp success */
+			if (smac_info->vid < 0x1000) {
+				smac_info->candidate_vid = 0xFFFF;
+				smac_info->update_vid = 1;
+			}
 		}
-	} else
+
+		/* get smac_index for RoCE use.
+		 * If no smac was yet assigned, register one.
+		 * If one was already assigned, but the new mac differs,
+		 * unregister the old one and register the new one.
+		*/
+		if (!smac_info->smac || smac_info->smac != smac) {
+			/* register candidate now, unreg if needed, after success */
+			smac_index = mlx4_register_mac(dev->dev, port, smac);
+			if (smac_index >= 0) {
+				smac_info->candidate_smac_index = smac_index;
+				smac_info->candidate_smac = smac;
+				smac_info->candidate_smac_port = port;
+			} else {
+				return -EINVAL;
+			}
+		} else {
+			smac_index = smac_info->smac_index;
+		}
+
+		memcpy(path->dmac, ah->dmac, 6);
+		path->ackto = MLX4_IB_LINK_TYPE_ETH;
+		/* put MAC table smac index for IBoE */
+		path->grh_mylmc = (u8) (smac_index) | 0x80;
+	} else {
 		path->sched_queue = MLX4_IB_DEFAULT_SCHED_QUEUE |
 			((port - 1) << 6) | ((ah->sl & 0xf) << 2);
+	}
 
 	return 0;
 }
 
 static int mlx4_set_path(struct mlx4_ib_dev *dev, const struct ib_qp_attr *qp,
 			 enum ib_qp_attr_mask qp_attr_mask,
+			 struct mlx4_ib_qp *mqp,
 			 struct mlx4_qp_path *path, u8 port)
 {
 	return _mlx4_set_path(dev, &qp->ah_attr,
 			      mlx4_mac_to_u64((u8 *)qp->smac),
 			      (qp_attr_mask & IB_QP_VID) ? qp->vlan_id : 0xffff,
-			      path, port);
+			      path, &mqp->pri, port);
 }
 
 static int mlx4_set_alt_path(struct mlx4_ib_dev *dev,
 			     const struct ib_qp_attr *qp,
 			     enum ib_qp_attr_mask qp_attr_mask,
+			     struct mlx4_ib_qp *mqp,
 			     struct mlx4_qp_path *path, u8 port)
 {
 	return _mlx4_set_path(dev, &qp->alt_ah_attr,
 			      mlx4_mac_to_u64((u8 *)qp->alt_smac),
 			      (qp_attr_mask & IB_QP_ALT_VID) ?
 			      qp->alt_vlan_id : 0xffff,
-			      path, port);
+			      path, &mqp->alt, port);
 }
 
 static void update_mcg_macs(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp)
@@ -1292,6 +1362,37 @@ static void update_mcg_macs(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp)
 	}
 }
 
+static int handle_eth_ud_smac_index(struct mlx4_ib_dev *dev, struct mlx4_ib_qp *qp, u8 *smac,
+				    struct mlx4_qp_context *context)
+{
+	struct net_device *ndev;
+	u64 u64_mac;
+	int smac_index;
+
+
+	ndev = dev->iboe.netdevs[qp->port - 1];
+	if (ndev) {
+		smac = ndev->dev_addr;
+		u64_mac = mlx4_mac_to_u64(smac);
+	} else {
+		u64_mac = dev->dev->caps.def_mac[qp->port];
+	}
+
+	context->pri_path.sched_queue = MLX4_IB_DEFAULT_SCHED_QUEUE | ((qp->port - 1) << 6);
+	if (!qp->pri.smac) {
+		smac_index = mlx4_register_mac(dev->dev, qp->port, u64_mac);
+		if (smac_index >= 0) {
+			qp->pri.candidate_smac_index = smac_index;
+			qp->pri.candidate_smac = u64_mac;
+			qp->pri.candidate_smac_port = qp->port;
+			context->pri_path.grh_mylmc = 0x80 | (u8) smac_index;
+		} else {
+			return -ENOENT;
+		}
+	}
+	return 0;
+}
+
 static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
 			       const struct ib_qp_attr *attr, int attr_mask,
 			       enum ib_qp_state cur_state, enum ib_qp_state new_state)
@@ -1403,7 +1504,7 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
 	}
 
 	if (attr_mask & IB_QP_AV) {
-		if (mlx4_set_path(dev, attr, attr_mask, &context->pri_path,
+		if (mlx4_set_path(dev, attr, attr_mask, qp, &context->pri_path,
 				  attr_mask & IB_QP_PORT ?
 				  attr->port_num : qp->port))
 			goto out;
@@ -1426,7 +1527,8 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
 		    dev->dev->caps.pkey_table_len[attr->alt_port_num])
 			goto out;
 
-		if (mlx4_set_alt_path(dev, attr, attr_mask, &context->alt_path,
+		if (mlx4_set_alt_path(dev, attr, attr_mask, qp,
+				      &context->alt_path,
 				      attr->alt_port_num))
 			goto out;
 
@@ -1532,6 +1634,20 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
 				context->pri_path.fl = 0x80;
 			context->pri_path.sched_queue |= MLX4_IB_DEFAULT_SCHED_QUEUE;
 		}
+		if (rdma_port_get_link_layer(&dev->ib_dev, qp->port) ==
+		    IB_LINK_LAYER_ETHERNET) {
+			if (qp->mlx4_ib_qp_type == MLX4_IB_QPT_TUN_GSI ||
+			    qp->mlx4_ib_qp_type == MLX4_IB_QPT_GSI)
+				context->pri_path.feup = 1 << 7; /* don't fsm */
+			/* handle smac_index */
+			if (qp->mlx4_ib_qp_type == MLX4_IB_QPT_UD ||
+			    qp->mlx4_ib_qp_type == MLX4_IB_QPT_PROXY_GSI ||
+			    qp->mlx4_ib_qp_type == MLX4_IB_QPT_TUN_GSI) {
+				err = handle_eth_ud_smac_index(dev, qp, (u8 *)attr->smac, context);
+				if (err)
+					return -EINVAL;
+			}
+		}
 	}
 
 	if (qp->ibqp.qp_type == IB_QPT_RAW_PACKET)
@@ -1619,28 +1735,113 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
 	 * If we moved a kernel QP to RESET, clean up all old CQ
 	 * entries and reinitialize the QP.
 	 */
-	if (new_state == IB_QPS_RESET && !ibqp->uobject) {
-		mlx4_ib_cq_clean(recv_cq, qp->mqp.qpn,
-				 ibqp->srq ? to_msrq(ibqp->srq): NULL);
-		if (send_cq != recv_cq)
-			mlx4_ib_cq_clean(send_cq, qp->mqp.qpn, NULL);
+	if (new_state == IB_QPS_RESET) {
+		if (!ibqp->uobject) {
+			mlx4_ib_cq_clean(recv_cq, qp->mqp.qpn,
+					 ibqp->srq ? to_msrq(ibqp->srq) : NULL);
+			if (send_cq != recv_cq)
+				mlx4_ib_cq_clean(send_cq, qp->mqp.qpn, NULL);
+
+			qp->rq.head = 0;
+			qp->rq.tail = 0;
+			qp->sq.head = 0;
+			qp->sq.tail = 0;
+			qp->sq_next_wqe = 0;
+			if (qp->rq.wqe_cnt)
+				*qp->db.db  = 0;
 
-		qp->rq.head = 0;
-		qp->rq.tail = 0;
-		qp->sq.head = 0;
-		qp->sq.tail = 0;
-		qp->sq_next_wqe = 0;
-		if (qp->rq.wqe_cnt)
-			*qp->db.db  = 0;
+			if (qp->flags & MLX4_IB_QP_NETIF)
+				mlx4_ib_steer_qp_reg(dev, qp, 0);
+		}
+		if (qp->pri.smac) {
+			mlx4_unregister_mac(dev->dev, qp->pri.smac_port, qp->pri.smac);
+			qp->pri.smac = 0;
+		}
+		if (qp->alt.smac) {
+			mlx4_unregister_mac(dev->dev, qp->alt.smac_port, qp->alt.smac);
+			qp->alt.smac = 0;
+		}
+		if (qp->pri.vid < 0x1000) {
+			mlx4_unregister_vlan(dev->dev, qp->pri.vlan_port, qp->pri.vid);
+			qp->pri.vid = 0xFFFF;
+			qp->pri.candidate_vid = 0xFFFF;
+			qp->pri.update_vid = 0;
+		}
 
-		if (qp->flags & MLX4_IB_QP_NETIF)
-			mlx4_ib_steer_qp_reg(dev, qp, 0);
+		if (qp->alt.vid < 0x1000) {
+			mlx4_unregister_vlan(dev->dev, qp->alt.vlan_port, qp->alt.vid);
+			qp->alt.vid = 0xFFFF;
+			qp->alt.candidate_vid = 0xFFFF;
+			qp->alt.update_vid = 0;
+		}
 	}
-
 out:
 	if (err && steer_qp)
 		mlx4_ib_steer_qp_reg(dev, qp, 0);
 	kfree(context);
+	if (qp->pri.candidate_smac) {
+		if (err) {
+			mlx4_unregister_mac(dev->dev, qp->pri.candidate_smac_port, qp->pri.candidate_smac);
+		} else {
+			if (qp->pri.smac)
+				mlx4_unregister_mac(dev->dev, qp->pri.smac_port, qp->pri.smac);
+			qp->pri.smac = qp->pri.candidate_smac;
+			qp->pri.smac_index = qp->pri.candidate_smac_index;
+			qp->pri.smac_port = qp->pri.candidate_smac_port;
+		}
+		qp->pri.candidate_smac = 0;
+		qp->pri.candidate_smac_index = 0;
+		qp->pri.candidate_smac_port = 0;
+	}
+	if (qp->alt.candidate_smac) {
+		if (err) {
+			mlx4_unregister_mac(dev->dev, qp->alt.candidate_smac_port, qp->alt.candidate_smac);
+		} else {
+			if (qp->alt.smac)
+				mlx4_unregister_mac(dev->dev, qp->alt.smac_port, qp->alt.smac);
+			qp->alt.smac = qp->alt.candidate_smac;
+			qp->alt.smac_index = qp->alt.candidate_smac_index;
+			qp->alt.smac_port = qp->alt.candidate_smac_port;
+		}
+		qp->alt.candidate_smac = 0;
+		qp->alt.candidate_smac_index = 0;
+		qp->alt.candidate_smac_port = 0;
+	}
+
+	if (qp->pri.update_vid) {
+		if (err) {
+			if (qp->pri.candidate_vid < 0x1000)
+				mlx4_unregister_vlan(dev->dev, qp->pri.candidate_vlan_port,
+						     qp->pri.candidate_vid);
+		} else {
+			if (qp->pri.vid < 0x1000)
+				mlx4_unregister_vlan(dev->dev, qp->pri.vlan_port,
+						     qp->pri.vid);
+			qp->pri.vid = qp->pri.candidate_vid;
+			qp->pri.vlan_port = qp->pri.candidate_vlan_port;
+			qp->pri.vlan_index =  qp->pri.candidate_vlan_index;
+		}
+		qp->pri.candidate_vid = 0xFFFF;
+		qp->pri.update_vid = 0;
+	}
+
+	if (qp->alt.update_vid) {
+		if (err) {
+			if (qp->alt.candidate_vid < 0x1000)
+				mlx4_unregister_vlan(dev->dev, qp->alt.candidate_vlan_port,
+						     qp->alt.candidate_vid);
+		} else {
+			if (qp->alt.vid < 0x1000)
+				mlx4_unregister_vlan(dev->dev, qp->alt.vlan_port,
+						     qp->alt.vid);
+			qp->alt.vid = qp->alt.candidate_vid;
+			qp->alt.vlan_port = qp->alt.candidate_vlan_port;
+			qp->alt.vlan_index =  qp->alt.candidate_vlan_index;
+		}
+		qp->alt.candidate_vid = 0xFFFF;
+		qp->alt.update_vid = 0;
+	}
+
 	return err;
 }
 
@@ -1842,9 +2043,9 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr,
 {
 	struct ib_device *ib_dev = sqp->qp.ibqp.device;
 	struct mlx4_wqe_mlx_seg *mlx = wqe;
+	struct mlx4_wqe_ctrl_seg *ctrl = wqe;
 	struct mlx4_wqe_inline_seg *inl = wqe + sizeof *mlx;
 	struct mlx4_ib_ah *ah = to_mah(wr->wr.ud.ah);
-	struct net_device *ndev;
 	union ib_gid sgid;
 	u16 pkey;
 	int send_size;
@@ -1868,12 +2069,11 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr,
 			/* When multi-function is enabled, the ib_core gid
 			 * indexes don't necessarily match the hw ones, so
 			 * we must use our own cache */
-			sgid.global.subnet_prefix =
-				to_mdev(ib_dev)->sriov.demux[sqp->qp.port - 1].
-				subnet_prefix;
-			sgid.global.interface_id =
-				to_mdev(ib_dev)->sriov.demux[sqp->qp.port - 1].
-				guid_cache[ah->av.ib.gid_index];
+			err = mlx4_get_roce_gid_from_slave(to_mdev(ib_dev)->dev,
+							   be32_to_cpu(ah->av.ib.port_pd) >> 24,
+							   ah->av.ib.gid_index, &sgid.raw[0]);
+			if (err)
+				return err;
 		} else  {
 			err = ib_get_cached_gid(ib_dev,
 						be32_to_cpu(ah->av.ib.port_pd) >> 24,
@@ -1902,6 +2102,9 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr,
 		sqp->ud_header.grh.flow_label    =
 			ah->av.ib.sl_tclass_flowlabel & cpu_to_be32(0xfffff);
 		sqp->ud_header.grh.hop_limit     = ah->av.ib.hop_limit;
+		if (is_eth)
+			memcpy(sqp->ud_header.grh.source_gid.raw, sgid.raw, 16);
+		else {
 		if (mlx4_is_mfunc(to_mdev(ib_dev)->dev)) {
 			/* When multi-function is enabled, the ib_core gid
 			 * indexes don't necessarily match the hw ones, so
@@ -1917,6 +2120,7 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr,
 					  be32_to_cpu(ah->av.ib.port_pd) >> 24,
 					  ah->av.ib.gid_index,
 					  &sqp->ud_header.grh.source_gid);
+		}
 		memcpy(sqp->ud_header.grh.destination_gid.raw,
 		       ah->av.ib.dgid, 16);
 	}
@@ -1949,16 +2153,23 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr,
 
 	if (is_eth) {
 		u8 *smac;
+		struct in6_addr in6;
+
 		u16 pcp = (be32_to_cpu(ah->av.ib.sl_tclass_flowlabel) >> 29) << 13;
 
 		mlx->sched_prio = cpu_to_be16(pcp);
 
 		memcpy(sqp->ud_header.eth.dmac_h, ah->av.eth.mac, 6);
 		/* FIXME: cache smac value? */
-		ndev = to_mdev(sqp->qp.ibqp.device)->iboe.netdevs[sqp->qp.port - 1];
-		if (!ndev)
-			return -ENODEV;
-		smac = ndev->dev_addr;
+		memcpy(&ctrl->srcrb_flags16[0], ah->av.eth.mac, 2);
+		memcpy(&ctrl->imm, ah->av.eth.mac + 2, 4);
+		memcpy(&in6, sgid.raw, sizeof(in6));
+
+		if (!mlx4_is_mfunc(to_mdev(ib_dev)->dev))
+			smac = to_mdev(sqp->qp.ibqp.device)->
+				iboe.netdevs[sqp->qp.port - 1]->dev_addr;
+		else	/* use the src mac of the tunnel */
+			smac = ah->av.eth.s_mac;
 		memcpy(sqp->ud_header.eth.smac_h, smac, 6);
 		if (!memcmp(sqp->ud_header.eth.smac_h, sqp->ud_header.eth.dmac_h, 6))
 			mlx->flags |= cpu_to_be32(MLX4_WQE_CTRL_FORCE_LOOPBACK);
@@ -2190,6 +2401,8 @@ static void build_tunnel_header(struct ib_send_wr *wr, void *wqe, unsigned *mlx_
 	hdr.remote_qpn = cpu_to_be32(wr->wr.ud.remote_qpn);
 	hdr.pkey_index = cpu_to_be16(wr->wr.ud.pkey_index);
 	hdr.qkey = cpu_to_be32(wr->wr.ud.remote_qkey);
+	memcpy(hdr.mac, ah->av.eth.mac, 6);
+	hdr.vlan = ah->av.eth.vlan;
 
 	spc = MLX4_INLINE_ALIGN -
 		((unsigned long) (inl + 1) & (MLX4_INLINE_ALIGN - 1));
diff --git a/drivers/infiniband/hw/mlx4/sysfs.c b/drivers/infiniband/hw/mlx4/sysfs.c
index db2ea31df832..5a38e43eca65 100644
--- a/drivers/infiniband/hw/mlx4/sysfs.c
+++ b/drivers/infiniband/hw/mlx4/sysfs.c
@@ -627,6 +627,7 @@ static int register_one_pkey_tree(struct mlx4_ib_dev *dev, int slave)
 	int port;
 	struct kobject *p, *t;
 	struct mlx4_port *mport;
+	struct mlx4_active_ports actv_ports;
 
 	get_name(dev, name, slave, sizeof name);
 
@@ -649,7 +650,11 @@ static int register_one_pkey_tree(struct mlx4_ib_dev *dev, int slave)
 		goto err_ports;
 	}
 
+	actv_ports = mlx4_get_active_ports(dev->dev, slave);
+
 	for (port = 1; port <= dev->dev->caps.num_ports; ++port) {
+		if (!test_bit(port - 1, actv_ports.ports))
+			continue;
 		err = add_port(dev, port, slave);
 		if (err)
 			goto err_add;