net/mlx4: Adapt code for N-Port VF

Adds support for N-Port VFs, this includes:
1. Adding support in the wrapped FW command
	In wrapped commands, we need to verify and convert
	the slave's port into the real physical port.
	Furthermore, when sending the response back to the slave,
	a reverse conversion should be made.
2. Adjusting sqpn for QP1 para-virtualization
	The slave assumes that sqpn is used for QP1 communication.
	If the slave is assigned to a port != (first port), we need
	to adjust the sqpn that will direct its QP1 packets into the
	correct endpoint.
3. Adjusting gid[5] to modify the port for raw ethernet
	In B0 steering, gid[5] contains the port. It needs
	to be adjusted into the physical port.
4. Adjusting number of ports in the query / ports caps in the FW commands
	When a slave queries the hardware, it needs to view only
	the physical ports it's assigned to.
5. Adjusting the sched_qp according to the port number
	The QP port is encoded in the sched_qp, thus in modify_qp we need
	to encode the correct port in sched_qp.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

12 files changed, 417 insertions, 79 deletions

diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c
index 2c572aed3f6f..fd36ec672632 100644
--- a/drivers/infiniband/hw/mlx4/mad.c
+++ b/drivers/infiniband/hw/mlx4/mad.c
@@ -1245,21 +1245,9 @@ out:
 
 static int get_slave_base_gid_ix(struct mlx4_ib_dev *dev, int slave, int port)
 {
-	int gids;
-	int vfs;
-
 	if (rdma_port_get_link_layer(&dev->ib_dev, port) == IB_LINK_LAYER_INFINIBAND)
 		return slave;
-
-	gids = MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS;
-	vfs = dev->dev->num_vfs;
-
-	if (slave == 0)
-		return 0;
-	if (slave <= gids % vfs)
-		return MLX4_ROCE_PF_GIDS + ((gids / vfs) + 1) * (slave - 1);
-
-	return MLX4_ROCE_PF_GIDS + (gids % vfs) + ((gids / vfs) * (slave - 1));
+	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,
@@ -1281,6 +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 ||
@@ -1360,6 +1349,10 @@ static void mlx4_ib_multiplex_mad(struct mlx4_ib_demux_pv_ctx *ctx, struct ib_wc
 	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 */
@@ -1949,7 +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 2ff428b17088..6cb85467dde7 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -2323,17 +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");
@@ -2345,9 +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/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;
diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index 59a1b2703281..516c1dd4963b 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -1643,8 +1643,16 @@ static int mlx4_master_activate_admin_state(struct mlx4_priv *priv, int slave)
 	int port, err;
 	struct mlx4_vport_state *vp_admin;
 	struct mlx4_vport_oper_state *vp_oper;
-
-	for (port = 1; port <= MLX4_MAX_PORTS; port++) {
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(
+			&priv->dev, slave);
+	int min_port = find_first_bit(actv_ports.ports,
+				      priv->dev.caps.num_ports) + 1;
+	int max_port = min_port - 1 +
+		bitmap_weight(actv_ports.ports, priv->dev.caps.num_ports);
+
+	for (port = min_port; port <= max_port; port++) {
+		if (!test_bit(port - 1, actv_ports.ports))
+			continue;
 		vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
 		vp_admin = &priv->mfunc.master.vf_admin[slave].vport[port];
 		vp_oper->state = *vp_admin;
@@ -1685,8 +1693,17 @@ static void mlx4_master_deactivate_admin_state(struct mlx4_priv *priv, int slave
 {
 	int port;
 	struct mlx4_vport_oper_state *vp_oper;
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(
+			&priv->dev, slave);
+	int min_port = find_first_bit(actv_ports.ports,
+				      priv->dev.caps.num_ports) + 1;
+	int max_port = min_port - 1 +
+		bitmap_weight(actv_ports.ports, priv->dev.caps.num_ports);
+
 
-	for (port = 1; port <= MLX4_MAX_PORTS; port++) {
+	for (port = min_port; port <= max_port; port++) {
+		if (!test_bit(port - 1, actv_ports.ports))
+			continue;
 		vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
 		if (NO_INDX != vp_oper->vlan_idx) {
 			__mlx4_unregister_vlan(&priv->dev,
diff --git a/drivers/net/ethernet/mellanox/mlx4/eq.c b/drivers/net/ethernet/mellanox/mlx4/eq.c
index 8992b38578d5..d501a2b0fb79 100644
--- a/drivers/net/ethernet/mellanox/mlx4/eq.c
+++ b/drivers/net/ethernet/mellanox/mlx4/eq.c
@@ -271,7 +271,10 @@ enum slave_port_state mlx4_get_slave_port_state(struct mlx4_dev *dev, int slave,
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
 	struct mlx4_slave_state *s_state = priv->mfunc.master.slave_state;
-	if (slave >= dev->num_slaves || port > MLX4_MAX_PORTS) {
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
+
+	if (slave >= dev->num_slaves || port > dev->caps.num_ports ||
+	    port <= 0 || !test_bit(port - 1, actv_ports.ports)) {
 		pr_err("%s: Error: asking for slave:%d, port:%d\n",
 		       __func__, slave, port);
 		return SLAVE_PORT_DOWN;
@@ -285,8 +288,10 @@ static int mlx4_set_slave_port_state(struct mlx4_dev *dev, int slave, u8 port,
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
 	struct mlx4_slave_state *s_state = priv->mfunc.master.slave_state;
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
 
-	if (slave >= dev->num_slaves || port > MLX4_MAX_PORTS || port == 0) {
+	if (slave >= dev->num_slaves || port > dev->caps.num_ports ||
+	    port <= 0 || !test_bit(port - 1, actv_ports.ports)) {
 		pr_err("%s: Error: asking for slave:%d, port:%d\n",
 		       __func__, slave, port);
 		return -1;
@@ -300,9 +305,13 @@ static void set_all_slave_state(struct mlx4_dev *dev, u8 port, int event)
 {
 	int i;
 	enum slave_port_gen_event gen_event;
+	struct mlx4_slaves_pport slaves_pport = mlx4_phys_to_slaves_pport(dev,
+									  port);
 
-	for (i = 0; i < dev->num_slaves; i++)
-		set_and_calc_slave_port_state(dev, i, port, event, &gen_event);
+	for (i = 0; i < dev->num_vfs + 1; i++)
+		if (test_bit(i, slaves_pport.slaves))
+			set_and_calc_slave_port_state(dev, i, port,
+						      event, &gen_event);
 }
 /**************************************************************************
 	The function get as input the new event to that port,
@@ -321,12 +330,14 @@ int set_and_calc_slave_port_state(struct mlx4_dev *dev, int slave,
 	struct mlx4_slave_state *ctx = NULL;
 	unsigned long flags;
 	int ret = -1;
+	struct mlx4_active_ports actv_ports = mlx4_get_active_ports(dev, slave);
 	enum slave_port_state cur_state =
 		mlx4_get_slave_port_state(dev, slave, port);
 
 	*gen_event = SLAVE_PORT_GEN_EVENT_NONE;
 
-	if (slave >= dev->num_slaves || port > MLX4_MAX_PORTS || port == 0) {
+	if (slave >= dev->num_slaves || port > dev->caps.num_ports ||
+	    port <= 0 || !test_bit(port - 1, actv_ports.ports)) {
 		pr_err("%s: Error: asking for slave:%d, port:%d\n",
 		       __func__, slave, port);
 		return ret;
@@ -542,15 +553,19 @@ static int mlx4_eq_int(struct mlx4_dev *dev, struct mlx4_eq *eq)
 				       be64_to_cpu(eqe->event.cmd.out_param));
 			break;
 
-		case MLX4_EVENT_TYPE_PORT_CHANGE:
+		case MLX4_EVENT_TYPE_PORT_CHANGE: {
+			struct mlx4_slaves_pport slaves_port;
 			port = be32_to_cpu(eqe->event.port_change.port) >> 28;
+			slaves_port = mlx4_phys_to_slaves_pport(dev, port);
 			if (eqe->subtype == MLX4_PORT_CHANGE_SUBTYPE_DOWN) {
 				mlx4_dispatch_event(dev, MLX4_DEV_EVENT_PORT_DOWN,
 						    port);
 				mlx4_priv(dev)->sense.do_sense_port[port] = 1;
 				if (!mlx4_is_master(dev))
 					break;
-				for (i = 0; i < dev->num_slaves; i++) {
+				for (i = 0; i < dev->num_vfs + 1; i++) {
+					if (!test_bit(i, slaves_port.slaves))
+						continue;
 					if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH) {
 						if (i == mlx4_master_func_num(dev))
 							continue;
@@ -558,8 +573,13 @@ static int mlx4_eq_int(struct mlx4_dev *dev, struct mlx4_eq *eq)
 							 " to slave: %d, port:%d\n",
 							 __func__, i, port);
 						s_info = &priv->mfunc.master.vf_oper[slave].vport[port].state;
-						if (IFLA_VF_LINK_STATE_AUTO == s_info->link_state)
+						if (IFLA_VF_LINK_STATE_AUTO == s_info->link_state) {
+							eqe->event.port_change.port =
+								cpu_to_be32(
+								(be32_to_cpu(eqe->event.port_change.port) & 0xFFFFFFF)
+								| (mlx4_phys_to_slave_port(dev, i, port) << 28));
 							mlx4_slave_event(dev, i, eqe);
+						}
 					} else {  /* IB port */
 						set_and_calc_slave_port_state(dev, i, port,
 									      MLX4_PORT_STATE_DEV_EVENT_PORT_DOWN,
@@ -580,12 +600,19 @@ static int mlx4_eq_int(struct mlx4_dev *dev, struct mlx4_eq *eq)
 				if (!mlx4_is_master(dev))
 					break;
 				if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH)
-					for (i = 0; i < dev->num_slaves; i++) {
+					for (i = 0; i < dev->num_vfs + 1; i++) {
+						if (!test_bit(i, slaves_port.slaves))
+							continue;
 						if (i == mlx4_master_func_num(dev))
 							continue;
 						s_info = &priv->mfunc.master.vf_oper[slave].vport[port].state;
-						if (IFLA_VF_LINK_STATE_AUTO == s_info->link_state)
+						if (IFLA_VF_LINK_STATE_AUTO == s_info->link_state) {
+							eqe->event.port_change.port =
+								cpu_to_be32(
+								(be32_to_cpu(eqe->event.port_change.port) & 0xFFFFFFF)
+								| (mlx4_phys_to_slave_port(dev, i, port) << 28));
 							mlx4_slave_event(dev, i, eqe);
+						}
 					}
 				else /* IB port */
 					/* port-up event will be sent to a slave when the
@@ -594,6 +621,7 @@ static int mlx4_eq_int(struct mlx4_dev *dev, struct mlx4_eq *eq)
 					set_all_slave_state(dev, port, MLX4_DEV_EVENT_PORT_UP);
 			}
 			break;
+		}
 
 		case MLX4_EVENT_TYPE_CQ_ERROR:
 			mlx4_warn(dev, "CQ %s on CQN %06x\n",
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.c b/drivers/net/ethernet/mellanox/mlx4/fw.c
index d0d8dd832557..6bd33e2fc17c 100644
--- a/drivers/net/ethernet/mellanox/mlx4/fw.c
+++ b/drivers/net/ethernet/mellanox/mlx4/fw.c
@@ -225,13 +225,25 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
 #define QUERY_FUNC_CAP_FLAGS0_FORCE_PHY_WQE_GID 0x80
 
 	if (vhcr->op_modifier == 1) {
+		struct mlx4_active_ports actv_ports =
+			mlx4_get_active_ports(dev, slave);
+		int converted_port = mlx4_slave_convert_port(
+				dev, slave, vhcr->in_modifier);
+
+		if (converted_port < 0)
+			return -EINVAL;
+
+		vhcr->in_modifier = converted_port;
 		/* Set nic_info bit to mark new fields support */
 		field  = QUERY_FUNC_CAP_FLAGS1_NIC_INFO;
 		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_FLAGS1_OFFSET);
 
-		field = vhcr->in_modifier; /* phys-port = logical-port */
+		/* phys-port = logical-port */
+		field = vhcr->in_modifier -
+			find_first_bit(actv_ports.ports, dev->caps.num_ports);
 		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_PHYS_PORT_OFFSET);
 
+		field = vhcr->in_modifier;
 		/* size is now the QP number */
 		size = dev->phys_caps.base_tunnel_sqpn + 8 * slave + field - 1;
 		MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP0_TUNNEL);
@@ -249,12 +261,16 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
 			 QUERY_FUNC_CAP_PHYS_PORT_ID);
 
 	} else if (vhcr->op_modifier == 0) {
+		struct mlx4_active_ports actv_ports =
+			mlx4_get_active_ports(dev, slave);
 		/* enable rdma and ethernet interfaces, and new quota locations */
 		field = (QUERY_FUNC_CAP_FLAG_ETH | QUERY_FUNC_CAP_FLAG_RDMA |
 			 QUERY_FUNC_CAP_FLAG_QUOTAS);
 		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_FLAGS_OFFSET);
 
-		field = dev->caps.num_ports;
+		field = min(
+			bitmap_weight(actv_ports.ports, dev->caps.num_ports),
+			dev->caps.num_ports);
 		MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_NUM_PORTS_OFFSET);
 
 		size = dev->caps.function_caps; /* set PF behaviours */
@@ -840,6 +856,10 @@ int mlx4_QUERY_DEV_CAP_wrapper(struct mlx4_dev *dev, int slave,
 	int	err = 0;
 	u8	field;
 	u32	bmme_flags;
+	int	real_port;
+	int	slave_port;
+	int	first_port;
+	struct mlx4_active_ports actv_ports;
 
 	err = mlx4_cmd_box(dev, 0, outbox->dma, 0, 0, MLX4_CMD_QUERY_DEV_CAP,
 			   MLX4_CMD_TIME_CLASS_A, MLX4_CMD_NATIVE);
@@ -852,8 +872,26 @@ int mlx4_QUERY_DEV_CAP_wrapper(struct mlx4_dev *dev, int slave,
 	MLX4_GET(flags, outbox->buf, QUERY_DEV_CAP_EXT_FLAGS_OFFSET);
 	flags |= MLX4_DEV_CAP_FLAG_PORT_MNG_CHG_EV;
 	flags &= ~MLX4_DEV_CAP_FLAG_MEM_WINDOW;
+	actv_ports = mlx4_get_active_ports(dev, slave);
+	first_port = find_first_bit(actv_ports.ports, dev->caps.num_ports);
+	for (slave_port = 0, real_port = first_port;
+	     real_port < first_port +
+	     bitmap_weight(actv_ports.ports, dev->caps.num_ports);
+	     ++real_port, ++slave_port) {
+		if (flags & (MLX4_DEV_CAP_FLAG_WOL_PORT1 << real_port))
+			flags |= MLX4_DEV_CAP_FLAG_WOL_PORT1 << slave_port;
+		else
+			flags &= ~(MLX4_DEV_CAP_FLAG_WOL_PORT1 << slave_port);
+	}
+	for (; slave_port < dev->caps.num_ports; ++slave_port)
+		flags &= ~(MLX4_DEV_CAP_FLAG_WOL_PORT1 << slave_port);
 	MLX4_PUT(outbox->buf, flags, QUERY_DEV_CAP_EXT_FLAGS_OFFSET);
 
+	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_VL_PORT_OFFSET);
+	field &= ~0x0F;
+	field |= bitmap_weight(actv_ports.ports, dev->caps.num_ports) & 0x0F;
+	MLX4_PUT(outbox->buf, field, QUERY_DEV_CAP_VL_PORT_OFFSET);
+
 	/* For guests, disable timestamp */
 	MLX4_GET(field, outbox->buf, QUERY_DEV_CAP_CQ_TS_SUPPORT_OFFSET);
 	field &= 0x7f;
@@ -903,12 +941,20 @@ int mlx4_QUERY_PORT_wrapper(struct mlx4_dev *dev, int slave,
 	u16 short_field;
 	int err;
 	int admin_link_state;
+	int port = mlx4_slave_convert_port(dev, slave,
+					   vhcr->in_modifier & 0xFF);
 
 #define MLX4_VF_PORT_NO_LINK_SENSE_MASK	0xE0
 #define MLX4_PORT_LINK_UP_MASK		0x80
 #define QUERY_PORT_CUR_MAX_PKEY_OFFSET	0x0c
 #define QUERY_PORT_CUR_MAX_GID_OFFSET	0x0e
 
+	if (port < 0)
+		return -EINVAL;
+
+	vhcr->in_modifier = (vhcr->in_modifier & ~0xFF) |
+			    (port & 0xFF);
+
 	err = mlx4_cmd_box(dev, 0, outbox->dma, vhcr->in_modifier, 0,
 			   MLX4_CMD_QUERY_PORT, MLX4_CMD_TIME_CLASS_B,
 			   MLX4_CMD_NATIVE);
@@ -936,7 +982,7 @@ int mlx4_QUERY_PORT_wrapper(struct mlx4_dev *dev, int slave,
 			 QUERY_PORT_SUPPORTED_TYPE_OFFSET);
 
 		if (dev->caps.port_type[vhcr->in_modifier] == MLX4_PORT_TYPE_ETH)
-			short_field = mlx4_get_slave_num_gids(dev, slave);
+			short_field = mlx4_get_slave_num_gids(dev, slave, port);
 		else
 			short_field = 1; /* slave max gids */
 		MLX4_PUT(outbox->buf, short_field,
@@ -1588,9 +1634,12 @@ int mlx4_INIT_PORT_wrapper(struct mlx4_dev *dev, int slave,
 			   struct mlx4_cmd_info *cmd)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
-	int port = vhcr->in_modifier;
+	int port = mlx4_slave_convert_port(dev, slave, vhcr->in_modifier);
 	int err;
 
+	if (port < 0)
+		return -EINVAL;
+
 	if (priv->mfunc.master.slave_state[slave].init_port_mask & (1 << port))
 		return 0;
 
@@ -1680,9 +1729,12 @@ int mlx4_CLOSE_PORT_wrapper(struct mlx4_dev *dev, int slave,
 			    struct mlx4_cmd_info *cmd)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
-	int port = vhcr->in_modifier;
+	int port = mlx4_slave_convert_port(dev, slave, vhcr->in_modifier);
 	int err;
 
+	if (port < 0)
+		return -EINVAL;
+
 	if (!(priv->mfunc.master.slave_state[slave].init_port_mask &
 	    (1 << port)))
 		return 0;
diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
index e1a55857af71..472925428de7 100644
--- a/drivers/net/ethernet/mellanox/mlx4/main.c
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -1471,7 +1471,7 @@ static void mlx4_parav_master_pf_caps(struct mlx4_dev *dev)
 	for (i = 1; i <= dev->caps.num_ports; i++) {
 		if (dev->caps.port_type[i] == MLX4_PORT_TYPE_ETH)
 			dev->caps.gid_table_len[i] =
-				mlx4_get_slave_num_gids(dev, 0);
+				mlx4_get_slave_num_gids(dev, 0, i);
 		else
 			dev->caps.gid_table_len[i] = 1;
 		dev->caps.pkey_table_len[i] =
@@ -1498,7 +1498,7 @@ static void choose_steering_mode(struct mlx4_dev *dev,
 	if (mlx4_log_num_mgm_entry_size == -1 &&
 	    dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_FS_EN &&
 	    (!mlx4_is_mfunc(dev) ||
-	     (dev_cap->fs_max_num_qp_per_entry >= (num_vfs + 1))) &&
+	     (dev_cap->fs_max_num_qp_per_entry >= (dev->num_vfs + 1))) &&
 	    choose_log_fs_mgm_entry_size(dev_cap->fs_max_num_qp_per_entry) >=
 		MLX4_MIN_MGM_LOG_ENTRY_SIZE) {
 		dev->oper_log_mgm_entry_size =
diff --git a/drivers/net/ethernet/mellanox/mlx4/mcg.c b/drivers/net/ethernet/mellanox/mlx4/mcg.c
index db7dc0b6667d..80ccb4edf825 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mcg.c
+++ b/drivers/net/ethernet/mellanox/mlx4/mcg.c
@@ -1387,9 +1387,12 @@ int mlx4_PROMISC_wrapper(struct mlx4_dev *dev, int slave,
 			 struct mlx4_cmd_info *cmd)
 {
 	u32 qpn = (u32) vhcr->in_param & 0xffffffff;
-	u8 port = vhcr->in_param >> 62;
+	int port = mlx4_slave_convert_port(dev, slave, vhcr->in_param >> 62);
 	enum mlx4_steer_type steer = vhcr->in_modifier;
 
+	if (port < 0)
+		return -EINVAL;
+
 	/* Promiscuous unicast is not allowed in mfunc */
 	if (mlx4_is_mfunc(dev) && steer == MLX4_UC_STEER)
 		return 0;
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index fe8715e35afa..9fca6c150de3 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -1287,8 +1287,7 @@ void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work);
 
 void mlx4_init_quotas(struct mlx4_dev *dev);
 
-int mlx4_get_slave_num_gids(struct mlx4_dev *dev, int slave);
-int mlx4_get_base_gid_ix(struct mlx4_dev *dev, int slave);
+int mlx4_get_slave_num_gids(struct mlx4_dev *dev, int slave, int port);
 /* Returns the VF index of slave */
 int mlx4_get_vf_indx(struct mlx4_dev *dev, int slave);
 
diff --git a/drivers/net/ethernet/mellanox/mlx4/port.c b/drivers/net/ethernet/mellanox/mlx4/port.c
index ece328166e94..2705b9ab9463 100644
--- a/drivers/net/ethernet/mellanox/mlx4/port.c
+++ b/drivers/net/ethernet/mellanox/mlx4/port.c
@@ -507,30 +507,82 @@ int mlx4_get_port_ib_caps(struct mlx4_dev *dev, u8 port, __be32 *caps)
 }
 static struct mlx4_roce_gid_entry zgid_entry;
 
-int mlx4_get_slave_num_gids(struct mlx4_dev *dev, int slave)
+int mlx4_get_slave_num_gids(struct mlx4_dev *dev, int slave, int port)
 {
+	int vfs;
+	int slave_gid = slave;
+	unsigned i;
+	struct mlx4_slaves_pport slaves_pport;
+	struct mlx4_active_ports actv_ports;
+	unsigned max_port_p_one;
+
 	if (slave == 0)
 		return MLX4_ROCE_PF_GIDS;
-	if (slave <= ((MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) % dev->num_vfs))
-		return ((MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) / dev->num_vfs) + 1;
-	return (MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) / dev->num_vfs;
+
+	/* Slave is a VF */
+	slaves_pport = mlx4_phys_to_slaves_pport(dev, port);
+	actv_ports = mlx4_get_active_ports(dev, slave);
+	max_port_p_one = find_first_bit(actv_ports.ports, dev->caps.num_ports) +
+		bitmap_weight(actv_ports.ports, dev->caps.num_ports) + 1;
+
+	for (i = 1; i < max_port_p_one; i++) {
+		struct mlx4_active_ports exclusive_ports;
+		struct mlx4_slaves_pport slaves_pport_actv;
+		bitmap_zero(exclusive_ports.ports, dev->caps.num_ports);
+		set_bit(i - 1, exclusive_ports.ports);
+		if (i == port)
+			continue;
+		slaves_pport_actv = mlx4_phys_to_slaves_pport_actv(
+				    dev, &exclusive_ports);
+		slave_gid -= bitmap_weight(slaves_pport_actv.slaves,
+					   dev->num_vfs + 1);
+	}
+	vfs = bitmap_weight(slaves_pport.slaves, dev->num_vfs + 1) - 1;
+	if (slave_gid <= ((MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) % vfs))
+		return ((MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) / vfs) + 1;
+	return (MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS) / vfs;
 }
 
-int mlx4_get_base_gid_ix(struct mlx4_dev *dev, int slave)
+int mlx4_get_base_gid_ix(struct mlx4_dev *dev, int slave, int port)
 {
 	int gids;
+	unsigned i;
+	int slave_gid = slave;
 	int vfs;
 
-	gids = MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS;
-	vfs = dev->num_vfs;
+	struct mlx4_slaves_pport slaves_pport;
+	struct mlx4_active_ports actv_ports;
+	unsigned max_port_p_one;
 
 	if (slave == 0)
 		return 0;
-	if (slave <= gids % vfs)
-		return MLX4_ROCE_PF_GIDS + ((gids / vfs) + 1) * (slave - 1);
 
-	return MLX4_ROCE_PF_GIDS + (gids % vfs) + ((gids / vfs) * (slave - 1));
+	slaves_pport = mlx4_phys_to_slaves_pport(dev, port);
+	actv_ports = mlx4_get_active_ports(dev, slave);
+	max_port_p_one = find_first_bit(actv_ports.ports, dev->caps.num_ports) +
+		bitmap_weight(actv_ports.ports, dev->caps.num_ports) + 1;
+
+	for (i = 1; i < max_port_p_one; i++) {
+		struct mlx4_active_ports exclusive_ports;
+		struct mlx4_slaves_pport slaves_pport_actv;
+		bitmap_zero(exclusive_ports.ports, dev->caps.num_ports);
+		set_bit(i - 1, exclusive_ports.ports);
+		if (i == port)
+			continue;
+		slaves_pport_actv = mlx4_phys_to_slaves_pport_actv(
+				    dev, &exclusive_ports);
+		slave_gid -= bitmap_weight(slaves_pport_actv.slaves,
+					   dev->num_vfs + 1);
+	}
+	gids = MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS;
+	vfs = bitmap_weight(slaves_pport.slaves, dev->num_vfs + 1) - 1;
+	if (slave_gid <= gids % vfs)
+		return MLX4_ROCE_PF_GIDS + ((gids / vfs) + 1) * (slave_gid - 1);
+
+	return MLX4_ROCE_PF_GIDS + (gids % vfs) +
+		((gids / vfs) * (slave_gid - 1));
 }
+EXPORT_SYMBOL_GPL(mlx4_get_base_gid_ix);
 
 static int mlx4_common_set_port(struct mlx4_dev *dev, int slave, u32 in_mod,
 				u8 op_mod, struct mlx4_cmd_mailbox *inbox)
@@ -617,8 +669,8 @@ static int mlx4_common_set_port(struct mlx4_dev *dev, int slave, u32 in_mod,
 			 * need a FOR-loop here over number of gids the guest has.
 			 * 1. Check no duplicates in gids passed by slave
 			 */
-			num_gids = mlx4_get_slave_num_gids(dev, slave);
-			base = mlx4_get_base_gid_ix(dev, slave);
+			num_gids = mlx4_get_slave_num_gids(dev, slave, port);
+			base = mlx4_get_base_gid_ix(dev, slave, port);
 			gid_entry_mbox = (struct mlx4_roce_gid_entry *)(inbox->buf);
 			for (i = 0; i < num_gids; gid_entry_mbox++, i++) {
 				if (!memcmp(gid_entry_mbox->raw, zgid_entry.raw,
@@ -738,6 +790,15 @@ int mlx4_SET_PORT_wrapper(struct mlx4_dev *dev, int slave,
 			  struct mlx4_cmd_mailbox *outbox,
 			  struct mlx4_cmd_info *cmd)
 {
+	int port = mlx4_slave_convert_port(
+			dev, slave, vhcr->in_modifier & 0xFF);
+
+	if (port < 0)
+		return -EINVAL;
+
+	vhcr->in_modifier = (vhcr->in_modifier & ~0xFF) |
+			    (port & 0xFF);
+
 	return mlx4_common_set_port(dev, slave, vhcr->in_modifier,
 				    vhcr->op_modifier, inbox);
 }
@@ -1026,10 +1087,16 @@ int mlx4_get_slave_from_roce_gid(struct mlx4_dev *dev, int port, u8 *gid,
 	struct mlx4_priv *priv = mlx4_priv(dev);
 	int i, found_ix = -1;
 	int vf_gids = MLX4_ROCE_MAX_GIDS - MLX4_ROCE_PF_GIDS;
+	struct mlx4_slaves_pport slaves_pport;
+	unsigned num_vfs;
+	int slave_gid;
 
 	if (!mlx4_is_mfunc(dev))
 		return -EINVAL;
 
+	slaves_pport = mlx4_phys_to_slaves_pport(dev, port);
+	num_vfs = bitmap_weight(slaves_pport.slaves, dev->num_vfs + 1) - 1;
+
 	for (i = 0; i < MLX4_ROCE_MAX_GIDS; i++) {
 		if (!memcmp(priv->roce_gids[port - 1][i].raw, gid, 16)) {
 			found_ix = i;
@@ -1039,16 +1106,67 @@ int mlx4_get_slave_from_roce_gid(struct mlx4_dev *dev, int port, u8 *gid,
 
 	if (found_ix >= 0) {
 		if (found_ix < MLX4_ROCE_PF_GIDS)
-			*slave_id = 0;
-		else if (found_ix < MLX4_ROCE_PF_GIDS + (vf_gids % dev->num_vfs) *
-			 (vf_gids / dev->num_vfs + 1))
-			*slave_id = ((found_ix - MLX4_ROCE_PF_GIDS) /
-				     (vf_gids / dev->num_vfs + 1)) + 1;
+			slave_gid = 0;
+		else if (found_ix < MLX4_ROCE_PF_GIDS + (vf_gids % num_vfs) *
+			 (vf_gids / num_vfs + 1))
+			slave_gid = ((found_ix - MLX4_ROCE_PF_GIDS) /
+				     (vf_gids / num_vfs + 1)) + 1;
 		else
-			*slave_id =
+			slave_gid =
 			((found_ix - MLX4_ROCE_PF_GIDS -
-			  ((vf_gids % dev->num_vfs) * ((vf_gids / dev->num_vfs + 1)))) /
-			 (vf_gids / dev->num_vfs)) + vf_gids % dev->num_vfs + 1;
+			  ((vf_gids % num_vfs) * ((vf_gids / num_vfs + 1)))) /
+			 (vf_gids / num_vfs)) + vf_gids % num_vfs + 1;
+
+		if (slave_gid) {
+			struct mlx4_active_ports exclusive_ports;
+			struct mlx4_active_ports actv_ports;
+			struct mlx4_slaves_pport slaves_pport_actv;
+			unsigned max_port_p_one;
+			int num_slaves_before = 1;
+
+			for (i = 1; i < port; i++) {
+				bitmap_zero(exclusive_ports.ports, dev->caps.num_ports);
+				set_bit(i, exclusive_ports.ports);
+				slaves_pport_actv =
+					mlx4_phys_to_slaves_pport_actv(
+							dev, &exclusive_ports);
+				num_slaves_before += bitmap_weight(
+						slaves_pport_actv.slaves,
+						dev->num_vfs + 1);
+			}
+
+			if (slave_gid < num_slaves_before) {
+				bitmap_zero(exclusive_ports.ports, dev->caps.num_ports);
+				set_bit(port - 1, exclusive_ports.ports);
+				slaves_pport_actv =
+					mlx4_phys_to_slaves_pport_actv(
+							dev, &exclusive_ports);
+				slave_gid += bitmap_weight(
+						slaves_pport_actv.slaves,
+						dev->num_vfs + 1) -
+						num_slaves_before;
+			}
+			actv_ports = mlx4_get_active_ports(dev, slave_gid);
+			max_port_p_one = find_first_bit(
+				actv_ports.ports, dev->caps.num_ports) +
+				bitmap_weight(actv_ports.ports,
+					      dev->caps.num_ports) + 1;
+
+			for (i = 1; i < max_port_p_one; i++) {
+				if (i == port)
+					continue;
+				bitmap_zero(exclusive_ports.ports,
+					    dev->caps.num_ports);
+				set_bit(i - 1, exclusive_ports.ports);
+				slaves_pport_actv =
+					mlx4_phys_to_slaves_pport_actv(
+						dev, &exclusive_ports);
+				slave_gid += bitmap_weight(
+						slaves_pport_actv.slaves,
+						dev->num_vfs + 1);
+			}
+		}
+		*slave_id = slave_gid;
 	}
 
 	return (found_ix >= 0) ? 0 : -EINVAL;
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index 74e490d70184..2a33513a0e31 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -468,6 +468,8 @@ int mlx4_init_resource_tracker(struct mlx4_dev *dev)
 
 		spin_lock_init(&res_alloc->alloc_lock);
 		for (t = 0; t < dev->num_vfs + 1; t++) {
+			struct mlx4_active_ports actv_ports =
+				mlx4_get_active_ports(dev, t);
 			switch (i) {
 			case RES_QP:
 				initialize_res_quotas(dev, res_alloc, RES_QP,
@@ -497,10 +499,27 @@ int mlx4_init_resource_tracker(struct mlx4_dev *dev)
 				break;
 			case RES_MAC:
 				if (t == mlx4_master_func_num(dev)) {
-					res_alloc->quota[t] = MLX4_MAX_MAC_NUM;
+					int max_vfs_pport = 0;
+					/* Calculate the max vfs per port for */
+					/* both ports.			      */
+					for (j = 0; j < dev->caps.num_ports;
+					     j++) {
+						struct mlx4_slaves_pport slaves_pport =
+							mlx4_phys_to_slaves_pport(dev, j + 1);
+						unsigned current_slaves =
+							bitmap_weight(slaves_pport.slaves,
+								      dev->caps.num_ports) - 1;
+						if (max_vfs_pport < current_slaves)
+							max_vfs_pport =
+								current_slaves;
+					}
+					res_alloc->quota[t] =
+						MLX4_MAX_MAC_NUM -
+						2 * max_vfs_pport;
 					res_alloc->guaranteed[t] = 2;
 					for (j = 0; j < MLX4_MAX_PORTS; j++)
-						res_alloc->res_port_free[j] = MLX4_MAX_MAC_NUM;
+						res_alloc->res_port_free[j] =
+							MLX4_MAX_MAC_NUM;
 				} else {
 					res_alloc->quota[t] = MLX4_MAX_MAC_NUM;
 					res_alloc->guaranteed[t] = 2;
@@ -528,9 +547,10 @@ int mlx4_init_resource_tracker(struct mlx4_dev *dev)
 				break;
 			}
 			if (i == RES_MAC || i == RES_VLAN) {
-				for (j = 0; j < MLX4_MAX_PORTS; j++)
-					res_alloc->res_port_rsvd[j] +=
-						res_alloc->guaranteed[t];
+				for (j = 0; j < dev->caps.num_ports; j++)
+					if (test_bit(j, actv_ports.ports))
+						res_alloc->res_port_rsvd[j] +=
+							res_alloc->guaranteed[t];
 			} else {
 				res_alloc->res_reserved += res_alloc->guaranteed[t];
 			}
@@ -612,7 +632,8 @@ static void update_gid(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *inbox,
 	if (MLX4_QP_ST_UD == ts) {
 		port = (qp_ctx->pri_path.sched_queue >> 6 & 1) + 1;
 		if (mlx4_is_eth(dev, port))
-			qp_ctx->pri_path.mgid_index = mlx4_get_base_gid_ix(dev, slave) | 0x80;
+			qp_ctx->pri_path.mgid_index =
+				mlx4_get_base_gid_ix(dev, slave, port) | 0x80;
 		else
 			qp_ctx->pri_path.mgid_index = slave | 0x80;
 
@@ -620,7 +641,8 @@ static void update_gid(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *inbox,
 		if (optpar & MLX4_QP_OPTPAR_PRIMARY_ADDR_PATH) {
 			port = (qp_ctx->pri_path.sched_queue >> 6 & 1) + 1;
 			if (mlx4_is_eth(dev, port)) {
-				qp_ctx->pri_path.mgid_index += mlx4_get_base_gid_ix(dev, slave);
+				qp_ctx->pri_path.mgid_index +=
+					mlx4_get_base_gid_ix(dev, slave, port);
 				qp_ctx->pri_path.mgid_index &= 0x7f;
 			} else {
 				qp_ctx->pri_path.mgid_index = slave & 0x7F;
@@ -629,7 +651,8 @@ static void update_gid(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *inbox,
 		if (optpar & MLX4_QP_OPTPAR_ALT_ADDR_PATH) {
 			port = (qp_ctx->alt_path.sched_queue >> 6 & 1) + 1;
 			if (mlx4_is_eth(dev, port)) {
-				qp_ctx->alt_path.mgid_index += mlx4_get_base_gid_ix(dev, slave);
+				qp_ctx->alt_path.mgid_index +=
+					mlx4_get_base_gid_ix(dev, slave, port);
 				qp_ctx->alt_path.mgid_index &= 0x7f;
 			} else {
 				qp_ctx->alt_path.mgid_index = slave & 0x7F;
@@ -1780,6 +1803,11 @@ static int mac_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
 		return err;
 
 	port = !in_port ? get_param_l(out_param) : in_port;
+	port = mlx4_slave_convert_port(
+			dev, slave, port);
+
+	if (port < 0)
+		return -EINVAL;
 	mac = in_param;
 
 	err = __mlx4_register_mac(dev, port, mac);
@@ -1887,6 +1915,11 @@ static int vlan_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
 	if (!port || op != RES_OP_RESERVE_AND_MAP)
 		return -EINVAL;
 
+	port = mlx4_slave_convert_port(
+			dev, slave, port);
+
+	if (port < 0)
+		return -EINVAL;
 	/* upstream kernels had NOP for reg/unreg vlan. Continue this. */
 	if (!in_port && port > 0 && port <= dev->caps.num_ports) {
 		slave_state[slave].old_vlan_api = true;
@@ -2184,6 +2217,11 @@ static int mac_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
 	switch (op) {
 	case RES_OP_RESERVE_AND_MAP:
 		port = !in_port ? get_param_l(out_param) : in_port;
+		port = mlx4_slave_convert_port(
+				dev, slave, port);
+
+		if (port < 0)
+			return -EINVAL;
 		mac_del_from_slave(dev, slave, in_param, port);
 		__mlx4_unregister_mac(dev, port, in_param);
 		break;
@@ -2203,6 +2241,11 @@ static int vlan_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
 	struct mlx4_slave_state *slave_state = priv->mfunc.master.slave_state;
 	int err = 0;
 
+	port = mlx4_slave_convert_port(
+			dev, slave, port);
+
+	if (port < 0)
+		return -EINVAL;
 	switch (op) {
 	case RES_OP_RESERVE_AND_MAP:
 		if (slave_state[slave].old_vlan_api)
@@ -2811,7 +2854,7 @@ static int verify_qp_parameters(struct mlx4_dev *dev,
 				if (optpar & MLX4_QP_OPTPAR_PRIMARY_ADDR_PATH) {
 					port = (qp_ctx->pri_path.sched_queue >> 6 & 1) + 1;
 					if (dev->caps.port_mask[port] != MLX4_PORT_TYPE_IB)
-						num_gids = mlx4_get_slave_num_gids(dev, slave);
+						num_gids = mlx4_get_slave_num_gids(dev, slave, port);
 					else
 						num_gids = 1;
 					if (qp_ctx->pri_path.mgid_index >= num_gids)
@@ -2820,7 +2863,7 @@ static int verify_qp_parameters(struct mlx4_dev *dev,
 				if (optpar & MLX4_QP_OPTPAR_ALT_ADDR_PATH) {
 					port = (qp_ctx->alt_path.sched_queue >> 6 & 1) + 1;
 					if (dev->caps.port_mask[port] != MLX4_PORT_TYPE_IB)
-						num_gids = mlx4_get_slave_num_gids(dev, slave);
+						num_gids = mlx4_get_slave_num_gids(dev, slave, port);
 					else
 						num_gids = 1;
 					if (qp_ctx->alt_path.mgid_index >= num_gids)
@@ -3338,6 +3381,39 @@ int mlx4_INIT2INIT_QP_wrapper(struct mlx4_dev *dev, int slave,
 	return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
 }
 
+static int adjust_qp_sched_queue(struct mlx4_dev *dev, int slave,
+				  struct mlx4_qp_context *qpc,
+				  struct mlx4_cmd_mailbox *inbox)
+{
+	enum mlx4_qp_optpar optpar = be32_to_cpu(*(__be32 *)inbox->buf);
+	u8 pri_sched_queue;
+	int port = mlx4_slave_convert_port(
+		   dev, slave, (qpc->pri_path.sched_queue >> 6 & 1) + 1) - 1;
+
+	if (port < 0)
+		return -EINVAL;
+
+	pri_sched_queue = (qpc->pri_path.sched_queue & ~(1 << 6)) |
+			  ((port & 1) << 6);
+
+	if (optpar & MLX4_QP_OPTPAR_PRIMARY_ADDR_PATH ||
+	    mlx4_is_eth(dev, port + 1)) {
+		qpc->pri_path.sched_queue = pri_sched_queue;
+	}
+
+	if (optpar & MLX4_QP_OPTPAR_ALT_ADDR_PATH) {
+		port = mlx4_slave_convert_port(
+				dev, slave, (qpc->alt_path.sched_queue >> 6 & 1)
+				+ 1) - 1;
+		if (port < 0)
+			return -EINVAL;
+		qpc->alt_path.sched_queue =
+			(qpc->alt_path.sched_queue & ~(1 << 6)) |
+			(port & 1) << 6;
+	}
+	return 0;
+}
+
 static int roce_verify_mac(struct mlx4_dev *dev, int slave,
 				struct mlx4_qp_context *qpc,
 				struct mlx4_cmd_mailbox *inbox)
@@ -3375,6 +3451,9 @@ int mlx4_INIT2RTR_QP_wrapper(struct mlx4_dev *dev, int slave,
 	u8 orig_vlan_index = qpc->pri_path.vlan_index;
 	u8 orig_feup = qpc->pri_path.feup;
 
+	err = adjust_qp_sched_queue(dev, slave, qpc, inbox);
+	if (err)
+		return err;
 	err = verify_qp_parameters(dev, inbox, QP_TRANS_INIT2RTR, slave);
 	if (err)
 		return err;
@@ -3426,6 +3505,9 @@ int mlx4_RTR2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
 	int err;
 	struct mlx4_qp_context *context = inbox->buf + 8;
 
+	err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
 	err = verify_qp_parameters(dev, inbox, QP_TRANS_RTR2RTS, slave);
 	if (err)
 		return err;
@@ -3445,6 +3527,9 @@ int mlx4_RTS2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
 	int err;
 	struct mlx4_qp_context *context = inbox->buf + 8;
 
+	err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
 	err = verify_qp_parameters(dev, inbox, QP_TRANS_RTS2RTS, slave);
 	if (err)
 		return err;
@@ -3463,6 +3548,9 @@ int mlx4_SQERR2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
 			      struct mlx4_cmd_info *cmd)
 {
 	struct mlx4_qp_context *context = inbox->buf + 8;
+	int err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
 	adjust_proxy_tun_qkey(dev, vhcr, context);
 	return mlx4_GEN_QP_wrapper(dev, slave, vhcr, inbox, outbox, cmd);
 }
@@ -3476,6 +3564,9 @@ int mlx4_SQD2SQD_QP_wrapper(struct mlx4_dev *dev, int slave,
 	int err;
 	struct mlx4_qp_context *context = inbox->buf + 8;
 
+	err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
 	err = verify_qp_parameters(dev, inbox, QP_TRANS_SQD2SQD, slave);
 	if (err)
 		return err;
@@ -3495,6 +3586,9 @@ int mlx4_SQD2RTS_QP_wrapper(struct mlx4_dev *dev, int slave,
 	int err;
 	struct mlx4_qp_context *context = inbox->buf + 8;
 
+	err = adjust_qp_sched_queue(dev, slave, context, inbox);
+	if (err)
+		return err;
 	err = verify_qp_parameters(dev, inbox, QP_TRANS_SQD2RTS, slave);
 	if (err)
 		return err;
@@ -3598,16 +3692,26 @@ static int rem_mcg_res(struct mlx4_dev *dev, int slave, struct res_qp *rqp,
 	return err;
 }
 
-static int qp_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
-		     int block_loopback, enum mlx4_protocol prot,
+static int qp_attach(struct mlx4_dev *dev, int slave, struct mlx4_qp *qp,
+		     u8 gid[16], int block_loopback, enum mlx4_protocol prot,
 		     enum mlx4_steer_type type, u64 *reg_id)
 {
 	switch (dev->caps.steering_mode) {
-	case MLX4_STEERING_MODE_DEVICE_MANAGED:
-		return mlx4_trans_to_dmfs_attach(dev, qp, gid, gid[5],
+	case MLX4_STEERING_MODE_DEVICE_MANAGED: {
+		int port = mlx4_slave_convert_port(dev, slave, gid[5]);
+		if (port < 0)
+			return port;
+		return mlx4_trans_to_dmfs_attach(dev, qp, gid, port,
 						block_loopback, prot,
 						reg_id);
+	}
 	case MLX4_STEERING_MODE_B0:
+		if (prot == MLX4_PROT_ETH) {
+			int port = mlx4_slave_convert_port(dev, slave, gid[5]);
+			if (port < 0)
+				return port;
+			gid[5] = port;
+		}
 		return mlx4_qp_attach_common(dev, qp, gid,
 					    block_loopback, prot, type);
 	default:
@@ -3615,9 +3719,9 @@ static int qp_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
 	}
 }
 
-static int qp_detach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16],
-		     enum mlx4_protocol prot, enum mlx4_steer_type type,
-		     u64 reg_id)
+static int qp_detach(struct mlx4_dev *dev, struct mlx4_qp *qp,
+		     u8 gid[16], enum mlx4_protocol prot,
+		     enum mlx4_steer_type type, u64 reg_id)
 {
 	switch (dev->caps.steering_mode) {
 	case MLX4_STEERING_MODE_DEVICE_MANAGED:
@@ -3654,7 +3758,7 @@ int mlx4_QP_ATTACH_wrapper(struct mlx4_dev *dev, int slave,
 
 	qp.qpn = qpn;
 	if (attach) {
-		err = qp_attach(dev, &qp, gid, block_loopback, prot,
+		err = qp_attach(dev, slave, &qp, gid, block_loopback, prot,
 				type, &reg_id);
 		if (err) {
 			pr_err("Fail to attach rule to qp 0x%x\n", qpn);
@@ -3790,6 +3894,9 @@ int mlx4_QP_FLOW_STEERING_ATTACH_wrapper(struct mlx4_dev *dev, int slave,
 		return -EOPNOTSUPP;
 
 	ctrl = (struct mlx4_net_trans_rule_hw_ctrl *)inbox->buf;
+	ctrl->port = mlx4_slave_convert_port(dev, slave, ctrl->port);
+	if (ctrl->port <= 0)
+		return -EINVAL;
 	qpn = be32_to_cpu(ctrl->qpn) & 0xffffff;
 	err = get_res(dev, slave, qpn, RES_QP, &rqp);
 	if (err) {
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 122c7cabaee7..6b3998396b99 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -1231,4 +1231,5 @@ struct mlx4_slaves_pport mlx4_phys_to_slaves_pport_actv(
 /* Returns the slave's virtual port that represents the physical port. */
 int mlx4_phys_to_slave_port(struct mlx4_dev *dev, int slave, int port);
 
+int mlx4_get_base_gid_ix(struct mlx4_dev *dev, int slave, int port);
 #endif /* MLX4_DEVICE_H */