mlx4: Add ref counting to port MAC table for RoCE

The IB side of RoCE requires the MAC table index of the
MAC address used by its QPs.

To obtain the real MAC index, the IB side registers the
MAC (increasing its ref count, and also returning the
real MAC index) during the modify-qp sequence.

This protects against the ETH side deleting or modifying
that MAC table entry while the QP is active.

Note that until the modify-qp command returns success,
the MAC and VLAN information only has "candidate" status.
If the modify-qp succeeds, the "candidate" info is promoted
to the operational MAC/VLAN info for the qp. If the modify fails,
the candidate MAC/VLAN is unregistered, and the old qp info
is preserved.

The patch is a bit complex, because there are multiple qp
transitions where the primary-path information may be
modified:  INIT-to-RTR, and SQD-to-SQD.

Similarly for the alternate path information.

Therefore the code must handle cases where path information
has already been entered into the QP context by previous
qp transitions.

For the MAC address, the success logic is as follows:
1. If there was no previous MAC, simply move the candidate
   MAC information to the operational information, and reset
   the candidate MAC info.
2. If there was a previous MAC, unregister it.  Then move
   the MAC information from candidate to operational, and
   reset the candidate info (as in 1. above).

The MAC address failure logic is the same for all cases:
 - Unregister the candidate MAC, and reset the candidate MAC info.

For Vlan registration, the logic is similar.

Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

2 files changed, 261 insertions, 43 deletions

diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index a230683af94..febc8f9bc59 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 {
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index c6ef2e7e304..11332f07402 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;
 }