IB/mlx4: Invoke UPDATE_QP for proxy QP1 on MAC changes
authorMatan Barak <matanb@mellanox.com>
Thu, 15 May 2014 12:29:28 +0000 (15:29 +0300)
committerDavid S. Miller <davem@davemloft.net>
Fri, 16 May 2014 19:12:45 +0000 (15:12 -0400)
When we receive a netdev event indicating a netdev change and/or
a netdev address change, we must change the MAC index used by the
proxy QP1 (in the QP context), otherwise RoCE CM packets sent by the
VF will not carry the same source MAC address as the non-CM packets.

We use the UPDATE_QP command to perform this change.

In order to avoid modifying a QP context based on netdev event,
while the driver attempts to destroy this QP (e.g either the mlx4_ib
or ib_mad modules are unloaded), we use mutex locking in both flows.

Since the relevant mlx4 proxy GSI QP is created indirectly by the
mad module when they create their GSI QP, the mlx4 didn't need to
keep track on that QP prior to this change.

Now, when QP modifications are needed to this QP from within the
driver, we added refernece to it.

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>
drivers/infiniband/hw/mlx4/main.c
drivers/infiniband/hw/mlx4/mlx4_ib.h
drivers/infiniband/hw/mlx4/qp.c

index 1b6dbe1..199c789 100644 (file)
@@ -48,6 +48,7 @@
 
 #include <linux/mlx4/driver.h>
 #include <linux/mlx4/cmd.h>
+#include <linux/mlx4/qp.h>
 
 #include "mlx4_ib.h"
 #include "user.h"
@@ -1614,6 +1615,53 @@ static int mlx4_ib_inet6_event(struct notifier_block *this, unsigned long event,
 }
 #endif
 
+#define MLX4_IB_INVALID_MAC    ((u64)-1)
+static void mlx4_ib_update_qps(struct mlx4_ib_dev *ibdev,
+                              struct net_device *dev,
+                              int port)
+{
+       u64 new_smac = 0;
+       u64 release_mac = MLX4_IB_INVALID_MAC;
+       struct mlx4_ib_qp *qp;
+
+       read_lock(&dev_base_lock);
+       new_smac = mlx4_mac_to_u64(dev->dev_addr);
+       read_unlock(&dev_base_lock);
+
+       mutex_lock(&ibdev->qp1_proxy_lock[port - 1]);
+       qp = ibdev->qp1_proxy[port - 1];
+       if (qp) {
+               int new_smac_index;
+               u64 old_smac = qp->pri.smac;
+               struct mlx4_update_qp_params update_params;
+
+               if (new_smac == old_smac)
+                       goto unlock;
+
+               new_smac_index = mlx4_register_mac(ibdev->dev, port, new_smac);
+
+               if (new_smac_index < 0)
+                       goto unlock;
+
+               update_params.smac_index = new_smac_index;
+               if (mlx4_update_qp(ibdev->dev, &qp->mqp, MLX4_UPDATE_QP_SMAC,
+                                  &update_params)) {
+                       release_mac = new_smac;
+                       goto unlock;
+               }
+
+               qp->pri.smac = new_smac;
+               qp->pri.smac_index = new_smac_index;
+
+               release_mac = old_smac;
+       }
+
+unlock:
+       mutex_unlock(&ibdev->qp1_proxy_lock[port - 1]);
+       if (release_mac != MLX4_IB_INVALID_MAC)
+               mlx4_unregister_mac(ibdev->dev, port, release_mac);
+}
+
 static void mlx4_ib_get_dev_addr(struct net_device *dev,
                                 struct mlx4_ib_dev *ibdev, u8 port)
 {
@@ -1689,9 +1737,13 @@ static int mlx4_ib_init_gid_table(struct mlx4_ib_dev *ibdev)
        return 0;
 }
 
-static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev)
+static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev,
+                                struct net_device *dev,
+                                unsigned long event)
+
 {
        struct mlx4_ib_iboe *iboe;
+       int update_qps_port = -1;
        int port;
 
        iboe = &ibdev->iboe;
@@ -1719,6 +1771,11 @@ static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev)
                }
                curr_master = iboe->masters[port - 1];
 
+               if (dev == iboe->netdevs[port - 1] &&
+                   (event == NETDEV_CHANGEADDR || event == NETDEV_REGISTER ||
+                    event == NETDEV_UP || event == NETDEV_CHANGE))
+                       update_qps_port = port;
+
                if (curr_netdev) {
                        port_state = (netif_running(curr_netdev) && netif_carrier_ok(curr_netdev)) ?
                                                IB_PORT_ACTIVE : IB_PORT_DOWN;
@@ -1752,6 +1809,9 @@ static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev)
        }
 
        spin_unlock(&iboe->lock);
+
+       if (update_qps_port > 0)
+               mlx4_ib_update_qps(ibdev, dev, update_qps_port);
 }
 
 static int mlx4_ib_netdev_event(struct notifier_block *this,
@@ -1764,7 +1824,7 @@ static int mlx4_ib_netdev_event(struct notifier_block *this,
                return NOTIFY_DONE;
 
        ibdev = container_of(this, struct mlx4_ib_dev, iboe.nb);
-       mlx4_ib_scan_netdevs(ibdev);
+       mlx4_ib_scan_netdevs(ibdev, dev, event);
 
        return NOTIFY_DONE;
 }
@@ -2043,6 +2103,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
                goto err_map;
 
        for (i = 0; i < ibdev->num_ports; ++i) {
+               mutex_init(&ibdev->qp1_proxy_lock[i]);
                if (mlx4_ib_port_link_layer(&ibdev->ib_dev, i + 1) ==
                                                IB_LINK_LAYER_ETHERNET) {
                        err = mlx4_counter_alloc(ibdev->dev, &ibdev->counters[i]);
@@ -2126,7 +2187,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
                for (i = 1 ; i <= ibdev->num_ports ; ++i)
                        reset_gid_table(ibdev, i);
                rtnl_lock();
-               mlx4_ib_scan_netdevs(ibdev);
+               mlx4_ib_scan_netdevs(ibdev, NULL, 0);
                rtnl_unlock();
                mlx4_ib_init_gid_table(ibdev);
        }
index f589522..66b0b7d 100644 (file)
@@ -522,6 +522,9 @@ struct mlx4_ib_dev {
        int steer_qpn_count;
        int steer_qpn_base;
        int steering_support;
+       struct mlx4_ib_qp      *qp1_proxy[MLX4_MAX_PORTS];
+       /* lock when destroying qp1_proxy and getting netdev events */
+       struct mutex            qp1_proxy_lock[MLX4_MAX_PORTS];
 };
 
 struct ib_event_work {
index 41308af..dc57482 100644 (file)
@@ -1132,6 +1132,12 @@ int mlx4_ib_destroy_qp(struct ib_qp *qp)
        if (is_qp0(dev, mqp))
                mlx4_CLOSE_PORT(dev->dev, mqp->port);
 
+       if (dev->qp1_proxy[mqp->port - 1] == mqp) {
+               mutex_lock(&dev->qp1_proxy_lock[mqp->port - 1]);
+               dev->qp1_proxy[mqp->port - 1] = NULL;
+               mutex_unlock(&dev->qp1_proxy_lock[mqp->port - 1]);
+       }
+
        pd = get_pd(mqp);
        destroy_qp_common(dev, mqp, !!pd->ibpd.uobject);
 
@@ -1646,6 +1652,8 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
                                err = handle_eth_ud_smac_index(dev, qp, (u8 *)attr->smac, context);
                                if (err)
                                        return -EINVAL;
+                               if (qp->mlx4_ib_qp_type == MLX4_IB_QPT_PROXY_GSI)
+                                       dev->qp1_proxy[qp->port - 1] = qp;
                        }
                }
        }