IB/mlx4: Create mirror flows in port aggregation mode
authorMoni Shoua <monis@mellanox.com>
Tue, 3 Feb 2015 14:48:38 +0000 (16:48 +0200)
committerDavid S. Miller <davem@davemloft.net>
Thu, 5 Feb 2015 00:14:25 +0000 (16:14 -0800)
In port aggregation mode flows for port #1 (the only port) should be mirrored
on port #2. This is because packets can arrive from either physical ports.

Signed-off-by: Moni Shoua <monis@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

index ed21ae68a9779fd3ebc79f52484d073586342627..ca522382dedc5b3c34898f193542a620e046a6f7 100644 (file)
@@ -851,7 +851,7 @@ int mlx4_ib_add_mc(struct mlx4_ib_dev *mdev, struct mlx4_ib_qp *mqp,
 
 struct mlx4_ib_steering {
        struct list_head list;
-       u64 reg_id;
+       struct mlx4_flow_reg_id reg_id;
        union ib_gid gid;
 };
 
@@ -1142,9 +1142,11 @@ static struct ib_flow *mlx4_ib_create_flow(struct ib_qp *qp,
                                    struct ib_flow_attr *flow_attr,
                                    int domain)
 {
-       int err = 0, i = 0;
+       int err = 0, i = 0, j = 0;
        struct mlx4_ib_flow *mflow;
        enum mlx4_net_trans_promisc_mode type[2];
+       struct mlx4_dev *dev = (to_mdev(qp->device))->dev;
+       int is_bonded = mlx4_is_bonded(dev);
 
        memset(type, 0, sizeof(type));
 
@@ -1179,26 +1181,55 @@ static struct ib_flow *mlx4_ib_create_flow(struct ib_qp *qp,
 
        while (i < ARRAY_SIZE(type) && type[i]) {
                err = __mlx4_ib_create_flow(qp, flow_attr, domain, type[i],
-                                           &mflow->reg_id[i]);
+                                           &mflow->reg_id[i].id);
                if (err)
                        goto err_create_flow;
                i++;
+               if (is_bonded) {
+                       flow_attr->port = 2;
+                       err = __mlx4_ib_create_flow(qp, flow_attr,
+                                                   domain, type[j],
+                                                   &mflow->reg_id[j].mirror);
+                       flow_attr->port = 1;
+                       if (err)
+                               goto err_create_flow;
+                       j++;
+               }
+
        }
 
        if (i < ARRAY_SIZE(type) && flow_attr->type == IB_FLOW_ATTR_NORMAL) {
-               err = mlx4_ib_tunnel_steer_add(qp, flow_attr, &mflow->reg_id[i]);
+               err = mlx4_ib_tunnel_steer_add(qp, flow_attr,
+                                              &mflow->reg_id[i].id);
                if (err)
                        goto err_create_flow;
                i++;
+               if (is_bonded) {
+                       flow_attr->port = 2;
+                       err = mlx4_ib_tunnel_steer_add(qp, flow_attr,
+                                                      &mflow->reg_id[j].mirror);
+                       flow_attr->port = 1;
+                       if (err)
+                               goto err_create_flow;
+                       j++;
+               }
+               /* function to create mirror rule */
        }
 
        return &mflow->ibflow;
 
 err_create_flow:
        while (i) {
-               (void)__mlx4_ib_destroy_flow(to_mdev(qp->device)->dev, mflow->reg_id[i]);
+               (void)__mlx4_ib_destroy_flow(to_mdev(qp->device)->dev,
+                                            mflow->reg_id[i].id);
                i--;
        }
+
+       while (j) {
+               (void)__mlx4_ib_destroy_flow(to_mdev(qp->device)->dev,
+                                            mflow->reg_id[j].mirror);
+               j--;
+       }
 err_free:
        kfree(mflow);
        return ERR_PTR(err);
@@ -1211,10 +1242,16 @@ static int mlx4_ib_destroy_flow(struct ib_flow *flow_id)
        struct mlx4_ib_dev *mdev = to_mdev(flow_id->qp->device);
        struct mlx4_ib_flow *mflow = to_mflow(flow_id);
 
-       while (i < ARRAY_SIZE(mflow->reg_id) && mflow->reg_id[i]) {
-               err = __mlx4_ib_destroy_flow(mdev->dev, mflow->reg_id[i]);
+       while (i < ARRAY_SIZE(mflow->reg_id) && mflow->reg_id[i].id) {
+               err = __mlx4_ib_destroy_flow(mdev->dev, mflow->reg_id[i].id);
                if (err)
                        ret = err;
+               if (mflow->reg_id[i].mirror) {
+                       err = __mlx4_ib_destroy_flow(mdev->dev,
+                                                    mflow->reg_id[i].mirror);
+                       if (err)
+                               ret = err;
+               }
                i++;
        }
 
@@ -1226,11 +1263,12 @@ static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
 {
        int err;
        struct mlx4_ib_dev *mdev = to_mdev(ibqp->device);
+       struct mlx4_dev *dev = mdev->dev;
        struct mlx4_ib_qp *mqp = to_mqp(ibqp);
-       u64 reg_id;
        struct mlx4_ib_steering *ib_steering = NULL;
        enum mlx4_protocol prot = (gid->raw[1] == 0x0e) ?
                MLX4_PROT_IB_IPV4 : MLX4_PROT_IB_IPV6;
+       struct mlx4_flow_reg_id reg_id;
 
        if (mdev->dev->caps.steering_mode ==
            MLX4_STEERING_MODE_DEVICE_MANAGED) {
@@ -1242,10 +1280,20 @@ static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
        err = mlx4_multicast_attach(mdev->dev, &mqp->mqp, gid->raw, mqp->port,
                                    !!(mqp->flags &
                                       MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK),
-                                   prot, &reg_id);
+                                   prot, &reg_id.id);
        if (err)
                goto err_malloc;
 
+       reg_id.mirror = 0;
+       if (mlx4_is_bonded(dev)) {
+               err = mlx4_multicast_attach(mdev->dev, &mqp->mqp, gid->raw, 2,
+                                           !!(mqp->flags &
+                                           MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK),
+                                           prot, &reg_id.mirror);
+               if (err)
+                       goto err_add;
+       }
+
        err = add_gid_entry(ibqp, gid);
        if (err)
                goto err_add;
@@ -1261,7 +1309,10 @@ static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
 
 err_add:
        mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw,
-                             prot, reg_id);
+                             prot, reg_id.id);
+       if (reg_id.mirror)
+               mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw,
+                                     prot, reg_id.mirror);
 err_malloc:
        kfree(ib_steering);
 
@@ -1288,10 +1339,12 @@ static int mlx4_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
 {
        int err;
        struct mlx4_ib_dev *mdev = to_mdev(ibqp->device);
+       struct mlx4_dev *dev = mdev->dev;
        struct mlx4_ib_qp *mqp = to_mqp(ibqp);
        struct net_device *ndev;
        struct mlx4_ib_gid_entry *ge;
-       u64 reg_id = 0;
+       struct mlx4_flow_reg_id reg_id = {0, 0};
+
        enum mlx4_protocol prot = (gid->raw[1] == 0x0e) ?
                MLX4_PROT_IB_IPV4 : MLX4_PROT_IB_IPV6;
 
@@ -1316,10 +1369,17 @@ static int mlx4_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
        }
 
        err = mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw,
-                                   prot, reg_id);
+                                   prot, reg_id.id);
        if (err)
                return err;
 
+       if (mlx4_is_bonded(dev)) {
+               err = mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw,
+                                           prot, reg_id.mirror);
+               if (err)
+                       return err;
+       }
+
        mutex_lock(&mqp->mutex);
        ge = find_gid_entry(mqp, gid->raw);
        if (ge) {
index 6eb743f65f6f5633eb126ae5fe7ac768299fe77d..2b49f9de255639922c14a29f8c1996ceb3ec603b 100644 (file)
@@ -134,10 +134,17 @@ struct mlx4_ib_fmr {
        struct mlx4_fmr         mfmr;
 };
 
+#define MAX_REGS_PER_FLOW 2
+
+struct mlx4_flow_reg_id {
+       u64 id;
+       u64 mirror;
+};
+
 struct mlx4_ib_flow {
        struct ib_flow ibflow;
        /* translating DMFS verbs sniffer rule to FW API requires two reg IDs */
-       u64 reg_id[2];
+       struct mlx4_flow_reg_id reg_id[MAX_REGS_PER_FLOW];
 };
 
 struct mlx4_ib_wq {