net/mlx5: E-Switch, Optimize HW steering tables in switchdev mode

Under switchdev mode we insert an eswitch miss rule causing any
unmatched traffic to be sent towards the PF vport. This miss rule can
be optimized if we break it to two, one case is for multicast traffic and
the other for unicast.

Breaking the miss rule into two (unicast and multicast) allows the firmware
to program the hardware in a more efficient way.

Using ConncetX-5 Ex with IXIA and testpmd (which use IB representors):

IXIA -> NIC -> PF -> IB representor -> NIC -> VF:
    - Without this optimization: 9.2 MPPS.
    - With this optimization: 18 MPPS.

VF -> NIC -> IB representor-> PF -> NIC -> IXIA:
    - Without this optimization: 17 MPPS.
    - With this optimization: 23.4 MPPS.

Signed-off-by: Mark Bloch <markb@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 9c1e1a2d02ef0c22a0afce19efa84ed7ae381cc7..98d2177d0806244f2d40acdee26359b74a3832b2 100644 (file)
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -129,7 +129,8 @@ struct mlx5_eswitch_fdb {
                        struct mlx5_flow_table *fdb;
                        struct mlx5_flow_group *send_to_vport_grp;
                        struct mlx5_flow_group *miss_grp;
-                       struct mlx5_flow_handle *miss_rule;
+                       struct mlx5_flow_handle *miss_rule_uni;
+                       struct mlx5_flow_handle *miss_rule_multi;
                        int vlan_push_pop_refcount;
                } offloads;
        };

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index a5f5339a4e88ea9fa24bad7018d49e411edd4c65..0692d280883c6454df75c5641e5a4d4a0f989a39 100644 (file)
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -351,7 +351,11 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
        struct mlx5_flow_destination dest = {};
        struct mlx5_flow_handle *flow_rule = NULL;
        struct mlx5_flow_spec *spec;
+       void *headers_c;
+       void *headers_v;
        int err = 0;
+       u8 *dmac_c;
+       u8 *dmac_v;
 
        spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
        if (!spec) {
@@ -359,6 +363,13 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
                goto out;
        }
 
+       spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+       headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+                                outer_headers);
+       dmac_c = MLX5_ADDR_OF(fte_match_param, headers_c,
+                             outer_headers.dmac_47_16);
+       dmac_c[0] = 0x01;
+
        dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
        dest.vport_num = 0;
        flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
@@ -367,11 +378,28 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
                                        &flow_act, &dest, 1);
        if (IS_ERR(flow_rule)) {
                err = PTR_ERR(flow_rule);
-               esw_warn(esw->dev,  "FDB: Failed to add miss flow rule err %d\n", err);
+               esw_warn(esw->dev,  "FDB: Failed to add unicast miss flow rule err %d\n", err);
                goto out;
        }
 
-       esw->fdb_table.offloads.miss_rule = flow_rule;
+       esw->fdb_table.offloads.miss_rule_uni = flow_rule;
+
+       headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+                                outer_headers);
+       dmac_v = MLX5_ADDR_OF(fte_match_param, headers_v,
+                             outer_headers.dmac_47_16);
+       dmac_v[0] = 0x01;
+       flow_rule = mlx5_add_flow_rules(esw->fdb_table.offloads.fdb, spec,
+                                       &flow_act, &dest, 1);
+       if (IS_ERR(flow_rule)) {
+               err = PTR_ERR(flow_rule);
+               esw_warn(esw->dev, "FDB: Failed to add multicast miss flow rule err %d\n", err);
+               mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_uni);
+               goto out;
+       }
+
+       esw->fdb_table.offloads.miss_rule_multi = flow_rule;
+
 out:
        kvfree(spec);
        return err;
@@ -440,6 +468,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
        struct mlx5_flow_group *g;
        void *match_criteria;
        u32 *flow_group_in;
+       u8 *dmac;
 
        esw_debug(esw->dev, "Create offloads FDB Tables\n");
        flow_group_in = kvzalloc(inlen, GFP_KERNEL);
@@ -457,7 +486,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
        if (err)
                goto fast_fdb_err;
 
-       table_size = nvports * MAX_SQ_NVPORTS + MAX_PF_SQ + 1;
+       table_size = nvports * MAX_SQ_NVPORTS + MAX_PF_SQ + 2;
 
        ft_attr.max_fte = table_size;
        ft_attr.prio = FDB_SLOW_PATH;
@@ -494,10 +523,16 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw, int nvports)
 
        /* create miss group */
        memset(flow_group_in, 0, inlen);
-       MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable, 0);
+       MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+                MLX5_MATCH_OUTER_HEADERS);
+       match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in,
+                                     match_criteria);
+       dmac = MLX5_ADDR_OF(fte_match_param, match_criteria,
+                           outer_headers.dmac_47_16);
+       dmac[0] = 0x01;
 
        MLX5_SET(create_flow_group_in, flow_group_in, start_flow_index, ix);
-       MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, ix + 1);
+       MLX5_SET(create_flow_group_in, flow_group_in, end_flow_index, ix + 2);
 
        g = mlx5_create_flow_group(fdb, flow_group_in);
        if (IS_ERR(g)) {
@@ -533,7 +568,8 @@ static void esw_destroy_offloads_fdb_tables(struct mlx5_eswitch *esw)
                return;
 
        esw_debug(esw->dev, "Destroy offloads FDB Tables\n");
-       mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule);
+       mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_multi);
+       mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_uni);
        mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_grp);
        mlx5_destroy_flow_group(esw->fdb_table.offloads.miss_grp);