net/mlx5e: Match recirculated packet miss in slow table using reg_c1

Previous patch in series that implements stack devices RX path implements
indirect table rules that match on tunnel VNI. After such rule is created
all tunnel traffic is recirculated to root table. However, recirculated
packet might not match on any rules installed in the table (for example,
when IP traffic follows ARP traffic). In that case packets appear on
representor of tunnel endpoint VF instead being redirected to the VF
itself.

Extend slow table with additional flow group that matches on reg_c0 (source
port value set by indirect tables implemented by previous patch in series)
and reg_c1 (special 0xFFF mark). When creating offloads fdb tables, install
one rule per VF vport to match on recirculated miss packets and redirect
them to appropriate VF vport. Modify indirect tables code to also rewrite
reg_c1 with special 0xFFF mark.

Implementation reuses reg_c1 tunnel id bits. This is safe to do because
recirculated packets are always matched before decapsulation.

Signed-off-by: Vlad Buslov <vladbu@nvidia.com>
Signed-off-by: Dmytro Linkin <dlinkin@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>

1 files changed, 117 insertions, 2 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index a44728595420..94cb0217b4f3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -1080,6 +1080,81 @@ void mlx5_eswitch_del_send_to_vport_rule(struct mlx5_flow_handle *rule)
 	mlx5_del_flow_rules(rule);
 }
 
+static void mlx5_eswitch_del_send_to_vport_meta_rules(struct mlx5_eswitch *esw)
+{
+	struct mlx5_flow_handle **flows = esw->fdb_table.offloads.send_to_vport_meta_rules;
+	int i = 0, num_vfs = esw->esw_funcs.num_vfs, vport_num;
+
+	if (!num_vfs || !flows)
+		return;
+
+	mlx5_esw_for_each_vf_vport_num(esw, vport_num, num_vfs)
+		mlx5_del_flow_rules(flows[i++]);
+
+	kvfree(flows);
+}
+
+static int
+mlx5_eswitch_add_send_to_vport_meta_rules(struct mlx5_eswitch *esw)
+{
+	int num_vfs, vport_num, rule_idx = 0, err = 0;
+	struct mlx5_flow_destination dest = {};
+	struct mlx5_flow_act flow_act = {0};
+	struct mlx5_flow_handle *flow_rule;
+	struct mlx5_flow_handle **flows;
+	struct mlx5_flow_spec *spec;
+
+	num_vfs = esw->esw_funcs.num_vfs;
+	flows = kvzalloc(num_vfs * sizeof(*flows), GFP_KERNEL);
+	if (!flows)
+		return -ENOMEM;
+
+	spec = kvzalloc(sizeof(*spec), GFP_KERNEL);
+	if (!spec) {
+		err = -ENOMEM;
+		goto alloc_err;
+	}
+
+	MLX5_SET(fte_match_param, spec->match_criteria,
+		 misc_parameters_2.metadata_reg_c_0, mlx5_eswitch_get_vport_metadata_mask());
+	MLX5_SET(fte_match_param, spec->match_criteria,
+		 misc_parameters_2.metadata_reg_c_1, ESW_TUN_MASK);
+	MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_c_1,
+		 ESW_TUN_SLOW_TABLE_GOTO_VPORT_MARK);
+
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS_2;
+	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
+	flow_act.action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+
+	mlx5_esw_for_each_vf_vport_num(esw, vport_num, num_vfs) {
+		MLX5_SET(fte_match_param, spec->match_value, misc_parameters_2.metadata_reg_c_0,
+			 mlx5_eswitch_get_vport_metadata_for_match(esw, vport_num));
+		dest.vport.num = vport_num;
+
+		flow_rule = mlx5_add_flow_rules(esw->fdb_table.offloads.slow_fdb,
+						spec, &flow_act, &dest, 1);
+		if (IS_ERR(flow_rule)) {
+			err = PTR_ERR(flow_rule);
+			esw_warn(esw->dev, "FDB: Failed to add send to vport meta rule idx %d, err %ld\n",
+				 rule_idx, PTR_ERR(flow_rule));
+			goto rule_err;
+		}
+		flows[rule_idx++] = flow_rule;
+	}
+
+	esw->fdb_table.offloads.send_to_vport_meta_rules = flows;
+	kvfree(spec);
+	return 0;
+
+rule_err:
+	while (--rule_idx >= 0)
+		mlx5_del_flow_rules(flows[rule_idx]);
+	kvfree(spec);
+alloc_err:
+	kvfree(flows);
+	return err;
+}
+
 static bool mlx5_eswitch_reg_c1_loopback_supported(struct mlx5_eswitch *esw)
 {
 	return MLX5_CAP_ESW_FLOWTABLE(esw->dev, fdb_to_vport_reg_c_id) &
@@ -1562,11 +1637,11 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw)
 {
 	int inlen = MLX5_ST_SZ_BYTES(create_flow_group_in);
 	struct mlx5_flow_table_attr ft_attr = {};
+	int num_vfs, table_size, ix, err = 0;
 	struct mlx5_core_dev *dev = esw->dev;
 	struct mlx5_flow_namespace *root_ns;
 	struct mlx5_flow_table *fdb = NULL;
 	u32 flags = 0, *flow_group_in;
-	int table_size, ix, err = 0;
 	struct mlx5_flow_group *g;
 	void *match_criteria;
 	u8 *dmac;
@@ -1592,7 +1667,7 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw)
 	}
 
 	table_size = esw->total_vports * MAX_SQ_NVPORTS + MAX_PF_SQ +
-		MLX5_ESW_MISS_FLOWS + esw->total_vports;
+		MLX5_ESW_MISS_FLOWS + esw->total_vports + esw->esw_funcs.num_vfs;
 
 	/* create the slow path fdb with encap set, so further table instances
 	 * can be created at run time while VFs are probed if the FW allows that.
@@ -1640,6 +1715,38 @@ static int esw_create_offloads_fdb_tables(struct mlx5_eswitch *esw)
 	}
 	esw->fdb_table.offloads.send_to_vport_grp = g;
 
+	/* meta send to vport */
+	memset(flow_group_in, 0, inlen);
+	MLX5_SET(create_flow_group_in, flow_group_in, match_criteria_enable,
+		 MLX5_MATCH_MISC_PARAMETERS_2);
+
+	match_criteria = MLX5_ADDR_OF(create_flow_group_in, flow_group_in, match_criteria);
+
+	MLX5_SET(fte_match_param, match_criteria,
+		 misc_parameters_2.metadata_reg_c_0, mlx5_eswitch_get_vport_metadata_mask());
+	MLX5_SET(fte_match_param, match_criteria,
+		 misc_parameters_2.metadata_reg_c_1, ESW_TUN_MASK);
+
+	num_vfs = esw->esw_funcs.num_vfs;
+	if (num_vfs) {
+		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 + num_vfs - 1);
+		ix += num_vfs;
+
+		g = mlx5_create_flow_group(fdb, flow_group_in);
+		if (IS_ERR(g)) {
+			err = PTR_ERR(g);
+			esw_warn(dev, "Failed to create send-to-vport meta flow group err(%d)\n",
+				 err);
+			goto send_vport_meta_err;
+		}
+		esw->fdb_table.offloads.send_to_vport_meta_grp = g;
+
+		err = mlx5_eswitch_add_send_to_vport_meta_rules(esw);
+		if (err)
+			goto meta_rule_err;
+	}
+
 	if (MLX5_CAP_ESW(esw->dev, merged_eswitch)) {
 		/* create peer esw miss group */
 		memset(flow_group_in, 0, inlen);
@@ -1707,6 +1814,11 @@ miss_err:
 	if (MLX5_CAP_ESW(esw->dev, merged_eswitch))
 		mlx5_destroy_flow_group(esw->fdb_table.offloads.peer_miss_grp);
 peer_miss_err:
+	mlx5_eswitch_del_send_to_vport_meta_rules(esw);
+meta_rule_err:
+	if (esw->fdb_table.offloads.send_to_vport_meta_grp)
+		mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_meta_grp);
+send_vport_meta_err:
 	mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_grp);
 send_vport_err:
 	esw_chains_destroy(esw, esw_chains(esw));
@@ -1728,7 +1840,10 @@ static void esw_destroy_offloads_fdb_tables(struct mlx5_eswitch *esw)
 	esw_debug(esw->dev, "Destroy offloads FDB Tables\n");
 	mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_multi);
 	mlx5_del_flow_rules(esw->fdb_table.offloads.miss_rule_uni);
+	mlx5_eswitch_del_send_to_vport_meta_rules(esw);
 	mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_grp);
+	if (esw->fdb_table.offloads.send_to_vport_meta_grp)
+		mlx5_destroy_flow_group(esw->fdb_table.offloads.send_to_vport_meta_grp);
 	if (MLX5_CAP_ESW(esw->dev, merged_eswitch))
 		mlx5_destroy_flow_group(esw->fdb_table.offloads.peer_miss_grp);
 	mlx5_destroy_flow_group(esw->fdb_table.offloads.miss_grp);