Merge branch 'mlx5-ntuple-steering'

Saeed Mahameed says:

====================
Mellanox 100G mlx5 ethtool ntuple steering

This series adds Ethernet ethtool ntuple steering 'ethtool -N|U' and exposes two more
counter sets to Ethtool statistics, RDMA vport and global flow control statistics.

We start from three refactoring patches of the flow steering infrastructure
    - mlx5_add_flow_rule will now receive mlx5 flow spec to simplify and reduce
      number of parameters
    - All low level steering objects are now wrapped in mlx5_flow_steering structure
      for better encapsulation
    - Flow steering object will now be removed properly and generically rather than
      traversing on a well-known steering tree objects

Patch#4 adds the infrastructure and the data structures needed for the ethtool ntuple
steering, all implemented in a new file 'en_fs_ethtool.c'.  Add the support for set_rxnfc
ethtool callback to add/remove/replace a flow spec of ethter type L2.

Patch#5 adds the support for L3/L4 flow specs and a higher priority in favor for L3/L4
rules when interleaving with L2 rules.

Patch#6 adds the support for get_rxnfc ethtool callback.

Patch#7,8 adds RDMA vport and global flow control statistics.

Applied on top: 8186f6e3 ('net-next: mediatek: fix compile error inside mtk_poll_controller()')
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/infiniband/hw/mlx5/main.c

@@ -1528,21 +1528,18 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
 {
 	struct mlx5_flow_table	*ft = ft_prio->flow_table;
 	struct mlx5_ib_flow_handler *handler;
+	struct mlx5_flow_spec *spec;
 	void *ib_flow = flow_attr + 1;
-	u8 match_criteria_enable = 0;
 	unsigned int spec_index;
-	u32 *match_c;
-	u32 *match_v;
 	u32 action;
 	int err = 0;
 
 	if (!is_valid_attr(flow_attr))
 		return ERR_PTR(-EINVAL);
 
-	match_c = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	match_v = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
+	spec = mlx5_vzalloc(sizeof(*spec));
 	handler = kzalloc(sizeof(*handler), GFP_KERNEL);
-	if (!handler || !match_c || !match_v) {
+	if (!handler || !spec) {
 		err = -ENOMEM;
 		goto free;
 	}
@@ -1550,7 +1547,8 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
 	INIT_LIST_HEAD(&handler->list);
 
 	for (spec_index = 0; spec_index < flow_attr->num_of_specs; spec_index++) {
-		err = parse_flow_attr(match_c, match_v, ib_flow);
+		err = parse_flow_attr(spec->match_criteria,
+				      spec->match_value, ib_flow);
 		if (err < 0)
 			goto free;
 
@@ -1558,11 +1556,11 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
 	}
 
 	/* Outer header support only */
-	match_criteria_enable = (!outer_header_zero(match_c)) << 0;
+	spec->match_criteria_enable = (!outer_header_zero(spec->match_criteria))
+		<< 0;
 	action = dst ? MLX5_FLOW_CONTEXT_ACTION_FWD_DEST :
 		MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO;
-	handler->rule = mlx5_add_flow_rule(ft, match_criteria_enable,
-					   match_c, match_v,
+	handler->rule = mlx5_add_flow_rule(ft, spec,
 					   action,
 					   MLX5_FS_DEFAULT_FLOW_TAG,
 					   dst);
@@ -1578,8 +1576,7 @@ static struct mlx5_ib_flow_handler *create_flow_rule(struct mlx5_ib_dev *dev,
 free:
 	if (err)
 		kfree(handler);
-	kfree(match_c);
-	kfree(match_v);
+	kvfree(spec);
 	return err ? ERR_PTR(err) : handler;
 }
 

drivers/net/ethernet/mellanox/mlx5/core/Makefile

@@ -8,6 +8,6 @@ mlx5_core-y :=	main.o cmd.o debugfs.o fw.o eq.o uar.o pagealloc.o \
 mlx5_core-$(CONFIG_MLX5_CORE_EN) += wq.o eswitch.o eswitch_offloads.o \
 		en_main.o en_common.o en_fs.o en_ethtool.o en_tx.o \
 		en_rx.o en_rx_am.o en_txrx.o en_clock.o vxlan.o \
-		en_tc.o en_arfs.o en_rep.o
+		en_tc.o en_arfs.o en_rep.o en_fs_ethtool.o
 
 mlx5_core-$(CONFIG_MLX5_CORE_EN_DCB) +=  en_dcbnl.o

drivers/net/ethernet/mellanox/mlx5/core/en.h

@@ -544,8 +544,24 @@ enum {
 	MLX5E_ARFS_FT_LEVEL
 };
 
+struct mlx5e_ethtool_table {
+	struct mlx5_flow_table *ft;
+	int                    num_rules;
+};
+
+#define ETHTOOL_NUM_L3_L4_FTS 7
+#define ETHTOOL_NUM_L2_FTS 4
+
+struct mlx5e_ethtool_steering {
+	struct mlx5e_ethtool_table      l3_l4_ft[ETHTOOL_NUM_L3_L4_FTS];
+	struct mlx5e_ethtool_table      l2_ft[ETHTOOL_NUM_L2_FTS];
+	struct list_head                rules;
+	int                             tot_num_rules;
+};
+
 struct mlx5e_flow_steering {
 	struct mlx5_flow_namespace      *ns;
+	struct mlx5e_ethtool_steering   ethtool;
 	struct mlx5e_tc_table           tc;
 	struct mlx5e_vlan_table         vlan;
 	struct mlx5e_l2_table           l2;
@@ -701,6 +717,16 @@ int mlx5e_create_flow_steering(struct mlx5e_priv *priv);
 void mlx5e_destroy_flow_steering(struct mlx5e_priv *priv);
 void mlx5e_init_l2_addr(struct mlx5e_priv *priv);
 void mlx5e_destroy_flow_table(struct mlx5e_flow_table *ft);
+int mlx5e_ethtool_get_flow(struct mlx5e_priv *priv, struct ethtool_rxnfc *info,
+			   int location);
+int mlx5e_ethtool_get_all_flows(struct mlx5e_priv *priv,
+				struct ethtool_rxnfc *info, u32 *rule_locs);
+int mlx5e_ethtool_flow_replace(struct mlx5e_priv *priv,
+			       struct ethtool_rx_flow_spec *fs);
+int mlx5e_ethtool_flow_remove(struct mlx5e_priv *priv,
+			      int location);
+void mlx5e_ethtool_init_steering(struct mlx5e_priv *priv);
+void mlx5e_ethtool_cleanup_steering(struct mlx5e_priv *priv);
 void mlx5e_set_rx_mode_work(struct work_struct *work);
 
 void mlx5e_fill_hwstamp(struct mlx5e_tstamp *clock, u64 timestamp,

drivers/net/ethernet/mellanox/mlx5/core/en_arfs.c

@@ -175,15 +175,12 @@ static int arfs_add_default_rule(struct mlx5e_priv *priv,
 {
 	struct arfs_table *arfs_t = &priv->fs.arfs.arfs_tables[type];
 	struct mlx5_flow_destination dest;
-	u8 match_criteria_enable = 0;
 	struct mlx5e_tir *tir = priv->indir_tir;
-	u32 *match_criteria;
-	u32 *match_value;
+	struct mlx5_flow_spec *spec;
 	int err = 0;
 
-	match_value	= mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	match_criteria	= mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	if (!match_value || !match_criteria) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
 		err = -ENOMEM;
 		goto out;
@@ -208,8 +205,7 @@ static int arfs_add_default_rule(struct mlx5e_priv *priv,
 		goto out;
 	}
 
-	arfs_t->default_rule = mlx5_add_flow_rule(arfs_t->ft.t, match_criteria_enable,
-						  match_criteria, match_value,
+	arfs_t->default_rule = mlx5_add_flow_rule(arfs_t->ft.t, spec,
 						  MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
 						  MLX5_FS_DEFAULT_FLOW_TAG,
 						  &dest);
@@ -220,8 +216,7 @@ static int arfs_add_default_rule(struct mlx5e_priv *priv,
 			   __func__, type);
 	}
 out:
-	kvfree(match_criteria);
-	kvfree(match_value);
+	kvfree(spec);
 	return err;
 }
 
@@ -475,23 +470,20 @@ static struct mlx5_flow_rule *arfs_add_rule(struct mlx5e_priv *priv,
 	struct mlx5_flow_rule *rule = NULL;
 	struct mlx5_flow_destination dest;
 	struct arfs_table *arfs_table;
-	u8 match_criteria_enable = 0;
+	struct mlx5_flow_spec *spec;
 	struct mlx5_flow_table *ft;
-	u32 *match_criteria;
-	u32 *match_value;
 	int err = 0;
 
-	match_value	= mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	match_criteria	= mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	if (!match_value || !match_criteria) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
 		err = -ENOMEM;
 		goto out;
 	}
-	match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
-	MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 			 outer_headers.ethertype);
-	MLX5_SET(fte_match_param, match_value, outer_headers.ethertype,
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.ethertype,
 		 ntohs(tuple->etype));
 	arfs_table = arfs_get_table(arfs, tuple->ip_proto, tuple->etype);
 	if (!arfs_table) {
@@ -501,59 +493,58 @@ static struct mlx5_flow_rule *arfs_add_rule(struct mlx5e_priv *priv,
 
 	ft = arfs_table->ft.t;
 	if (tuple->ip_proto == IPPROTO_TCP) {
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.tcp_dport);
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.tcp_sport);
-		MLX5_SET(fte_match_param, match_value, outer_headers.tcp_dport,
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.tcp_dport,
 			 ntohs(tuple->dst_port));
-		MLX5_SET(fte_match_param, match_value, outer_headers.tcp_sport,
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.tcp_sport,
 			 ntohs(tuple->src_port));
 	} else {
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.udp_dport);
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.udp_sport);
-		MLX5_SET(fte_match_param, match_value, outer_headers.udp_dport,
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.udp_dport,
 			 ntohs(tuple->dst_port));
-		MLX5_SET(fte_match_param, match_value, outer_headers.udp_sport,
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.udp_sport,
 			 ntohs(tuple->src_port));
 	}
 	if (tuple->etype == htons(ETH_P_IP)) {
-		memcpy(MLX5_ADDR_OF(fte_match_param, match_value,
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
 				    outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4),
 		       &tuple->src_ipv4,
 		       4);
-		memcpy(MLX5_ADDR_OF(fte_match_param, match_value,
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
 				    outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
 		       &tuple->dst_ipv4,
 		       4);
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.src_ipv4_src_ipv6.ipv4_layout.ipv4);
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria,
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
 				 outer_headers.dst_ipv4_dst_ipv6.ipv4_layout.ipv4);
 	} else {
-		memcpy(MLX5_ADDR_OF(fte_match_param, match_value,
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
 				    outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
 		       &tuple->src_ipv6,
 		       16);
-		memcpy(MLX5_ADDR_OF(fte_match_param, match_value,
+		memcpy(MLX5_ADDR_OF(fte_match_param, spec->match_value,
 				    outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
 		       &tuple->dst_ipv6,
 		       16);
-		memset(MLX5_ADDR_OF(fte_match_param, match_criteria,
+		memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
 				    outer_headers.src_ipv4_src_ipv6.ipv6_layout.ipv6),
 		       0xff,
 		       16);
-		memset(MLX5_ADDR_OF(fte_match_param, match_criteria,
+		memset(MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
 				    outer_headers.dst_ipv4_dst_ipv6.ipv6_layout.ipv6),
 		       0xff,
 		       16);
 	}
 	dest.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
 	dest.tir_num = priv->direct_tir[arfs_rule->rxq].tirn;
-	rule = mlx5_add_flow_rule(ft, match_criteria_enable, match_criteria,
-				  match_value, MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
+	rule = mlx5_add_flow_rule(ft, spec, MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
 				  MLX5_FS_DEFAULT_FLOW_TAG,
 				  &dest);
 	if (IS_ERR(rule)) {
@@ -563,8 +554,7 @@ static struct mlx5_flow_rule *arfs_add_rule(struct mlx5e_priv *priv,
 	}
 
 out:
-	kvfree(match_criteria);
-	kvfree(match_value);
+	kvfree(spec);
 	return err ? ERR_PTR(err) : rule;
 }
 

drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c

@@ -139,6 +139,18 @@ static unsigned long mlx5e_query_pfc_combined(struct mlx5e_priv *priv)
 	return err ? 0 : pfc_en_tx | pfc_en_rx;
 }
 
+static bool mlx5e_query_global_pause_combined(struct mlx5e_priv *priv)
+{
+	struct mlx5_core_dev *mdev = priv->mdev;
+	u32 rx_pause;
+	u32 tx_pause;
+	int err;
+
+	err = mlx5_query_port_pause(mdev, &rx_pause, &tx_pause);
+
+	return err ? false : rx_pause | tx_pause;
+}
+
 #define MLX5E_NUM_Q_CNTRS(priv) (NUM_Q_COUNTERS * (!!priv->q_counter))
 #define MLX5E_NUM_RQ_STATS(priv) \
 	(NUM_RQ_STATS * priv->params.num_channels * \
@@ -147,8 +159,8 @@ static unsigned long mlx5e_query_pfc_combined(struct mlx5e_priv *priv)
 	(NUM_SQ_STATS * priv->params.num_channels * priv->params.num_tc * \
 	 test_bit(MLX5E_STATE_OPENED, &priv->state))
 #define MLX5E_NUM_PFC_COUNTERS(priv) \
-	(hweight8(mlx5e_query_pfc_combined(priv)) * \
-	 NUM_PPORT_PER_PRIO_PFC_COUNTERS)
+	((mlx5e_query_global_pause_combined(priv) + hweight8(mlx5e_query_pfc_combined(priv))) * \
+	  NUM_PPORT_PER_PRIO_PFC_COUNTERS)
 
 static int mlx5e_get_sset_count(struct net_device *dev, int sset)
 {
@@ -210,8 +222,18 @@ static void mlx5e_fill_stats_strings(struct mlx5e_priv *priv, uint8_t *data)
 	pfc_combined = mlx5e_query_pfc_combined(priv);
 	for_each_set_bit(prio, &pfc_combined, NUM_PPORT_PRIO) {
 		for (i = 0; i < NUM_PPORT_PER_PRIO_PFC_COUNTERS; i++) {
+			char pfc_string[ETH_GSTRING_LEN];
+
+			snprintf(pfc_string, sizeof(pfc_string), "prio%d", prio);
 			sprintf(data + (idx++) * ETH_GSTRING_LEN,
-				pport_per_prio_pfc_stats_desc[i].format, prio);
+				pport_per_prio_pfc_stats_desc[i].format, pfc_string);
+		}
+	}
+
+	if (mlx5e_query_global_pause_combined(priv)) {
+		for (i = 0; i < NUM_PPORT_PER_PRIO_PFC_COUNTERS; i++) {
+			sprintf(data + (idx++) * ETH_GSTRING_LEN,
+				pport_per_prio_pfc_stats_desc[i].format, "global");
 		}
 	}
 
@@ -306,6 +328,13 @@ static void mlx5e_get_ethtool_stats(struct net_device *dev,
 		}
 	}
 
+	if (mlx5e_query_global_pause_combined(priv)) {
+		for (i = 0; i < NUM_PPORT_PER_PRIO_PFC_COUNTERS; i++) {
+			data[idx++] = MLX5E_READ_CTR64_BE(&priv->stats.pport.per_prio_counters[0],
+							  pport_per_prio_pfc_stats_desc, 0);
+		}
+	}
+
 	if (!test_bit(MLX5E_STATE_OPENED, &priv->state))
 		return;
 
@@ -931,6 +960,15 @@ static int mlx5e_get_rxnfc(struct net_device *netdev,
 	case ETHTOOL_GRXRINGS:
 		info->data = priv->params.num_channels;
 		break;
+	case ETHTOOL_GRXCLSRLCNT:
+		info->rule_cnt = priv->fs.ethtool.tot_num_rules;
+		break;
+	case ETHTOOL_GRXCLSRULE:
+		err = mlx5e_ethtool_get_flow(priv, info, info->fs.location);
+		break;
+	case ETHTOOL_GRXCLSRLALL:
+		err = mlx5e_ethtool_get_all_flows(priv, info, rule_locs);
+		break;
 	default:
 		err = -EOPNOTSUPP;
 		break;
@@ -1368,6 +1406,26 @@ static u32 mlx5e_get_priv_flags(struct net_device *netdev)
 	return priv->pflags;
 }
 
+static int mlx5e_set_rxnfc(struct net_device *dev, struct ethtool_rxnfc *cmd)
+{
+	int err = 0;
+	struct mlx5e_priv *priv = netdev_priv(dev);
+
+	switch (cmd->cmd) {
+	case ETHTOOL_SRXCLSRLINS:
+		err = mlx5e_ethtool_flow_replace(priv, &cmd->fs);
+		break;
+	case ETHTOOL_SRXCLSRLDEL:
+		err = mlx5e_ethtool_flow_remove(priv, cmd->fs.location);
+		break;
+	default:
+		err = -EOPNOTSUPP;
+		break;
+	}
+
+	return err;
+}
+
 const struct ethtool_ops mlx5e_ethtool_ops = {
 	.get_drvinfo       = mlx5e_get_drvinfo,
 	.get_link          = ethtool_op_get_link,
@@ -1387,6 +1445,7 @@ const struct ethtool_ops mlx5e_ethtool_ops = {
 	.get_rxfh          = mlx5e_get_rxfh,
 	.set_rxfh          = mlx5e_set_rxfh,
 	.get_rxnfc         = mlx5e_get_rxnfc,
+	.set_rxnfc         = mlx5e_set_rxnfc,
 	.get_tunable       = mlx5e_get_tunable,
 	.set_tunable       = mlx5e_set_tunable,
 	.get_pauseparam    = mlx5e_get_pauseparam,

drivers/net/ethernet/mellanox/mlx5/core/en_fs.c

@@ -156,19 +156,18 @@ enum mlx5e_vlan_rule_type {
 
 static int __mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
 				 enum mlx5e_vlan_rule_type rule_type,
-				 u16 vid, u32 *mc, u32 *mv)
+				 u16 vid, struct mlx5_flow_spec *spec)
 {
 	struct mlx5_flow_table *ft = priv->fs.vlan.ft.t;
 	struct mlx5_flow_destination dest;
-	u8 match_criteria_enable = 0;
 	struct mlx5_flow_rule **rule_p;
 	int err = 0;
 
 	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
 	dest.ft = priv->fs.l2.ft.t;
 
-	match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
-	MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.vlan_tag);
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.vlan_tag);
 
 	switch (rule_type) {
 	case MLX5E_VLAN_RULE_TYPE_UNTAGGED:
@@ -176,17 +175,19 @@ static int __mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
 		break;
 	case MLX5E_VLAN_RULE_TYPE_ANY_VID:
 		rule_p = &priv->fs.vlan.any_vlan_rule;
-		MLX5_SET(fte_match_param, mv, outer_headers.vlan_tag, 1);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.vlan_tag, 1);
 		break;
 	default: /* MLX5E_VLAN_RULE_TYPE_MATCH_VID */
 		rule_p = &priv->fs.vlan.active_vlans_rule[vid];
-		MLX5_SET(fte_match_param, mv, outer_headers.vlan_tag, 1);
-		MLX5_SET_TO_ONES(fte_match_param, mc, outer_headers.first_vid);
-		MLX5_SET(fte_match_param, mv, outer_headers.first_vid, vid);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.vlan_tag, 1);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria,
+				 outer_headers.first_vid);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.first_vid,
+			 vid);
 		break;
 	}
 
-	*rule_p = mlx5_add_flow_rule(ft, match_criteria_enable, mc, mv,
+	*rule_p = mlx5_add_flow_rule(ft, spec,
 				     MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
 				     MLX5_FS_DEFAULT_FLOW_TAG,
 				     &dest);
@@ -203,27 +204,21 @@ static int __mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
 static int mlx5e_add_vlan_rule(struct mlx5e_priv *priv,
 			       enum mlx5e_vlan_rule_type rule_type, u16 vid)
 {
-	u32 *match_criteria;
-	u32 *match_value;
+	struct mlx5_flow_spec *spec;
 	int err = 0;
 
-	match_value	= mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	match_criteria	= mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	if (!match_value || !match_criteria) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
-		err = -ENOMEM;
-		goto add_vlan_rule_out;
+		return -ENOMEM;
 	}
 
 	if (rule_type == MLX5E_VLAN_RULE_TYPE_MATCH_VID)
 		mlx5e_vport_context_update_vlans(priv);
 
-	err = __mlx5e_add_vlan_rule(priv, rule_type, vid, match_criteria,
-				    match_value);
+	err = __mlx5e_add_vlan_rule(priv, rule_type, vid, spec);
 
-add_vlan_rule_out:
-	kvfree(match_criteria);
-	kvfree(match_value);
+	kvfree(spec);
 
 	return err;
 }
@@ -598,32 +593,27 @@ static struct mlx5_flow_rule *mlx5e_generate_ttc_rule(struct mlx5e_priv *priv,
 						      u8 proto)
 {
 	struct mlx5_flow_rule *rule;
-	u8 match_criteria_enable = 0;
-	u32 *match_criteria;
-	u32 *match_value;
+	struct mlx5_flow_spec *spec;
 	int err = 0;
 
-	match_value	= mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	match_criteria	= mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	if (!match_value || !match_criteria) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
-		err = -ENOMEM;
-		goto out;
+		return ERR_PTR(-ENOMEM);
 	}
 
 	if (proto) {
-		match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.ip_protocol);
-		MLX5_SET(fte_match_param, match_value, outer_headers.ip_protocol, proto);
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ip_protocol);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.ip_protocol, proto);
 	}
 	if (etype) {
-		match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
-		MLX5_SET_TO_ONES(fte_match_param, match_criteria, outer_headers.ethertype);
-		MLX5_SET(fte_match_param, match_value, outer_headers.ethertype, etype);
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.ethertype);
+		MLX5_SET(fte_match_param, spec->match_value, outer_headers.ethertype, etype);
 	}
 
-	rule = mlx5_add_flow_rule(ft, match_criteria_enable,
-				  match_criteria, match_value,
+	rule = mlx5_add_flow_rule(ft, spec,
 				  MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
 				  MLX5_FS_DEFAULT_FLOW_TAG,
 				  dest);
@@ -631,9 +621,8 @@ static struct mlx5_flow_rule *mlx5e_generate_ttc_rule(struct mlx5e_priv *priv,
 		err = PTR_ERR(rule);
 		netdev_err(priv->netdev, "%s: add rule failed\n", __func__);
 	}
-out:
-	kvfree(match_criteria);
-	kvfree(match_value);
+
+	kvfree(spec);
 	return err ? ERR_PTR(err) : rule;
 }
 
@@ -792,24 +781,20 @@ static int mlx5e_add_l2_flow_rule(struct mlx5e_priv *priv,
 {
 	struct mlx5_flow_table *ft = priv->fs.l2.ft.t;
 	struct mlx5_flow_destination dest;
-	u8 match_criteria_enable = 0;
-	u32 *match_criteria;
-	u32 *match_value;
+	struct mlx5_flow_spec *spec;
 	int err = 0;
 	u8 *mc_dmac;
 	u8 *mv_dmac;
 
-	match_value    = mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	match_criteria = mlx5_vzalloc(MLX5_ST_SZ_BYTES(fte_match_param));
-	if (!match_value || !match_criteria) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		netdev_err(priv->netdev, "%s: alloc failed\n", __func__);
-		err = -ENOMEM;
-		goto add_l2_rule_out;
+		return -ENOMEM;
 	}
 
-	mc_dmac = MLX5_ADDR_OF(fte_match_param, match_criteria,
+	mc_dmac = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
 			       outer_headers.dmac_47_16);
-	mv_dmac = MLX5_ADDR_OF(fte_match_param, match_value,
+	mv_dmac = MLX5_ADDR_OF(fte_match_param, spec->match_value,
 			       outer_headers.dmac_47_16);
 
 	dest.type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE;
@@ -817,13 +802,13 @@ static int mlx5e_add_l2_flow_rule(struct mlx5e_priv *priv,
 
 	switch (type) {
 	case MLX5E_FULLMATCH:
-		match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
 		eth_broadcast_addr(mc_dmac);
 		ether_addr_copy(mv_dmac, ai->addr);
 		break;
 
 	case MLX5E_ALLMULTI:
-		match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+		spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
 		mc_dmac[0] = 0x01;
 		mv_dmac[0] = 0x01;
 		break;
@@ -832,8 +817,7 @@ static int mlx5e_add_l2_flow_rule(struct mlx5e_priv *priv,
 		break;
 	}
 
-	ai->rule = mlx5_add_flow_rule(ft, match_criteria_enable, match_criteria,
-				      match_value,
+	ai->rule = mlx5_add_flow_rule(ft, spec,
 				      MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
 				      MLX5_FS_DEFAULT_FLOW_TAG, &dest);
 	if (IS_ERR(ai->rule)) {
@@ -843,9 +827,7 @@ static int mlx5e_add_l2_flow_rule(struct mlx5e_priv *priv,
 		ai->rule = NULL;
 	}
 
-add_l2_rule_out:
-	kvfree(match_criteria);
-	kvfree(match_value);
+	kvfree(spec);
 
 	return err;
 }
@@ -1102,6 +1084,8 @@ int mlx5e_create_flow_steering(struct mlx5e_priv *priv)
 		goto err_destroy_l2_table;
 	}
 
+	mlx5e_ethtool_init_steering(priv);
+
 	return 0;
 
 err_destroy_l2_table:
@@ -1121,4 +1105,5 @@ void mlx5e_destroy_flow_steering(struct mlx5e_priv *priv)
 	mlx5e_destroy_l2_table(priv);
 	mlx5e_destroy_ttc_table(priv);
 	mlx5e_arfs_destroy_tables(priv);
+	mlx5e_ethtool_cleanup_steering(priv);
 }

drivers/net/ethernet/mellanox/mlx5/core/en_fs_ethtool.c

+/*
+ * Copyright (c) 2016, Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ *     Redistribution and use in source and binary forms, with or
+ *     without modification, are permitted provided that the following
+ *     conditions are met:
+ *
+ *      - Redistributions of source code must retain the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer.
+ *
+ *      - Redistributions in binary form must reproduce the above
+ *        copyright notice, this list of conditions and the following
+ *        disclaimer in the documentation and/or other materials
+ *        provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/mlx5/fs.h>
+#include "en.h"
+
+struct mlx5e_ethtool_rule {
+	struct list_head             list;
+	struct ethtool_rx_flow_spec  flow_spec;
+	struct mlx5_flow_rule        *rule;
+	struct mlx5e_ethtool_table   *eth_ft;
+};
+
+static void put_flow_table(struct mlx5e_ethtool_table *eth_ft)
+{
+	if (!--eth_ft->num_rules) {
+		mlx5_destroy_flow_table(eth_ft->ft);
+		eth_ft->ft = NULL;
+	}
+}
+
+#define MLX5E_ETHTOOL_L3_L4_PRIO 0
+#define MLX5E_ETHTOOL_L2_PRIO (MLX5E_ETHTOOL_L3_L4_PRIO + ETHTOOL_NUM_L3_L4_FTS)
+#define MLX5E_ETHTOOL_NUM_ENTRIES 64000
+#define MLX5E_ETHTOOL_NUM_GROUPS  10
+static struct mlx5e_ethtool_table *get_flow_table(struct mlx5e_priv *priv,
+						  struct ethtool_rx_flow_spec *fs,
+						  int num_tuples)
+{
+	struct mlx5e_ethtool_table *eth_ft;
+	struct mlx5_flow_namespace *ns;
+	struct mlx5_flow_table *ft;
+	int max_tuples;
+	int table_size;
+	int prio;
+
+	switch (fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+		max_tuples = ETHTOOL_NUM_L3_L4_FTS;
+		prio = MLX5E_ETHTOOL_L3_L4_PRIO + (max_tuples - num_tuples);
+		eth_ft = &priv->fs.ethtool.l3_l4_ft[prio];
+		break;
+	case IP_USER_FLOW:
+		max_tuples = ETHTOOL_NUM_L3_L4_FTS;
+		prio = MLX5E_ETHTOOL_L3_L4_PRIO + (max_tuples - num_tuples);
+		eth_ft = &priv->fs.ethtool.l3_l4_ft[prio];
+		break;
+	case ETHER_FLOW:
+		max_tuples = ETHTOOL_NUM_L2_FTS;
+		prio = max_tuples - num_tuples;
+		eth_ft = &priv->fs.ethtool.l2_ft[prio];
+		prio += MLX5E_ETHTOOL_L2_PRIO;
+		break;
+	default:
+		return ERR_PTR(-EINVAL);
+	}
+
+	eth_ft->num_rules++;
+	if (eth_ft->ft)
+		return eth_ft;
+
+	ns = mlx5_get_flow_namespace(priv->mdev,
+				     MLX5_FLOW_NAMESPACE_ETHTOOL);
+	if (!ns)
+		return ERR_PTR(-ENOTSUPP);
+
+	table_size = min_t(u32, BIT(MLX5_CAP_FLOWTABLE(priv->mdev,
+						       flow_table_properties_nic_receive.log_max_ft_size)),
+			   MLX5E_ETHTOOL_NUM_ENTRIES);
+	ft = mlx5_create_auto_grouped_flow_table(ns, prio,
+						 table_size,
+						 MLX5E_ETHTOOL_NUM_GROUPS, 0);
+	if (IS_ERR(ft))
+		return (void *)ft;
+
+	eth_ft->ft = ft;
+	return eth_ft;
+}
+
+static void mask_spec(u8 *mask, u8 *val, size_t size)
+{
+	unsigned int i;
+
+	for (i = 0; i < size; i++, mask++, val++)
+		*((u8 *)val) = *((u8 *)mask) & *((u8 *)val);
+}
+
+static void set_ips(void *outer_headers_v, void *outer_headers_c, __be32 ip4src_m,
+		    __be32 ip4src_v, __be32 ip4dst_m, __be32 ip4dst_v)
+{
+	if (ip4src_m) {
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v,
+				    src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       &ip4src_v, sizeof(ip4src_v));
+		memset(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c,
+				    src_ipv4_src_ipv6.ipv4_layout.ipv4),
+		       0xff, sizeof(ip4src_m));
+	}
+	if (ip4dst_m) {
+		memcpy(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_v,
+				    dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       &ip4dst_v, sizeof(ip4dst_v));
+		memset(MLX5_ADDR_OF(fte_match_set_lyr_2_4, outer_headers_c,
+				    dst_ipv4_dst_ipv6.ipv4_layout.ipv4),
+		       0xff, sizeof(ip4dst_m));
+	}
+	MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v,
+		 ethertype, ETH_P_IP);
+	MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c,
+		 ethertype, 0xffff);
+}
+
+static int set_flow_attrs(u32 *match_c, u32 *match_v,
+			  struct ethtool_rx_flow_spec *fs)
+{
+	void *outer_headers_c = MLX5_ADDR_OF(fte_match_param, match_c,
+					     outer_headers);
+	void *outer_headers_v = MLX5_ADDR_OF(fte_match_param, match_v,
+					     outer_headers);
+	u32 flow_type = fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT);
+	struct ethtool_tcpip4_spec *l4_mask;
+	struct ethtool_tcpip4_spec *l4_val;
+	struct ethtool_usrip4_spec *l3_mask;
+	struct ethtool_usrip4_spec *l3_val;
+	struct ethhdr *eth_val;
+	struct ethhdr *eth_mask;
+
+	switch (flow_type) {
+	case TCP_V4_FLOW:
+		l4_mask = &fs->m_u.tcp_ip4_spec;
+		l4_val = &fs->h_u.tcp_ip4_spec;
+		set_ips(outer_headers_v, outer_headers_c, l4_mask->ip4src,
+			l4_val->ip4src, l4_mask->ip4dst, l4_val->ip4dst);
+
+		if (l4_mask->psrc) {
+			MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, tcp_sport,
+				 0xffff);
+			MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, tcp_sport,
+				 ntohs(l4_val->psrc));
+		}
+		if (l4_mask->pdst) {
+			MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, tcp_dport,
+				 0xffff);
+			MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, tcp_dport,
+				 ntohs(l4_val->pdst));
+		}
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, ip_protocol,
+			 0xffff);
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, ip_protocol,
+			 IPPROTO_TCP);
+		break;
+	case UDP_V4_FLOW:
+		l4_mask = &fs->m_u.tcp_ip4_spec;
+		l4_val = &fs->h_u.tcp_ip4_spec;
+		set_ips(outer_headers_v, outer_headers_c, l4_mask->ip4src,
+			l4_val->ip4src, l4_mask->ip4dst, l4_val->ip4dst);
+
+		if (l4_mask->psrc) {
+			MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, udp_sport,
+				 0xffff);
+			MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, udp_sport,
+				 ntohs(l4_val->psrc));
+		}
+		if (l4_mask->pdst) {
+			MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, udp_dport,
+				 0xffff);
+			MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, udp_dport,
+				 ntohs(l4_val->pdst));
+		}
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, ip_protocol,
+			 0xffff);
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, ip_protocol,
+			 IPPROTO_UDP);
+		break;
+	case IP_USER_FLOW:
+		l3_mask = &fs->m_u.usr_ip4_spec;
+		l3_val = &fs->h_u.usr_ip4_spec;
+		set_ips(outer_headers_v, outer_headers_c, l3_mask->ip4src,
+			l3_val->ip4src, l3_mask->ip4dst, l3_val->ip4dst);
+		break;
+	case ETHER_FLOW:
+		eth_mask = &fs->m_u.ether_spec;
+		eth_val = &fs->h_u.ether_spec;
+
+		mask_spec((u8 *)eth_mask, (u8 *)eth_val, sizeof(*eth_mask));
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4,
+					     outer_headers_c, smac_47_16),
+				eth_mask->h_source);
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4,
+					     outer_headers_v, smac_47_16),
+				eth_val->h_source);
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4,
+					     outer_headers_c, dmac_47_16),
+				eth_mask->h_dest);
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4,
+					     outer_headers_v, dmac_47_16),
+				eth_val->h_dest);
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c, ethertype,
+			 ntohs(eth_mask->h_proto));
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v, ethertype,
+			 ntohs(eth_val->h_proto));
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if ((fs->flow_type & FLOW_EXT) &&
+	    (fs->m_ext.vlan_tci & cpu_to_be16(VLAN_VID_MASK))) {
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c,
+			 vlan_tag, 1);
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v,
+			 vlan_tag, 1);
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_c,
+			 first_vid, 0xfff);
+		MLX5_SET(fte_match_set_lyr_2_4, outer_headers_v,
+			 first_vid, ntohs(fs->h_ext.vlan_tci));
+	}
+	if (fs->flow_type & FLOW_MAC_EXT &&
+	    !is_zero_ether_addr(fs->m_ext.h_dest)) {
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4,
+					     outer_headers_c, dmac_47_16),
+				fs->m_ext.h_dest);
+		ether_addr_copy(MLX5_ADDR_OF(fte_match_set_lyr_2_4,
+					     outer_headers_v, dmac_47_16),
+				fs->h_ext.h_dest);
+	}
+
+	return 0;
+}
+
+static void add_rule_to_list(struct mlx5e_priv *priv,
+			     struct mlx5e_ethtool_rule *rule)
+{
+	struct mlx5e_ethtool_rule *iter;
+	struct list_head *head = &priv->fs.ethtool.rules;
+
+	list_for_each_entry(iter, &priv->fs.ethtool.rules, list) {
+		if (iter->flow_spec.location > rule->flow_spec.location)
+			break;
+		head = &iter->list;
+	}
+	priv->fs.ethtool.tot_num_rules++;
+	list_add(&rule->list, head);
+}
+
+static bool outer_header_zero(u32 *match_criteria)
+{
+	int size = MLX5_ST_SZ_BYTES(fte_match_param);
+	char *outer_headers_c = MLX5_ADDR_OF(fte_match_param, match_criteria,
+					     outer_headers);
+
+	return outer_headers_c[0] == 0 && !memcmp(outer_headers_c,
+						  outer_headers_c + 1,
+						  size - 1);
+}
+
+static struct mlx5_flow_rule *add_ethtool_flow_rule(struct mlx5e_priv *priv,
+						    struct mlx5_flow_table *ft,
+						    struct ethtool_rx_flow_spec *fs)
+{
+	struct mlx5_flow_destination *dst = NULL;
+	struct mlx5_flow_spec *spec;
+	struct mlx5_flow_rule *rule;
+	int err = 0;
+	u32 action;
+
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec)
+		return ERR_PTR(-ENOMEM);
+	err = set_flow_attrs(spec->match_criteria, spec->match_value,
+			     fs);
+	if (err)
+		goto free;
+
+	if (fs->ring_cookie == RX_CLS_FLOW_DISC) {
+		action = MLX5_FLOW_CONTEXT_ACTION_DROP;
+	} else {
+		dst = kzalloc(sizeof(*dst), GFP_KERNEL);
+		if (!dst) {
+			err = -ENOMEM;
+			goto free;
+		}
+
+		dst->type = MLX5_FLOW_DESTINATION_TYPE_TIR;
+		dst->tir_num = priv->direct_tir[fs->ring_cookie].tirn;
+		action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST;
+	}
+
+	spec->match_criteria_enable = (!outer_header_zero(spec->match_criteria));
+	rule = mlx5_add_flow_rule(ft, spec, action,
+				  MLX5_FS_DEFAULT_FLOW_TAG, dst);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		netdev_err(priv->netdev, "%s: failed to add ethtool steering rule: %d\n",
+			   __func__, err);
+		goto free;
+	}
+free:
+	kvfree(spec);
+	kfree(dst);
+	return err ? ERR_PTR(err) : rule;
+}
+
+static void del_ethtool_rule(struct mlx5e_priv *priv,
+			     struct mlx5e_ethtool_rule *eth_rule)
+{
+	if (eth_rule->rule)
+		mlx5_del_flow_rule(eth_rule->rule);
+	list_del(&eth_rule->list);
+	priv->fs.ethtool.tot_num_rules--;
+	put_flow_table(eth_rule->eth_ft);
+	kfree(eth_rule);
+}
+
+static struct mlx5e_ethtool_rule *find_ethtool_rule(struct mlx5e_priv *priv,
+						    int location)
+{
+	struct mlx5e_ethtool_rule *iter;
+
+	list_for_each_entry(iter, &priv->fs.ethtool.rules, list) {
+		if (iter->flow_spec.location == location)
+			return iter;
+	}
+	return NULL;
+}
+
+static struct mlx5e_ethtool_rule *get_ethtool_rule(struct mlx5e_priv *priv,
+						   int location)
+{
+	struct mlx5e_ethtool_rule *eth_rule;
+
+	eth_rule = find_ethtool_rule(priv, location);
+	if (eth_rule)
+		del_ethtool_rule(priv, eth_rule);
+
+	eth_rule = kzalloc(sizeof(*eth_rule), GFP_KERNEL);
+	if (!eth_rule)
+		return ERR_PTR(-ENOMEM);
+
+	add_rule_to_list(priv, eth_rule);
+	return eth_rule;
+}
+
+#define MAX_NUM_OF_ETHTOOL_RULES BIT(10)
+
+#define all_ones(field) (field == (__force typeof(field))-1)
+#define all_zeros_or_all_ones(field)		\
+	((field) == 0 || (field) == (__force typeof(field))-1)
+
+static int validate_flow(struct mlx5e_priv *priv,
+			 struct ethtool_rx_flow_spec *fs)
+{
+	struct ethtool_tcpip4_spec *l4_mask;
+	struct ethtool_usrip4_spec *l3_mask;
+	struct ethhdr *eth_mask;
+	int num_tuples = 0;
+
+	if (fs->location >= MAX_NUM_OF_ETHTOOL_RULES)
+		return -EINVAL;
+
+	if (fs->ring_cookie >= priv->params.num_channels &&
+	    fs->ring_cookie != RX_CLS_FLOW_DISC)
+		return -EINVAL;
+
+	switch (fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+	case ETHER_FLOW:
+		eth_mask = &fs->m_u.ether_spec;
+		if (!is_zero_ether_addr(eth_mask->h_dest))
+			num_tuples++;
+		if (!is_zero_ether_addr(eth_mask->h_source))
+			num_tuples++;
+		if (eth_mask->h_proto)
+			num_tuples++;
+		break;
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+		if (fs->m_u.tcp_ip4_spec.tos)
+			return -EINVAL;
+		l4_mask = &fs->m_u.tcp_ip4_spec;
+		if (l4_mask->ip4src) {
+			if (!all_ones(l4_mask->ip4src))
+				return -EINVAL;
+			num_tuples++;
+		}
+		if (l4_mask->ip4dst) {
+			if (!all_ones(l4_mask->ip4dst))
+				return -EINVAL;
+			num_tuples++;
+		}
+		if (l4_mask->psrc) {
+			if (!all_ones(l4_mask->psrc))
+				return -EINVAL;
+			num_tuples++;
+		}
+		if (l4_mask->pdst) {
+			if (!all_ones(l4_mask->pdst))
+				return -EINVAL;
+			num_tuples++;
+		}
+		/* Flow is TCP/UDP */
+		num_tuples++;
+		break;
+	case IP_USER_FLOW:
+		l3_mask = &fs->m_u.usr_ip4_spec;
+		if (l3_mask->l4_4_bytes || l3_mask->tos || l3_mask->proto ||
+		    fs->h_u.usr_ip4_spec.ip_ver != ETH_RX_NFC_IP4)
+			return -EINVAL;
+		if (l3_mask->ip4src) {
+			if (!all_ones(l3_mask->ip4src))
+				return -EINVAL;
+			num_tuples++;
+		}
+		if (l3_mask->ip4dst) {
+			if (!all_ones(l3_mask->ip4dst))
+				return -EINVAL;
+			num_tuples++;
+		}
+		/* Flow is IPv4 */
+		num_tuples++;
+		break;
+	default:
+		return -EINVAL;
+	}
+	if ((fs->flow_type & FLOW_EXT)) {
+		if (fs->m_ext.vlan_etype ||
+		    (fs->m_ext.vlan_tci != cpu_to_be16(VLAN_VID_MASK)))
+			return -EINVAL;
+
+		if (fs->m_ext.vlan_tci) {
+			if (be16_to_cpu(fs->h_ext.vlan_tci) >= VLAN_N_VID)
+				return -EINVAL;
+		}
+		num_tuples++;
+	}
+
+	if (fs->flow_type & FLOW_MAC_EXT &&
+	    !is_zero_ether_addr(fs->m_ext.h_dest))
+		num_tuples++;
+
+	return num_tuples;
+}
+
+int mlx5e_ethtool_flow_replace(struct mlx5e_priv *priv,
+			       struct ethtool_rx_flow_spec *fs)
+{
+	struct mlx5e_ethtool_table *eth_ft;
+	struct mlx5e_ethtool_rule *eth_rule;
+	struct mlx5_flow_rule *rule;
+	int num_tuples;
+	int err;
+
+	num_tuples = validate_flow(priv, fs);
+	if (num_tuples <= 0) {
+		netdev_warn(priv->netdev, "%s: flow is not valid\n",  __func__);
+		return -EINVAL;
+	}
+
+	eth_ft = get_flow_table(priv, fs, num_tuples);
+	if (IS_ERR(eth_ft))
+		return PTR_ERR(eth_ft);
+
+	eth_rule = get_ethtool_rule(priv, fs->location);
+	if (IS_ERR(eth_rule)) {
+		put_flow_table(eth_ft);
+		return PTR_ERR(eth_rule);
+	}
+
+	eth_rule->flow_spec = *fs;
+	eth_rule->eth_ft = eth_ft;
+	if (!eth_ft->ft) {
+		err = -EINVAL;
+		goto del_ethtool_rule;
+	}
+	rule = add_ethtool_flow_rule(priv, eth_ft->ft, fs);
+	if (IS_ERR(rule)) {
+		err = PTR_ERR(rule);
+		goto del_ethtool_rule;
+	}
+
+	eth_rule->rule = rule;
+
+	return 0;
+
+del_ethtool_rule:
+	del_ethtool_rule(priv, eth_rule);
+
+	return err;
+}
+
+int mlx5e_ethtool_flow_remove(struct mlx5e_priv *priv,
+			      int location)
+{
+	struct mlx5e_ethtool_rule *eth_rule;
+	int err = 0;
+
+	if (location >= MAX_NUM_OF_ETHTOOL_RULES)
+		return -ENOSPC;
+
+	eth_rule = find_ethtool_rule(priv, location);
+	if (!eth_rule) {
+		err =  -ENOENT;
+		goto out;
+	}
+
+	del_ethtool_rule(priv, eth_rule);
+out:
+	return err;
+}
+
+int mlx5e_ethtool_get_flow(struct mlx5e_priv *priv, struct ethtool_rxnfc *info,
+			   int location)
+{
+	struct mlx5e_ethtool_rule *eth_rule;
+
+	if (location < 0 || location >= MAX_NUM_OF_ETHTOOL_RULES)
+		return -EINVAL;
+
+	list_for_each_entry(eth_rule, &priv->fs.ethtool.rules, list) {
+		if (eth_rule->flow_spec.location == location) {
+			info->fs = eth_rule->flow_spec;
+			return 0;
+		}
+	}
+
+	return -ENOENT;
+}
+
+int mlx5e_ethtool_get_all_flows(struct mlx5e_priv *priv, struct ethtool_rxnfc *info,
+				u32 *rule_locs)
+{
+	int location = 0;
+	int idx = 0;
+	int err = 0;
+
+	while ((!err || err == -ENOENT) && idx < info->rule_cnt) {
+		err = mlx5e_ethtool_get_flow(priv, info, location);
+		if (!err)
+			rule_locs[idx++] = location;
+		location++;
+	}
+	return err;
+}
+
+void mlx5e_ethtool_cleanup_steering(struct mlx5e_priv *priv)
+{
+	struct mlx5e_ethtool_rule *iter;
+	struct mlx5e_ethtool_rule *temp;
+
+	list_for_each_entry_safe(iter, temp, &priv->fs.ethtool.rules, list)
+		del_ethtool_rule(priv, iter);
+}
+
+void mlx5e_ethtool_init_steering(struct mlx5e_priv *priv)
+{
+	INIT_LIST_HEAD(&priv->fs.ethtool.rules);
+}

drivers/net/ethernet/mellanox/mlx5/core/en_stats.h

@@ -151,6 +151,22 @@ static const struct counter_desc vport_stats_desc[] = {
 		VPORT_COUNTER_OFF(transmitted_eth_broadcast.packets) },
 	{ "tx_vport_broadcast_bytes",
 		VPORT_COUNTER_OFF(transmitted_eth_broadcast.octets) },
+	{ "rx_vport_rdma_unicast_packets",
+		VPORT_COUNTER_OFF(received_ib_unicast.packets) },
+	{ "rx_vport_rdma_unicast_bytes",
+		VPORT_COUNTER_OFF(received_ib_unicast.octets) },
+	{ "tx_vport_rdma_unicast_packets",
+		VPORT_COUNTER_OFF(transmitted_ib_unicast.packets) },
+	{ "tx_vport_rdma_unicast_bytes",
+		VPORT_COUNTER_OFF(transmitted_ib_unicast.octets) },
+	{ "rx_vport_rdma_multicast_packets",
+		VPORT_COUNTER_OFF(received_ib_multicast.packets) },
+	{ "rx_vport_rdma_multicast_bytes",
+		VPORT_COUNTER_OFF(received_ib_multicast.octets) },
+	{ "tx_vport_rdma_multicast_packets",
+		VPORT_COUNTER_OFF(transmitted_ib_multicast.packets) },
+	{ "tx_vport_rdma_multicast_bytes",
+		VPORT_COUNTER_OFF(transmitted_ib_multicast.octets) },
 };
 
 #define PPORT_802_3_OFF(c) \
@@ -238,11 +254,12 @@ static const struct counter_desc pport_per_prio_traffic_stats_desc[] = {
 };
 
 static const struct counter_desc pport_per_prio_pfc_stats_desc[] = {
-	{ "rx_prio%d_pause", PPORT_PER_PRIO_OFF(rx_pause) },
-	{ "rx_prio%d_pause_duration", PPORT_PER_PRIO_OFF(rx_pause_duration) },
-	{ "tx_prio%d_pause", PPORT_PER_PRIO_OFF(tx_pause) },
-	{ "tx_prio%d_pause_duration", PPORT_PER_PRIO_OFF(tx_pause_duration) },
-	{ "rx_prio%d_pause_transition", PPORT_PER_PRIO_OFF(rx_pause_transition) },
+	/* %s is "global" or "prio{i}" */
+	{ "rx_%s_pause", PPORT_PER_PRIO_OFF(rx_pause) },
+	{ "rx_%s_pause_duration", PPORT_PER_PRIO_OFF(rx_pause_duration) },
+	{ "tx_%s_pause", PPORT_PER_PRIO_OFF(tx_pause) },
+	{ "tx_%s_pause_duration", PPORT_PER_PRIO_OFF(tx_pause_duration) },
+	{ "rx_%s_pause_transition", PPORT_PER_PRIO_OFF(rx_pause_transition) },
 };
 
 struct mlx5e_rq_stats {

drivers/net/ethernet/mellanox/mlx5/core/en_tc.c

@@ -50,7 +50,7 @@ struct mlx5e_tc_flow {
 #define MLX5E_TC_TABLE_NUM_GROUPS 4
 
 static struct mlx5_flow_rule *mlx5e_tc_add_flow(struct mlx5e_priv *priv,
-						u32 *match_c, u32 *match_v,
+						struct mlx5_flow_spec *spec,
 						u32 action, u32 flow_tag)
 {
 	struct mlx5_core_dev *dev = priv->mdev;
@@ -88,8 +88,8 @@ static struct mlx5_flow_rule *mlx5e_tc_add_flow(struct mlx5e_priv *priv,
 		table_created = true;
 	}
 
-	rule = mlx5_add_flow_rule(priv->fs.tc.t, MLX5_MATCH_OUTER_HEADERS,
-				  match_c, match_v,
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
+	rule = mlx5_add_flow_rule(priv->fs.tc.t, spec,
 				  action, flow_tag,
 				  &dest);
 
@@ -126,12 +126,13 @@ static void mlx5e_tc_del_flow(struct mlx5e_priv *priv,
 	}
 }
 
-static int parse_cls_flower(struct mlx5e_priv *priv,
-			    u32 *match_c, u32 *match_v,
+static int parse_cls_flower(struct mlx5e_priv *priv, struct mlx5_flow_spec *spec,
 			    struct tc_cls_flower_offload *f)
 {
-	void *headers_c = MLX5_ADDR_OF(fte_match_param, match_c, outer_headers);
-	void *headers_v = MLX5_ADDR_OF(fte_match_param, match_v, outer_headers);
+	void *headers_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+				       outer_headers);
+	void *headers_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+				       outer_headers);
 	u16 addr_type = 0;
 	u8 ip_proto = 0;
 
@@ -342,12 +343,11 @@ int mlx5e_configure_flower(struct mlx5e_priv *priv, __be16 protocol,
 			   struct tc_cls_flower_offload *f)
 {
 	struct mlx5e_tc_table *tc = &priv->fs.tc;
-	u32 *match_c;
-	u32 *match_v;
 	int err = 0;
 	u32 flow_tag;
 	u32 action;
 	struct mlx5e_tc_flow *flow;
+	struct mlx5_flow_spec *spec;
 	struct mlx5_flow_rule *old = NULL;
 
 	flow = rhashtable_lookup_fast(&tc->ht, &f->cookie,
@@ -357,16 +357,15 @@ int mlx5e_configure_flower(struct mlx5e_priv *priv, __be16 protocol,
 	else
 		flow = kzalloc(sizeof(*flow), GFP_KERNEL);
 
-	match_c = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	match_v = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	if (!match_c || !match_v || !flow) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec || !flow) {
 		err = -ENOMEM;
 		goto err_free;
 	}
 
 	flow->cookie = f->cookie;
 
-	err = parse_cls_flower(priv, match_c, match_v, f);
+	err = parse_cls_flower(priv, spec, f);
 	if (err < 0)
 		goto err_free;
 
@@ -379,8 +378,7 @@ int mlx5e_configure_flower(struct mlx5e_priv *priv, __be16 protocol,
 	if (err)
 		goto err_free;
 
-	flow->rule = mlx5e_tc_add_flow(priv, match_c, match_v, action,
-				       flow_tag);
+	flow->rule = mlx5e_tc_add_flow(priv, spec, action, flow_tag);
 	if (IS_ERR(flow->rule)) {
 		err = PTR_ERR(flow->rule);
 		goto err_hash_del;
@@ -398,8 +396,7 @@ int mlx5e_configure_flower(struct mlx5e_priv *priv, __be16 protocol,
 	if (!old)
 		kfree(flow);
 out:
-	kfree(match_c);
-	kfree(match_v);
+	kvfree(spec);
 	return err;
 }
 

drivers/net/ethernet/mellanox/mlx5/core/eswitch.c

@@ -329,25 +329,23 @@ __esw_fdb_set_vport_rule(struct mlx5_eswitch *esw, u32 vport, bool rx_rule,
 			    MLX5_MATCH_OUTER_HEADERS);
 	struct mlx5_flow_rule *flow_rule = NULL;
 	struct mlx5_flow_destination dest;
+	struct mlx5_flow_spec *spec;
 	void *mv_misc = NULL;
 	void *mc_misc = NULL;
 	u8 *dmac_v = NULL;
 	u8 *dmac_c = NULL;
-	u32 *match_v;
-	u32 *match_c;
 
 	if (rx_rule)
 		match_header |= MLX5_MATCH_MISC_PARAMETERS;
-	match_v = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	match_c = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	if (!match_v || !match_c) {
+
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		pr_warn("FDB: Failed to alloc match parameters\n");
-		goto out;
+		return NULL;
 	}
-
-	dmac_v = MLX5_ADDR_OF(fte_match_param, match_v,
+	dmac_v = MLX5_ADDR_OF(fte_match_param, spec->match_value,
 			      outer_headers.dmac_47_16);
-	dmac_c = MLX5_ADDR_OF(fte_match_param, match_c,
+	dmac_c = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
 			      outer_headers.dmac_47_16);
 
 	if (match_header & MLX5_MATCH_OUTER_HEADERS) {
@@ -356,8 +354,10 @@ __esw_fdb_set_vport_rule(struct mlx5_eswitch *esw, u32 vport, bool rx_rule,
 	}
 
 	if (match_header & MLX5_MATCH_MISC_PARAMETERS) {
-		mv_misc  = MLX5_ADDR_OF(fte_match_param, match_v, misc_parameters);
-		mc_misc  = MLX5_ADDR_OF(fte_match_param, match_c, misc_parameters);
+		mv_misc  = MLX5_ADDR_OF(fte_match_param, spec->match_value,
+					misc_parameters);
+		mc_misc  = MLX5_ADDR_OF(fte_match_param, spec->match_criteria,
+					misc_parameters);
 		MLX5_SET(fte_match_set_misc, mv_misc, source_port, UPLINK_VPORT);
 		MLX5_SET_TO_ONES(fte_match_set_misc, mc_misc, source_port);
 	}
@@ -368,11 +368,9 @@ __esw_fdb_set_vport_rule(struct mlx5_eswitch *esw, u32 vport, bool rx_rule,
 	esw_debug(esw->dev,
 		  "\tFDB add rule dmac_v(%pM) dmac_c(%pM) -> vport(%d)\n",
 		  dmac_v, dmac_c, vport);
+	spec->match_criteria_enable = match_header;
 	flow_rule =
-		mlx5_add_flow_rule(esw->fdb_table.fdb,
-				   match_header,
-				   match_c,
-				   match_v,
+		mlx5_add_flow_rule(esw->fdb_table.fdb, spec,
 				   MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
 				   0, &dest);
 	if (IS_ERR(flow_rule)) {
@@ -381,9 +379,8 @@ __esw_fdb_set_vport_rule(struct mlx5_eswitch *esw, u32 vport, bool rx_rule,
 			 dmac_v, dmac_c, vport, PTR_ERR(flow_rule));
 		flow_rule = NULL;
 	}
-out:
-	kfree(match_v);
-	kfree(match_c);
+
+	kvfree(spec);
 	return flow_rule;
 }
 
@@ -1293,9 +1290,8 @@ static void esw_vport_disable_ingress_acl(struct mlx5_eswitch *esw,
 static int esw_vport_ingress_config(struct mlx5_eswitch *esw,
 				    struct mlx5_vport *vport)
 {
+	struct mlx5_flow_spec *spec;
 	u8 smac[ETH_ALEN];
-	u32 *match_v;
-	u32 *match_c;
 	int err = 0;
 	u8 *smac_v;
 
@@ -1329,9 +1325,8 @@ static int esw_vport_ingress_config(struct mlx5_eswitch *esw,
 		  "vport[%d] configure ingress rules, vlan(%d) qos(%d)\n",
 		  vport->vport, vport->vlan, vport->qos);
 
-	match_v = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	match_c = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	if (!match_v || !match_c) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		err = -ENOMEM;
 		esw_warn(esw->dev, "vport[%d] configure ingress rules failed, err(%d)\n",
 			 vport->vport, err);
@@ -1339,22 +1334,20 @@ static int esw_vport_ingress_config(struct mlx5_eswitch *esw,
 	}
 
 	if (vport->vlan || vport->qos)
-		MLX5_SET_TO_ONES(fte_match_param, match_c, outer_headers.vlan_tag);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.vlan_tag);
 
 	if (vport->spoofchk) {
-		MLX5_SET_TO_ONES(fte_match_param, match_c, outer_headers.smac_47_16);
-		MLX5_SET_TO_ONES(fte_match_param, match_c, outer_headers.smac_15_0);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.smac_47_16);
+		MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.smac_15_0);
 		smac_v = MLX5_ADDR_OF(fte_match_param,
-				      match_v,
+				      spec->match_value,
 				      outer_headers.smac_47_16);
 		ether_addr_copy(smac_v, smac);
 	}
 
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
 	vport->ingress.allow_rule =
-		mlx5_add_flow_rule(vport->ingress.acl,
-				   MLX5_MATCH_OUTER_HEADERS,
-				   match_c,
-				   match_v,
+		mlx5_add_flow_rule(vport->ingress.acl, spec,
 				   MLX5_FLOW_CONTEXT_ACTION_ALLOW,
 				   0, NULL);
 	if (IS_ERR(vport->ingress.allow_rule)) {
@@ -1365,13 +1358,9 @@ static int esw_vport_ingress_config(struct mlx5_eswitch *esw,
 		goto out;
 	}
 
-	memset(match_c, 0, MLX5_ST_SZ_BYTES(fte_match_param));
-	memset(match_v, 0, MLX5_ST_SZ_BYTES(fte_match_param));
+	memset(spec, 0, sizeof(*spec));
 	vport->ingress.drop_rule =
-		mlx5_add_flow_rule(vport->ingress.acl,
-				   0,
-				   match_c,
-				   match_v,
+		mlx5_add_flow_rule(vport->ingress.acl, spec,
 				   MLX5_FLOW_CONTEXT_ACTION_DROP,
 				   0, NULL);
 	if (IS_ERR(vport->ingress.drop_rule)) {
@@ -1385,17 +1374,14 @@ static int esw_vport_ingress_config(struct mlx5_eswitch *esw,
 out:
 	if (err)
 		esw_vport_cleanup_ingress_rules(esw, vport);
-
-	kfree(match_v);
-	kfree(match_c);
+	kvfree(spec);
 	return err;
 }
 
 static int esw_vport_egress_config(struct mlx5_eswitch *esw,
 				   struct mlx5_vport *vport)
 {
-	u32 *match_v;
-	u32 *match_c;
+	struct mlx5_flow_spec *spec;
 	int err = 0;
 
 	esw_vport_cleanup_egress_rules(esw, vport);
@@ -1411,9 +1397,8 @@ static int esw_vport_egress_config(struct mlx5_eswitch *esw,
 		  "vport[%d] configure egress rules, vlan(%d) qos(%d)\n",
 		  vport->vport, vport->vlan, vport->qos);
 
-	match_v = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	match_c = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	if (!match_v || !match_c) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		err = -ENOMEM;
 		esw_warn(esw->dev, "vport[%d] configure egress rules failed, err(%d)\n",
 			 vport->vport, err);
@@ -1421,16 +1406,14 @@ static int esw_vport_egress_config(struct mlx5_eswitch *esw,
 	}
 
 	/* Allowed vlan rule */
-	MLX5_SET_TO_ONES(fte_match_param, match_c, outer_headers.vlan_tag);
-	MLX5_SET_TO_ONES(fte_match_param, match_v, outer_headers.vlan_tag);
-	MLX5_SET_TO_ONES(fte_match_param, match_c, outer_headers.first_vid);
-	MLX5_SET(fte_match_param, match_v, outer_headers.first_vid, vport->vlan);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.vlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_value, outer_headers.vlan_tag);
+	MLX5_SET_TO_ONES(fte_match_param, spec->match_criteria, outer_headers.first_vid);
+	MLX5_SET(fte_match_param, spec->match_value, outer_headers.first_vid, vport->vlan);
 
+	spec->match_criteria_enable = MLX5_MATCH_OUTER_HEADERS;
 	vport->egress.allowed_vlan =
-		mlx5_add_flow_rule(vport->egress.acl,
-				   MLX5_MATCH_OUTER_HEADERS,
-				   match_c,
-				   match_v,
+		mlx5_add_flow_rule(vport->egress.acl, spec,
 				   MLX5_FLOW_CONTEXT_ACTION_ALLOW,
 				   0, NULL);
 	if (IS_ERR(vport->egress.allowed_vlan)) {
@@ -1442,13 +1425,9 @@ static int esw_vport_egress_config(struct mlx5_eswitch *esw,
 	}
 
 	/* Drop others rule (star rule) */
-	memset(match_c, 0, MLX5_ST_SZ_BYTES(fte_match_param));
-	memset(match_v, 0, MLX5_ST_SZ_BYTES(fte_match_param));
+	memset(spec, 0, sizeof(*spec));
 	vport->egress.drop_rule =
-		mlx5_add_flow_rule(vport->egress.acl,
-				   0,
-				   match_c,
-				   match_v,
+		mlx5_add_flow_rule(vport->egress.acl, spec,
 				   MLX5_FLOW_CONTEXT_ACTION_DROP,
 				   0, NULL);
 	if (IS_ERR(vport->egress.drop_rule)) {
@@ -1458,8 +1437,7 @@ static int esw_vport_egress_config(struct mlx5_eswitch *esw,
 		vport->egress.drop_rule = NULL;
 	}
 out:
-	kfree(match_v);
-	kfree(match_c);
+	kvfree(spec);
 	return err;
 }
 

drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c

@@ -43,37 +43,35 @@ mlx5_eswitch_add_send_to_vport_rule(struct mlx5_eswitch *esw, int vport, u32 sqn
 {
 	struct mlx5_flow_destination dest;
 	struct mlx5_flow_rule *flow_rule;
-	int match_header = MLX5_MATCH_MISC_PARAMETERS;
-	u32 *match_v, *match_c;
+	struct mlx5_flow_spec *spec;
 	void *misc;
 
-	match_v = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	match_c = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	if (!match_v || !match_c) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		esw_warn(esw->dev, "FDB: Failed to alloc match parameters\n");
 		flow_rule = ERR_PTR(-ENOMEM);
 		goto out;
 	}
 
-	misc = MLX5_ADDR_OF(fte_match_param, match_v, misc_parameters);
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
 	MLX5_SET(fte_match_set_misc, misc, source_sqn, sqn);
 	MLX5_SET(fte_match_set_misc, misc, source_port, 0x0); /* source vport is 0 */
 
-	misc = MLX5_ADDR_OF(fte_match_param, match_c, misc_parameters);
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
 	MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_sqn);
 	MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
 
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS;
 	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
 	dest.vport_num = vport;
 
-	flow_rule = mlx5_add_flow_rule(esw->fdb_table.fdb, match_header, match_c,
-				       match_v, MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
+	flow_rule = mlx5_add_flow_rule(esw->fdb_table.fdb, spec,
+				       MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
 				       0, &dest);
 	if (IS_ERR(flow_rule))
 		esw_warn(esw->dev, "FDB: Failed to add send to vport rule err %ld\n", PTR_ERR(flow_rule));
 out:
-	kfree(match_v);
-	kfree(match_c);
+	kvfree(spec);
 	return flow_rule;
 }
 
@@ -138,12 +136,11 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
 {
 	struct mlx5_flow_destination dest;
 	struct mlx5_flow_rule *flow_rule = NULL;
-	u32 *match_v, *match_c;
+	struct mlx5_flow_spec *spec;
 	int err = 0;
 
-	match_v = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	match_c = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	if (!match_v || !match_c) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		esw_warn(esw->dev, "FDB: Failed to alloc match parameters\n");
 		err = -ENOMEM;
 		goto out;
@@ -152,8 +149,9 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
 	dest.type = MLX5_FLOW_DESTINATION_TYPE_VPORT;
 	dest.vport_num = 0;
 
-	flow_rule = mlx5_add_flow_rule(esw->fdb_table.fdb, 0, match_c, match_v,
-				       MLX5_FLOW_CONTEXT_ACTION_FWD_DEST, 0, &dest);
+	flow_rule = mlx5_add_flow_rule(esw->fdb_table.fdb, spec,
+				       MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
+				       0, &dest);
 	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);
@@ -162,8 +160,7 @@ static int esw_add_fdb_miss_rule(struct mlx5_eswitch *esw)
 
 	esw->fdb_table.offloads.miss_rule = flow_rule;
 out:
-	kfree(match_v);
-	kfree(match_c);
+	kvfree(spec);
 	return err;
 }
 
@@ -351,29 +348,28 @@ mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, int vport, u32 tirn)
 {
 	struct mlx5_flow_destination dest;
 	struct mlx5_flow_rule *flow_rule;
-	int match_header = MLX5_MATCH_MISC_PARAMETERS;
-	u32 *match_v, *match_c;
+	struct mlx5_flow_spec *spec;
 	void *misc;
 
-	match_v = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	match_c = kzalloc(MLX5_ST_SZ_BYTES(fte_match_param), GFP_KERNEL);
-	if (!match_v || !match_c) {
+	spec = mlx5_vzalloc(sizeof(*spec));
+	if (!spec) {
 		esw_warn(esw->dev, "Failed to alloc match parameters\n");
 		flow_rule = ERR_PTR(-ENOMEM);
 		goto out;
 	}
 
-	misc = MLX5_ADDR_OF(fte_match_param, match_v, misc_parameters);
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_value, misc_parameters);
 	MLX5_SET(fte_match_set_misc, misc, source_port, vport);
 
-	misc = MLX5_ADDR_OF(fte_match_param, match_c, misc_parameters);
+	misc = MLX5_ADDR_OF(fte_match_param, spec->match_criteria, misc_parameters);
 	MLX5_SET_TO_ONES(fte_match_set_misc, misc, source_port);
 
+	spec->match_criteria_enable = MLX5_MATCH_MISC_PARAMETERS;
 	dest.type = MLX5_FLOW_DESTINATION_TYPE_TIR;
 	dest.tir_num = tirn;
 
-	flow_rule = mlx5_add_flow_rule(esw->offloads.ft_offloads, match_header, match_c,
-				       match_v, MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
+	flow_rule = mlx5_add_flow_rule(esw->offloads.ft_offloads, spec,
+				       MLX5_FLOW_CONTEXT_ACTION_FWD_DEST,
 				       0, &dest);
 	if (IS_ERR(flow_rule)) {
 		esw_warn(esw->dev, "fs offloads: Failed to add vport rx rule err %ld\n", PTR_ERR(flow_rule));
@@ -381,8 +377,7 @@ mlx5_eswitch_create_vport_rx_rule(struct mlx5_eswitch *esw, int vport, u32 tirn)
 	}
 
 out:
-	kfree(match_v);
-	kfree(match_c);
+	kvfree(spec);
 	return flow_rule;
 }
 

drivers/net/ethernet/mellanox/mlx5/core/fs_core.c

@@ -67,13 +67,21 @@
 #define FS_REQUIRED_CAPS(...) {.arr_sz = INIT_CAPS_ARRAY_SIZE(__VA_ARGS__), \
 			       .caps = (long[]) {__VA_ARGS__} }
 
+#define FS_CHAINING_CAPS  FS_REQUIRED_CAPS(FS_CAP(flow_table_properties_nic_receive.flow_modify_en), \
+					   FS_CAP(flow_table_properties_nic_receive.modify_root), \
+					   FS_CAP(flow_table_properties_nic_receive.identified_miss_table_mode), \
+					   FS_CAP(flow_table_properties_nic_receive.flow_table_modify))
+
 #define LEFTOVERS_NUM_LEVELS 1
 #define LEFTOVERS_NUM_PRIOS 1
 
 #define BY_PASS_PRIO_NUM_LEVELS 1
-#define BY_PASS_MIN_LEVEL (KERNEL_MIN_LEVEL + MLX5_BY_PASS_NUM_PRIOS +\
+#define BY_PASS_MIN_LEVEL (ETHTOOL_MIN_LEVEL + MLX5_BY_PASS_NUM_PRIOS +\
 			   LEFTOVERS_NUM_PRIOS)
 
+#define ETHTOOL_PRIO_NUM_LEVELS 1
+#define ETHTOOL_NUM_PRIOS 10
+#define ETHTOOL_MIN_LEVEL (KERNEL_MIN_LEVEL + ETHTOOL_NUM_PRIOS)
 /* Vlan, mac, ttc, aRFS */
 #define KERNEL_NIC_PRIO_NUM_LEVELS 4
 #define KERNEL_NIC_NUM_PRIOS 1
@@ -103,27 +111,24 @@ static struct init_tree_node {
 	int num_levels;
 } root_fs = {
 	.type = FS_TYPE_NAMESPACE,
-	.ar_size = 5,
+	.ar_size = 6,
 	.children = (struct init_tree_node[]) {
 		ADD_PRIO(0, BY_PASS_MIN_LEVEL, 0,
-			 FS_REQUIRED_CAPS(FS_CAP(flow_table_properties_nic_receive.flow_modify_en),
-					  FS_CAP(flow_table_properties_nic_receive.modify_root),
-					  FS_CAP(flow_table_properties_nic_receive.identified_miss_table_mode),
-					  FS_CAP(flow_table_properties_nic_receive.flow_table_modify)),
+			 FS_CHAINING_CAPS,
 			 ADD_NS(ADD_MULTIPLE_PRIO(MLX5_BY_PASS_NUM_PRIOS,
 						  BY_PASS_PRIO_NUM_LEVELS))),
 		ADD_PRIO(0, OFFLOADS_MIN_LEVEL, 0, {},
 			 ADD_NS(ADD_MULTIPLE_PRIO(OFFLOADS_NUM_PRIOS, OFFLOADS_MAX_FT))),
-
+		ADD_PRIO(0, ETHTOOL_MIN_LEVEL, 0,
+			 FS_CHAINING_CAPS,
+			 ADD_NS(ADD_MULTIPLE_PRIO(ETHTOOL_NUM_PRIOS,
+						  ETHTOOL_PRIO_NUM_LEVELS))),
 		ADD_PRIO(0, KERNEL_MIN_LEVEL, 0, {},
 			 ADD_NS(ADD_MULTIPLE_PRIO(1, 1),
 				ADD_MULTIPLE_PRIO(KERNEL_NIC_NUM_PRIOS,
 						  KERNEL_NIC_PRIO_NUM_LEVELS))),
 		ADD_PRIO(0, BY_PASS_MIN_LEVEL, 0,
-			 FS_REQUIRED_CAPS(FS_CAP(flow_table_properties_nic_receive.flow_modify_en),
-					  FS_CAP(flow_table_properties_nic_receive.modify_root),
-					  FS_CAP(flow_table_properties_nic_receive.identified_miss_table_mode),
-					  FS_CAP(flow_table_properties_nic_receive.flow_table_modify)),
+			 FS_CHAINING_CAPS,
 			 ADD_NS(ADD_MULTIPLE_PRIO(LEFTOVERS_NUM_PRIOS, LEFTOVERS_NUM_LEVELS))),
 		ADD_PRIO(0, ANCHOR_MIN_LEVEL, 0, {},
 			 ADD_NS(ADD_MULTIPLE_PRIO(ANCHOR_NUM_PRIOS, ANCHOR_NUM_LEVELS))),
@@ -1160,9 +1165,7 @@ static bool dest_is_valid(struct mlx5_flow_destination *dest,
 
 static struct mlx5_flow_rule *
 _mlx5_add_flow_rule(struct mlx5_flow_table *ft,
-		    u8 match_criteria_enable,
-		    u32 *match_criteria,
-		    u32 *match_value,
+		   struct mlx5_flow_spec *spec,
 		    u32 action,
 		    u32 flow_tag,
 		    struct mlx5_flow_destination *dest)
@@ -1176,22 +1179,23 @@ _mlx5_add_flow_rule(struct mlx5_flow_table *ft,
 	nested_lock_ref_node(&ft->node, FS_MUTEX_GRANDPARENT);
 	fs_for_each_fg(g, ft)
 		if (compare_match_criteria(g->mask.match_criteria_enable,
-					   match_criteria_enable,
+					   spec->match_criteria_enable,
 					   g->mask.match_criteria,
-					   match_criteria)) {
-			rule = add_rule_fg(g, match_value,
+					   spec->match_criteria)) {
+			rule = add_rule_fg(g, spec->match_value,
 					   action, flow_tag, dest);
 			if (!IS_ERR(rule) || PTR_ERR(rule) != -ENOSPC)
 				goto unlock;
 		}
 
-	g = create_autogroup(ft, match_criteria_enable, match_criteria);
+	g = create_autogroup(ft, spec->match_criteria_enable,
+			     spec->match_criteria);
 	if (IS_ERR(g)) {
 		rule = (void *)g;
 		goto unlock;
 	}
 
-	rule = add_rule_fg(g, match_value,
+	rule = add_rule_fg(g, spec->match_value,
 			   action, flow_tag, dest);
 	if (IS_ERR(rule)) {
 		/* Remove assumes refcount > 0 and autogroup creates a group
@@ -1215,9 +1219,7 @@ static bool fwd_next_prio_supported(struct mlx5_flow_table *ft)
 
 struct mlx5_flow_rule *
 mlx5_add_flow_rule(struct mlx5_flow_table *ft,
-		   u8 match_criteria_enable,
-		   u32 *match_criteria,
-		   u32 *match_value,
+		   struct mlx5_flow_spec *spec,
 		   u32 action,
 		   u32 flow_tag,
 		   struct mlx5_flow_destination *dest)
@@ -1248,8 +1250,7 @@ mlx5_add_flow_rule(struct mlx5_flow_table *ft,
 		}
 	}
 
-	rule =	_mlx5_add_flow_rule(ft, match_criteria_enable, match_criteria,
-				    match_value, action, flow_tag, dest);
+	rule = _mlx5_add_flow_rule(ft, spec, action, flow_tag, dest);
 
 	if (sw_action == MLX5_FLOW_CONTEXT_ACTION_FWD_NEXT_PRIO) {
 		if (!IS_ERR_OR_NULL(rule) &&
@@ -1367,41 +1368,47 @@ void mlx5_destroy_flow_group(struct mlx5_flow_group *fg)
 struct mlx5_flow_namespace *mlx5_get_flow_namespace(struct mlx5_core_dev *dev,
 						    enum mlx5_flow_namespace_type type)
 {
-	struct mlx5_flow_root_namespace *root_ns = dev->priv.root_ns;
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+	struct mlx5_flow_root_namespace *root_ns;
 	int prio;
 	struct fs_prio *fs_prio;
 	struct mlx5_flow_namespace *ns;
 
-	if (!root_ns)
+	if (!steering)
 		return NULL;
 
 	switch (type) {
 	case MLX5_FLOW_NAMESPACE_BYPASS:
 	case MLX5_FLOW_NAMESPACE_OFFLOADS:
+	case MLX5_FLOW_NAMESPACE_ETHTOOL:
 	case MLX5_FLOW_NAMESPACE_KERNEL:
 	case MLX5_FLOW_NAMESPACE_LEFTOVERS:
 	case MLX5_FLOW_NAMESPACE_ANCHOR:
 		prio = type;
 		break;
 	case MLX5_FLOW_NAMESPACE_FDB:
-		if (dev->priv.fdb_root_ns)
-			return &dev->priv.fdb_root_ns->ns;
+		if (steering->fdb_root_ns)
+			return &steering->fdb_root_ns->ns;
 		else
 			return NULL;
 	case MLX5_FLOW_NAMESPACE_ESW_EGRESS:
-		if (dev->priv.esw_egress_root_ns)
-			return &dev->priv.esw_egress_root_ns->ns;
+		if (steering->esw_egress_root_ns)
+			return &steering->esw_egress_root_ns->ns;
 		else
 			return NULL;
 	case MLX5_FLOW_NAMESPACE_ESW_INGRESS:
-		if (dev->priv.esw_ingress_root_ns)
-			return &dev->priv.esw_ingress_root_ns->ns;
+		if (steering->esw_ingress_root_ns)
+			return &steering->esw_ingress_root_ns->ns;
 		else
 			return NULL;
 	default:
 		return NULL;
 	}
 
+	root_ns = steering->root_ns;
+	if (!root_ns)
+		return NULL;
+
 	fs_prio = find_prio(&root_ns->ns, prio);
 	if (!fs_prio)
 		return NULL;
@@ -1487,13 +1494,13 @@ static bool has_required_caps(struct mlx5_core_dev *dev, struct node_caps *caps)
 	return true;
 }
 
-static int init_root_tree_recursive(struct mlx5_core_dev *dev,
+static int init_root_tree_recursive(struct mlx5_flow_steering *steering,
 				    struct init_tree_node *init_node,
 				    struct fs_node *fs_parent_node,
 				    struct init_tree_node *init_parent_node,
 				    int prio)
 {
-	int max_ft_level = MLX5_CAP_FLOWTABLE(dev,
+	int max_ft_level = MLX5_CAP_FLOWTABLE(steering->dev,
 					      flow_table_properties_nic_receive.
 					      max_ft_level);
 	struct mlx5_flow_namespace *fs_ns;
@@ -1504,7 +1511,7 @@ static int init_root_tree_recursive(struct mlx5_core_dev *dev,
 
 	if (init_node->type == FS_TYPE_PRIO) {
 		if ((init_node->min_ft_level > max_ft_level) ||
-		    !has_required_caps(dev, &init_node->caps))
+		    !has_required_caps(steering->dev, &init_node->caps))
 			return 0;
 
 		fs_get_obj(fs_ns, fs_parent_node);
@@ -1525,7 +1532,7 @@ static int init_root_tree_recursive(struct mlx5_core_dev *dev,
 	}
 	prio = 0;
 	for (i = 0; i < init_node->ar_size; i++) {
-		err = init_root_tree_recursive(dev, &init_node->children[i],
+		err = init_root_tree_recursive(steering, &init_node->children[i],
 					       base, init_node, prio);
 		if (err)
 			return err;
@@ -1538,7 +1545,7 @@ static int init_root_tree_recursive(struct mlx5_core_dev *dev,
 	return 0;
 }
 
-static int init_root_tree(struct mlx5_core_dev *dev,
+static int init_root_tree(struct mlx5_flow_steering *steering,
 			  struct init_tree_node *init_node,
 			  struct fs_node *fs_parent_node)
 {
@@ -1548,7 +1555,7 @@ static int init_root_tree(struct mlx5_core_dev *dev,
 
 	fs_get_obj(fs_ns, fs_parent_node);
 	for (i = 0; i < init_node->ar_size; i++) {
-		err = init_root_tree_recursive(dev, &init_node->children[i],
+		err = init_root_tree_recursive(steering, &init_node->children[i],
 					       &fs_ns->node,
 					       init_node, i);
 		if (err)
@@ -1557,7 +1564,7 @@ static int init_root_tree(struct mlx5_core_dev *dev,
 	return 0;
 }
 
-static struct mlx5_flow_root_namespace *create_root_ns(struct mlx5_core_dev *dev,
+static struct mlx5_flow_root_namespace *create_root_ns(struct mlx5_flow_steering *steering,
 						       enum fs_flow_table_type
 						       table_type)
 {
@@ -1569,7 +1576,7 @@ static struct mlx5_flow_root_namespace *create_root_ns(struct mlx5_core_dev *dev
 	if (!root_ns)
 		return NULL;
 
-	root_ns->dev = dev;
+	root_ns->dev = steering->dev;
 	root_ns->table_type = table_type;
 
 	ns = &root_ns->ns;
@@ -1624,212 +1631,126 @@ static void set_prio_attrs(struct mlx5_flow_root_namespace *root_ns)
 #define ANCHOR_PRIO 0
 #define ANCHOR_SIZE 1
 #define ANCHOR_LEVEL 0
-static int create_anchor_flow_table(struct mlx5_core_dev
-							*dev)
+static int create_anchor_flow_table(struct mlx5_flow_steering *steering)
 {
 	struct mlx5_flow_namespace *ns = NULL;
 	struct mlx5_flow_table *ft;
 
-	ns = mlx5_get_flow_namespace(dev, MLX5_FLOW_NAMESPACE_ANCHOR);
+	ns = mlx5_get_flow_namespace(steering->dev, MLX5_FLOW_NAMESPACE_ANCHOR);
 	if (!ns)
 		return -EINVAL;
 	ft = mlx5_create_flow_table(ns, ANCHOR_PRIO, ANCHOR_SIZE, ANCHOR_LEVEL);
 	if (IS_ERR(ft)) {
-		mlx5_core_err(dev, "Failed to create last anchor flow table");
+		mlx5_core_err(steering->dev, "Failed to create last anchor flow table");
 		return PTR_ERR(ft);
 	}
 	return 0;
 }
 
-static int init_root_ns(struct mlx5_core_dev *dev)
+static int init_root_ns(struct mlx5_flow_steering *steering)
 {
 
-	dev->priv.root_ns = create_root_ns(dev, FS_FT_NIC_RX);
-	if (IS_ERR_OR_NULL(dev->priv.root_ns))
+	steering->root_ns = create_root_ns(steering, FS_FT_NIC_RX);
+	if (IS_ERR_OR_NULL(steering->root_ns))
 		goto cleanup;
 
-	if (init_root_tree(dev, &root_fs, &dev->priv.root_ns->ns.node))
+	if (init_root_tree(steering, &root_fs, &steering->root_ns->ns.node))
 		goto cleanup;
 
-	set_prio_attrs(dev->priv.root_ns);
+	set_prio_attrs(steering->root_ns);
 
-	if (create_anchor_flow_table(dev))
+	if (create_anchor_flow_table(steering))
 		goto cleanup;
 
 	return 0;
 
 cleanup:
-	mlx5_cleanup_fs(dev);
+	mlx5_cleanup_fs(steering->dev);
 	return -ENOMEM;
 }
 
-static void cleanup_single_prio_root_ns(struct mlx5_core_dev *dev,
-					struct mlx5_flow_root_namespace *root_ns)
+static void clean_tree(struct fs_node *node)
 {
-	struct fs_node *prio;
-
-	if (!root_ns)
-		return;
+	if (node) {
+		struct fs_node *iter;
+		struct fs_node *temp;
 
-	if (!list_empty(&root_ns->ns.node.children)) {
-		prio = list_first_entry(&root_ns->ns.node.children,
-					struct fs_node,
-				 list);
-		if (tree_remove_node(prio))
-			mlx5_core_warn(dev,
-				       "Flow steering priority wasn't destroyed, refcount > 1\n");
+		list_for_each_entry_safe(iter, temp, &node->children, list)
+			clean_tree(iter);
+		tree_remove_node(node);
 	}
-	if (tree_remove_node(&root_ns->ns.node))
-		mlx5_core_warn(dev,
-			       "Flow steering namespace wasn't destroyed, refcount > 1\n");
-	root_ns = NULL;
 }
 
-static void destroy_flow_tables(struct fs_prio *prio)
+static void cleanup_root_ns(struct mlx5_flow_root_namespace *root_ns)
 {
-	struct mlx5_flow_table *iter;
-	struct mlx5_flow_table *tmp;
-
-	fs_for_each_ft_safe(iter, tmp, prio)
-		mlx5_destroy_flow_table(iter);
-}
-
-static void cleanup_root_ns(struct mlx5_core_dev *dev)
-{
-	struct mlx5_flow_root_namespace *root_ns = dev->priv.root_ns;
-	struct fs_prio *iter_prio;
-
-	if (!MLX5_CAP_GEN(dev, nic_flow_table))
-		return;
-
 	if (!root_ns)
 		return;
 
-	/* stage 1 */
-	fs_for_each_prio(iter_prio, &root_ns->ns) {
-		struct fs_node *node;
-		struct mlx5_flow_namespace *iter_ns;
-
-		fs_for_each_ns_or_ft(node, iter_prio) {
-			if (node->type == FS_TYPE_FLOW_TABLE)
-				continue;
-			fs_get_obj(iter_ns, node);
-			while (!list_empty(&iter_ns->node.children)) {
-				struct fs_prio *obj_iter_prio2;
-				struct fs_node *iter_prio2 =
-					list_first_entry(&iter_ns->node.children,
-							 struct fs_node,
-							 list);
-
-				fs_get_obj(obj_iter_prio2, iter_prio2);
-				destroy_flow_tables(obj_iter_prio2);
-				if (tree_remove_node(iter_prio2)) {
-					mlx5_core_warn(dev,
-						       "Priority %d wasn't destroyed, refcount > 1\n",
-						       obj_iter_prio2->prio);
-					return;
-				}
-			}
-		}
-	}
-
-	/* stage 2 */
-	fs_for_each_prio(iter_prio, &root_ns->ns) {
-		while (!list_empty(&iter_prio->node.children)) {
-			struct fs_node *iter_ns =
-				list_first_entry(&iter_prio->node.children,
-						 struct fs_node,
-						 list);
-			if (tree_remove_node(iter_ns)) {
-				mlx5_core_warn(dev,
-					       "Namespace wasn't destroyed, refcount > 1\n");
-				return;
-			}
-		}
-	}
-
-	/* stage 3 */
-	while (!list_empty(&root_ns->ns.node.children)) {
-		struct fs_prio *obj_prio_node;
-		struct fs_node *prio_node =
-			list_first_entry(&root_ns->ns.node.children,
-					 struct fs_node,
-					 list);
-
-		fs_get_obj(obj_prio_node, prio_node);
-		if (tree_remove_node(prio_node)) {
-			mlx5_core_warn(dev,
-				       "Priority %d wasn't destroyed, refcount > 1\n",
-				       obj_prio_node->prio);
-			return;
-		}
-	}
-
-	if (tree_remove_node(&root_ns->ns.node)) {
-		mlx5_core_warn(dev,
-			       "root namespace wasn't destroyed, refcount > 1\n");
-		return;
-	}
-
-	dev->priv.root_ns = NULL;
+	clean_tree(&root_ns->ns.node);
 }
 
 void mlx5_cleanup_fs(struct mlx5_core_dev *dev)
 {
+	struct mlx5_flow_steering *steering = dev->priv.steering;
+
 	if (MLX5_CAP_GEN(dev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
 		return;
 
-	cleanup_root_ns(dev);
-	cleanup_single_prio_root_ns(dev, dev->priv.fdb_root_ns);
-	cleanup_single_prio_root_ns(dev, dev->priv.esw_egress_root_ns);
-	cleanup_single_prio_root_ns(dev, dev->priv.esw_ingress_root_ns);
+	cleanup_root_ns(steering->root_ns);
+	cleanup_root_ns(steering->esw_egress_root_ns);
+	cleanup_root_ns(steering->esw_ingress_root_ns);
+	cleanup_root_ns(steering->fdb_root_ns);
 	mlx5_cleanup_fc_stats(dev);
+	kfree(steering);
 }
 
-static int init_fdb_root_ns(struct mlx5_core_dev *dev)
+static int init_fdb_root_ns(struct mlx5_flow_steering *steering)
 {
 	struct fs_prio *prio;
 
-	dev->priv.fdb_root_ns = create_root_ns(dev, FS_FT_FDB);
-	if (!dev->priv.fdb_root_ns)
+	steering->fdb_root_ns = create_root_ns(steering, FS_FT_FDB);
+	if (!steering->fdb_root_ns)
 		return -ENOMEM;
 
 	/* Create single prio */
-	prio = fs_create_prio(&dev->priv.fdb_root_ns->ns, 0, 1);
+	prio = fs_create_prio(&steering->fdb_root_ns->ns, 0, 1);
 	if (IS_ERR(prio)) {
-		cleanup_single_prio_root_ns(dev, dev->priv.fdb_root_ns);
+		cleanup_root_ns(steering->fdb_root_ns);
+		steering->fdb_root_ns = NULL;
 		return PTR_ERR(prio);
 	} else {
 		return 0;
 	}
 }
 
-static int init_egress_acl_root_ns(struct mlx5_core_dev *dev)
+static int init_ingress_acl_root_ns(struct mlx5_flow_steering *steering)
 {
 	struct fs_prio *prio;
 
-	dev->priv.esw_egress_root_ns = create_root_ns(dev, FS_FT_ESW_EGRESS_ACL);
-	if (!dev->priv.esw_egress_root_ns)
+	steering->esw_egress_root_ns = create_root_ns(steering, FS_FT_ESW_EGRESS_ACL);
+	if (!steering->esw_egress_root_ns)
 		return -ENOMEM;
 
 	/* create 1 prio*/
-	prio = fs_create_prio(&dev->priv.esw_egress_root_ns->ns, 0, MLX5_TOTAL_VPORTS(dev));
+	prio = fs_create_prio(&steering->esw_egress_root_ns->ns, 0,
+			      MLX5_TOTAL_VPORTS(steering->dev));
 	if (IS_ERR(prio))
 		return PTR_ERR(prio);
 	else
 		return 0;
 }
 
-static int init_ingress_acl_root_ns(struct mlx5_core_dev *dev)
+static int init_egress_acl_root_ns(struct mlx5_flow_steering *steering)
 {
 	struct fs_prio *prio;
 
-	dev->priv.esw_ingress_root_ns = create_root_ns(dev, FS_FT_ESW_INGRESS_ACL);
-	if (!dev->priv.esw_ingress_root_ns)
+	steering->esw_ingress_root_ns = create_root_ns(steering, FS_FT_ESW_INGRESS_ACL);
+	if (!steering->esw_ingress_root_ns)
 		return -ENOMEM;
 
 	/* create 1 prio*/
-	prio = fs_create_prio(&dev->priv.esw_ingress_root_ns->ns, 0, MLX5_TOTAL_VPORTS(dev));
+	prio = fs_create_prio(&steering->esw_ingress_root_ns->ns, 0,
+			      MLX5_TOTAL_VPORTS(steering->dev));
 	if (IS_ERR(prio))
 		return PTR_ERR(prio);
 	else
@@ -1838,6 +1759,7 @@ static int init_ingress_acl_root_ns(struct mlx5_core_dev *dev)
 
 int mlx5_init_fs(struct mlx5_core_dev *dev)
 {
+	struct mlx5_flow_steering *steering;
 	int err = 0;
 
 	if (MLX5_CAP_GEN(dev, port_type) != MLX5_CAP_PORT_TYPE_ETH)
@@ -1847,26 +1769,32 @@ int mlx5_init_fs(struct mlx5_core_dev *dev)
 	if (err)
 		return err;
 
+	steering = kzalloc(sizeof(*steering), GFP_KERNEL);
+	if (!steering)
+		return -ENOMEM;
+	steering->dev = dev;
+	dev->priv.steering = steering;
+
 	if (MLX5_CAP_GEN(dev, nic_flow_table) &&
 	    MLX5_CAP_FLOWTABLE_NIC_RX(dev, ft_support)) {
-		err = init_root_ns(dev);
+		err = init_root_ns(steering);
 		if (err)
 			goto err;
 	}
 
 	if (MLX5_CAP_GEN(dev, eswitch_flow_table)) {
 		if (MLX5_CAP_ESW_FLOWTABLE_FDB(dev, ft_support)) {
-			err = init_fdb_root_ns(dev);
+			err = init_fdb_root_ns(steering);
 			if (err)
 				goto err;
 		}
 		if (MLX5_CAP_ESW_EGRESS_ACL(dev, ft_support)) {
-			err = init_egress_acl_root_ns(dev);
+			err = init_egress_acl_root_ns(steering);
 			if (err)
 				goto err;
 		}
 		if (MLX5_CAP_ESW_INGRESS_ACL(dev, ft_support)) {
-			err = init_ingress_acl_root_ns(dev);
+			err = init_ingress_acl_root_ns(steering);
 			if (err)
 				goto err;
 		}

drivers/net/ethernet/mellanox/mlx5/core/fs_core.h

@@ -55,6 +55,14 @@ enum fs_fte_status {
 	FS_FTE_STATUS_EXISTING = 1UL << 0,
 };
 
+struct mlx5_flow_steering {
+	struct mlx5_core_dev *dev;
+	struct mlx5_flow_root_namespace *root_ns;
+	struct mlx5_flow_root_namespace *fdb_root_ns;
+	struct mlx5_flow_root_namespace *esw_egress_root_ns;
+	struct mlx5_flow_root_namespace *esw_ingress_root_ns;
+};
+
 struct fs_node {
 	struct list_head	list;
 	struct list_head	children;

include/linux/mlx5/driver.h

@@ -550,14 +550,10 @@ struct mlx5_priv {
 	struct list_head        ctx_list;
 	spinlock_t              ctx_lock;
 
+	struct mlx5_flow_steering *steering;
 	struct mlx5_eswitch     *eswitch;
 	struct mlx5_core_sriov	sriov;
 	unsigned long		pci_dev_data;
-	struct mlx5_flow_root_namespace *root_ns;
-	struct mlx5_flow_root_namespace *fdb_root_ns;
-	struct mlx5_flow_root_namespace *esw_egress_root_ns;
-	struct mlx5_flow_root_namespace *esw_ingress_root_ns;
-
 	struct mlx5_fc_stats		fc_stats;
 	struct mlx5_rl_table            rl_table;
 };

include/linux/mlx5/fs.h

@@ -55,6 +55,7 @@ static inline void build_leftovers_ft_param(int *priority,
 enum mlx5_flow_namespace_type {
 	MLX5_FLOW_NAMESPACE_BYPASS,
 	MLX5_FLOW_NAMESPACE_OFFLOADS,
+	MLX5_FLOW_NAMESPACE_ETHTOOL,
 	MLX5_FLOW_NAMESPACE_KERNEL,
 	MLX5_FLOW_NAMESPACE_LEFTOVERS,
 	MLX5_FLOW_NAMESPACE_ANCHOR,
@@ -68,6 +69,12 @@ struct mlx5_flow_group;
 struct mlx5_flow_rule;
 struct mlx5_flow_namespace;
 
+struct mlx5_flow_spec {
+	u8   match_criteria_enable;
+	u32  match_criteria[MLX5_ST_SZ_DW(fte_match_param)];
+	u32  match_value[MLX5_ST_SZ_DW(fte_match_param)];
+};
+
 struct mlx5_flow_destination {
 	enum mlx5_flow_destination_type	type;
 	union {
@@ -116,9 +123,7 @@ void mlx5_destroy_flow_group(struct mlx5_flow_group *fg);
  */
 struct mlx5_flow_rule *
 mlx5_add_flow_rule(struct mlx5_flow_table *ft,
-		   u8 match_criteria_enable,
-		   u32 *match_criteria,
-		   u32 *match_value,
+		   struct mlx5_flow_spec *spec,
 		   u32 action,
 		   u32 flow_tag,
 		   struct mlx5_flow_destination *dest);