Merge branch 'hns3-flow-director'

Huazhong Tan says:

====================
net: hns3: refactor and new features for flow director

This patchset refactor some functions and add some new features for
flow director.

patch 1~3: refactor large functions
patch 4, 7: add traffic class and user-def field support for ethtool
patch 5: refactor flow director configuration
patch 6: clean up for hns3_del_all_fd_entries()

change log:
V1->V2: modifies patch 5 as Jakub suggested, keep configuring
	ethtool/tc flower rules synchronously while aRFS
	asynchronously.
	changes the usecnt of user-def rule checking in patch 7.
	removes previous patches 8 and 9 from this series, since
	there are issues that need further discussion.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/hisilicon/hns3/hnae3.h

@@ -612,8 +612,6 @@ struct hnae3_ae_ops {
 			    struct ethtool_rxnfc *cmd);
 	int (*del_fd_entry)(struct hnae3_handle *handle,
 			    struct ethtool_rxnfc *cmd);
-	void (*del_all_fd_entries)(struct hnae3_handle *handle,
-				   bool clear_list);
 	int (*get_fd_rule_cnt)(struct hnae3_handle *handle,
 			       struct ethtool_rxnfc *cmd);
 	int (*get_fd_rule_info)(struct hnae3_handle *handle,

drivers/net/ethernet/hisilicon/hns3/hns3_enet.c

@@ -4143,14 +4143,6 @@ static void hns3_uninit_phy(struct net_device *netdev)
 		h->ae_algo->ops->mac_disconnect_phy(h);
 }
 
-static void hns3_del_all_fd_rules(struct net_device *netdev, bool clear_list)
-{
-	struct hnae3_handle *h = hns3_get_handle(netdev);
-
-	if (h->ae_algo->ops->del_all_fd_entries)
-		h->ae_algo->ops->del_all_fd_entries(h, clear_list);
-}
-
 static int hns3_client_start(struct hnae3_handle *handle)
 {
 	if (!handle->ae_algo->ops->client_start)
@@ -4337,8 +4329,6 @@ static void hns3_client_uninit(struct hnae3_handle *handle, bool reset)
 
 	hns3_nic_uninit_irq(priv);
 
-	hns3_del_all_fd_rules(netdev, true);
-
 	hns3_clear_all_ring(handle, true);
 
 	hns3_nic_uninit_vector_data(priv);

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h

@@ -243,6 +243,7 @@ enum hclge_opcode_type {
 	HCLGE_OPC_FD_KEY_CONFIG		= 0x1202,
 	HCLGE_OPC_FD_TCAM_OP		= 0x1203,
 	HCLGE_OPC_FD_AD_OP		= 0x1204,
+	HCLGE_OPC_FD_USER_DEF_OP	= 0x1207,
 
 	/* MDIO command */
 	HCLGE_OPC_MDIO_CONFIG		= 0x1900,
@@ -1082,6 +1083,19 @@ struct hclge_fd_ad_config_cmd {
 	u8 rsv2[8];
 };
 
+#define HCLGE_FD_USER_DEF_OFT_S		0
+#define HCLGE_FD_USER_DEF_OFT_M		GENMASK(14, 0)
+#define HCLGE_FD_USER_DEF_EN_B		15
+struct hclge_fd_user_def_cfg_cmd {
+	__le16 ol2_cfg;
+	__le16 l2_cfg;
+	__le16 ol3_cfg;
+	__le16 l3_cfg;
+	__le16 ol4_cfg;
+	__le16 l4_cfg;
+	u8 rsv[12];
+};
+
 struct hclge_get_m7_bd_cmd {
 	__le32 bd_num;
 	u8 rsv[20];

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c

@@ -62,7 +62,7 @@ static void hclge_sync_vlan_filter(struct hclge_dev *hdev);
 static int hclge_reset_ae_dev(struct hnae3_ae_dev *ae_dev);
 static bool hclge_get_hw_reset_stat(struct hnae3_handle *handle);
 static void hclge_rfs_filter_expire(struct hclge_dev *hdev);
-static void hclge_clear_arfs_rules(struct hnae3_handle *handle);
+static int hclge_clear_arfs_rules(struct hclge_dev *hdev);
 static enum hnae3_reset_type hclge_get_reset_level(struct hnae3_ae_dev *ae_dev,
 						   unsigned long *addr);
 static int hclge_set_default_loopback(struct hclge_dev *hdev);
@@ -70,6 +70,7 @@ static int hclge_set_default_loopback(struct hclge_dev *hdev);
 static void hclge_sync_mac_table(struct hclge_dev *hdev);
 static void hclge_restore_hw_table(struct hclge_dev *hdev);
 static void hclge_sync_promisc_mode(struct hclge_dev *hdev);
+static void hclge_sync_fd_table(struct hclge_dev *hdev);
 
 static struct hnae3_ae_algo ae_algo;
 
@@ -384,36 +385,62 @@ static const struct key_info meta_data_key_info[] = {
 };
 
 static const struct key_info tuple_key_info[] = {
-	{ OUTER_DST_MAC, 48},
-	{ OUTER_SRC_MAC, 48},
-	{ OUTER_VLAN_TAG_FST, 16},
-	{ OUTER_VLAN_TAG_SEC, 16},
-	{ OUTER_ETH_TYPE, 16},
-	{ OUTER_L2_RSV, 16},
-	{ OUTER_IP_TOS, 8},
-	{ OUTER_IP_PROTO, 8},
-	{ OUTER_SRC_IP, 32},
-	{ OUTER_DST_IP, 32},
-	{ OUTER_L3_RSV, 16},
-	{ OUTER_SRC_PORT, 16},
-	{ OUTER_DST_PORT, 16},
-	{ OUTER_L4_RSV, 32},
-	{ OUTER_TUN_VNI, 24},
-	{ OUTER_TUN_FLOW_ID, 8},
-	{ INNER_DST_MAC, 48},
-	{ INNER_SRC_MAC, 48},
-	{ INNER_VLAN_TAG_FST, 16},
-	{ INNER_VLAN_TAG_SEC, 16},
-	{ INNER_ETH_TYPE, 16},
-	{ INNER_L2_RSV, 16},
-	{ INNER_IP_TOS, 8},
-	{ INNER_IP_PROTO, 8},
-	{ INNER_SRC_IP, 32},
-	{ INNER_DST_IP, 32},
-	{ INNER_L3_RSV, 16},
-	{ INNER_SRC_PORT, 16},
-	{ INNER_DST_PORT, 16},
-	{ INNER_L4_RSV, 32},
+	{ OUTER_DST_MAC, 48, KEY_OPT_MAC, -1, -1 },
+	{ OUTER_SRC_MAC, 48, KEY_OPT_MAC, -1, -1 },
+	{ OUTER_VLAN_TAG_FST, 16, KEY_OPT_LE16, -1, -1 },
+	{ OUTER_VLAN_TAG_SEC, 16, KEY_OPT_LE16, -1, -1 },
+	{ OUTER_ETH_TYPE, 16, KEY_OPT_LE16, -1, -1 },
+	{ OUTER_L2_RSV, 16, KEY_OPT_LE16, -1, -1 },
+	{ OUTER_IP_TOS, 8, KEY_OPT_U8, -1, -1 },
+	{ OUTER_IP_PROTO, 8, KEY_OPT_U8, -1, -1 },
+	{ OUTER_SRC_IP, 32, KEY_OPT_IP, -1, -1 },
+	{ OUTER_DST_IP, 32, KEY_OPT_IP, -1, -1 },
+	{ OUTER_L3_RSV, 16, KEY_OPT_LE16, -1, -1 },
+	{ OUTER_SRC_PORT, 16, KEY_OPT_LE16, -1, -1 },
+	{ OUTER_DST_PORT, 16, KEY_OPT_LE16, -1, -1 },
+	{ OUTER_L4_RSV, 32, KEY_OPT_LE32, -1, -1 },
+	{ OUTER_TUN_VNI, 24, KEY_OPT_VNI, -1, -1 },
+	{ OUTER_TUN_FLOW_ID, 8, KEY_OPT_U8, -1, -1 },
+	{ INNER_DST_MAC, 48, KEY_OPT_MAC,
+	  offsetof(struct hclge_fd_rule, tuples.dst_mac),
+	  offsetof(struct hclge_fd_rule, tuples_mask.dst_mac) },
+	{ INNER_SRC_MAC, 48, KEY_OPT_MAC,
+	  offsetof(struct hclge_fd_rule, tuples.src_mac),
+	  offsetof(struct hclge_fd_rule, tuples_mask.src_mac) },
+	{ INNER_VLAN_TAG_FST, 16, KEY_OPT_LE16,
+	  offsetof(struct hclge_fd_rule, tuples.vlan_tag1),
+	  offsetof(struct hclge_fd_rule, tuples_mask.vlan_tag1) },
+	{ INNER_VLAN_TAG_SEC, 16, KEY_OPT_LE16, -1, -1 },
+	{ INNER_ETH_TYPE, 16, KEY_OPT_LE16,
+	  offsetof(struct hclge_fd_rule, tuples.ether_proto),
+	  offsetof(struct hclge_fd_rule, tuples_mask.ether_proto) },
+	{ INNER_L2_RSV, 16, KEY_OPT_LE16,
+	  offsetof(struct hclge_fd_rule, tuples.l2_user_def),
+	  offsetof(struct hclge_fd_rule, tuples_mask.l2_user_def) },
+	{ INNER_IP_TOS, 8, KEY_OPT_U8,
+	  offsetof(struct hclge_fd_rule, tuples.ip_tos),
+	  offsetof(struct hclge_fd_rule, tuples_mask.ip_tos) },
+	{ INNER_IP_PROTO, 8, KEY_OPT_U8,
+	  offsetof(struct hclge_fd_rule, tuples.ip_proto),
+	  offsetof(struct hclge_fd_rule, tuples_mask.ip_proto) },
+	{ INNER_SRC_IP, 32, KEY_OPT_IP,
+	  offsetof(struct hclge_fd_rule, tuples.src_ip),
+	  offsetof(struct hclge_fd_rule, tuples_mask.src_ip) },
+	{ INNER_DST_IP, 32, KEY_OPT_IP,
+	  offsetof(struct hclge_fd_rule, tuples.dst_ip),
+	  offsetof(struct hclge_fd_rule, tuples_mask.dst_ip) },
+	{ INNER_L3_RSV, 16, KEY_OPT_LE16,
+	  offsetof(struct hclge_fd_rule, tuples.l3_user_def),
+	  offsetof(struct hclge_fd_rule, tuples_mask.l3_user_def) },
+	{ INNER_SRC_PORT, 16, KEY_OPT_LE16,
+	  offsetof(struct hclge_fd_rule, tuples.src_port),
+	  offsetof(struct hclge_fd_rule, tuples_mask.src_port) },
+	{ INNER_DST_PORT, 16, KEY_OPT_LE16,
+	  offsetof(struct hclge_fd_rule, tuples.dst_port),
+	  offsetof(struct hclge_fd_rule, tuples_mask.dst_port) },
+	{ INNER_L4_RSV, 32, KEY_OPT_LE32,
+	  offsetof(struct hclge_fd_rule, tuples.l4_user_def),
+	  offsetof(struct hclge_fd_rule, tuples_mask.l4_user_def) },
 };
 
 static int hclge_mac_update_stats_defective(struct hclge_dev *hdev)
@@ -4241,6 +4268,7 @@ static void hclge_periodic_service_task(struct hclge_dev *hdev)
 	hclge_update_link_status(hdev);
 	hclge_sync_mac_table(hdev);
 	hclge_sync_promisc_mode(hdev);
+	hclge_sync_fd_table(hdev);
 
 	if (time_is_after_jiffies(hdev->last_serv_processed + HZ)) {
 		delta = jiffies - hdev->last_serv_processed;
@@ -5142,6 +5170,285 @@ static void hclge_request_update_promisc_mode(struct hnae3_handle *handle)
 	set_bit(HCLGE_STATE_PROMISC_CHANGED, &hdev->state);
 }
 
+static void hclge_sync_fd_state(struct hclge_dev *hdev)
+{
+	if (hlist_empty(&hdev->fd_rule_list))
+		hdev->fd_active_type = HCLGE_FD_RULE_NONE;
+}
+
+static void hclge_fd_inc_rule_cnt(struct hclge_dev *hdev, u16 location)
+{
+	if (!test_bit(location, hdev->fd_bmap)) {
+		set_bit(location, hdev->fd_bmap);
+		hdev->hclge_fd_rule_num++;
+	}
+}
+
+static void hclge_fd_dec_rule_cnt(struct hclge_dev *hdev, u16 location)
+{
+	if (test_bit(location, hdev->fd_bmap)) {
+		clear_bit(location, hdev->fd_bmap);
+		hdev->hclge_fd_rule_num--;
+	}
+}
+
+static void hclge_fd_free_node(struct hclge_dev *hdev,
+			       struct hclge_fd_rule *rule)
+{
+	hlist_del(&rule->rule_node);
+	kfree(rule);
+	hclge_sync_fd_state(hdev);
+}
+
+static void hclge_update_fd_rule_node(struct hclge_dev *hdev,
+				      struct hclge_fd_rule *old_rule,
+				      struct hclge_fd_rule *new_rule,
+				      enum HCLGE_FD_NODE_STATE state)
+{
+	switch (state) {
+	case HCLGE_FD_TO_ADD:
+	case HCLGE_FD_ACTIVE:
+		/* 1) if the new state is TO_ADD, just replace the old rule
+		 * with the same location, no matter its state, because the
+		 * new rule will be configured to the hardware.
+		 * 2) if the new state is ACTIVE, it means the new rule
+		 * has been configured to the hardware, so just replace
+		 * the old rule node with the same location.
+		 * 3) for it doesn't add a new node to the list, so it's
+		 * unnecessary to update the rule number and fd_bmap.
+		 */
+		new_rule->rule_node.next = old_rule->rule_node.next;
+		new_rule->rule_node.pprev = old_rule->rule_node.pprev;
+		memcpy(old_rule, new_rule, sizeof(*old_rule));
+		kfree(new_rule);
+		break;
+	case HCLGE_FD_DELETED:
+		hclge_fd_dec_rule_cnt(hdev, old_rule->location);
+		hclge_fd_free_node(hdev, old_rule);
+		break;
+	case HCLGE_FD_TO_DEL:
+		/* if new request is TO_DEL, and old rule is existent
+		 * 1) the state of old rule is TO_DEL, we need do nothing,
+		 * because we delete rule by location, other rule content
+		 * is unncessary.
+		 * 2) the state of old rule is ACTIVE, we need to change its
+		 * state to TO_DEL, so the rule will be deleted when periodic
+		 * task being scheduled.
+		 * 3) the state of old rule is TO_ADD, it means the rule hasn't
+		 * been added to hardware, so we just delete the rule node from
+		 * fd_rule_list directly.
+		 */
+		if (old_rule->state == HCLGE_FD_TO_ADD) {
+			hclge_fd_dec_rule_cnt(hdev, old_rule->location);
+			hclge_fd_free_node(hdev, old_rule);
+			return;
+		}
+		old_rule->state = HCLGE_FD_TO_DEL;
+		break;
+	}
+}
+
+static struct hclge_fd_rule *hclge_find_fd_rule(struct hlist_head *hlist,
+						u16 location,
+						struct hclge_fd_rule **parent)
+{
+	struct hclge_fd_rule *rule;
+	struct hlist_node *node;
+
+	hlist_for_each_entry_safe(rule, node, hlist, rule_node) {
+		if (rule->location == location)
+			return rule;
+		else if (rule->location > location)
+			return NULL;
+		/* record the parent node, use to keep the nodes in fd_rule_list
+		 * in ascend order.
+		 */
+		*parent = rule;
+	}
+
+	return NULL;
+}
+
+/* insert fd rule node in ascend order according to rule->location */
+static void hclge_fd_insert_rule_node(struct hlist_head *hlist,
+				      struct hclge_fd_rule *rule,
+				      struct hclge_fd_rule *parent)
+{
+	INIT_HLIST_NODE(&rule->rule_node);
+
+	if (parent)
+		hlist_add_behind(&rule->rule_node, &parent->rule_node);
+	else
+		hlist_add_head(&rule->rule_node, hlist);
+}
+
+static int hclge_fd_set_user_def_cmd(struct hclge_dev *hdev,
+				     struct hclge_fd_user_def_cfg *cfg)
+{
+	struct hclge_fd_user_def_cfg_cmd *req;
+	struct hclge_desc desc;
+	u16 data = 0;
+	int ret;
+
+	hclge_cmd_setup_basic_desc(&desc, HCLGE_OPC_FD_USER_DEF_OP, false);
+
+	req = (struct hclge_fd_user_def_cfg_cmd *)desc.data;
+
+	hnae3_set_bit(data, HCLGE_FD_USER_DEF_EN_B, cfg[0].ref_cnt > 0);
+	hnae3_set_field(data, HCLGE_FD_USER_DEF_OFT_M,
+			HCLGE_FD_USER_DEF_OFT_S, cfg[0].offset);
+	req->ol2_cfg = cpu_to_le16(data);
+
+	data = 0;
+	hnae3_set_bit(data, HCLGE_FD_USER_DEF_EN_B, cfg[1].ref_cnt > 0);
+	hnae3_set_field(data, HCLGE_FD_USER_DEF_OFT_M,
+			HCLGE_FD_USER_DEF_OFT_S, cfg[1].offset);
+	req->ol3_cfg = cpu_to_le16(data);
+
+	data = 0;
+	hnae3_set_bit(data, HCLGE_FD_USER_DEF_EN_B, cfg[2].ref_cnt > 0);
+	hnae3_set_field(data, HCLGE_FD_USER_DEF_OFT_M,
+			HCLGE_FD_USER_DEF_OFT_S, cfg[2].offset);
+	req->ol4_cfg = cpu_to_le16(data);
+
+	ret = hclge_cmd_send(&hdev->hw, &desc, 1);
+	if (ret)
+		dev_err(&hdev->pdev->dev,
+			"failed to set fd user def data, ret= %d\n", ret);
+	return ret;
+}
+
+static void hclge_sync_fd_user_def_cfg(struct hclge_dev *hdev, bool locked)
+{
+	int ret;
+
+	if (!test_and_clear_bit(HCLGE_STATE_FD_USER_DEF_CHANGED, &hdev->state))
+		return;
+
+	if (!locked)
+		spin_lock_bh(&hdev->fd_rule_lock);
+
+	ret = hclge_fd_set_user_def_cmd(hdev, hdev->fd_cfg.user_def_cfg);
+	if (ret)
+		set_bit(HCLGE_STATE_FD_USER_DEF_CHANGED, &hdev->state);
+
+	if (!locked)
+		spin_unlock_bh(&hdev->fd_rule_lock);
+}
+
+static int hclge_fd_check_user_def_refcnt(struct hclge_dev *hdev,
+					  struct hclge_fd_rule *rule)
+{
+	struct hlist_head *hlist = &hdev->fd_rule_list;
+	struct hclge_fd_rule *fd_rule, *parent = NULL;
+	struct hclge_fd_user_def_info *info, *old_info;
+	struct hclge_fd_user_def_cfg *cfg;
+
+	if (!rule || rule->rule_type != HCLGE_FD_EP_ACTIVE ||
+	    rule->ep.user_def.layer == HCLGE_FD_USER_DEF_NONE)
+		return 0;
+
+	/* for valid layer is start from 1, so need minus 1 to get the cfg */
+	cfg = &hdev->fd_cfg.user_def_cfg[rule->ep.user_def.layer - 1];
+	info = &rule->ep.user_def;
+
+	if (!cfg->ref_cnt || cfg->offset == info->offset)
+		return 0;
+
+	if (cfg->ref_cnt > 1)
+		goto error;
+
+	fd_rule = hclge_find_fd_rule(hlist, rule->location, &parent);
+	if (fd_rule) {
+		old_info = &fd_rule->ep.user_def;
+		if (info->layer == old_info->layer)
+			return 0;
+	}
+
+error:
+	dev_err(&hdev->pdev->dev,
+		"No available offset for layer%d fd rule, each layer only support one user def offset.\n",
+		info->layer + 1);
+	return -ENOSPC;
+}
+
+static void hclge_fd_inc_user_def_refcnt(struct hclge_dev *hdev,
+					 struct hclge_fd_rule *rule)
+{
+	struct hclge_fd_user_def_cfg *cfg;
+
+	if (!rule || rule->rule_type != HCLGE_FD_EP_ACTIVE ||
+	    rule->ep.user_def.layer == HCLGE_FD_USER_DEF_NONE)
+		return;
+
+	cfg = &hdev->fd_cfg.user_def_cfg[rule->ep.user_def.layer - 1];
+	if (!cfg->ref_cnt) {
+		cfg->offset = rule->ep.user_def.offset;
+		set_bit(HCLGE_STATE_FD_USER_DEF_CHANGED, &hdev->state);
+	}
+	cfg->ref_cnt++;
+}
+
+static void hclge_fd_dec_user_def_refcnt(struct hclge_dev *hdev,
+					 struct hclge_fd_rule *rule)
+{
+	struct hclge_fd_user_def_cfg *cfg;
+
+	if (!rule || rule->rule_type != HCLGE_FD_EP_ACTIVE ||
+	    rule->ep.user_def.layer == HCLGE_FD_USER_DEF_NONE)
+		return;
+
+	cfg = &hdev->fd_cfg.user_def_cfg[rule->ep.user_def.layer - 1];
+	if (!cfg->ref_cnt)
+		return;
+
+	cfg->ref_cnt--;
+	if (!cfg->ref_cnt) {
+		cfg->offset = 0;
+		set_bit(HCLGE_STATE_FD_USER_DEF_CHANGED, &hdev->state);
+	}
+}
+
+static void hclge_update_fd_list(struct hclge_dev *hdev,
+				 enum HCLGE_FD_NODE_STATE state, u16 location,
+				 struct hclge_fd_rule *new_rule)
+{
+	struct hlist_head *hlist = &hdev->fd_rule_list;
+	struct hclge_fd_rule *fd_rule, *parent = NULL;
+
+	fd_rule = hclge_find_fd_rule(hlist, location, &parent);
+	if (fd_rule) {
+		hclge_fd_dec_user_def_refcnt(hdev, fd_rule);
+		if (state == HCLGE_FD_ACTIVE)
+			hclge_fd_inc_user_def_refcnt(hdev, new_rule);
+		hclge_sync_fd_user_def_cfg(hdev, true);
+
+		hclge_update_fd_rule_node(hdev, fd_rule, new_rule, state);
+		return;
+	}
+
+	/* it's unlikely to fail here, because we have checked the rule
+	 * exist before.
+	 */
+	if (unlikely(state == HCLGE_FD_TO_DEL || state == HCLGE_FD_DELETED)) {
+		dev_warn(&hdev->pdev->dev,
+			 "failed to delete fd rule %u, it's inexistent\n",
+			 location);
+		return;
+	}
+
+	hclge_fd_inc_user_def_refcnt(hdev, new_rule);
+	hclge_sync_fd_user_def_cfg(hdev, true);
+
+	hclge_fd_insert_rule_node(hlist, new_rule, parent);
+	hclge_fd_inc_rule_cnt(hdev, new_rule->location);
+
+	if (state == HCLGE_FD_TO_ADD) {
+		set_bit(HCLGE_STATE_FD_TBL_CHANGED, &hdev->state);
+		hclge_task_schedule(hdev, 0);
+	}
+}
+
 static int hclge_get_fd_mode(struct hclge_dev *hdev, u8 *fd_mode)
 {
 	struct hclge_get_fd_mode_cmd *req;
@@ -5220,6 +5527,17 @@ static int hclge_set_fd_key_config(struct hclge_dev *hdev,
 	return ret;
 }
 
+static void hclge_fd_disable_user_def(struct hclge_dev *hdev)
+{
+	struct hclge_fd_user_def_cfg *cfg = hdev->fd_cfg.user_def_cfg;
+
+	spin_lock_bh(&hdev->fd_rule_lock);
+	memset(cfg, 0, sizeof(hdev->fd_cfg.user_def_cfg));
+	spin_unlock_bh(&hdev->fd_rule_lock);
+
+	hclge_fd_set_user_def_cmd(hdev, cfg);
+}
+
 static int hclge_init_fd_config(struct hclge_dev *hdev)
 {
 #define LOW_2_WORDS		0x03
@@ -5260,9 +5578,12 @@ static int hclge_init_fd_config(struct hclge_dev *hdev)
 				BIT(INNER_SRC_PORT) | BIT(INNER_DST_PORT);
 
 	/* If use max 400bit key, we can support tuples for ether type */
-	if (hdev->fd_cfg.fd_mode == HCLGE_FD_MODE_DEPTH_2K_WIDTH_400B_STAGE_1)
+	if (hdev->fd_cfg.fd_mode == HCLGE_FD_MODE_DEPTH_2K_WIDTH_400B_STAGE_1) {
 		key_cfg->tuple_active |=
 				BIT(INNER_DST_MAC) | BIT(INNER_SRC_MAC);
+		if (hdev->ae_dev->dev_version >= HNAE3_DEVICE_VERSION_V3)
+			key_cfg->tuple_active |= HCLGE_FD_TUPLE_USER_DEF_TUPLES;
+	}
 
 	/* roce_type is used to filter roce frames
 	 * dst_vport is used to specify the rule
@@ -5371,96 +5692,57 @@ static int hclge_fd_ad_config(struct hclge_dev *hdev, u8 stage, int loc,
 static bool hclge_fd_convert_tuple(u32 tuple_bit, u8 *key_x, u8 *key_y,
 				   struct hclge_fd_rule *rule)
 {
+	int offset, moffset, ip_offset;
+	enum HCLGE_FD_KEY_OPT key_opt;
 	u16 tmp_x_s, tmp_y_s;
 	u32 tmp_x_l, tmp_y_l;
+	u8 *p = (u8 *)rule;
 	int i;
 
-	if (rule->unused_tuple & tuple_bit)
+	if (rule->unused_tuple & BIT(tuple_bit))
 		return true;
 
-	switch (tuple_bit) {
-	case BIT(INNER_DST_MAC):
-		for (i = 0; i < ETH_ALEN; i++) {
-			calc_x(key_x[ETH_ALEN - 1 - i], rule->tuples.dst_mac[i],
-			       rule->tuples_mask.dst_mac[i]);
-			calc_y(key_y[ETH_ALEN - 1 - i], rule->tuples.dst_mac[i],
-			       rule->tuples_mask.dst_mac[i]);
-		}
+	key_opt = tuple_key_info[tuple_bit].key_opt;
+	offset = tuple_key_info[tuple_bit].offset;
+	moffset = tuple_key_info[tuple_bit].moffset;
 
-		return true;
-	case BIT(INNER_SRC_MAC):
-		for (i = 0; i < ETH_ALEN; i++) {
-			calc_x(key_x[ETH_ALEN - 1 - i], rule->tuples.src_mac[i],
-			       rule->tuples_mask.src_mac[i]);
-			calc_y(key_y[ETH_ALEN - 1 - i], rule->tuples.src_mac[i],
-			       rule->tuples_mask.src_mac[i]);
-		}
-
-		return true;
-	case BIT(INNER_VLAN_TAG_FST):
-		calc_x(tmp_x_s, rule->tuples.vlan_tag1,
-		       rule->tuples_mask.vlan_tag1);
-		calc_y(tmp_y_s, rule->tuples.vlan_tag1,
-		       rule->tuples_mask.vlan_tag1);
-		*(__le16 *)key_x = cpu_to_le16(tmp_x_s);
-		*(__le16 *)key_y = cpu_to_le16(tmp_y_s);
+	switch (key_opt) {
+	case KEY_OPT_U8:
+		calc_x(*key_x, p[offset], p[moffset]);
+		calc_y(*key_y, p[offset], p[moffset]);
 
 		return true;
-	case BIT(INNER_ETH_TYPE):
-		calc_x(tmp_x_s, rule->tuples.ether_proto,
-		       rule->tuples_mask.ether_proto);
-		calc_y(tmp_y_s, rule->tuples.ether_proto,
-		       rule->tuples_mask.ether_proto);
+	case KEY_OPT_LE16:
+		calc_x(tmp_x_s, *(u16 *)(&p[offset]), *(u16 *)(&p[moffset]));
+		calc_y(tmp_y_s, *(u16 *)(&p[offset]), *(u16 *)(&p[moffset]));
 		*(__le16 *)key_x = cpu_to_le16(tmp_x_s);
 		*(__le16 *)key_y = cpu_to_le16(tmp_y_s);
 
 		return true;
-	case BIT(INNER_IP_TOS):
-		calc_x(*key_x, rule->tuples.ip_tos, rule->tuples_mask.ip_tos);
-		calc_y(*key_y, rule->tuples.ip_tos, rule->tuples_mask.ip_tos);
-
-		return true;
-	case BIT(INNER_IP_PROTO):
-		calc_x(*key_x, rule->tuples.ip_proto,
-		       rule->tuples_mask.ip_proto);
-		calc_y(*key_y, rule->tuples.ip_proto,
-		       rule->tuples_mask.ip_proto);
-
-		return true;
-	case BIT(INNER_SRC_IP):
-		calc_x(tmp_x_l, rule->tuples.src_ip[IPV4_INDEX],
-		       rule->tuples_mask.src_ip[IPV4_INDEX]);
-		calc_y(tmp_y_l, rule->tuples.src_ip[IPV4_INDEX],
-		       rule->tuples_mask.src_ip[IPV4_INDEX]);
+	case KEY_OPT_LE32:
+		calc_x(tmp_x_l, *(u32 *)(&p[offset]), *(u32 *)(&p[moffset]));
+		calc_y(tmp_y_l, *(u32 *)(&p[offset]), *(u32 *)(&p[moffset]));
 		*(__le32 *)key_x = cpu_to_le32(tmp_x_l);
 		*(__le32 *)key_y = cpu_to_le32(tmp_y_l);
 
 		return true;
-	case BIT(INNER_DST_IP):
-		calc_x(tmp_x_l, rule->tuples.dst_ip[IPV4_INDEX],
-		       rule->tuples_mask.dst_ip[IPV4_INDEX]);
-		calc_y(tmp_y_l, rule->tuples.dst_ip[IPV4_INDEX],
-		       rule->tuples_mask.dst_ip[IPV4_INDEX]);
-		*(__le32 *)key_x = cpu_to_le32(tmp_x_l);
-		*(__le32 *)key_y = cpu_to_le32(tmp_y_l);
-
-		return true;
-	case BIT(INNER_SRC_PORT):
-		calc_x(tmp_x_s, rule->tuples.src_port,
-		       rule->tuples_mask.src_port);
-		calc_y(tmp_y_s, rule->tuples.src_port,
-		       rule->tuples_mask.src_port);
-		*(__le16 *)key_x = cpu_to_le16(tmp_x_s);
-		*(__le16 *)key_y = cpu_to_le16(tmp_y_s);
+	case KEY_OPT_MAC:
+		for (i = 0; i < ETH_ALEN; i++) {
+			calc_x(key_x[ETH_ALEN - 1 - i], p[offset + i],
+			       p[moffset + i]);
+			calc_y(key_y[ETH_ALEN - 1 - i], p[offset + i],
+			       p[moffset + i]);
+		}
 
 		return true;
-	case BIT(INNER_DST_PORT):
-		calc_x(tmp_x_s, rule->tuples.dst_port,
-		       rule->tuples_mask.dst_port);
-		calc_y(tmp_y_s, rule->tuples.dst_port,
-		       rule->tuples_mask.dst_port);
-		*(__le16 *)key_x = cpu_to_le16(tmp_x_s);
-		*(__le16 *)key_y = cpu_to_le16(tmp_y_s);
+	case KEY_OPT_IP:
+		ip_offset = IPV4_INDEX * sizeof(u32);
+		calc_x(tmp_x_l, *(u32 *)(&p[offset + ip_offset]),
+		       *(u32 *)(&p[moffset + ip_offset]));
+		calc_y(tmp_y_l, *(u32 *)(&p[offset + ip_offset]),
+		       *(u32 *)(&p[moffset + ip_offset]));
+		*(__le32 *)key_x = cpu_to_le32(tmp_x_l);
+		*(__le32 *)key_y = cpu_to_le32(tmp_y_l);
 
 		return true;
 	default:
@@ -5548,12 +5830,12 @@ static int hclge_config_key(struct hclge_dev *hdev, u8 stage,
 
 	for (i = 0 ; i < MAX_TUPLE; i++) {
 		bool tuple_valid;
-		u32 check_tuple;
 
 		tuple_size = tuple_key_info[i].key_length / 8;
-		check_tuple = key_cfg->tuple_active & BIT(i);
+		if (!(key_cfg->tuple_active & BIT(i)))
+			continue;
 
-		tuple_valid = hclge_fd_convert_tuple(check_tuple, cur_key_x,
+		tuple_valid = hclge_fd_convert_tuple(i, cur_key_x,
 						     cur_key_y, rule);
 		if (tuple_valid) {
 			cur_key_x += tuple_size;
@@ -5684,8 +5966,7 @@ static int hclge_fd_check_tcpip6_tuple(struct ethtool_tcpip6_spec *spec,
 	if (!spec || !unused_tuple)
 		return -EINVAL;
 
-	*unused_tuple |= BIT(INNER_SRC_MAC) | BIT(INNER_DST_MAC) |
-		BIT(INNER_IP_TOS);
+	*unused_tuple |= BIT(INNER_SRC_MAC) | BIT(INNER_DST_MAC);
 
 	/* check whether src/dst ip address used */
 	if (ipv6_addr_any((struct in6_addr *)spec->ip6src))
@@ -5700,8 +5981,8 @@ static int hclge_fd_check_tcpip6_tuple(struct ethtool_tcpip6_spec *spec,
 	if (!spec->pdst)
 		*unused_tuple |= BIT(INNER_DST_PORT);
 
-	if (spec->tclass)
-		return -EOPNOTSUPP;
+	if (!spec->tclass)
+		*unused_tuple |= BIT(INNER_IP_TOS);
 
 	return 0;
 }
@@ -5713,7 +5994,7 @@ static int hclge_fd_check_ip6_tuple(struct ethtool_usrip6_spec *spec,
 		return -EINVAL;
 
 	*unused_tuple |= BIT(INNER_SRC_MAC) | BIT(INNER_DST_MAC) |
-		BIT(INNER_IP_TOS) | BIT(INNER_SRC_PORT) | BIT(INNER_DST_PORT);
+			BIT(INNER_SRC_PORT) | BIT(INNER_DST_PORT);
 
 	/* check whether src/dst ip address used */
 	if (ipv6_addr_any((struct in6_addr *)spec->ip6src))
@@ -5725,8 +6006,8 @@ static int hclge_fd_check_ip6_tuple(struct ethtool_usrip6_spec *spec,
 	if (!spec->l4_proto)
 		*unused_tuple |= BIT(INNER_IP_PROTO);
 
-	if (spec->tclass)
-		return -EOPNOTSUPP;
+	if (!spec->tclass)
+		*unused_tuple |= BIT(INNER_IP_TOS);
 
 	if (spec->l4_4_bytes)
 		return -EOPNOTSUPP;
@@ -5796,9 +6077,98 @@ static int hclge_fd_check_ext_tuple(struct hclge_dev *hdev,
 	return 0;
 }
 
+static int hclge_fd_get_user_def_layer(u32 flow_type, u32 *unused_tuple,
+				       struct hclge_fd_user_def_info *info)
+{
+	switch (flow_type) {
+	case ETHER_FLOW:
+		info->layer = HCLGE_FD_USER_DEF_L2;
+		*unused_tuple &= ~BIT(INNER_L2_RSV);
+		break;
+	case IP_USER_FLOW:
+	case IPV6_USER_FLOW:
+		info->layer = HCLGE_FD_USER_DEF_L3;
+		*unused_tuple &= ~BIT(INNER_L3_RSV);
+		break;
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+	case TCP_V6_FLOW:
+	case UDP_V6_FLOW:
+		info->layer = HCLGE_FD_USER_DEF_L4;
+		*unused_tuple &= ~BIT(INNER_L4_RSV);
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	return 0;
+}
+
+static bool hclge_fd_is_user_def_all_masked(struct ethtool_rx_flow_spec *fs)
+{
+	return be32_to_cpu(fs->m_ext.data[1] | fs->m_ext.data[0]) == 0;
+}
+
+static int hclge_fd_parse_user_def_field(struct hclge_dev *hdev,
+					 struct ethtool_rx_flow_spec *fs,
+					 u32 *unused_tuple,
+					 struct hclge_fd_user_def_info *info)
+{
+	u32 tuple_active = hdev->fd_cfg.key_cfg[HCLGE_FD_STAGE_1].tuple_active;
+	u32 flow_type = fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT);
+	u16 data, offset, data_mask, offset_mask;
+	int ret;
+
+	info->layer = HCLGE_FD_USER_DEF_NONE;
+	*unused_tuple |= HCLGE_FD_TUPLE_USER_DEF_TUPLES;
+
+	if (!(fs->flow_type & FLOW_EXT) || hclge_fd_is_user_def_all_masked(fs))
+		return 0;
+
+	/* user-def data from ethtool is 64 bit value, the bit0~15 is used
+	 * for data, and bit32~47 is used for offset.
+	 */
+	data = be32_to_cpu(fs->h_ext.data[1]) & HCLGE_FD_USER_DEF_DATA;
+	data_mask = be32_to_cpu(fs->m_ext.data[1]) & HCLGE_FD_USER_DEF_DATA;
+	offset = be32_to_cpu(fs->h_ext.data[0]) & HCLGE_FD_USER_DEF_OFFSET;
+	offset_mask = be32_to_cpu(fs->m_ext.data[0]) & HCLGE_FD_USER_DEF_OFFSET;
+
+	if (!(tuple_active & HCLGE_FD_TUPLE_USER_DEF_TUPLES)) {
+		dev_err(&hdev->pdev->dev, "user-def bytes are not supported\n");
+		return -EOPNOTSUPP;
+	}
+
+	if (offset > HCLGE_FD_MAX_USER_DEF_OFFSET) {
+		dev_err(&hdev->pdev->dev,
+			"user-def offset[%u] should be no more than %u\n",
+			offset, HCLGE_FD_MAX_USER_DEF_OFFSET);
+		return -EINVAL;
+	}
+
+	if (offset_mask != HCLGE_FD_USER_DEF_OFFSET_UNMASK) {
+		dev_err(&hdev->pdev->dev, "user-def offset can't be masked\n");
+		return -EINVAL;
+	}
+
+	ret = hclge_fd_get_user_def_layer(flow_type, unused_tuple, info);
+	if (ret) {
+		dev_err(&hdev->pdev->dev,
+			"unsupported flow type for user-def bytes, ret = %d\n",
+			ret);
+		return ret;
+	}
+
+	info->data = data;
+	info->data_mask = data_mask;
+	info->offset = offset;
+
+	return 0;
+}
+
 static int hclge_fd_check_spec(struct hclge_dev *hdev,
 			       struct ethtool_rx_flow_spec *fs,
-			       u32 *unused_tuple)
+			       u32 *unused_tuple,
+			       struct hclge_fd_user_def_info *info)
 {
 	u32 flow_type;
 	int ret;
@@ -5811,11 +6181,9 @@ static int hclge_fd_check_spec(struct hclge_dev *hdev,
 		return -EINVAL;
 	}
 
-	if ((fs->flow_type & FLOW_EXT) &&
-	    (fs->h_ext.data[0] != 0 || fs->h_ext.data[1] != 0)) {
-		dev_err(&hdev->pdev->dev, "user-def bytes are not supported\n");
-		return -EOPNOTSUPP;
-	}
+	ret = hclge_fd_parse_user_def_field(hdev, fs, unused_tuple, info);
+	if (ret)
+		return ret;
 
 	flow_type = fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT);
 	switch (flow_type) {
@@ -5867,217 +6235,194 @@ static int hclge_fd_check_spec(struct hclge_dev *hdev,
 	return hclge_fd_check_ext_tuple(hdev, fs, unused_tuple);
 }
 
-static bool hclge_fd_rule_exist(struct hclge_dev *hdev, u16 location)
-{
-	struct hclge_fd_rule *rule = NULL;
-	struct hlist_node *node2;
-
-	spin_lock_bh(&hdev->fd_rule_lock);
-	hlist_for_each_entry_safe(rule, node2, &hdev->fd_rule_list, rule_node) {
-		if (rule->location >= location)
-			break;
-	}
-
-	spin_unlock_bh(&hdev->fd_rule_lock);
-
-	return  rule && rule->location == location;
-}
-
-/* make sure being called after lock up with fd_rule_lock */
-static int hclge_fd_update_rule_list(struct hclge_dev *hdev,
-				     struct hclge_fd_rule *new_rule,
-				     u16 location,
-				     bool is_add)
+static void hclge_fd_get_tcpip4_tuple(struct hclge_dev *hdev,
+				      struct ethtool_rx_flow_spec *fs,
+				      struct hclge_fd_rule *rule, u8 ip_proto)
 {
-	struct hclge_fd_rule *rule = NULL, *parent = NULL;
-	struct hlist_node *node2;
-
-	if (is_add && !new_rule)
-		return -EINVAL;
+	rule->tuples.src_ip[IPV4_INDEX] =
+			be32_to_cpu(fs->h_u.tcp_ip4_spec.ip4src);
+	rule->tuples_mask.src_ip[IPV4_INDEX] =
+			be32_to_cpu(fs->m_u.tcp_ip4_spec.ip4src);
 
-	hlist_for_each_entry_safe(rule, node2,
-				  &hdev->fd_rule_list, rule_node) {
-		if (rule->location >= location)
-			break;
-		parent = rule;
-	}
+	rule->tuples.dst_ip[IPV4_INDEX] =
+			be32_to_cpu(fs->h_u.tcp_ip4_spec.ip4dst);
+	rule->tuples_mask.dst_ip[IPV4_INDEX] =
+			be32_to_cpu(fs->m_u.tcp_ip4_spec.ip4dst);
 
-	if (rule && rule->location == location) {
-		hlist_del(&rule->rule_node);
-		kfree(rule);
-		hdev->hclge_fd_rule_num--;
-
-		if (!is_add) {
-			if (!hdev->hclge_fd_rule_num)
-				hdev->fd_active_type = HCLGE_FD_RULE_NONE;
-			clear_bit(location, hdev->fd_bmap);
-
-			return 0;
-		}
-	} else if (!is_add) {
-		dev_err(&hdev->pdev->dev,
-			"delete fail, rule %u is inexistent\n",
-			location);
-		return -EINVAL;
-	}
+	rule->tuples.src_port = be16_to_cpu(fs->h_u.tcp_ip4_spec.psrc);
+	rule->tuples_mask.src_port = be16_to_cpu(fs->m_u.tcp_ip4_spec.psrc);
 
-	INIT_HLIST_NODE(&new_rule->rule_node);
+	rule->tuples.dst_port = be16_to_cpu(fs->h_u.tcp_ip4_spec.pdst);
+	rule->tuples_mask.dst_port = be16_to_cpu(fs->m_u.tcp_ip4_spec.pdst);
 
-	if (parent)
-		hlist_add_behind(&new_rule->rule_node, &parent->rule_node);
-	else
-		hlist_add_head(&new_rule->rule_node, &hdev->fd_rule_list);
+	rule->tuples.ip_tos = fs->h_u.tcp_ip4_spec.tos;
+	rule->tuples_mask.ip_tos = fs->m_u.tcp_ip4_spec.tos;
 
-	set_bit(location, hdev->fd_bmap);
-	hdev->hclge_fd_rule_num++;
-	hdev->fd_active_type = new_rule->rule_type;
+	rule->tuples.ether_proto = ETH_P_IP;
+	rule->tuples_mask.ether_proto = 0xFFFF;
 
-	return 0;
+	rule->tuples.ip_proto = ip_proto;
+	rule->tuples_mask.ip_proto = 0xFF;
 }
 
-static int hclge_fd_get_tuple(struct hclge_dev *hdev,
-			      struct ethtool_rx_flow_spec *fs,
-			      struct hclge_fd_rule *rule)
+static void hclge_fd_get_ip4_tuple(struct hclge_dev *hdev,
+				   struct ethtool_rx_flow_spec *fs,
+				   struct hclge_fd_rule *rule)
 {
-	u32 flow_type = fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT);
-
-	switch (flow_type) {
-	case SCTP_V4_FLOW:
-	case TCP_V4_FLOW:
-	case UDP_V4_FLOW:
-		rule->tuples.src_ip[IPV4_INDEX] =
-				be32_to_cpu(fs->h_u.tcp_ip4_spec.ip4src);
-		rule->tuples_mask.src_ip[IPV4_INDEX] =
-				be32_to_cpu(fs->m_u.tcp_ip4_spec.ip4src);
-
-		rule->tuples.dst_ip[IPV4_INDEX] =
-				be32_to_cpu(fs->h_u.tcp_ip4_spec.ip4dst);
-		rule->tuples_mask.dst_ip[IPV4_INDEX] =
-				be32_to_cpu(fs->m_u.tcp_ip4_spec.ip4dst);
-
-		rule->tuples.src_port = be16_to_cpu(fs->h_u.tcp_ip4_spec.psrc);
-		rule->tuples_mask.src_port =
-				be16_to_cpu(fs->m_u.tcp_ip4_spec.psrc);
+	rule->tuples.src_ip[IPV4_INDEX] =
+			be32_to_cpu(fs->h_u.usr_ip4_spec.ip4src);
+	rule->tuples_mask.src_ip[IPV4_INDEX] =
+			be32_to_cpu(fs->m_u.usr_ip4_spec.ip4src);
 
-		rule->tuples.dst_port = be16_to_cpu(fs->h_u.tcp_ip4_spec.pdst);
-		rule->tuples_mask.dst_port =
-				be16_to_cpu(fs->m_u.tcp_ip4_spec.pdst);
+	rule->tuples.dst_ip[IPV4_INDEX] =
+			be32_to_cpu(fs->h_u.usr_ip4_spec.ip4dst);
+	rule->tuples_mask.dst_ip[IPV4_INDEX] =
+			be32_to_cpu(fs->m_u.usr_ip4_spec.ip4dst);
 
-		rule->tuples.ip_tos = fs->h_u.tcp_ip4_spec.tos;
-		rule->tuples_mask.ip_tos = fs->m_u.tcp_ip4_spec.tos;
+	rule->tuples.ip_tos = fs->h_u.usr_ip4_spec.tos;
+	rule->tuples_mask.ip_tos = fs->m_u.usr_ip4_spec.tos;
 
-		rule->tuples.ether_proto = ETH_P_IP;
-		rule->tuples_mask.ether_proto = 0xFFFF;
+	rule->tuples.ip_proto = fs->h_u.usr_ip4_spec.proto;
+	rule->tuples_mask.ip_proto = fs->m_u.usr_ip4_spec.proto;
 
-		break;
-	case IP_USER_FLOW:
-		rule->tuples.src_ip[IPV4_INDEX] =
-				be32_to_cpu(fs->h_u.usr_ip4_spec.ip4src);
-		rule->tuples_mask.src_ip[IPV4_INDEX] =
-				be32_to_cpu(fs->m_u.usr_ip4_spec.ip4src);
-
-		rule->tuples.dst_ip[IPV4_INDEX] =
-				be32_to_cpu(fs->h_u.usr_ip4_spec.ip4dst);
-		rule->tuples_mask.dst_ip[IPV4_INDEX] =
-				be32_to_cpu(fs->m_u.usr_ip4_spec.ip4dst);
+	rule->tuples.ether_proto = ETH_P_IP;
+	rule->tuples_mask.ether_proto = 0xFFFF;
+}
 
-		rule->tuples.ip_tos = fs->h_u.usr_ip4_spec.tos;
-		rule->tuples_mask.ip_tos = fs->m_u.usr_ip4_spec.tos;
+static void hclge_fd_get_tcpip6_tuple(struct hclge_dev *hdev,
+				      struct ethtool_rx_flow_spec *fs,
+				      struct hclge_fd_rule *rule, u8 ip_proto)
+{
+	be32_to_cpu_array(rule->tuples.src_ip, fs->h_u.tcp_ip6_spec.ip6src,
+			  IPV6_SIZE);
+	be32_to_cpu_array(rule->tuples_mask.src_ip, fs->m_u.tcp_ip6_spec.ip6src,
+			  IPV6_SIZE);
 
-		rule->tuples.ip_proto = fs->h_u.usr_ip4_spec.proto;
-		rule->tuples_mask.ip_proto = fs->m_u.usr_ip4_spec.proto;
+	be32_to_cpu_array(rule->tuples.dst_ip, fs->h_u.tcp_ip6_spec.ip6dst,
+			  IPV6_SIZE);
+	be32_to_cpu_array(rule->tuples_mask.dst_ip, fs->m_u.tcp_ip6_spec.ip6dst,
+			  IPV6_SIZE);
 
-		rule->tuples.ether_proto = ETH_P_IP;
-		rule->tuples_mask.ether_proto = 0xFFFF;
+	rule->tuples.src_port = be16_to_cpu(fs->h_u.tcp_ip6_spec.psrc);
+	rule->tuples_mask.src_port = be16_to_cpu(fs->m_u.tcp_ip6_spec.psrc);
 
-		break;
-	case SCTP_V6_FLOW:
-	case TCP_V6_FLOW:
-	case UDP_V6_FLOW:
-		be32_to_cpu_array(rule->tuples.src_ip,
-				  fs->h_u.tcp_ip6_spec.ip6src, IPV6_SIZE);
-		be32_to_cpu_array(rule->tuples_mask.src_ip,
-				  fs->m_u.tcp_ip6_spec.ip6src, IPV6_SIZE);
+	rule->tuples.dst_port = be16_to_cpu(fs->h_u.tcp_ip6_spec.pdst);
+	rule->tuples_mask.dst_port = be16_to_cpu(fs->m_u.tcp_ip6_spec.pdst);
 
-		be32_to_cpu_array(rule->tuples.dst_ip,
-				  fs->h_u.tcp_ip6_spec.ip6dst, IPV6_SIZE);
-		be32_to_cpu_array(rule->tuples_mask.dst_ip,
-				  fs->m_u.tcp_ip6_spec.ip6dst, IPV6_SIZE);
+	rule->tuples.ether_proto = ETH_P_IPV6;
+	rule->tuples_mask.ether_proto = 0xFFFF;
 
-		rule->tuples.src_port = be16_to_cpu(fs->h_u.tcp_ip6_spec.psrc);
-		rule->tuples_mask.src_port =
-				be16_to_cpu(fs->m_u.tcp_ip6_spec.psrc);
+	rule->tuples.ip_tos = fs->h_u.tcp_ip6_spec.tclass;
+	rule->tuples_mask.ip_tos = fs->m_u.tcp_ip6_spec.tclass;
 
-		rule->tuples.dst_port = be16_to_cpu(fs->h_u.tcp_ip6_spec.pdst);
-		rule->tuples_mask.dst_port =
-				be16_to_cpu(fs->m_u.tcp_ip6_spec.pdst);
+	rule->tuples.ip_proto = ip_proto;
+	rule->tuples_mask.ip_proto = 0xFF;
+}
 
-		rule->tuples.ether_proto = ETH_P_IPV6;
-		rule->tuples_mask.ether_proto = 0xFFFF;
+static void hclge_fd_get_ip6_tuple(struct hclge_dev *hdev,
+				   struct ethtool_rx_flow_spec *fs,
+				   struct hclge_fd_rule *rule)
+{
+	be32_to_cpu_array(rule->tuples.src_ip, fs->h_u.usr_ip6_spec.ip6src,
+			  IPV6_SIZE);
+	be32_to_cpu_array(rule->tuples_mask.src_ip, fs->m_u.usr_ip6_spec.ip6src,
+			  IPV6_SIZE);
 
-		break;
-	case IPV6_USER_FLOW:
-		be32_to_cpu_array(rule->tuples.src_ip,
-				  fs->h_u.usr_ip6_spec.ip6src, IPV6_SIZE);
-		be32_to_cpu_array(rule->tuples_mask.src_ip,
-				  fs->m_u.usr_ip6_spec.ip6src, IPV6_SIZE);
+	be32_to_cpu_array(rule->tuples.dst_ip, fs->h_u.usr_ip6_spec.ip6dst,
+			  IPV6_SIZE);
+	be32_to_cpu_array(rule->tuples_mask.dst_ip, fs->m_u.usr_ip6_spec.ip6dst,
+			  IPV6_SIZE);
 
-		be32_to_cpu_array(rule->tuples.dst_ip,
-				  fs->h_u.usr_ip6_spec.ip6dst, IPV6_SIZE);
-		be32_to_cpu_array(rule->tuples_mask.dst_ip,
-				  fs->m_u.usr_ip6_spec.ip6dst, IPV6_SIZE);
+	rule->tuples.ip_proto = fs->h_u.usr_ip6_spec.l4_proto;
+	rule->tuples_mask.ip_proto = fs->m_u.usr_ip6_spec.l4_proto;
 
-		rule->tuples.ip_proto = fs->h_u.usr_ip6_spec.l4_proto;
-		rule->tuples_mask.ip_proto = fs->m_u.usr_ip6_spec.l4_proto;
+	rule->tuples.ip_tos = fs->h_u.tcp_ip6_spec.tclass;
+	rule->tuples_mask.ip_tos = fs->m_u.tcp_ip6_spec.tclass;
 
-		rule->tuples.ether_proto = ETH_P_IPV6;
-		rule->tuples_mask.ether_proto = 0xFFFF;
+	rule->tuples.ether_proto = ETH_P_IPV6;
+	rule->tuples_mask.ether_proto = 0xFFFF;
+}
 
-		break;
-	case ETHER_FLOW:
-		ether_addr_copy(rule->tuples.src_mac,
-				fs->h_u.ether_spec.h_source);
-		ether_addr_copy(rule->tuples_mask.src_mac,
-				fs->m_u.ether_spec.h_source);
+static void hclge_fd_get_ether_tuple(struct hclge_dev *hdev,
+				     struct ethtool_rx_flow_spec *fs,
+				     struct hclge_fd_rule *rule)
+{
+	ether_addr_copy(rule->tuples.src_mac, fs->h_u.ether_spec.h_source);
+	ether_addr_copy(rule->tuples_mask.src_mac, fs->m_u.ether_spec.h_source);
 
-		ether_addr_copy(rule->tuples.dst_mac,
-				fs->h_u.ether_spec.h_dest);
-		ether_addr_copy(rule->tuples_mask.dst_mac,
-				fs->m_u.ether_spec.h_dest);
+	ether_addr_copy(rule->tuples.dst_mac, fs->h_u.ether_spec.h_dest);
+	ether_addr_copy(rule->tuples_mask.dst_mac, fs->m_u.ether_spec.h_dest);
 
-		rule->tuples.ether_proto =
-				be16_to_cpu(fs->h_u.ether_spec.h_proto);
-		rule->tuples_mask.ether_proto =
-				be16_to_cpu(fs->m_u.ether_spec.h_proto);
+	rule->tuples.ether_proto = be16_to_cpu(fs->h_u.ether_spec.h_proto);
+	rule->tuples_mask.ether_proto = be16_to_cpu(fs->m_u.ether_spec.h_proto);
+}
 
+static void hclge_fd_get_user_def_tuple(struct hclge_fd_user_def_info *info,
+					struct hclge_fd_rule *rule)
+{
+	switch (info->layer) {
+	case HCLGE_FD_USER_DEF_L2:
+		rule->tuples.l2_user_def = info->data;
+		rule->tuples_mask.l2_user_def = info->data_mask;
+		break;
+	case HCLGE_FD_USER_DEF_L3:
+		rule->tuples.l3_user_def = info->data;
+		rule->tuples_mask.l3_user_def = info->data_mask;
+		break;
+	case HCLGE_FD_USER_DEF_L4:
+		rule->tuples.l4_user_def = (u32)info->data << 16;
+		rule->tuples_mask.l4_user_def = (u32)info->data_mask << 16;
 		break;
 	default:
-		return -EOPNOTSUPP;
+		break;
 	}
 
+	rule->ep.user_def = *info;
+}
+
+static int hclge_fd_get_tuple(struct hclge_dev *hdev,
+			      struct ethtool_rx_flow_spec *fs,
+			      struct hclge_fd_rule *rule,
+			      struct hclge_fd_user_def_info *info)
+{
+	u32 flow_type = fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT);
+
 	switch (flow_type) {
 	case SCTP_V4_FLOW:
-	case SCTP_V6_FLOW:
-		rule->tuples.ip_proto = IPPROTO_SCTP;
-		rule->tuples_mask.ip_proto = 0xFF;
+		hclge_fd_get_tcpip4_tuple(hdev, fs, rule, IPPROTO_SCTP);
 		break;
 	case TCP_V4_FLOW:
-	case TCP_V6_FLOW:
-		rule->tuples.ip_proto = IPPROTO_TCP;
-		rule->tuples_mask.ip_proto = 0xFF;
+		hclge_fd_get_tcpip4_tuple(hdev, fs, rule, IPPROTO_TCP);
 		break;
 	case UDP_V4_FLOW:
+		hclge_fd_get_tcpip4_tuple(hdev, fs, rule, IPPROTO_UDP);
+		break;
+	case IP_USER_FLOW:
+		hclge_fd_get_ip4_tuple(hdev, fs, rule);
+		break;
+	case SCTP_V6_FLOW:
+		hclge_fd_get_tcpip6_tuple(hdev, fs, rule, IPPROTO_SCTP);
+		break;
+	case TCP_V6_FLOW:
+		hclge_fd_get_tcpip6_tuple(hdev, fs, rule, IPPROTO_TCP);
+		break;
 	case UDP_V6_FLOW:
-		rule->tuples.ip_proto = IPPROTO_UDP;
-		rule->tuples_mask.ip_proto = 0xFF;
+		hclge_fd_get_tcpip6_tuple(hdev, fs, rule, IPPROTO_UDP);
 		break;
-	default:
+	case IPV6_USER_FLOW:
+		hclge_fd_get_ip6_tuple(hdev, fs, rule);
 		break;
+	case ETHER_FLOW:
+		hclge_fd_get_ether_tuple(hdev, fs, rule);
+		break;
+	default:
+		return -EOPNOTSUPP;
 	}
 
 	if (fs->flow_type & FLOW_EXT) {
 		rule->tuples.vlan_tag1 = be16_to_cpu(fs->h_ext.vlan_tci);
 		rule->tuples_mask.vlan_tag1 = be16_to_cpu(fs->m_ext.vlan_tci);
+		hclge_fd_get_user_def_tuple(info, rule);
 	}
 
 	if (fs->flow_type & FLOW_MAC_EXT) {
@@ -6088,33 +6433,52 @@ static int hclge_fd_get_tuple(struct hclge_dev *hdev,
 	return 0;
 }
 
-/* make sure being called after lock up with fd_rule_lock */
 static int hclge_fd_config_rule(struct hclge_dev *hdev,
 				struct hclge_fd_rule *rule)
 {
 	int ret;
 
-	if (!rule) {
+	ret = hclge_config_action(hdev, HCLGE_FD_STAGE_1, rule);
+	if (ret)
+		return ret;
+
+	return hclge_config_key(hdev, HCLGE_FD_STAGE_1, rule);
+}
+
+static int hclge_add_fd_entry_common(struct hclge_dev *hdev,
+				     struct hclge_fd_rule *rule)
+{
+	int ret;
+
+	spin_lock_bh(&hdev->fd_rule_lock);
+
+	if (hdev->fd_active_type != rule->rule_type &&
+	    (hdev->fd_active_type == HCLGE_FD_TC_FLOWER_ACTIVE ||
+	     hdev->fd_active_type == HCLGE_FD_EP_ACTIVE)) {
 		dev_err(&hdev->pdev->dev,
-			"The flow director rule is NULL\n");
+			"mode conflict(new type %d, active type %d), please delete existent rules first\n",
+			rule->rule_type, hdev->fd_active_type);
+		spin_unlock_bh(&hdev->fd_rule_lock);
 		return -EINVAL;
 	}
 
-	/* it will never fail here, so needn't to check return value */
-	hclge_fd_update_rule_list(hdev, rule, rule->location, true);
+	ret = hclge_fd_check_user_def_refcnt(hdev, rule);
+	if (ret)
+		goto out;
 
-	ret = hclge_config_action(hdev, HCLGE_FD_STAGE_1, rule);
+	ret = hclge_clear_arfs_rules(hdev);
 	if (ret)
-		goto clear_rule;
+		goto out;
 
-	ret = hclge_config_key(hdev, HCLGE_FD_STAGE_1, rule);
+	ret = hclge_fd_config_rule(hdev, rule);
 	if (ret)
-		goto clear_rule;
+		goto out;
 
-	return 0;
+	hclge_update_fd_list(hdev, HCLGE_FD_ACTIVE, rule->location, rule);
+	hdev->fd_active_type = rule->rule_type;
 
-clear_rule:
-	hclge_fd_update_rule_list(hdev, rule, rule->location, false);
+out:
+	spin_unlock_bh(&hdev->fd_rule_lock);
 	return ret;
 }
 
@@ -6126,11 +6490,48 @@ static bool hclge_is_cls_flower_active(struct hnae3_handle *handle)
 	return hdev->fd_active_type == HCLGE_FD_TC_FLOWER_ACTIVE;
 }
 
+static int hclge_fd_parse_ring_cookie(struct hclge_dev *hdev, u64 ring_cookie,
+				      u16 *vport_id, u8 *action, u16 *queue_id)
+{
+	struct hclge_vport *vport = hdev->vport;
+
+	if (ring_cookie == RX_CLS_FLOW_DISC) {
+		*action = HCLGE_FD_ACTION_DROP_PACKET;
+	} else {
+		u32 ring = ethtool_get_flow_spec_ring(ring_cookie);
+		u8 vf = ethtool_get_flow_spec_ring_vf(ring_cookie);
+		u16 tqps;
+
+		if (vf > hdev->num_req_vfs) {
+			dev_err(&hdev->pdev->dev,
+				"Error: vf id (%u) > max vf num (%u)\n",
+				vf, hdev->num_req_vfs);
+			return -EINVAL;
+		}
+
+		*vport_id = vf ? hdev->vport[vf].vport_id : vport->vport_id;
+		tqps = hdev->vport[vf].nic.kinfo.num_tqps;
+
+		if (ring >= tqps) {
+			dev_err(&hdev->pdev->dev,
+				"Error: queue id (%u) > max tqp num (%u)\n",
+				ring, tqps - 1);
+			return -EINVAL;
+		}
+
+		*action = HCLGE_FD_ACTION_SELECT_QUEUE;
+		*queue_id = ring;
+	}
+
+	return 0;
+}
+
 static int hclge_add_fd_entry(struct hnae3_handle *handle,
 			      struct ethtool_rxnfc *cmd)
 {
 	struct hclge_vport *vport = hclge_get_vport(handle);
 	struct hclge_dev *hdev = vport->back;
+	struct hclge_fd_user_def_info info;
 	u16 dst_vport_id = 0, q_index = 0;
 	struct ethtool_rx_flow_spec *fs;
 	struct hclge_fd_rule *rule;
@@ -6150,51 +6551,22 @@ static int hclge_add_fd_entry(struct hnae3_handle *handle,
 		return -EOPNOTSUPP;
 	}
 
-	if (hclge_is_cls_flower_active(handle)) {
-		dev_err(&hdev->pdev->dev,
-			"please delete all exist cls flower rules first\n");
-		return -EINVAL;
-	}
-
 	fs = (struct ethtool_rx_flow_spec *)&cmd->fs;
 
-	ret = hclge_fd_check_spec(hdev, fs, &unused);
+	ret = hclge_fd_check_spec(hdev, fs, &unused, &info);
 	if (ret)
 		return ret;
 
-	if (fs->ring_cookie == RX_CLS_FLOW_DISC) {
-		action = HCLGE_FD_ACTION_DROP_PACKET;
-	} else {
-		u32 ring = ethtool_get_flow_spec_ring(fs->ring_cookie);
-		u8 vf = ethtool_get_flow_spec_ring_vf(fs->ring_cookie);
-		u16 tqps;
-
-		if (vf > hdev->num_req_vfs) {
-			dev_err(&hdev->pdev->dev,
-				"Error: vf id (%u) > max vf num (%u)\n",
-				vf, hdev->num_req_vfs);
-			return -EINVAL;
-		}
-
-		dst_vport_id = vf ? hdev->vport[vf].vport_id : vport->vport_id;
-		tqps = vf ? hdev->vport[vf].alloc_tqps : vport->alloc_tqps;
-
-		if (ring >= tqps) {
-			dev_err(&hdev->pdev->dev,
-				"Error: queue id (%u) > max tqp num (%u)\n",
-				ring, tqps - 1);
-			return -EINVAL;
-		}
-
-		action = HCLGE_FD_ACTION_SELECT_QUEUE;
-		q_index = ring;
-	}
+	ret = hclge_fd_parse_ring_cookie(hdev, fs->ring_cookie, &dst_vport_id,
+					 &action, &q_index);
+	if (ret)
+		return ret;
 
 	rule = kzalloc(sizeof(*rule), GFP_KERNEL);
 	if (!rule)
 		return -ENOMEM;
 
-	ret = hclge_fd_get_tuple(hdev, fs, rule);
+	ret = hclge_fd_get_tuple(hdev, fs, rule, &info);
 	if (ret) {
 		kfree(rule);
 		return ret;
@@ -6208,15 +6580,9 @@ static int hclge_add_fd_entry(struct hnae3_handle *handle,
 	rule->action = action;
 	rule->rule_type = HCLGE_FD_EP_ACTIVE;
 
-	/* to avoid rule conflict, when user configure rule by ethtool,
-	 * we need to clear all arfs rules
-	 */
-	spin_lock_bh(&hdev->fd_rule_lock);
-	hclge_clear_arfs_rules(handle);
-
-	ret = hclge_fd_config_rule(hdev, rule);
-
-	spin_unlock_bh(&hdev->fd_rule_lock);
+	ret = hclge_add_fd_entry_common(hdev, rule);
+	if (ret)
+		kfree(rule);
 
 	return ret;
 }
@@ -6237,32 +6603,30 @@ static int hclge_del_fd_entry(struct hnae3_handle *handle,
 	if (fs->location >= hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_1])
 		return -EINVAL;
 
-	if (hclge_is_cls_flower_active(handle) || !hdev->hclge_fd_rule_num ||
-	    !hclge_fd_rule_exist(hdev, fs->location)) {
+	spin_lock_bh(&hdev->fd_rule_lock);
+	if (hdev->fd_active_type == HCLGE_FD_TC_FLOWER_ACTIVE ||
+	    !test_bit(fs->location, hdev->fd_bmap)) {
 		dev_err(&hdev->pdev->dev,
 			"Delete fail, rule %u is inexistent\n", fs->location);
+		spin_unlock_bh(&hdev->fd_rule_lock);
 		return -ENOENT;
 	}
 
 	ret = hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true, fs->location,
 				   NULL, false);
 	if (ret)
-		return ret;
+		goto out;
 
-	spin_lock_bh(&hdev->fd_rule_lock);
-	ret = hclge_fd_update_rule_list(hdev, NULL, fs->location, false);
+	hclge_update_fd_list(hdev, HCLGE_FD_DELETED, fs->location, NULL);
 
+out:
 	spin_unlock_bh(&hdev->fd_rule_lock);
-
 	return ret;
 }
 
-/* make sure being called after lock up with fd_rule_lock */
-static void hclge_del_all_fd_entries(struct hnae3_handle *handle,
-				     bool clear_list)
+static void hclge_clear_fd_rules_in_list(struct hclge_dev *hdev,
+					 bool clear_list)
 {
-	struct hclge_vport *vport = hclge_get_vport(handle);
-	struct hclge_dev *hdev = vport->back;
 	struct hclge_fd_rule *rule;
 	struct hlist_node *node;
 	u16 location;
@@ -6270,6 +6634,8 @@ static void hclge_del_all_fd_entries(struct hnae3_handle *handle,
 	if (!hnae3_dev_fd_supported(hdev))
 		return;
 
+	spin_lock_bh(&hdev->fd_rule_lock);
+
 	for_each_set_bit(location, hdev->fd_bmap,
 			 hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_1])
 		hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true, location,
@@ -6286,6 +6652,14 @@ static void hclge_del_all_fd_entries(struct hnae3_handle *handle,
 		bitmap_zero(hdev->fd_bmap,
 			    hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_1]);
 	}
+
+	spin_unlock_bh(&hdev->fd_rule_lock);
+}
+
+static void hclge_del_all_fd_entries(struct hclge_dev *hdev)
+{
+	hclge_clear_fd_rules_in_list(hdev, true);
+	hclge_fd_disable_user_def(hdev);
 }
 
 static int hclge_restore_fd_entries(struct hnae3_handle *handle)
@@ -6294,7 +6668,6 @@ static int hclge_restore_fd_entries(struct hnae3_handle *handle)
 	struct hclge_dev *hdev = vport->back;
 	struct hclge_fd_rule *rule;
 	struct hlist_node *node;
-	int ret;
 
 	/* Return ok here, because reset error handling will check this
 	 * return value. If error is returned here, the reset process will
@@ -6309,25 +6682,11 @@ static int hclge_restore_fd_entries(struct hnae3_handle *handle)
 
 	spin_lock_bh(&hdev->fd_rule_lock);
 	hlist_for_each_entry_safe(rule, node, &hdev->fd_rule_list, rule_node) {
-		ret = hclge_config_action(hdev, HCLGE_FD_STAGE_1, rule);
-		if (!ret)
-			ret = hclge_config_key(hdev, HCLGE_FD_STAGE_1, rule);
-
-		if (ret) {
-			dev_warn(&hdev->pdev->dev,
-				 "Restore rule %u failed, remove it\n",
-				 rule->location);
-			clear_bit(rule->location, hdev->fd_bmap);
-			hlist_del(&rule->rule_node);
-			kfree(rule);
-			hdev->hclge_fd_rule_num--;
-		}
+		if (rule->state == HCLGE_FD_ACTIVE)
+			rule->state = HCLGE_FD_TO_ADD;
 	}
-
-	if (hdev->hclge_fd_rule_num)
-		hdev->fd_active_type = HCLGE_FD_EP_ACTIVE;
-
 	spin_unlock_bh(&hdev->fd_rule_lock);
+	set_bit(HCLGE_STATE_FD_TBL_CHANGED, &hdev->state);
 
 	return 0;
 }
@@ -6415,6 +6774,10 @@ static void hclge_fd_get_tcpip6_info(struct hclge_fd_rule *rule,
 		cpu_to_be32_array(spec_mask->ip6dst, rule->tuples_mask.dst_ip,
 				  IPV6_SIZE);
 
+	spec->tclass = rule->tuples.ip_tos;
+	spec_mask->tclass = rule->unused_tuple & BIT(INNER_IP_TOS) ?
+			0 : rule->tuples_mask.ip_tos;
+
 	spec->psrc = cpu_to_be16(rule->tuples.src_port);
 	spec_mask->psrc = rule->unused_tuple & BIT(INNER_SRC_PORT) ?
 			0 : cpu_to_be16(rule->tuples_mask.src_port);
@@ -6442,6 +6805,10 @@ static void hclge_fd_get_ip6_info(struct hclge_fd_rule *rule,
 		cpu_to_be32_array(spec_mask->ip6dst,
 				  rule->tuples_mask.dst_ip, IPV6_SIZE);
 
+	spec->tclass = rule->tuples.ip_tos;
+	spec_mask->tclass = rule->unused_tuple & BIT(INNER_IP_TOS) ?
+			0 : rule->tuples_mask.ip_tos;
+
 	spec->l4_proto = rule->tuples.ip_proto;
 	spec_mask->l4_proto = rule->unused_tuple & BIT(INNER_IP_PROTO) ?
 			0 : rule->tuples_mask.ip_proto;
@@ -6469,6 +6836,24 @@ static void hclge_fd_get_ether_info(struct hclge_fd_rule *rule,
 			0 : cpu_to_be16(rule->tuples_mask.ether_proto);
 }
 
+static void hclge_fd_get_user_def_info(struct ethtool_rx_flow_spec *fs,
+				       struct hclge_fd_rule *rule)
+{
+	if ((rule->unused_tuple & HCLGE_FD_TUPLE_USER_DEF_TUPLES) ==
+	    HCLGE_FD_TUPLE_USER_DEF_TUPLES) {
+		fs->h_ext.data[0] = 0;
+		fs->h_ext.data[1] = 0;
+		fs->m_ext.data[0] = 0;
+		fs->m_ext.data[1] = 0;
+	} else {
+		fs->h_ext.data[0] = cpu_to_be32(rule->ep.user_def.offset);
+		fs->h_ext.data[1] = cpu_to_be32(rule->ep.user_def.data);
+		fs->m_ext.data[0] =
+				cpu_to_be32(HCLGE_FD_USER_DEF_OFFSET_UNMASK);
+		fs->m_ext.data[1] = cpu_to_be32(rule->ep.user_def.data_mask);
+	}
+}
+
 static void hclge_fd_get_ext_info(struct ethtool_rx_flow_spec *fs,
 				  struct hclge_fd_rule *rule)
 {
@@ -6477,6 +6862,8 @@ static void hclge_fd_get_ext_info(struct ethtool_rx_flow_spec *fs,
 		fs->m_ext.vlan_tci =
 				rule->unused_tuple & BIT(INNER_VLAN_TAG_FST) ?
 				0 : cpu_to_be16(rule->tuples_mask.vlan_tag1);
+
+		hclge_fd_get_user_def_info(fs, rule);
 	}
 
 	if (fs->flow_type & FLOW_MAC_EXT) {
@@ -6588,6 +6975,9 @@ static int hclge_get_all_rules(struct hnae3_handle *handle,
 			return -EMSGSIZE;
 		}
 
+		if (rule->state == HCLGE_FD_TO_DEL)
+			continue;
+
 		rule_locs[cnt] = rule->location;
 		cnt++;
 	}
@@ -6669,9 +7059,7 @@ static int hclge_add_fd_entry_by_arfs(struct hnae3_handle *handle, u16 queue_id,
 	struct hclge_fd_rule_tuples new_tuples = {};
 	struct hclge_dev *hdev = vport->back;
 	struct hclge_fd_rule *rule;
-	u16 tmp_queue_id;
 	u16 bit_id;
-	int ret;
 
 	if (!hnae3_dev_fd_supported(hdev))
 		return -EOPNOTSUPP;
@@ -6707,34 +7095,20 @@ static int hclge_add_fd_entry_by_arfs(struct hnae3_handle *handle, u16 queue_id,
 			return -ENOMEM;
 		}
 
-		set_bit(bit_id, hdev->fd_bmap);
 		rule->location = bit_id;
 		rule->arfs.flow_id = flow_id;
 		rule->queue_id = queue_id;
 		hclge_fd_build_arfs_rule(&new_tuples, rule);
-		ret = hclge_fd_config_rule(hdev, rule);
-
-		spin_unlock_bh(&hdev->fd_rule_lock);
-
-		if (ret)
-			return ret;
-
-		return rule->location;
+		hclge_update_fd_list(hdev, HCLGE_FD_TO_ADD, rule->location,
+				     rule);
+		hdev->fd_active_type = HCLGE_FD_ARFS_ACTIVE;
+	} else if (rule->queue_id != queue_id) {
+		rule->queue_id = queue_id;
+		rule->state = HCLGE_FD_TO_ADD;
+		set_bit(HCLGE_STATE_FD_TBL_CHANGED, &hdev->state);
+		hclge_task_schedule(hdev, 0);
 	}
-
 	spin_unlock_bh(&hdev->fd_rule_lock);
-
-	if (rule->queue_id == queue_id)
-		return rule->location;
-
-	tmp_queue_id = rule->queue_id;
-	rule->queue_id = queue_id;
-	ret = hclge_config_action(hdev, HCLGE_FD_STAGE_1, rule);
-	if (ret) {
-		rule->queue_id = tmp_queue_id;
-		return ret;
-	}
-
 	return rule->location;
 }
 
@@ -6744,7 +7118,6 @@ static void hclge_rfs_filter_expire(struct hclge_dev *hdev)
 	struct hnae3_handle *handle = &hdev->vport[0].nic;
 	struct hclge_fd_rule *rule;
 	struct hlist_node *node;
-	HLIST_HEAD(del_list);
 
 	spin_lock_bh(&hdev->fd_rule_lock);
 	if (hdev->fd_active_type != HCLGE_FD_ARFS_ACTIVE) {
@@ -6752,33 +7125,50 @@ static void hclge_rfs_filter_expire(struct hclge_dev *hdev)
 		return;
 	}
 	hlist_for_each_entry_safe(rule, node, &hdev->fd_rule_list, rule_node) {
+		if (rule->state != HCLGE_FD_ACTIVE)
+			continue;
 		if (rps_may_expire_flow(handle->netdev, rule->queue_id,
 					rule->arfs.flow_id, rule->location)) {
-			hlist_del_init(&rule->rule_node);
-			hlist_add_head(&rule->rule_node, &del_list);
-			hdev->hclge_fd_rule_num--;
-			clear_bit(rule->location, hdev->fd_bmap);
+			rule->state = HCLGE_FD_TO_DEL;
+			set_bit(HCLGE_STATE_FD_TBL_CHANGED, &hdev->state);
 		}
 	}
 	spin_unlock_bh(&hdev->fd_rule_lock);
-
-	hlist_for_each_entry_safe(rule, node, &del_list, rule_node) {
-		hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true,
-				     rule->location, NULL, false);
-		kfree(rule);
-	}
 #endif
 }
 
 /* make sure being called after lock up with fd_rule_lock */
-static void hclge_clear_arfs_rules(struct hnae3_handle *handle)
+static int hclge_clear_arfs_rules(struct hclge_dev *hdev)
 {
 #ifdef CONFIG_RFS_ACCEL
-	struct hclge_vport *vport = hclge_get_vport(handle);
-	struct hclge_dev *hdev = vport->back;
+	struct hclge_fd_rule *rule;
+	struct hlist_node *node;
+	int ret;
 
-	if (hdev->fd_active_type == HCLGE_FD_ARFS_ACTIVE)
-		hclge_del_all_fd_entries(handle, true);
+	if (hdev->fd_active_type != HCLGE_FD_ARFS_ACTIVE)
+		return 0;
+
+	hlist_for_each_entry_safe(rule, node, &hdev->fd_rule_list, rule_node) {
+		switch (rule->state) {
+		case HCLGE_FD_TO_DEL:
+		case HCLGE_FD_ACTIVE:
+			ret = hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true,
+						   rule->location, NULL, false);
+			if (ret)
+				return ret;
+			fallthrough;
+		case HCLGE_FD_TO_ADD:
+			hclge_fd_dec_rule_cnt(hdev, rule->location);
+			hlist_del(&rule->rule_node);
+			kfree(rule);
+			break;
+		default:
+			break;
+		}
+	}
+	hclge_sync_fd_state(hdev);
+
+	return 0;
 #endif
 }
 
@@ -6961,12 +7351,6 @@ static int hclge_add_cls_flower(struct hnae3_handle *handle,
 	struct hclge_fd_rule *rule;
 	int ret;
 
-	if (hdev->fd_active_type == HCLGE_FD_EP_ACTIVE) {
-		dev_err(&hdev->pdev->dev,
-			"please remove all exist fd rules via ethtool first\n");
-		return -EINVAL;
-	}
-
 	ret = hclge_check_cls_flower(hdev, cls_flower, tc);
 	if (ret) {
 		dev_err(&hdev->pdev->dev,
@@ -6979,8 +7363,10 @@ static int hclge_add_cls_flower(struct hnae3_handle *handle,
 		return -ENOMEM;
 
 	ret = hclge_parse_cls_flower(hdev, cls_flower, rule);
-	if (ret)
-		goto err;
+	if (ret) {
+		kfree(rule);
+		return ret;
+	}
 
 	rule->action = HCLGE_FD_ACTION_SELECT_TC;
 	rule->cls_flower.tc = tc;
@@ -6989,22 +7375,10 @@ static int hclge_add_cls_flower(struct hnae3_handle *handle,
 	rule->cls_flower.cookie = cls_flower->cookie;
 	rule->rule_type = HCLGE_FD_TC_FLOWER_ACTIVE;
 
-	spin_lock_bh(&hdev->fd_rule_lock);
-	hclge_clear_arfs_rules(handle);
-
-	ret = hclge_fd_config_rule(hdev, rule);
-
-	spin_unlock_bh(&hdev->fd_rule_lock);
-
-	if (ret) {
-		dev_err(&hdev->pdev->dev,
-			"failed to add cls flower rule, ret = %d\n", ret);
-		goto err;
-	}
+	ret = hclge_add_fd_entry_common(hdev, rule);
+	if (ret)
+		kfree(rule);
 
-	return 0;
-err:
-	kfree(rule);
 	return ret;
 }
 
@@ -7041,25 +7415,66 @@ static int hclge_del_cls_flower(struct hnae3_handle *handle,
 	ret = hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true, rule->location,
 				   NULL, false);
 	if (ret) {
-		dev_err(&hdev->pdev->dev,
-			"failed to delete cls flower rule %u, ret = %d\n",
-			rule->location, ret);
 		spin_unlock_bh(&hdev->fd_rule_lock);
 		return ret;
 	}
 
-	ret = hclge_fd_update_rule_list(hdev, NULL, rule->location, false);
-	if (ret) {
-		dev_err(&hdev->pdev->dev,
-			"failed to delete cls flower rule %u in list, ret = %d\n",
-			rule->location, ret);
-		spin_unlock_bh(&hdev->fd_rule_lock);
-		return ret;
+	hclge_update_fd_list(hdev, HCLGE_FD_DELETED, rule->location, NULL);
+	spin_unlock_bh(&hdev->fd_rule_lock);
+
+	return 0;
+}
+
+static void hclge_sync_fd_list(struct hclge_dev *hdev, struct hlist_head *hlist)
+{
+	struct hclge_fd_rule *rule;
+	struct hlist_node *node;
+	int ret = 0;
+
+	if (!test_and_clear_bit(HCLGE_STATE_FD_TBL_CHANGED, &hdev->state))
+		return;
+
+	spin_lock_bh(&hdev->fd_rule_lock);
+
+	hlist_for_each_entry_safe(rule, node, hlist, rule_node) {
+		switch (rule->state) {
+		case HCLGE_FD_TO_ADD:
+			ret = hclge_fd_config_rule(hdev, rule);
+			if (ret)
+				goto out;
+			rule->state = HCLGE_FD_ACTIVE;
+			break;
+		case HCLGE_FD_TO_DEL:
+			ret = hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true,
+						   rule->location, NULL, false);
+			if (ret)
+				goto out;
+			hclge_fd_dec_rule_cnt(hdev, rule->location);
+			hclge_fd_free_node(hdev, rule);
+			break;
+		default:
+			break;
+		}
 	}
 
+out:
+	if (ret)
+		set_bit(HCLGE_STATE_FD_TBL_CHANGED, &hdev->state);
+
 	spin_unlock_bh(&hdev->fd_rule_lock);
+}
 
-	return 0;
+static void hclge_sync_fd_table(struct hclge_dev *hdev)
+{
+	if (test_and_clear_bit(HCLGE_STATE_FD_CLEAR_ALL, &hdev->state)) {
+		bool clear_list = hdev->fd_active_type == HCLGE_FD_ARFS_ACTIVE;
+
+		hclge_clear_fd_rules_in_list(hdev, clear_list);
+	}
+
+	hclge_sync_fd_user_def_cfg(hdev, false);
+
+	hclge_sync_fd_list(hdev, &hdev->fd_rule_list);
 }
 
 static bool hclge_get_hw_reset_stat(struct hnae3_handle *handle)
@@ -7099,18 +7514,15 @@ static void hclge_enable_fd(struct hnae3_handle *handle, bool enable)
 {
 	struct hclge_vport *vport = hclge_get_vport(handle);
 	struct hclge_dev *hdev = vport->back;
-	bool clear;
 
 	hdev->fd_en = enable;
-	clear = hdev->fd_active_type == HCLGE_FD_ARFS_ACTIVE;
 
-	if (!enable) {
-		spin_lock_bh(&hdev->fd_rule_lock);
-		hclge_del_all_fd_entries(handle, clear);
-		spin_unlock_bh(&hdev->fd_rule_lock);
-	} else {
+	if (!enable)
+		set_bit(HCLGE_STATE_FD_CLEAR_ALL, &hdev->state);
+	else
 		hclge_restore_fd_entries(handle);
-	}
+
+	hclge_task_schedule(hdev, 0);
 }
 
 static void hclge_cfg_mac_mode(struct hclge_dev *hdev, bool enable)
@@ -7581,7 +7993,7 @@ static void hclge_ae_stop(struct hnae3_handle *handle)
 
 	set_bit(HCLGE_STATE_DOWN, &hdev->state);
 	spin_lock_bh(&hdev->fd_rule_lock);
-	hclge_clear_arfs_rules(handle);
+	hclge_clear_arfs_rules(hdev);
 	spin_unlock_bh(&hdev->fd_rule_lock);
 
 	/* If it is not PF reset, the firmware will disable the MAC,
@@ -9617,7 +10029,7 @@ static void hclge_restore_hw_table(struct hclge_dev *hdev)
 	hclge_restore_mac_table_common(vport);
 	hclge_restore_vport_vlan_table(vport);
 	set_bit(HCLGE_STATE_PROMISC_CHANGED, &hdev->state);
-
+	set_bit(HCLGE_STATE_FD_USER_DEF_CHANGED, &hdev->state);
 	hclge_restore_fd_entries(handle);
 }
 
@@ -11306,6 +11718,7 @@ static void hclge_uninit_ae_dev(struct hnae3_ae_dev *ae_dev)
 	hclge_misc_affinity_teardown(hdev);
 	hclge_state_uninit(hdev);
 	hclge_uninit_mac_table(hdev);
+	hclge_del_all_fd_entries(hdev);
 
 	if (mac->phydev)
 		mdiobus_unregister(mac->mdio_bus);
@@ -12129,7 +12542,6 @@ static const struct hnae3_ae_ops hclge_ops = {
 	.get_link_mode = hclge_get_link_mode,
 	.add_fd_entry = hclge_add_fd_entry,
 	.del_fd_entry = hclge_del_fd_entry,
-	.del_all_fd_entries = hclge_del_all_fd_entries,
 	.get_fd_rule_cnt = hclge_get_fd_rule_cnt,
 	.get_fd_rule_info = hclge_get_fd_rule_info,
 	.get_fd_all_rules = hclge_get_all_rules,

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h

@@ -223,6 +223,9 @@ enum HCLGE_DEV_STATE {
 	HCLGE_STATE_LINK_UPDATING,
 	HCLGE_STATE_PROMISC_CHANGED,
 	HCLGE_STATE_RST_FAIL,
+	HCLGE_STATE_FD_TBL_CHANGED,
+	HCLGE_STATE_FD_CLEAR_ALL,
+	HCLGE_STATE_FD_USER_DEF_CHANGED,
 	HCLGE_STATE_MAX
 };
 
@@ -536,6 +539,9 @@ enum HCLGE_FD_TUPLE {
 	MAX_TUPLE,
 };
 
+#define HCLGE_FD_TUPLE_USER_DEF_TUPLES \
+	(BIT(INNER_L2_RSV) | BIT(INNER_L3_RSV) | BIT(INNER_L4_RSV))
+
 enum HCLGE_FD_META_DATA {
 	PACKET_TYPE_ID,
 	IP_FRAGEMENT,
@@ -548,9 +554,21 @@ enum HCLGE_FD_META_DATA {
 	MAX_META_DATA,
 };
 
+enum HCLGE_FD_KEY_OPT {
+	KEY_OPT_U8,
+	KEY_OPT_LE16,
+	KEY_OPT_LE32,
+	KEY_OPT_MAC,
+	KEY_OPT_IP,
+	KEY_OPT_VNI,
+};
+
 struct key_info {
 	u8 key_type;
 	u8 key_length; /* use bit as unit */
+	enum HCLGE_FD_KEY_OPT key_opt;
+	int offset;
+	int moffset;
 };
 
 #define MAX_KEY_LENGTH	400
@@ -558,6 +576,11 @@ struct key_info {
 #define MAX_KEY_BYTES	(MAX_KEY_DWORDS * 4)
 #define MAX_META_DATA_LENGTH	32
 
+#define HCLGE_FD_MAX_USER_DEF_OFFSET	9000
+#define HCLGE_FD_USER_DEF_DATA		GENMASK(15, 0)
+#define HCLGE_FD_USER_DEF_OFFSET	GENMASK(15, 0)
+#define HCLGE_FD_USER_DEF_OFFSET_UNMASK	GENMASK(15, 0)
+
 /* assigned by firmware, the real filter number for each pf may be less */
 #define MAX_FD_FILTER_NUM	4096
 #define HCLGE_ARFS_EXPIRE_INTERVAL	5UL
@@ -580,6 +603,33 @@ enum HCLGE_FD_ACTION {
 	HCLGE_FD_ACTION_SELECT_TC,
 };
 
+enum HCLGE_FD_NODE_STATE {
+	HCLGE_FD_TO_ADD,
+	HCLGE_FD_TO_DEL,
+	HCLGE_FD_ACTIVE,
+	HCLGE_FD_DELETED,
+};
+
+enum HCLGE_FD_USER_DEF_LAYER {
+	HCLGE_FD_USER_DEF_NONE,
+	HCLGE_FD_USER_DEF_L2,
+	HCLGE_FD_USER_DEF_L3,
+	HCLGE_FD_USER_DEF_L4,
+};
+
+#define HCLGE_FD_USER_DEF_LAYER_NUM 3
+struct hclge_fd_user_def_cfg {
+	u16 ref_cnt;
+	u16 offset;
+};
+
+struct hclge_fd_user_def_info {
+	enum HCLGE_FD_USER_DEF_LAYER layer;
+	u16 data;
+	u16 data_mask;
+	u16 offset;
+};
+
 struct hclge_fd_key_cfg {
 	u8 key_sel;
 	u8 inner_sipv6_word_en;
@@ -596,6 +646,7 @@ struct hclge_fd_cfg {
 	u32 rule_num[MAX_STAGE_NUM]; /* rule entry number */
 	u16 cnt_num[MAX_STAGE_NUM]; /* rule hit counter number */
 	struct hclge_fd_key_cfg key_cfg[MAX_STAGE_NUM];
+	struct hclge_fd_user_def_cfg user_def_cfg[HCLGE_FD_USER_DEF_LAYER_NUM];
 };
 
 #define IPV4_INDEX	3
@@ -612,6 +663,9 @@ struct hclge_fd_rule_tuples {
 	u16 dst_port;
 	u16 vlan_tag1;
 	u16 ether_proto;
+	u16 l2_user_def;
+	u16 l3_user_def;
+	u32 l4_user_def;
 	u8 ip_tos;
 	u8 ip_proto;
 };
@@ -630,11 +684,15 @@ struct hclge_fd_rule {
 		struct {
 			u16 flow_id; /* only used for arfs */
 		} arfs;
+		struct {
+			struct hclge_fd_user_def_info user_def;
+		} ep;
 	};
 	u16 queue_id;
 	u16 vf_id;
 	u16 location;
 	enum HCLGE_FD_ACTIVE_RULE_TYPE rule_type;
+	enum HCLGE_FD_NODE_STATE state;
 	u8 action;
 };