Merge branch 'Support-of-Flow-Director-in-HNS3-Ethernet-Driver-for-HiP08-Rev2-SoC'

Salil Mehta says:

====================
Support of Flow Director in HNS3 Ethernet Driver for HiP08 Rev2 SoC

This patch-set adds the support of FD(Flow Director) in the HNS3 PF driver
for HiP08 Rev2(0x21) SoC of Hisilicon. FD can be used in filtering the flows
and deciding to drop the flow or forward it to paricular queue.

Configuration consists of rules with input keys and actions. The rules are
stored in TCAM.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

7 files changed, 1721 insertions, 9 deletions

diff --git a/drivers/net/ethernet/hisilicon/hns3/hnae3.h b/drivers/net/ethernet/hisilicon/hns3/hnae3.h
index dea422c53738..591ee2ee4bf6 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hnae3.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hnae3.h
@@ -51,6 +51,7 @@
 #define HNAE3_KNIC_CLIENT_INITED_B		0x3
 #define HNAE3_UNIC_CLIENT_INITED_B		0x4
 #define HNAE3_ROCE_CLIENT_INITED_B		0x5
+#define HNAE3_DEV_SUPPORT_FD_B			0x6
 
 #define HNAE3_DEV_SUPPORT_ROCE_DCB_BITS (BIT(HNAE3_DEV_SUPPORT_DCB_B) |\
 		BIT(HNAE3_DEV_SUPPORT_ROCE_B))
@@ -61,6 +62,9 @@
 #define hnae3_dev_dcb_supported(hdev) \
 	hnae3_get_bit(hdev->ae_dev->flag, HNAE3_DEV_SUPPORT_DCB_B)
 
+#define hnae3_dev_fd_supported(hdev) \
+	hnae3_get_bit((hdev)->ae_dev->flag, HNAE3_DEV_SUPPORT_FD_B)
+
 #define ring_ptr_move_fw(ring, p) \
 	((ring)->p = ((ring)->p + 1) % (ring)->desc_num)
 #define ring_ptr_move_bw(ring, p) \
@@ -175,6 +179,7 @@ struct hnae3_ae_dev {
 	struct list_head node;
 	u32 flag;
 	enum hnae3_dev_type dev_type;
+	enum hnae3_reset_type reset_type;
 	void *priv;
 };
 
@@ -412,6 +417,20 @@ struct hnae3_ae_ops {
 	void (*get_link_mode)(struct hnae3_handle *handle,
 			      unsigned long *supported,
 			      unsigned long *advertising);
+	int (*add_fd_entry)(struct hnae3_handle *handle,
+			    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,
+				struct ethtool_rxnfc *cmd);
+	int (*get_fd_all_rules)(struct hnae3_handle *handle,
+				struct ethtool_rxnfc *cmd, u32 *rule_locs);
+	int (*restore_fd_rules)(struct hnae3_handle *handle);
+	void (*enable_fd)(struct hnae3_handle *handle, bool enable);
 };
 
 struct hnae3_dcb_ops {
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
index 1050652274a0..c2692563a4d9 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.c
@@ -1285,6 +1285,13 @@ static int hns3_nic_set_features(struct net_device *netdev,
 			return ret;
 	}
 
+	if ((changed & NETIF_F_NTUPLE) && h->ae_algo->ops->enable_fd) {
+		if (features & NETIF_F_NTUPLE)
+			h->ae_algo->ops->enable_fd(h, true);
+		else
+			h->ae_algo->ops->enable_fd(h, false);
+	}
+
 	netdev->features = features;
 	return 0;
 }
@@ -1622,6 +1629,13 @@ static void hns3_disable_sriov(struct pci_dev *pdev)
 	pci_disable_sriov(pdev);
 }
 
+static void hns3_get_dev_capability(struct pci_dev *pdev,
+				    struct hnae3_ae_dev *ae_dev)
+{
+	if (pdev->revision >= 0x21)
+		hnae3_set_bit(ae_dev->flag, HNAE3_DEV_SUPPORT_FD_B, 1);
+}
+
 /* hns3_probe - Device initialization routine
  * @pdev: PCI device information struct
  * @ent: entry in hns3_pci_tbl
@@ -1647,6 +1661,8 @@ static int hns3_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	ae_dev->pdev = pdev;
 	ae_dev->flag = ent->driver_data;
 	ae_dev->dev_type = HNAE3_DEV_KNIC;
+	ae_dev->reset_type = HNAE3_NONE_RESET;
+	hns3_get_dev_capability(pdev, ae_dev);
 	pci_set_drvdata(pdev, ae_dev);
 
 	hnae3_register_ae_dev(ae_dev);
@@ -1761,8 +1777,14 @@ static void hns3_set_default_feature(struct net_device *netdev)
 		NETIF_F_GSO_GRE_CSUM | NETIF_F_GSO_UDP_TUNNEL |
 		NETIF_F_GSO_UDP_TUNNEL_CSUM | NETIF_F_SCTP_CRC;
 
-	if (pdev->revision != 0x20)
+	if (pdev->revision >= 0x21) {
 		netdev->hw_features |= NETIF_F_HW_VLAN_CTAG_FILTER;
+
+		if (!(h->flags & HNAE3_SUPPORT_VF)) {
+			netdev->hw_features |= NETIF_F_NTUPLE;
+			netdev->features |= NETIF_F_NTUPLE;
+		}
+	}
 }
 
 static int hns3_alloc_buffer(struct hns3_enet_ring *ring,
@@ -3142,6 +3164,25 @@ static void hns3_uninit_mac_addr(struct net_device *netdev)
 		h->ae_algo->ops->rm_uc_addr(h, netdev->dev_addr);
 }
 
+static int hns3_restore_fd_rules(struct net_device *netdev)
+{
+	struct hnae3_handle *h = hns3_get_handle(netdev);
+	int ret = 0;
+
+	if (h->ae_algo->ops->restore_fd_rules)
+		ret = h->ae_algo->ops->restore_fd_rules(h);
+
+	return ret;
+}
+
+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 void hns3_nic_set_priv_ops(struct net_device *netdev)
 {
 	struct hns3_nic_priv *priv = netdev_priv(netdev);
@@ -3258,6 +3299,8 @@ static void hns3_client_uninit(struct hnae3_handle *handle, bool reset)
 	if (netdev->reg_state != NETREG_UNINITIALIZED)
 		unregister_netdev(netdev);
 
+	hns3_del_all_fd_rules(netdev, true);
+
 	hns3_force_clear_all_rx_ring(handle);
 
 	ret = hns3_nic_uninit_vector_data(priv);
@@ -3553,6 +3596,8 @@ static int hns3_reset_notify_init_enet(struct hnae3_handle *handle)
 	if (!(handle->flags & HNAE3_SUPPORT_VF))
 		hns3_restore_vlan(netdev);
 
+	hns3_restore_fd_rules(netdev);
+
 	/* Carrier off reporting is important to ethtool even BEFORE open */
 	netif_carrier_off(netdev);
 
@@ -3573,6 +3618,7 @@ static int hns3_reset_notify_init_enet(struct hnae3_handle *handle)
 
 static int hns3_reset_notify_uninit_enet(struct hnae3_handle *handle)
 {
+	struct hnae3_ae_dev *ae_dev = pci_get_drvdata(handle->pdev);
 	struct net_device *netdev = handle->kinfo.netdev;
 	struct hns3_nic_priv *priv = netdev_priv(netdev);
 	int ret;
@@ -3593,6 +3639,13 @@ static int hns3_reset_notify_uninit_enet(struct hnae3_handle *handle)
 
 	hns3_uninit_mac_addr(netdev);
 
+	/* it is cumbersome for hardware to pick-and-choose rules for deletion
+	 * from TCAM. Hence, for function reset software intervention is
+	 * required to delete the rules
+	 */
+	if (hns3_dev_ongoing_func_reset(ae_dev))
+		hns3_del_all_fd_rules(netdev, false);
+
 	return ret;
 }
 
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
index 27d704f4b306..ac881e8fc05d 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_enet.h
@@ -585,6 +585,11 @@ static inline void hns3_write_reg(void __iomem *base, u32 reg, u32 value)
 	writel(value, reg_addr + reg);
 }
 
+static inline bool hns3_dev_ongoing_func_reset(struct hnae3_ae_dev *ae_dev)
+{
+	return (ae_dev && (ae_dev->reset_type == HNAE3_FUNC_RESET));
+}
+
 #define hns3_write_dev(a, reg, value) \
 	hns3_write_reg((a)->io_base, (reg), (value))
 
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c b/drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c
index 8ad2c3eeb46b..7d79a074a214 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3_ethtool.c
@@ -699,20 +699,33 @@ static int hns3_get_rxnfc(struct net_device *netdev,
 {
 	struct hnae3_handle *h = hns3_get_handle(netdev);
 
-	if (!h->ae_algo || !h->ae_algo->ops || !h->ae_algo->ops->get_rss_tuple)
+	if (!h->ae_algo || !h->ae_algo->ops)
 		return -EOPNOTSUPP;
 
 	switch (cmd->cmd) {
 	case ETHTOOL_GRXRINGS:
-		cmd->data = h->kinfo.rss_size;
-		break;
+		cmd->data = h->kinfo.num_tqps;
+		return 0;
 	case ETHTOOL_GRXFH:
-		return h->ae_algo->ops->get_rss_tuple(h, cmd);
+		if (h->ae_algo->ops->get_rss_tuple)
+			return h->ae_algo->ops->get_rss_tuple(h, cmd);
+		return -EOPNOTSUPP;
+	case ETHTOOL_GRXCLSRLCNT:
+		if (h->ae_algo->ops->get_fd_rule_cnt)
+			return h->ae_algo->ops->get_fd_rule_cnt(h, cmd);
+		return -EOPNOTSUPP;
+	case ETHTOOL_GRXCLSRULE:
+		if (h->ae_algo->ops->get_fd_rule_info)
+			return h->ae_algo->ops->get_fd_rule_info(h, cmd);
+		return -EOPNOTSUPP;
+	case ETHTOOL_GRXCLSRLALL:
+		if (h->ae_algo->ops->get_fd_all_rules)
+			return h->ae_algo->ops->get_fd_all_rules(h, cmd,
+								 rule_locs);
+		return -EOPNOTSUPP;
 	default:
 		return -EOPNOTSUPP;
 	}
-
-	return 0;
 }
 
 static int hns3_change_all_ring_bd_num(struct hns3_nic_priv *priv,
@@ -795,12 +808,22 @@ static int hns3_set_rxnfc(struct net_device *netdev, struct ethtool_rxnfc *cmd)
 {
 	struct hnae3_handle *h = hns3_get_handle(netdev);
 
-	if (!h->ae_algo || !h->ae_algo->ops || !h->ae_algo->ops->set_rss_tuple)
+	if (!h->ae_algo || !h->ae_algo->ops)
 		return -EOPNOTSUPP;
 
 	switch (cmd->cmd) {
 	case ETHTOOL_SRXFH:
-		return h->ae_algo->ops->set_rss_tuple(h, cmd);
+		if (h->ae_algo->ops->set_rss_tuple)
+			return h->ae_algo->ops->set_rss_tuple(h, cmd);
+		return -EOPNOTSUPP;
+	case ETHTOOL_SRXCLSRLINS:
+		if (h->ae_algo->ops->add_fd_entry)
+			return h->ae_algo->ops->add_fd_entry(h, cmd);
+		return -EOPNOTSUPP;
+	case ETHTOOL_SRXCLSRLDEL:
+		if (h->ae_algo->ops->del_fd_entry)
+			return h->ae_algo->ops->del_fd_entry(h, cmd);
+		return -EOPNOTSUPP;
 	default:
 		return -EOPNOTSUPP;
 	}
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h
index 842dc3f05247..e5e66b27e03e 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_cmd.h
@@ -190,6 +190,13 @@ enum hclge_opcode_type {
 	HCLGE_OPC_VLAN_FILTER_PF_CFG	= 0x1101,
 	HCLGE_OPC_VLAN_FILTER_VF_CFG	= 0x1102,
 
+	/* Flow Director commands */
+	HCLGE_OPC_FD_MODE_CTRL		= 0x1200,
+	HCLGE_OPC_FD_GET_ALLOCATION	= 0x1201,
+	HCLGE_OPC_FD_KEY_CONFIG		= 0x1202,
+	HCLGE_OPC_FD_TCAM_OP		= 0x1203,
+	HCLGE_OPC_FD_AD_OP		= 0x1204,
+
 	/* MDIO command */
 	HCLGE_OPC_MDIO_CONFIG		= 0x1900,
 
@@ -819,6 +826,76 @@ struct hclge_set_led_state_cmd {
 	u8 rsv2[20];
 };
 
+struct hclge_get_fd_mode_cmd {
+	u8 mode;
+	u8 enable;
+	u8 rsv[22];
+};
+
+struct hclge_get_fd_allocation_cmd {
+	__le32 stage1_entry_num;
+	__le32 stage2_entry_num;
+	__le16 stage1_counter_num;
+	__le16 stage2_counter_num;
+	u8 rsv[12];
+};
+
+struct hclge_set_fd_key_config_cmd {
+	u8 stage;
+	u8 key_select;
+	u8 inner_sipv6_word_en;
+	u8 inner_dipv6_word_en;
+	u8 outer_sipv6_word_en;
+	u8 outer_dipv6_word_en;
+	u8 rsv1[2];
+	__le32 tuple_mask;
+	__le32 meta_data_mask;
+	u8 rsv2[8];
+};
+
+#define HCLGE_FD_EPORT_SW_EN_B		0
+struct hclge_fd_tcam_config_1_cmd {
+	u8 stage;
+	u8 xy_sel;
+	u8 port_info;
+	u8 rsv1[1];
+	__le32 index;
+	u8 entry_vld;
+	u8 rsv2[7];
+	u8 tcam_data[8];
+};
+
+struct hclge_fd_tcam_config_2_cmd {
+	u8 tcam_data[24];
+};
+
+struct hclge_fd_tcam_config_3_cmd {
+	u8 tcam_data[20];
+	u8 rsv[4];
+};
+
+#define HCLGE_FD_AD_DROP_B		0
+#define HCLGE_FD_AD_DIRECT_QID_B	1
+#define HCLGE_FD_AD_QID_S		2
+#define HCLGE_FD_AD_QID_M		GENMASK(12, 2)
+#define HCLGE_FD_AD_USE_COUNTER_B	12
+#define HCLGE_FD_AD_COUNTER_NUM_S	13
+#define HCLGE_FD_AD_COUNTER_NUM_M	GENMASK(20, 13)
+#define HCLGE_FD_AD_NXT_STEP_B		20
+#define HCLGE_FD_AD_NXT_KEY_S		21
+#define HCLGE_FD_AD_NXT_KEY_M		GENMASK(26, 21)
+#define HCLGE_FD_AD_WR_RULE_ID_B	0
+#define HCLGE_FD_AD_RULE_ID_S		1
+#define HCLGE_FD_AD_RULE_ID_M		GENMASK(13, 1)
+
+struct hclge_fd_ad_config_cmd {
+	u8 stage;
+	u8 rsv1[3];
+	__le32 index;
+	__le64 ad_data;
+	u8 rsv2[8];
+};
+
 int hclge_cmd_init(struct hclge_dev *hdev);
 static inline void hclge_write_reg(void __iomem *base, u32 reg, u32 value)
 {
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index c17ceeefa453..21ca4af3b37a 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -2513,8 +2513,13 @@ static void hclge_clear_reset_cause(struct hclge_dev *hdev)
 
 static void hclge_reset(struct hclge_dev *hdev)
 {
+	struct hnae3_ae_dev *ae_dev = pci_get_drvdata(hdev->pdev);
 	struct hnae3_handle *handle;
 
+	/* Initialize ae_dev reset status as well, in case enet layer wants to
+	 * know if device is undergoing reset
+	 */
+	ae_dev->reset_type = hdev->reset_type;
 	/* perform reset of the stack & ae device for a client */
 	handle = &hdev->vport[0].nic;
 	rtnl_lock();
@@ -2535,6 +2540,7 @@ static void hclge_reset(struct hclge_dev *hdev)
 	hclge_notify_client(hdev, HNAE3_UP_CLIENT);
 	handle->last_reset_time = jiffies;
 	rtnl_unlock();
+	ae_dev->reset_type = HNAE3_NONE_RESET;
 }
 
 static void hclge_reset_event(struct hnae3_handle *handle)
@@ -3328,6 +3334,1281 @@ static void hclge_set_promisc_mode(struct hnae3_handle *handle, bool en_uc_pmc,
 	hclge_cmd_set_promisc_mode(hdev, &param);
 }
 
+static int hclge_get_fd_mode(struct hclge_dev *hdev, u8 *fd_mode)
+{
+	struct hclge_get_fd_mode_cmd *req;
+	struct hclge_desc desc;
+	int ret;
+
+	hclge_cmd_setup_basic_desc(&desc, HCLGE_OPC_FD_MODE_CTRL, true);
+
+	req = (struct hclge_get_fd_mode_cmd *)desc.data;
+
+	ret = hclge_cmd_send(&hdev->hw, &desc, 1);
+	if (ret) {
+		dev_err(&hdev->pdev->dev, "get fd mode fail, ret=%d\n", ret);
+		return ret;
+	}
+
+	*fd_mode = req->mode;
+
+	return ret;
+}
+
+static int hclge_get_fd_allocation(struct hclge_dev *hdev,
+				   u32 *stage1_entry_num,
+				   u32 *stage2_entry_num,
+				   u16 *stage1_counter_num,
+				   u16 *stage2_counter_num)
+{
+	struct hclge_get_fd_allocation_cmd *req;
+	struct hclge_desc desc;
+	int ret;
+
+	hclge_cmd_setup_basic_desc(&desc, HCLGE_OPC_FD_GET_ALLOCATION, true);
+
+	req = (struct hclge_get_fd_allocation_cmd *)desc.data;
+
+	ret = hclge_cmd_send(&hdev->hw, &desc, 1);
+	if (ret) {
+		dev_err(&hdev->pdev->dev, "query fd allocation fail, ret=%d\n",
+			ret);
+		return ret;
+	}
+
+	*stage1_entry_num = le32_to_cpu(req->stage1_entry_num);
+	*stage2_entry_num = le32_to_cpu(req->stage2_entry_num);
+	*stage1_counter_num = le16_to_cpu(req->stage1_counter_num);
+	*stage2_counter_num = le16_to_cpu(req->stage2_counter_num);
+
+	return ret;
+}
+
+static int hclge_set_fd_key_config(struct hclge_dev *hdev, int stage_num)
+{
+	struct hclge_set_fd_key_config_cmd *req;
+	struct hclge_fd_key_cfg *stage;
+	struct hclge_desc desc;
+	int ret;
+
+	hclge_cmd_setup_basic_desc(&desc, HCLGE_OPC_FD_KEY_CONFIG, false);
+
+	req = (struct hclge_set_fd_key_config_cmd *)desc.data;
+	stage = &hdev->fd_cfg.key_cfg[stage_num];
+	req->stage = stage_num;
+	req->key_select = stage->key_sel;
+	req->inner_sipv6_word_en = stage->inner_sipv6_word_en;
+	req->inner_dipv6_word_en = stage->inner_dipv6_word_en;
+	req->outer_sipv6_word_en = stage->outer_sipv6_word_en;
+	req->outer_dipv6_word_en = stage->outer_dipv6_word_en;
+	req->tuple_mask = cpu_to_le32(~stage->tuple_active);
+	req->meta_data_mask = cpu_to_le32(~stage->meta_data_active);
+
+	ret = hclge_cmd_send(&hdev->hw, &desc, 1);
+	if (ret)
+		dev_err(&hdev->pdev->dev, "set fd key fail, ret=%d\n", ret);
+
+	return ret;
+}
+
+static int hclge_init_fd_config(struct hclge_dev *hdev)
+{
+#define LOW_2_WORDS		0x03
+	struct hclge_fd_key_cfg *key_cfg;
+	int ret;
+
+	if (!hnae3_dev_fd_supported(hdev))
+		return 0;
+
+	ret = hclge_get_fd_mode(hdev, &hdev->fd_cfg.fd_mode);
+	if (ret)
+		return ret;
+
+	switch (hdev->fd_cfg.fd_mode) {
+	case HCLGE_FD_MODE_DEPTH_2K_WIDTH_400B_STAGE_1:
+		hdev->fd_cfg.max_key_length = MAX_KEY_LENGTH;
+		break;
+	case HCLGE_FD_MODE_DEPTH_4K_WIDTH_200B_STAGE_1:
+		hdev->fd_cfg.max_key_length = MAX_KEY_LENGTH / 2;
+		break;
+	default:
+		dev_err(&hdev->pdev->dev,
+			"Unsupported flow director mode %d\n",
+			hdev->fd_cfg.fd_mode);
+		return -EOPNOTSUPP;
+	}
+
+	hdev->fd_cfg.fd_en = true;
+	hdev->fd_cfg.proto_support =
+		TCP_V4_FLOW | UDP_V4_FLOW | SCTP_V4_FLOW | TCP_V6_FLOW |
+		UDP_V6_FLOW | SCTP_V6_FLOW | IPV4_USER_FLOW | IPV6_USER_FLOW;
+	key_cfg = &hdev->fd_cfg.key_cfg[HCLGE_FD_STAGE_1];
+	key_cfg->key_sel = HCLGE_FD_KEY_BASE_ON_TUPLE,
+	key_cfg->inner_sipv6_word_en = LOW_2_WORDS;
+	key_cfg->inner_dipv6_word_en = LOW_2_WORDS;
+	key_cfg->outer_sipv6_word_en = 0;
+	key_cfg->outer_dipv6_word_en = 0;
+
+	key_cfg->tuple_active = BIT(INNER_VLAN_TAG_FST) | BIT(INNER_ETH_TYPE) |
+				BIT(INNER_IP_PROTO) | BIT(INNER_IP_TOS) |
+				BIT(INNER_SRC_IP) | BIT(INNER_DST_IP) |
+				BIT(INNER_SRC_PORT) | BIT(INNER_DST_PORT);
+
+	/* If use max 400bit key, we can support tuples for ether type */
+	if (hdev->fd_cfg.max_key_length == MAX_KEY_LENGTH) {
+		hdev->fd_cfg.proto_support |= ETHER_FLOW;
+		key_cfg->tuple_active |=
+				BIT(INNER_DST_MAC) | BIT(INNER_SRC_MAC);
+	}
+
+	/* roce_type is used to filter roce frames
+	 * dst_vport is used to specify the rule
+	 */
+	key_cfg->meta_data_active = BIT(ROCE_TYPE) | BIT(DST_VPORT);
+
+	ret = hclge_get_fd_allocation(hdev,
+				      &hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_1],
+				      &hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_2],
+				      &hdev->fd_cfg.cnt_num[HCLGE_FD_STAGE_1],
+				      &hdev->fd_cfg.cnt_num[HCLGE_FD_STAGE_2]);
+	if (ret)
+		return ret;
+
+	return hclge_set_fd_key_config(hdev, HCLGE_FD_STAGE_1);
+}
+
+static int hclge_fd_tcam_config(struct hclge_dev *hdev, u8 stage, bool sel_x,
+				int loc, u8 *key, bool is_add)
+{
+	struct hclge_fd_tcam_config_1_cmd *req1;
+	struct hclge_fd_tcam_config_2_cmd *req2;
+	struct hclge_fd_tcam_config_3_cmd *req3;
+	struct hclge_desc desc[3];
+	int ret;
+
+	hclge_cmd_setup_basic_desc(&desc[0], HCLGE_OPC_FD_TCAM_OP, false);
+	desc[0].flag |= cpu_to_le16(HCLGE_CMD_FLAG_NEXT);
+	hclge_cmd_setup_basic_desc(&desc[1], HCLGE_OPC_FD_TCAM_OP, false);
+	desc[1].flag |= cpu_to_le16(HCLGE_CMD_FLAG_NEXT);
+	hclge_cmd_setup_basic_desc(&desc[2], HCLGE_OPC_FD_TCAM_OP, false);
+
+	req1 = (struct hclge_fd_tcam_config_1_cmd *)desc[0].data;
+	req2 = (struct hclge_fd_tcam_config_2_cmd *)desc[1].data;
+	req3 = (struct hclge_fd_tcam_config_3_cmd *)desc[2].data;
+
+	req1->stage = stage;
+	req1->xy_sel = sel_x ? 1 : 0;
+	hnae3_set_bit(req1->port_info, HCLGE_FD_EPORT_SW_EN_B, 0);
+	req1->index = cpu_to_le32(loc);
+	req1->entry_vld = sel_x ? is_add : 0;
+
+	if (key) {
+		memcpy(req1->tcam_data, &key[0], sizeof(req1->tcam_data));
+		memcpy(req2->tcam_data, &key[sizeof(req1->tcam_data)],
+		       sizeof(req2->tcam_data));
+		memcpy(req3->tcam_data, &key[sizeof(req1->tcam_data) +
+		       sizeof(req2->tcam_data)], sizeof(req3->tcam_data));
+	}
+
+	ret = hclge_cmd_send(&hdev->hw, desc, 3);
+	if (ret)
+		dev_err(&hdev->pdev->dev,
+			"config tcam key fail, ret=%d\n",
+			ret);
+
+	return ret;
+}
+
+static int hclge_fd_ad_config(struct hclge_dev *hdev, u8 stage, int loc,
+			      struct hclge_fd_ad_data *action)
+{
+	struct hclge_fd_ad_config_cmd *req;
+	struct hclge_desc desc;
+	u64 ad_data = 0;
+	int ret;
+
+	hclge_cmd_setup_basic_desc(&desc, HCLGE_OPC_FD_AD_OP, false);
+
+	req = (struct hclge_fd_ad_config_cmd *)desc.data;
+	req->index = cpu_to_le32(loc);
+	req->stage = stage;
+
+	hnae3_set_bit(ad_data, HCLGE_FD_AD_WR_RULE_ID_B,
+		      action->write_rule_id_to_bd);
+	hnae3_set_field(ad_data, HCLGE_FD_AD_RULE_ID_M, HCLGE_FD_AD_RULE_ID_S,
+			action->rule_id);
+	ad_data <<= 32;
+	hnae3_set_bit(ad_data, HCLGE_FD_AD_DROP_B, action->drop_packet);
+	hnae3_set_bit(ad_data, HCLGE_FD_AD_DIRECT_QID_B,
+		      action->forward_to_direct_queue);
+	hnae3_set_field(ad_data, HCLGE_FD_AD_QID_M, HCLGE_FD_AD_QID_S,
+			action->queue_id);
+	hnae3_set_bit(ad_data, HCLGE_FD_AD_USE_COUNTER_B, action->use_counter);
+	hnae3_set_field(ad_data, HCLGE_FD_AD_COUNTER_NUM_M,
+			HCLGE_FD_AD_COUNTER_NUM_S, action->counter_id);
+	hnae3_set_bit(ad_data, HCLGE_FD_AD_NXT_STEP_B, action->use_next_stage);
+	hnae3_set_field(ad_data, HCLGE_FD_AD_NXT_KEY_M, HCLGE_FD_AD_NXT_KEY_S,
+			action->counter_id);
+
+	req->ad_data = cpu_to_le64(ad_data);
+	ret = hclge_cmd_send(&hdev->hw, &desc, 1);
+	if (ret)
+		dev_err(&hdev->pdev->dev, "fd ad config fail, ret=%d\n", ret);
+
+	return ret;
+}
+
+static bool hclge_fd_convert_tuple(u32 tuple_bit, u8 *key_x, u8 *key_y,
+				   struct hclge_fd_rule *rule)
+{
+	u16 tmp_x_s, tmp_y_s;
+	u32 tmp_x_l, tmp_y_l;
+	int i;
+
+	if (rule->unused_tuple & tuple_bit)
+		return true;
+
+	switch (tuple_bit) {
+	case 0:
+		return false;
+	case BIT(INNER_DST_MAC):
+		for (i = 0; i < 6; i++) {
+			calc_x(key_x[5 - i], rule->tuples.dst_mac[i],
+			       rule->tuples_mask.dst_mac[i]);
+			calc_y(key_y[5 - i], rule->tuples.dst_mac[i],
+			       rule->tuples_mask.dst_mac[i]);
+		}
+
+		return true;
+	case BIT(INNER_SRC_MAC):
+		for (i = 0; i < 6; i++) {
+			calc_x(key_x[5 - i], rule->tuples.src_mac[i],
+			       rule->tuples.src_mac[i]);
+			calc_y(key_y[5 - i], rule->tuples.src_mac[i],
+			       rule->tuples.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);
+
+		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);
+		*(__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[3],
+		       rule->tuples_mask.src_ip[3]);
+		calc_y(tmp_y_l, rule->tuples.src_ip[3],
+		       rule->tuples_mask.src_ip[3]);
+		*(__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[3],
+		       rule->tuples_mask.dst_ip[3]);
+		calc_y(tmp_y_l, rule->tuples.dst_ip[3],
+		       rule->tuples_mask.dst_ip[3]);
+		*(__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);
+
+		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);
+
+		return true;
+	default:
+		return false;
+	}
+}
+
+static u32 hclge_get_port_number(enum HLCGE_PORT_TYPE port_type, u8 pf_id,
+				 u8 vf_id, u8 network_port_id)
+{
+	u32 port_number = 0;
+
+	if (port_type == HOST_PORT) {
+		hnae3_set_field(port_number, HCLGE_PF_ID_M, HCLGE_PF_ID_S,
+				pf_id);
+		hnae3_set_field(port_number, HCLGE_VF_ID_M, HCLGE_VF_ID_S,
+				vf_id);
+		hnae3_set_bit(port_number, HCLGE_PORT_TYPE_B, HOST_PORT);
+	} else {
+		hnae3_set_field(port_number, HCLGE_NETWORK_PORT_ID_M,
+				HCLGE_NETWORK_PORT_ID_S, network_port_id);
+		hnae3_set_bit(port_number, HCLGE_PORT_TYPE_B, NETWORK_PORT);
+	}
+
+	return port_number;
+}
+
+static void hclge_fd_convert_meta_data(struct hclge_fd_key_cfg *key_cfg,
+				       __le32 *key_x, __le32 *key_y,
+				       struct hclge_fd_rule *rule)
+{
+	u32 tuple_bit, meta_data = 0, tmp_x, tmp_y, port_number;
+	u8 cur_pos = 0, tuple_size, shift_bits;
+	int i;
+
+	for (i = 0; i < MAX_META_DATA; i++) {
+		tuple_size = meta_data_key_info[i].key_length;
+		tuple_bit = key_cfg->meta_data_active & BIT(i);
+
+		switch (tuple_bit) {
+		case BIT(ROCE_TYPE):
+			hnae3_set_bit(meta_data, cur_pos, NIC_PACKET);
+			cur_pos += tuple_size;
+			break;
+		case BIT(DST_VPORT):
+			port_number = hclge_get_port_number(HOST_PORT, 0,
+							    rule->vf_id, 0);
+			hnae3_set_field(meta_data,
+					GENMASK(cur_pos + tuple_size, cur_pos),
+					cur_pos, port_number);
+			cur_pos += tuple_size;
+			break;
+		default:
+			break;
+		}
+	}
+
+	calc_x(tmp_x, meta_data, 0xFFFFFFFF);
+	calc_y(tmp_y, meta_data, 0xFFFFFFFF);
+	shift_bits = sizeof(meta_data) * 8 - cur_pos;
+
+	*key_x = cpu_to_le32(tmp_x << shift_bits);
+	*key_y = cpu_to_le32(tmp_y << shift_bits);
+}
+
+/* A complete key is combined with meta data key and tuple key.
+ * Meta data key is stored at the MSB region, and tuple key is stored at
+ * the LSB region, unused bits will be filled 0.
+ */
+static int hclge_config_key(struct hclge_dev *hdev, u8 stage,
+			    struct hclge_fd_rule *rule)
+{
+	struct hclge_fd_key_cfg *key_cfg = &hdev->fd_cfg.key_cfg[stage];
+	u8 key_x[MAX_KEY_BYTES], key_y[MAX_KEY_BYTES];
+	u8 *cur_key_x, *cur_key_y;
+	int i, ret, tuple_size;
+	u8 meta_data_region;
+
+	memset(key_x, 0, sizeof(key_x));
+	memset(key_y, 0, sizeof(key_y));
+	cur_key_x = key_x;
+	cur_key_y = key_y;
+
+	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);
+
+		tuple_valid = hclge_fd_convert_tuple(check_tuple, cur_key_x,
+						     cur_key_y, rule);
+		if (tuple_valid) {
+			cur_key_x += tuple_size;
+			cur_key_y += tuple_size;
+		}
+	}
+
+	meta_data_region = hdev->fd_cfg.max_key_length / 8 -
+			MAX_META_DATA_LENGTH / 8;
+
+	hclge_fd_convert_meta_data(key_cfg,
+				   (__le32 *)(key_x + meta_data_region),
+				   (__le32 *)(key_y + meta_data_region),
+				   rule);
+
+	ret = hclge_fd_tcam_config(hdev, stage, false, rule->location, key_y,
+				   true);
+	if (ret) {
+		dev_err(&hdev->pdev->dev,
+			"fd key_y config fail, loc=%d, ret=%d\n",
+			rule->queue_id, ret);
+		return ret;
+	}
+
+	ret = hclge_fd_tcam_config(hdev, stage, true, rule->location, key_x,
+				   true);
+	if (ret)
+		dev_err(&hdev->pdev->dev,
+			"fd key_x config fail, loc=%d, ret=%d\n",
+			rule->queue_id, ret);
+	return ret;
+}
+
+static int hclge_config_action(struct hclge_dev *hdev, u8 stage,
+			       struct hclge_fd_rule *rule)
+{
+	struct hclge_fd_ad_data ad_data;
+
+	ad_data.ad_id = rule->location;
+
+	if (rule->action == HCLGE_FD_ACTION_DROP_PACKET) {
+		ad_data.drop_packet = true;
+		ad_data.forward_to_direct_queue = false;
+		ad_data.queue_id = 0;
+	} else {
+		ad_data.drop_packet = false;
+		ad_data.forward_to_direct_queue = true;
+		ad_data.queue_id = rule->queue_id;
+	}
+
+	ad_data.use_counter = false;
+	ad_data.counter_id = 0;
+
+	ad_data.use_next_stage = false;
+	ad_data.next_input_key = 0;
+
+	ad_data.write_rule_id_to_bd = true;
+	ad_data.rule_id = rule->location;
+
+	return hclge_fd_ad_config(hdev, stage, ad_data.ad_id, &ad_data);
+}
+
+static int hclge_fd_check_spec(struct hclge_dev *hdev,
+			       struct ethtool_rx_flow_spec *fs, u32 *unused)
+{
+	struct ethtool_tcpip4_spec *tcp_ip4_spec;
+	struct ethtool_usrip4_spec *usr_ip4_spec;
+	struct ethtool_tcpip6_spec *tcp_ip6_spec;
+	struct ethtool_usrip6_spec *usr_ip6_spec;
+	struct ethhdr *ether_spec;
+
+	if (fs->location >= hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_1])
+		return -EINVAL;
+
+	if (!(fs->flow_type & hdev->fd_cfg.proto_support))
+		return -EOPNOTSUPP;
+
+	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;
+	}
+
+	switch (fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+	case SCTP_V4_FLOW:
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+		tcp_ip4_spec = &fs->h_u.tcp_ip4_spec;
+		*unused |= BIT(INNER_SRC_MAC) | BIT(INNER_DST_MAC);
+
+		if (!tcp_ip4_spec->ip4src)
+			*unused |= BIT(INNER_SRC_IP);
+
+		if (!tcp_ip4_spec->ip4dst)
+			*unused |= BIT(INNER_DST_IP);
+
+		if (!tcp_ip4_spec->psrc)
+			*unused |= BIT(INNER_SRC_PORT);
+
+		if (!tcp_ip4_spec->pdst)
+			*unused |= BIT(INNER_DST_PORT);
+
+		if (!tcp_ip4_spec->tos)
+			*unused |= BIT(INNER_IP_TOS);
+
+		break;
+	case IP_USER_FLOW:
+		usr_ip4_spec = &fs->h_u.usr_ip4_spec;
+		*unused |= BIT(INNER_SRC_MAC) | BIT(INNER_DST_MAC) |
+			BIT(INNER_SRC_PORT) | BIT(INNER_DST_PORT);
+
+		if (!usr_ip4_spec->ip4src)
+			*unused |= BIT(INNER_SRC_IP);
+
+		if (!usr_ip4_spec->ip4dst)
+			*unused |= BIT(INNER_DST_IP);
+
+		if (!usr_ip4_spec->tos)
+			*unused |= BIT(INNER_IP_TOS);
+
+		if (!usr_ip4_spec->proto)
+			*unused |= BIT(INNER_IP_PROTO);
+
+		if (usr_ip4_spec->l4_4_bytes)
+			return -EOPNOTSUPP;
+
+		if (usr_ip4_spec->ip_ver != ETH_RX_NFC_IP4)
+			return -EOPNOTSUPP;
+
+		break;
+	case SCTP_V6_FLOW:
+	case TCP_V6_FLOW:
+	case UDP_V6_FLOW:
+		tcp_ip6_spec = &fs->h_u.tcp_ip6_spec;
+		*unused |= BIT(INNER_SRC_MAC) | BIT(INNER_DST_MAC) |
+			BIT(INNER_IP_TOS);
+
+		if (!tcp_ip6_spec->ip6src[0] && !tcp_ip6_spec->ip6src[1] &&
+		    !tcp_ip6_spec->ip6src[2] && !tcp_ip6_spec->ip6src[3])
+			*unused |= BIT(INNER_SRC_IP);
+
+		if (!tcp_ip6_spec->ip6dst[0] && !tcp_ip6_spec->ip6dst[1] &&
+		    !tcp_ip6_spec->ip6dst[2] && !tcp_ip6_spec->ip6dst[3])
+			*unused |= BIT(INNER_DST_IP);
+
+		if (!tcp_ip6_spec->psrc)
+			*unused |= BIT(INNER_SRC_PORT);
+
+		if (!tcp_ip6_spec->pdst)
+			*unused |= BIT(INNER_DST_PORT);
+
+		if (tcp_ip6_spec->tclass)
+			return -EOPNOTSUPP;
+
+		break;
+	case IPV6_USER_FLOW:
+		usr_ip6_spec = &fs->h_u.usr_ip6_spec;
+		*unused |= BIT(INNER_SRC_MAC) | BIT(INNER_DST_MAC) |
+			BIT(INNER_IP_TOS) | BIT(INNER_SRC_PORT) |
+			BIT(INNER_DST_PORT);
+
+		if (!usr_ip6_spec->ip6src[0] && !usr_ip6_spec->ip6src[1] &&
+		    !usr_ip6_spec->ip6src[2] && !usr_ip6_spec->ip6src[3])
+			*unused |= BIT(INNER_SRC_IP);
+
+		if (!usr_ip6_spec->ip6dst[0] && !usr_ip6_spec->ip6dst[1] &&
+		    !usr_ip6_spec->ip6dst[2] && !usr_ip6_spec->ip6dst[3])
+			*unused |= BIT(INNER_DST_IP);
+
+		if (!usr_ip6_spec->l4_proto)
+			*unused |= BIT(INNER_IP_PROTO);
+
+		if (usr_ip6_spec->tclass)
+			return -EOPNOTSUPP;
+
+		if (usr_ip6_spec->l4_4_bytes)
+			return -EOPNOTSUPP;
+
+		break;
+	case ETHER_FLOW:
+		ether_spec = &fs->h_u.ether_spec;
+		*unused |= BIT(INNER_SRC_IP) | BIT(INNER_DST_IP) |
+			BIT(INNER_SRC_PORT) | BIT(INNER_DST_PORT) |
+			BIT(INNER_IP_TOS) | BIT(INNER_IP_PROTO);
+
+		if (is_zero_ether_addr(ether_spec->h_source))
+			*unused |= BIT(INNER_SRC_MAC);
+
+		if (is_zero_ether_addr(ether_spec->h_dest))
+			*unused |= BIT(INNER_DST_MAC);
+
+		if (!ether_spec->h_proto)
+			*unused |= BIT(INNER_ETH_TYPE);
+
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	if ((fs->flow_type & FLOW_EXT)) {
+		if (fs->h_ext.vlan_etype)
+			return -EOPNOTSUPP;
+		if (!fs->h_ext.vlan_tci)
+			*unused |= BIT(INNER_VLAN_TAG_FST);
+
+		if (fs->m_ext.vlan_tci) {
+			if (be16_to_cpu(fs->h_ext.vlan_tci) >= VLAN_N_VID)
+				return -EINVAL;
+		}
+	} else {
+		*unused |= BIT(INNER_VLAN_TAG_FST);
+	}
+
+	if (fs->flow_type & FLOW_MAC_EXT) {
+		if (!(hdev->fd_cfg.proto_support & ETHER_FLOW))
+			return -EOPNOTSUPP;
+
+		if (is_zero_ether_addr(fs->h_ext.h_dest))
+			*unused |= BIT(INNER_DST_MAC);
+		else
+			*unused &= ~(BIT(INNER_DST_MAC));
+	}
+
+	return 0;
+}
+
+static bool hclge_fd_rule_exist(struct hclge_dev *hdev, u16 location)
+{
+	struct hclge_fd_rule *rule = NULL;
+	struct hlist_node *node2;
+
+	hlist_for_each_entry_safe(rule, node2, &hdev->fd_rule_list, rule_node) {
+		if (rule->location >= location)
+			break;
+	}
+
+	return  rule && rule->location == location;
+}
+
+static int hclge_fd_update_rule_list(struct hclge_dev *hdev,
+				     struct hclge_fd_rule *new_rule,
+				     u16 location,
+				     bool is_add)
+{
+	struct hclge_fd_rule *rule = NULL, *parent = NULL;
+	struct hlist_node *node2;
+
+	if (is_add && !new_rule)
+		return -EINVAL;
+
+	hlist_for_each_entry_safe(rule, node2,
+				  &hdev->fd_rule_list, rule_node) {
+		if (rule->location >= location)
+			break;
+		parent = rule;
+	}
+
+	if (rule && rule->location == location) {
+		hlist_del(&rule->rule_node);
+		kfree(rule);
+		hdev->hclge_fd_rule_num--;
+
+		if (!is_add)
+			return 0;
+
+	} else if (!is_add) {
+		dev_err(&hdev->pdev->dev,
+			"delete fail, rule %d is inexistent\n",
+			location);
+		return -EINVAL;
+	}
+
+	INIT_HLIST_NODE(&new_rule->rule_node);
+
+	if (parent)
+		hlist_add_behind(&new_rule->rule_node, &parent->rule_node);
+	else
+		hlist_add_head(&new_rule->rule_node, &hdev->fd_rule_list);
+
+	hdev->hclge_fd_rule_num++;
+
+	return 0;
+}
+
+static int hclge_fd_get_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[3] =
+				be32_to_cpu(fs->h_u.tcp_ip4_spec.ip4src);
+		rule->tuples_mask.src_ip[3] =
+				be32_to_cpu(fs->m_u.tcp_ip4_spec.ip4src);
+
+		rule->tuples.dst_ip[3] =
+				be32_to_cpu(fs->h_u.tcp_ip4_spec.ip4dst);
+		rule->tuples_mask.dst_ip[3] =
+				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.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.ip_tos = fs->h_u.tcp_ip4_spec.tos;
+		rule->tuples_mask.ip_tos = fs->m_u.tcp_ip4_spec.tos;
+
+		rule->tuples.ether_proto = ETH_P_IP;
+		rule->tuples_mask.ether_proto = 0xFFFF;
+
+		break;
+	case IP_USER_FLOW:
+		rule->tuples.src_ip[3] =
+				be32_to_cpu(fs->h_u.usr_ip4_spec.ip4src);
+		rule->tuples_mask.src_ip[3] =
+				be32_to_cpu(fs->m_u.usr_ip4_spec.ip4src);
+
+		rule->tuples.dst_ip[3] =
+				be32_to_cpu(fs->h_u.usr_ip4_spec.ip4dst);
+		rule->tuples_mask.dst_ip[3] =
+				be32_to_cpu(fs->m_u.usr_ip4_spec.ip4dst);
+
+		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.ip_proto = fs->h_u.usr_ip4_spec.proto;
+		rule->tuples_mask.ip_proto = fs->m_u.usr_ip4_spec.proto;
+
+		rule->tuples.ether_proto = ETH_P_IP;
+		rule->tuples_mask.ether_proto = 0xFFFF;
+
+		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, 4);
+		be32_to_cpu_array(rule->tuples_mask.src_ip,
+				  fs->m_u.tcp_ip6_spec.ip6src, 4);
+
+		be32_to_cpu_array(rule->tuples.dst_ip,
+				  fs->h_u.tcp_ip6_spec.ip6dst, 4);
+		be32_to_cpu_array(rule->tuples_mask.dst_ip,
+				  fs->m_u.tcp_ip6_spec.ip6dst, 4);
+
+		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.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.ether_proto = ETH_P_IPV6;
+		rule->tuples_mask.ether_proto = 0xFFFF;
+
+		break;
+	case IPV6_USER_FLOW:
+		be32_to_cpu_array(rule->tuples.src_ip,
+				  fs->h_u.usr_ip6_spec.ip6src, 4);
+		be32_to_cpu_array(rule->tuples_mask.src_ip,
+				  fs->m_u.usr_ip6_spec.ip6src, 4);
+
+		be32_to_cpu_array(rule->tuples.dst_ip,
+				  fs->h_u.usr_ip6_spec.ip6dst, 4);
+		be32_to_cpu_array(rule->tuples_mask.dst_ip,
+				  fs->m_u.usr_ip6_spec.ip6dst, 4);
+
+		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.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);
+
+		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);
+
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	switch (flow_type) {
+	case SCTP_V4_FLOW:
+	case SCTP_V6_FLOW:
+		rule->tuples.ip_proto = IPPROTO_SCTP;
+		rule->tuples_mask.ip_proto = 0xFF;
+		break;
+	case TCP_V4_FLOW:
+	case TCP_V6_FLOW:
+		rule->tuples.ip_proto = IPPROTO_TCP;
+		rule->tuples_mask.ip_proto = 0xFF;
+		break;
+	case UDP_V4_FLOW:
+	case UDP_V6_FLOW:
+		rule->tuples.ip_proto = IPPROTO_UDP;
+		rule->tuples_mask.ip_proto = 0xFF;
+		break;
+	default:
+		break;
+	}
+
+	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);
+	}
+
+	if (fs->flow_type & FLOW_MAC_EXT) {
+		ether_addr_copy(rule->tuples.dst_mac, fs->h_ext.h_dest);
+		ether_addr_copy(rule->tuples_mask.dst_mac, fs->m_ext.h_dest);
+	}
+
+	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;
+	u16 dst_vport_id = 0, q_index = 0;
+	struct ethtool_rx_flow_spec *fs;
+	struct hclge_fd_rule *rule;
+	u32 unused = 0;
+	u8 action;
+	int ret;
+
+	if (!hnae3_dev_fd_supported(hdev))
+		return -EOPNOTSUPP;
+
+	if (!hdev->fd_cfg.fd_en) {
+		dev_warn(&hdev->pdev->dev,
+			 "Please enable flow director first\n");
+		return -EOPNOTSUPP;
+	}
+
+	fs = (struct ethtool_rx_flow_spec *)&cmd->fs;
+
+	ret = hclge_fd_check_spec(hdev, fs, &unused);
+	if (ret) {
+		dev_err(&hdev->pdev->dev, "Check fd spec failed\n");
+		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;
+
+		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 (%d) > max tqp num (%d)\n",
+				ring, tqps - 1);
+			return -EINVAL;
+		}
+
+		if (vf > hdev->num_req_vfs) {
+			dev_err(&hdev->pdev->dev,
+				"Error: vf id (%d) > max vf num (%d)\n",
+				vf, hdev->num_req_vfs);
+			return -EINVAL;
+		}
+
+		action = HCLGE_FD_ACTION_ACCEPT_PACKET;
+		q_index = ring;
+	}
+
+	rule = kzalloc(sizeof(*rule), GFP_KERNEL);
+	if (!rule)
+		return -ENOMEM;
+
+	ret = hclge_fd_get_tuple(hdev, fs, rule);
+	if (ret)
+		goto free_rule;
+
+	rule->flow_type = fs->flow_type;
+
+	rule->location = fs->location;
+	rule->unused_tuple = unused;
+	rule->vf_id = dst_vport_id;
+	rule->queue_id = q_index;
+	rule->action = action;
+
+	ret = hclge_config_action(hdev, HCLGE_FD_STAGE_1, rule);
+	if (ret)
+		goto free_rule;
+
+	ret = hclge_config_key(hdev, HCLGE_FD_STAGE_1, rule);
+	if (ret)
+		goto free_rule;
+
+	ret = hclge_fd_update_rule_list(hdev, rule, fs->location, true);
+	if (ret)
+		goto free_rule;
+
+	return ret;
+
+free_rule:
+	kfree(rule);
+	return ret;
+}
+
+static int hclge_del_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 ethtool_rx_flow_spec *fs;
+	int ret;
+
+	if (!hnae3_dev_fd_supported(hdev))
+		return -EOPNOTSUPP;
+
+	fs = (struct ethtool_rx_flow_spec *)&cmd->fs;
+
+	if (fs->location >= hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_1])
+		return -EINVAL;
+
+	if (!hclge_fd_rule_exist(hdev, fs->location)) {
+		dev_err(&hdev->pdev->dev,
+			"Delete fail, rule %d is inexistent\n",
+			fs->location);
+		return -ENOENT;
+	}
+
+	ret = hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true,
+				   fs->location, NULL, false);
+	if (ret)
+		return ret;
+
+	return hclge_fd_update_rule_list(hdev, NULL, fs->location,
+					 false);
+}
+
+static void hclge_del_all_fd_entries(struct hnae3_handle *handle,
+				     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;
+
+	if (!hnae3_dev_fd_supported(hdev))
+		return;
+
+	if (clear_list) {
+		hlist_for_each_entry_safe(rule, node, &hdev->fd_rule_list,
+					  rule_node) {
+			hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true,
+					     rule->location, NULL, false);
+			hlist_del(&rule->rule_node);
+			kfree(rule);
+			hdev->hclge_fd_rule_num--;
+		}
+	} else {
+		hlist_for_each_entry_safe(rule, node, &hdev->fd_rule_list,
+					  rule_node)
+			hclge_fd_tcam_config(hdev, HCLGE_FD_STAGE_1, true,
+					     rule->location, NULL, false);
+	}
+}
+
+static int hclge_restore_fd_entries(struct hnae3_handle *handle)
+{
+	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 (!hnae3_dev_fd_supported(hdev))
+		return -EOPNOTSUPP;
+
+	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 %d failed, remove it\n",
+				 rule->location);
+			hlist_del(&rule->rule_node);
+			kfree(rule);
+			hdev->hclge_fd_rule_num--;
+		}
+	}
+	return 0;
+}
+
+static int hclge_get_fd_rule_cnt(struct hnae3_handle *handle,
+				 struct ethtool_rxnfc *cmd)
+{
+	struct hclge_vport *vport = hclge_get_vport(handle);
+	struct hclge_dev *hdev = vport->back;
+
+	if (!hnae3_dev_fd_supported(hdev))
+		return -EOPNOTSUPP;
+
+	cmd->rule_cnt = hdev->hclge_fd_rule_num;
+	cmd->data = hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_1];
+
+	return 0;
+}
+
+static int hclge_get_fd_rule_info(struct hnae3_handle *handle,
+				  struct ethtool_rxnfc *cmd)
+{
+	struct hclge_vport *vport = hclge_get_vport(handle);
+	struct hclge_fd_rule *rule = NULL;
+	struct hclge_dev *hdev = vport->back;
+	struct ethtool_rx_flow_spec *fs;
+	struct hlist_node *node2;
+
+	if (!hnae3_dev_fd_supported(hdev))
+		return -EOPNOTSUPP;
+
+	fs = (struct ethtool_rx_flow_spec *)&cmd->fs;
+
+	hlist_for_each_entry_safe(rule, node2, &hdev->fd_rule_list, rule_node) {
+		if (rule->location >= fs->location)
+			break;
+	}
+
+	if (!rule || fs->location != rule->location)
+		return -ENOENT;
+
+	fs->flow_type = rule->flow_type;
+	switch (fs->flow_type & ~(FLOW_EXT | FLOW_MAC_EXT)) {
+	case SCTP_V4_FLOW:
+	case TCP_V4_FLOW:
+	case UDP_V4_FLOW:
+		fs->h_u.tcp_ip4_spec.ip4src =
+				cpu_to_be32(rule->tuples.src_ip[3]);
+		fs->m_u.tcp_ip4_spec.ip4src =
+				rule->unused_tuple & BIT(INNER_SRC_IP) ?
+				0 : cpu_to_be32(rule->tuples_mask.src_ip[3]);
+
+		fs->h_u.tcp_ip4_spec.ip4dst =
+				cpu_to_be32(rule->tuples.dst_ip[3]);
+		fs->m_u.tcp_ip4_spec.ip4dst =
+				rule->unused_tuple & BIT(INNER_DST_IP) ?
+				0 : cpu_to_be32(rule->tuples_mask.dst_ip[3]);
+
+		fs->h_u.tcp_ip4_spec.psrc = cpu_to_be16(rule->tuples.src_port);
+		fs->m_u.tcp_ip4_spec.psrc =
+				rule->unused_tuple & BIT(INNER_SRC_PORT) ?
+				0 : cpu_to_be16(rule->tuples_mask.src_port);
+
+		fs->h_u.tcp_ip4_spec.pdst = cpu_to_be16(rule->tuples.dst_port);
+		fs->m_u.tcp_ip4_spec.pdst =
+				rule->unused_tuple & BIT(INNER_DST_PORT) ?
+				0 : cpu_to_be16(rule->tuples_mask.dst_port);
+
+		fs->h_u.tcp_ip4_spec.tos = rule->tuples.ip_tos;
+		fs->m_u.tcp_ip4_spec.tos =
+				rule->unused_tuple & BIT(INNER_IP_TOS) ?
+				0 : rule->tuples_mask.ip_tos;
+
+		break;
+	case IP_USER_FLOW:
+		fs->h_u.usr_ip4_spec.ip4src =
+				cpu_to_be32(rule->tuples.src_ip[3]);
+		fs->m_u.tcp_ip4_spec.ip4src =
+				rule->unused_tuple & BIT(INNER_SRC_IP) ?
+				0 : cpu_to_be32(rule->tuples_mask.src_ip[3]);
+
+		fs->h_u.usr_ip4_spec.ip4dst =
+				cpu_to_be32(rule->tuples.dst_ip[3]);
+		fs->m_u.usr_ip4_spec.ip4dst =
+				rule->unused_tuple & BIT(INNER_DST_IP) ?
+				0 : cpu_to_be32(rule->tuples_mask.dst_ip[3]);
+
+		fs->h_u.usr_ip4_spec.tos = rule->tuples.ip_tos;
+		fs->m_u.usr_ip4_spec.tos =
+				rule->unused_tuple & BIT(INNER_IP_TOS) ?
+				0 : rule->tuples_mask.ip_tos;
+
+		fs->h_u.usr_ip4_spec.proto = rule->tuples.ip_proto;
+		fs->m_u.usr_ip4_spec.proto =
+				rule->unused_tuple & BIT(INNER_IP_PROTO) ?
+				0 : rule->tuples_mask.ip_proto;
+
+		fs->h_u.usr_ip4_spec.ip_ver = ETH_RX_NFC_IP4;
+
+		break;
+	case SCTP_V6_FLOW:
+	case TCP_V6_FLOW:
+	case UDP_V6_FLOW:
+		cpu_to_be32_array(fs->h_u.tcp_ip6_spec.ip6src,
+				  rule->tuples.src_ip, 4);
+		if (rule->unused_tuple & BIT(INNER_SRC_IP))
+			memset(fs->m_u.tcp_ip6_spec.ip6src, 0, sizeof(int) * 4);
+		else
+			cpu_to_be32_array(fs->m_u.tcp_ip6_spec.ip6src,
+					  rule->tuples_mask.src_ip, 4);
+
+		cpu_to_be32_array(fs->h_u.tcp_ip6_spec.ip6dst,
+				  rule->tuples.dst_ip, 4);
+		if (rule->unused_tuple & BIT(INNER_DST_IP))
+			memset(fs->m_u.tcp_ip6_spec.ip6dst, 0, sizeof(int) * 4);
+		else
+			cpu_to_be32_array(fs->m_u.tcp_ip6_spec.ip6dst,
+					  rule->tuples_mask.dst_ip, 4);
+
+		fs->h_u.tcp_ip6_spec.psrc = cpu_to_be16(rule->tuples.src_port);
+		fs->m_u.tcp_ip6_spec.psrc =
+				rule->unused_tuple & BIT(INNER_SRC_PORT) ?
+				0 : cpu_to_be16(rule->tuples_mask.src_port);
+
+		fs->h_u.tcp_ip6_spec.pdst = cpu_to_be16(rule->tuples.dst_port);
+		fs->m_u.tcp_ip6_spec.pdst =
+				rule->unused_tuple & BIT(INNER_DST_PORT) ?
+				0 : cpu_to_be16(rule->tuples_mask.dst_port);
+
+		break;
+	case IPV6_USER_FLOW:
+		cpu_to_be32_array(fs->h_u.usr_ip6_spec.ip6src,
+				  rule->tuples.src_ip, 4);
+		if (rule->unused_tuple & BIT(INNER_SRC_IP))
+			memset(fs->m_u.usr_ip6_spec.ip6src, 0, sizeof(int) * 4);
+		else
+			cpu_to_be32_array(fs->m_u.usr_ip6_spec.ip6src,
+					  rule->tuples_mask.src_ip, 4);
+
+		cpu_to_be32_array(fs->h_u.usr_ip6_spec.ip6dst,
+				  rule->tuples.dst_ip, 4);
+		if (rule->unused_tuple & BIT(INNER_DST_IP))
+			memset(fs->m_u.usr_ip6_spec.ip6dst, 0, sizeof(int) * 4);
+		else
+			cpu_to_be32_array(fs->m_u.usr_ip6_spec.ip6dst,
+					  rule->tuples_mask.dst_ip, 4);
+
+		fs->h_u.usr_ip6_spec.l4_proto = rule->tuples.ip_proto;
+		fs->m_u.usr_ip6_spec.l4_proto =
+				rule->unused_tuple & BIT(INNER_IP_PROTO) ?
+				0 : rule->tuples_mask.ip_proto;
+
+		break;
+	case ETHER_FLOW:
+		ether_addr_copy(fs->h_u.ether_spec.h_source,
+				rule->tuples.src_mac);
+		if (rule->unused_tuple & BIT(INNER_SRC_MAC))
+			eth_zero_addr(fs->m_u.ether_spec.h_source);
+		else
+			ether_addr_copy(fs->m_u.ether_spec.h_source,
+					rule->tuples_mask.src_mac);
+
+		ether_addr_copy(fs->h_u.ether_spec.h_dest,
+				rule->tuples.dst_mac);
+		if (rule->unused_tuple & BIT(INNER_DST_MAC))
+			eth_zero_addr(fs->m_u.ether_spec.h_dest);
+		else
+			ether_addr_copy(fs->m_u.ether_spec.h_dest,
+					rule->tuples_mask.dst_mac);
+
+		fs->h_u.ether_spec.h_proto =
+				cpu_to_be16(rule->tuples.ether_proto);
+		fs->m_u.ether_spec.h_proto =
+				rule->unused_tuple & BIT(INNER_ETH_TYPE) ?
+				0 : cpu_to_be16(rule->tuples_mask.ether_proto);
+
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	if (fs->flow_type & FLOW_EXT) {
+		fs->h_ext.vlan_tci = cpu_to_be16(rule->tuples.vlan_tag1);
+		fs->m_ext.vlan_tci =
+				rule->unused_tuple & BIT(INNER_VLAN_TAG_FST) ?
+				cpu_to_be16(VLAN_VID_MASK) :
+				cpu_to_be16(rule->tuples_mask.vlan_tag1);
+	}
+
+	if (fs->flow_type & FLOW_MAC_EXT) {
+		ether_addr_copy(fs->h_ext.h_dest, rule->tuples.dst_mac);
+		if (rule->unused_tuple & BIT(INNER_DST_MAC))
+			eth_zero_addr(fs->m_u.ether_spec.h_dest);
+		else
+			ether_addr_copy(fs->m_u.ether_spec.h_dest,
+					rule->tuples_mask.dst_mac);
+	}
+
+	if (rule->action == HCLGE_FD_ACTION_DROP_PACKET) {
+		fs->ring_cookie = RX_CLS_FLOW_DISC;
+	} else {
+		u64 vf_id;
+
+		fs->ring_cookie = rule->queue_id;
+		vf_id = rule->vf_id;
+		vf_id <<= ETHTOOL_RX_FLOW_SPEC_RING_VF_OFF;
+		fs->ring_cookie |= vf_id;
+	}
+
+	return 0;
+}
+
+static int hclge_get_all_rules(struct hnae3_handle *handle,
+			       struct ethtool_rxnfc *cmd, u32 *rule_locs)
+{
+	struct hclge_vport *vport = hclge_get_vport(handle);
+	struct hclge_dev *hdev = vport->back;
+	struct hclge_fd_rule *rule;
+	struct hlist_node *node2;
+	int cnt = 0;
+
+	if (!hnae3_dev_fd_supported(hdev))
+		return -EOPNOTSUPP;
+
+	cmd->data = hdev->fd_cfg.rule_num[HCLGE_FD_STAGE_1];
+
+	hlist_for_each_entry_safe(rule, node2,
+				  &hdev->fd_rule_list, rule_node) {
+		if (cnt == cmd->rule_cnt)
+			return -EMSGSIZE;
+
+		rule_locs[cnt] = rule->location;
+		cnt++;
+	}
+
+	cmd->rule_cnt = cnt;
+
+	return 0;
+}
+
+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;
+
+	hdev->fd_cfg.fd_en = enable;
+	if (!enable)
+		hclge_del_all_fd_entries(handle, false);
+	else
+		hclge_restore_fd_entries(handle);
+}
+
 static void hclge_cfg_mac_mode(struct hclge_dev *hdev, bool enable)
 {
 	struct hclge_desc desc;
@@ -5502,6 +6783,13 @@ static int hclge_init_ae_dev(struct hnae3_ae_dev *ae_dev)
 		goto err_mdiobus_unreg;
 	}
 
+	ret = hclge_init_fd_config(hdev);
+	if (ret) {
+		dev_err(&pdev->dev,
+			"fd table init fail, ret=%d\n", ret);
+		goto err_mdiobus_unreg;
+	}
+
 	hclge_dcb_ops_set(hdev);
 
 	timer_setup(&hdev->service_timer, hclge_service_timer, 0);
@@ -5608,6 +6896,13 @@ static int hclge_reset_ae_dev(struct hnae3_ae_dev *ae_dev)
 		return ret;
 	}
 
+	ret = hclge_init_fd_config(hdev);
+	if (ret) {
+		dev_err(&pdev->dev,
+			"fd table init fail, ret=%d\n", ret);
+		return ret;
+	}
+
 	dev_info(&pdev->dev, "Reset done, %s driver initialization finished.\n",
 		 HCLGE_DRIVER_NAME);
 
@@ -6047,6 +7342,14 @@ static const struct hnae3_ae_ops hclge_ops = {
 	.get_regs = hclge_get_regs,
 	.set_led_id = hclge_set_led_id,
 	.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,
+	.restore_fd_rules = hclge_restore_fd_entries,
+	.enable_fd = hclge_enable_fd,
 };
 
 static struct hnae3_ae_algo ae_algo = {
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
index 7841b830a716..06adbdd27b95 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
@@ -79,6 +79,19 @@
 #define HCLGE_VF_NUM_PER_CMD           64
 #define HCLGE_VF_NUM_PER_BYTE          8
 
+enum HLCGE_PORT_TYPE {
+	HOST_PORT,
+	NETWORK_PORT
+};
+
+#define HCLGE_PF_ID_S			0
+#define HCLGE_PF_ID_M			GENMASK(2, 0)
+#define HCLGE_VF_ID_S			3
+#define HCLGE_VF_ID_M			GENMASK(10, 3)
+#define HCLGE_PORT_TYPE_B		11
+#define HCLGE_NETWORK_PORT_ID_S		0
+#define HCLGE_NETWORK_PORT_ID_M		GENMASK(3, 0)
+
 /* Reset related Registers */
 #define HCLGE_MISC_RESET_STS_REG	0x20700
 #define HCLGE_MISC_VECTOR_INT_STS	0x20800
@@ -359,6 +372,221 @@ struct hclge_vlan_type_cfg {
 	u16 tx_in_vlan_type;
 };
 
+enum HCLGE_FD_MODE {
+	HCLGE_FD_MODE_DEPTH_2K_WIDTH_400B_STAGE_1,
+	HCLGE_FD_MODE_DEPTH_1K_WIDTH_400B_STAGE_2,
+	HCLGE_FD_MODE_DEPTH_4K_WIDTH_200B_STAGE_1,
+	HCLGE_FD_MODE_DEPTH_2K_WIDTH_200B_STAGE_2,
+};
+
+enum HCLGE_FD_KEY_TYPE {
+	HCLGE_FD_KEY_BASE_ON_PTYPE,
+	HCLGE_FD_KEY_BASE_ON_TUPLE,
+};
+
+enum HCLGE_FD_STAGE {
+	HCLGE_FD_STAGE_1,
+	HCLGE_FD_STAGE_2,
+};
+
+/* OUTER_XXX indicates tuples in tunnel header of tunnel packet
+ * INNER_XXX indicate tuples in tunneled header of tunnel packet or
+ *           tuples of non-tunnel packet
+ */
+enum HCLGE_FD_TUPLE {
+	OUTER_DST_MAC,
+	OUTER_SRC_MAC,
+	OUTER_VLAN_TAG_FST,
+	OUTER_VLAN_TAG_SEC,
+	OUTER_ETH_TYPE,
+	OUTER_L2_RSV,
+	OUTER_IP_TOS,
+	OUTER_IP_PROTO,
+	OUTER_SRC_IP,
+	OUTER_DST_IP,
+	OUTER_L3_RSV,
+	OUTER_SRC_PORT,
+	OUTER_DST_PORT,
+	OUTER_L4_RSV,
+	OUTER_TUN_VNI,
+	OUTER_TUN_FLOW_ID,
+	INNER_DST_MAC,
+	INNER_SRC_MAC,
+	INNER_VLAN_TAG_FST,
+	INNER_VLAN_TAG_SEC,
+	INNER_ETH_TYPE,
+	INNER_L2_RSV,
+	INNER_IP_TOS,
+	INNER_IP_PROTO,
+	INNER_SRC_IP,
+	INNER_DST_IP,
+	INNER_L3_RSV,
+	INNER_SRC_PORT,
+	INNER_DST_PORT,
+	INNER_L4_RSV,
+	MAX_TUPLE,
+};
+
+enum HCLGE_FD_META_DATA {
+	PACKET_TYPE_ID,
+	IP_FRAGEMENT,
+	ROCE_TYPE,
+	NEXT_KEY,
+	VLAN_NUMBER,
+	SRC_VPORT,
+	DST_VPORT,
+	TUNNEL_PACKET,
+	MAX_META_DATA,
+};
+
+struct key_info {
+	u8 key_type;
+	u8 key_length;
+};
+
+static const struct key_info meta_data_key_info[] = {
+	{ PACKET_TYPE_ID, 6},
+	{ IP_FRAGEMENT, 1},
+	{ ROCE_TYPE, 1},
+	{ NEXT_KEY, 5},
+	{ VLAN_NUMBER, 2},
+	{ SRC_VPORT, 12},
+	{ DST_VPORT, 12},
+	{ TUNNEL_PACKET, 1},
+};
+
+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},
+};
+
+#define MAX_KEY_LENGTH	400
+#define MAX_KEY_DWORDS	DIV_ROUND_UP(MAX_KEY_LENGTH / 8, 4)
+#define MAX_KEY_BYTES	(MAX_KEY_DWORDS * 4)
+#define MAX_META_DATA_LENGTH	32
+
+enum HCLGE_FD_PACKET_TYPE {
+	NIC_PACKET,
+	ROCE_PACKET,
+};
+
+enum HCLGE_FD_ACTION {
+	HCLGE_FD_ACTION_ACCEPT_PACKET,
+	HCLGE_FD_ACTION_DROP_PACKET,
+};
+
+struct hclge_fd_key_cfg {
+	u8 key_sel;
+	u8 inner_sipv6_word_en;
+	u8 inner_dipv6_word_en;
+	u8 outer_sipv6_word_en;
+	u8 outer_dipv6_word_en;
+	u32 tuple_active;
+	u32 meta_data_active;
+};
+
+struct hclge_fd_cfg {
+	u8 fd_mode;
+	u8 fd_en;
+	u16 max_key_length;
+	u32 proto_support;
+	u32 rule_num[2]; /* rule entry number */
+	u16 cnt_num[2]; /* rule hit counter number */
+	struct hclge_fd_key_cfg key_cfg[2];
+};
+
+struct hclge_fd_rule_tuples {
+	u8 src_mac[6];
+	u8 dst_mac[6];
+	u32 src_ip[4];
+	u32 dst_ip[4];
+	u16 src_port;
+	u16 dst_port;
+	u16 vlan_tag1;
+	u16 ether_proto;
+	u8 ip_tos;
+	u8 ip_proto;
+};
+
+struct hclge_fd_rule {
+	struct hlist_node rule_node;
+	struct hclge_fd_rule_tuples tuples;
+	struct hclge_fd_rule_tuples tuples_mask;
+	u32 unused_tuple;
+	u32 flow_type;
+	u8 action;
+	u16 vf_id;
+	u16 queue_id;
+	u16 location;
+};
+
+struct hclge_fd_ad_data {
+	u16 ad_id;
+	u8 drop_packet;
+	u8 forward_to_direct_queue;
+	u16 queue_id;
+	u8 use_counter;
+	u8 counter_id;
+	u8 use_next_stage;
+	u8 write_rule_id_to_bd;
+	u8 next_input_key;
+	u16 rule_id;
+};
+
+/* For each bit of TCAM entry, it uses a pair of 'x' and
+ * 'y' to indicate which value to match, like below:
+ * ----------------------------------
+ * | bit x | bit y |  search value  |
+ * ----------------------------------
+ * |   0   |   0   |   always hit   |
+ * ----------------------------------
+ * |   1   |   0   |   match '0'    |
+ * ----------------------------------
+ * |   0   |   1   |   match '1'    |
+ * ----------------------------------
+ * |   1   |   1   |   invalid      |
+ * ----------------------------------
+ * Then for input key(k) and mask(v), we can calculate the value by
+ * the formulae:
+ *	x = (~k) & v
+ *	y = (k ^ ~v) & k
+ */
+#define calc_x(x, k, v) ((x) = (~(k) & (v)))
+#define calc_y(y, k, v) \
+	do { \
+		const typeof(k) _k_ = (k); \
+		const typeof(v) _v_ = (v); \
+		(y) = (_k_ ^ ~_v_) & (_k_); \
+	} while (0)
+
 #define HCLGE_VPORT_NUM 256
 struct hclge_dev {
 	struct pci_dev *pdev;
@@ -448,6 +676,10 @@ struct hclge_dev {
 	struct hclge_vlan_type_cfg vlan_type_cfg;
 
 	unsigned long vlan_table[VLAN_N_VID][BITS_TO_LONGS(HCLGE_VPORT_NUM)];
+
+	struct hclge_fd_cfg fd_cfg;
+	struct hlist_head fd_rule_list;
+	u16 hclge_fd_rule_num;
 };
 
 /* VPort level vlan tag configuration for TX direction */