net/hns3: support VF
authorWei Hu (Xavier) <xavier.huwei@huawei.com>
Thu, 26 Sep 2019 14:02:01 +0000 (22:02 +0800)
committerFerruh Yigit <ferruh.yigit@intel.com>
Mon, 7 Oct 2019 13:00:57 +0000 (15:00 +0200)
This patch adds support for hns3 VF PMD driver.

In current version, we only support VF device is bound to vfio_pci or
igb_uio and then driven by DPDK driver when PF is driven by kernel mode
hns3 ethdev driver, VF is not supported when PF is driven by DPDK
driver.

Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Signed-off-by: Chunsong Feng <fengchunsong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Signed-off-by: Hao Chen <chenhao164@huawei.com>
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Chengchang Tang <tangchengchang@hisilicon.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
doc/guides/nics/features/hns3_vf.ini [new file with mode: 0644]
drivers/net/hns3/Makefile
drivers/net/hns3/hns3_ethdev_vf.c [new file with mode: 0644]
drivers/net/hns3/meson.build

diff --git a/doc/guides/nics/features/hns3_vf.ini b/doc/guides/nics/features/hns3_vf.ini
new file mode 100644 (file)
index 0000000..e77a3fb
--- /dev/null
@@ -0,0 +1,21 @@
+;
+; Supported features of the 'hns3' network poll mode driver.
+;
+; Refer to default.ini for the full list of available PMD features.
+;
+[Features]
+Link status          = Y
+MTU update           = Y
+Jumbo frame          = Y
+Unicast MAC filter   = Y
+Multicast MAC filter = Y
+RSS hash             = Y
+RSS key update       = Y
+RSS reta update      = Y
+VLAN filter          = Y
+Flow director        = Y
+Flow API             = Y
+VLAN offload         = Y
+Linux UIO            = Y
+Linux VFIO           = Y
+ARMv8                = Y
index fc39279..a8890f0 100644 (file)
@@ -23,6 +23,7 @@ LIBABIVER := 1
 # all source are stored in SRCS-y
 #
 SRCS-$(CONFIG_RTE_LIBRTE_HNS3_PMD) += hns3_ethdev.c
+SRCS-$(CONFIG_RTE_LIBRTE_HNS3_PMD) += hns3_ethdev_vf.c
 SRCS-$(CONFIG_RTE_LIBRTE_HNS3_PMD) += hns3_cmd.c
 SRCS-$(CONFIG_RTE_LIBRTE_HNS3_PMD) += hns3_mbx.c
 SRCS-$(CONFIG_RTE_LIBRTE_HNS3_PMD) += hns3_rss.c
diff --git a/drivers/net/hns3/hns3_ethdev_vf.c b/drivers/net/hns3/hns3_ethdev_vf.c
new file mode 100644 (file)
index 0000000..230a0cb
--- /dev/null
@@ -0,0 +1,1237 @@
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2018-2019 Hisilicon Limited.
+ */
+
+#include <errno.h>
+#include <stdio.h>
+#include <stdbool.h>
+#include <string.h>
+#include <inttypes.h>
+#include <unistd.h>
+#include <arpa/inet.h>
+#include <rte_alarm.h>
+#include <rte_atomic.h>
+#include <rte_bus_pci.h>
+#include <rte_byteorder.h>
+#include <rte_common.h>
+#include <rte_cycles.h>
+#include <rte_dev.h>
+#include <rte_eal.h>
+#include <rte_ether.h>
+#include <rte_ethdev_driver.h>
+#include <rte_ethdev_pci.h>
+#include <rte_interrupts.h>
+#include <rte_io.h>
+#include <rte_log.h>
+#include <rte_pci.h>
+
+#include "hns3_ethdev.h"
+#include "hns3_logs.h"
+#include "hns3_regs.h"
+#include "hns3_dcb.h"
+
+#define HNS3VF_KEEP_ALIVE_INTERVAL     2000000 /* us */
+#define HNS3VF_SERVICE_INTERVAL                1000000 /* us */
+
+#define HNS3VF_RESET_WAIT_MS   20
+#define HNS3VF_RESET_WAIT_CNT  2000
+
+enum hns3vf_evt_cause {
+       HNS3VF_VECTOR0_EVENT_RST,
+       HNS3VF_VECTOR0_EVENT_MBX,
+       HNS3VF_VECTOR0_EVENT_OTHER,
+};
+
+static int hns3vf_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu);
+static int hns3vf_dev_configure_vlan(struct rte_eth_dev *dev);
+
+static int
+hns3vf_add_mac_addr(struct rte_eth_dev *dev, struct rte_ether_addr *mac_addr,
+                   __attribute__ ((unused)) uint32_t idx,
+                   __attribute__ ((unused)) uint32_t pool)
+{
+       struct hns3_hw *hw = HNS3_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+       char mac_str[RTE_ETHER_ADDR_FMT_SIZE];
+       int ret;
+
+       rte_spinlock_lock(&hw->lock);
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_SET_UNICAST,
+                               HNS3_MBX_MAC_VLAN_UC_ADD, mac_addr->addr_bytes,
+                               RTE_ETHER_ADDR_LEN, false, NULL, 0);
+       rte_spinlock_unlock(&hw->lock);
+       if (ret) {
+               rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE,
+                                     mac_addr);
+               hns3_err(hw, "Failed to add mac addr(%s) for vf: %d", mac_str,
+                        ret);
+       }
+
+       return ret;
+}
+
+static void
+hns3vf_remove_mac_addr(struct rte_eth_dev *dev, uint32_t idx)
+{
+       struct hns3_hw *hw = HNS3_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+       /* index will be checked by upper level rte interface */
+       struct rte_ether_addr *mac_addr = &dev->data->mac_addrs[idx];
+       char mac_str[RTE_ETHER_ADDR_FMT_SIZE];
+       int ret;
+
+       rte_spinlock_lock(&hw->lock);
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_SET_UNICAST,
+                               HNS3_MBX_MAC_VLAN_UC_REMOVE,
+                               mac_addr->addr_bytes, RTE_ETHER_ADDR_LEN, false,
+                               NULL, 0);
+       rte_spinlock_unlock(&hw->lock);
+       if (ret) {
+               rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE,
+                                     mac_addr);
+               hns3_err(hw, "Failed to remove mac addr(%s) for vf: %d",
+                        mac_str, ret);
+       }
+}
+
+static int
+hns3vf_set_default_mac_addr(struct rte_eth_dev *dev,
+                           struct rte_ether_addr *mac_addr)
+{
+#define HNS3_TWO_ETHER_ADDR_LEN (RTE_ETHER_ADDR_LEN * 2)
+       struct hns3_hw *hw = HNS3_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+       struct rte_ether_addr *old_addr;
+       uint8_t addr_bytes[HNS3_TWO_ETHER_ADDR_LEN]; /* for 2 MAC addresses */
+       char mac_str[RTE_ETHER_ADDR_FMT_SIZE];
+       int ret;
+
+       if (!rte_is_valid_assigned_ether_addr(mac_addr)) {
+               rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE,
+                                     mac_addr);
+               hns3_err(hw, "Failed to set mac addr, addr(%s) invalid.",
+                        mac_str);
+               return -EINVAL;
+       }
+
+       old_addr = (struct rte_ether_addr *)hw->mac.mac_addr;
+       rte_spinlock_lock(&hw->lock);
+       memcpy(addr_bytes, mac_addr->addr_bytes, RTE_ETHER_ADDR_LEN);
+       memcpy(&addr_bytes[RTE_ETHER_ADDR_LEN], old_addr->addr_bytes,
+              RTE_ETHER_ADDR_LEN);
+
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_SET_UNICAST,
+                               HNS3_MBX_MAC_VLAN_UC_MODIFY, addr_bytes,
+                               HNS3_TWO_ETHER_ADDR_LEN, false, NULL, 0);
+       if (ret) {
+               rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE,
+                                     mac_addr);
+               hns3_err(hw, "Failed to set mac addr(%s) for vf: %d", mac_str,
+                        ret);
+       }
+
+       rte_ether_addr_copy(mac_addr,
+                           (struct rte_ether_addr *)hw->mac.mac_addr);
+       rte_spinlock_unlock(&hw->lock);
+
+       return ret;
+}
+
+static int
+hns3vf_configure_mac_addr(struct hns3_adapter *hns, bool del)
+{
+       struct hns3_hw *hw = &hns->hw;
+       struct rte_ether_addr *addr;
+       enum hns3_mbx_mac_vlan_subcode opcode;
+       char mac_str[RTE_ETHER_ADDR_FMT_SIZE];
+       int ret = 0;
+       int i;
+
+       if (del)
+               opcode = HNS3_MBX_MAC_VLAN_UC_REMOVE;
+       else
+               opcode = HNS3_MBX_MAC_VLAN_UC_ADD;
+       for (i = 0; i < HNS3_VF_UC_MACADDR_NUM; i++) {
+               addr = &hw->data->mac_addrs[i];
+               if (!rte_is_valid_assigned_ether_addr(addr))
+                       continue;
+               rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE, addr);
+               hns3_dbg(hw, "rm mac addr: %s", mac_str);
+               ret = hns3_send_mbx_msg(hw, HNS3_MBX_SET_UNICAST, opcode,
+                                       addr->addr_bytes, RTE_ETHER_ADDR_LEN,
+                                       false, NULL, 0);
+               if (ret) {
+                       hns3_err(hw, "Failed to remove mac addr for vf: %d",
+                                ret);
+                       break;
+               }
+       }
+       return ret;
+}
+
+static int
+hns3vf_add_mc_mac_addr(struct hns3_adapter *hns,
+                      struct rte_ether_addr *mac_addr)
+{
+       char mac_str[RTE_ETHER_ADDR_FMT_SIZE];
+       struct hns3_hw *hw = &hns->hw;
+       int ret;
+
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_SET_MULTICAST,
+                               HNS3_MBX_MAC_VLAN_MC_ADD,
+                               mac_addr->addr_bytes, RTE_ETHER_ADDR_LEN, false,
+                               NULL, 0);
+       if (ret) {
+               rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE,
+                                     mac_addr);
+               hns3_err(hw, "Failed to add mc mac addr(%s) for vf: %d",
+                        mac_str, ret);
+               return ret;
+       }
+
+       return 0;
+}
+
+static int
+hns3vf_remove_mc_mac_addr(struct hns3_adapter *hns,
+                         struct rte_ether_addr *mac_addr)
+{
+       char mac_str[RTE_ETHER_ADDR_FMT_SIZE];
+       struct hns3_hw *hw = &hns->hw;
+       int ret;
+
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_SET_MULTICAST,
+                               HNS3_MBX_MAC_VLAN_MC_REMOVE,
+                               mac_addr->addr_bytes, RTE_ETHER_ADDR_LEN, false,
+                               NULL, 0);
+       if (ret) {
+               rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE,
+                                     mac_addr);
+               hns3_err(hw, "Failed to remove mc mac addr(%s) for vf: %d",
+                        mac_str, ret);
+               return ret;
+       }
+
+       return 0;
+}
+
+static int
+hns3vf_set_mc_mac_addr_list(struct rte_eth_dev *dev,
+                           struct rte_ether_addr *mc_addr_set,
+                           uint32_t nb_mc_addr)
+{
+       struct hns3_adapter *hns = dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+       struct rte_ether_addr *addr;
+       char mac_str[RTE_ETHER_ADDR_FMT_SIZE];
+       int cur_addr_num;
+       int set_addr_num;
+       int num;
+       int ret;
+       int i;
+
+       if (nb_mc_addr > HNS3_MC_MACADDR_NUM) {
+               hns3_err(hw, "Failed to set mc mac addr, nb_mc_addr(%d) "
+                        "invalid. valid range: 0~%d",
+                        nb_mc_addr, HNS3_MC_MACADDR_NUM);
+               return -EINVAL;
+       }
+
+       set_addr_num = (int)nb_mc_addr;
+       for (i = 0; i < set_addr_num; i++) {
+               addr = &mc_addr_set[i];
+               if (!rte_is_multicast_ether_addr(addr)) {
+                       rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE,
+                                             addr);
+                       hns3_err(hw,
+                                "Failed to set mc mac addr, addr(%s) invalid.",
+                                mac_str);
+                       return -EINVAL;
+               }
+       }
+       rte_spinlock_lock(&hw->lock);
+       cur_addr_num = hw->mc_addrs_num;
+       for (i = 0; i < cur_addr_num; i++) {
+               num = cur_addr_num - i - 1;
+               addr = &hw->mc_addrs[num];
+               ret = hns3vf_remove_mc_mac_addr(hns, addr);
+               if (ret) {
+                       rte_spinlock_unlock(&hw->lock);
+                       return ret;
+               }
+
+               hw->mc_addrs_num--;
+       }
+
+       for (i = 0; i < set_addr_num; i++) {
+               addr = &mc_addr_set[i];
+               ret = hns3vf_add_mc_mac_addr(hns, addr);
+               if (ret) {
+                       rte_spinlock_unlock(&hw->lock);
+                       return ret;
+               }
+
+               rte_ether_addr_copy(addr, &hw->mc_addrs[hw->mc_addrs_num]);
+               hw->mc_addrs_num++;
+       }
+       rte_spinlock_unlock(&hw->lock);
+
+       return 0;
+}
+
+static int
+hns3vf_configure_all_mc_mac_addr(struct hns3_adapter *hns, bool del)
+{
+       char mac_str[RTE_ETHER_ADDR_FMT_SIZE];
+       struct hns3_hw *hw = &hns->hw;
+       struct rte_ether_addr *addr;
+       int err = 0;
+       int ret;
+       int i;
+
+       for (i = 0; i < hw->mc_addrs_num; i++) {
+               addr = &hw->mc_addrs[i];
+               if (!rte_is_multicast_ether_addr(addr))
+                       continue;
+               if (del)
+                       ret = hns3vf_remove_mc_mac_addr(hns, addr);
+               else
+                       ret = hns3vf_add_mc_mac_addr(hns, addr);
+               if (ret) {
+                       err = ret;
+                       rte_ether_format_addr(mac_str, RTE_ETHER_ADDR_FMT_SIZE,
+                                             addr);
+                       hns3_err(hw, "Failed to %s mc mac addr: %s for vf: %d",
+                                del ? "Remove" : "Restore", mac_str, ret);
+               }
+       }
+       return err;
+}
+
+static int
+hns3vf_set_promisc_mode(struct hns3_hw *hw, bool en_bc_pmc)
+{
+       struct hns3_mbx_vf_to_pf_cmd *req;
+       struct hns3_cmd_desc desc;
+       int ret;
+
+       req = (struct hns3_mbx_vf_to_pf_cmd *)desc.data;
+
+       hns3_cmd_setup_basic_desc(&desc, HNS3_OPC_MBX_VF_TO_PF, false);
+       req->msg[0] = HNS3_MBX_SET_PROMISC_MODE;
+       req->msg[1] = en_bc_pmc ? 1 : 0;
+
+       ret = hns3_cmd_send(hw, &desc, 1);
+       if (ret)
+               hns3_err(hw, "Set promisc mode fail, status is %d", ret);
+
+       return ret;
+}
+
+static int
+hns3vf_dev_configure(struct rte_eth_dev *dev)
+{
+       struct hns3_hw *hw = HNS3_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+       struct hns3_rss_conf *rss_cfg = &hw->rss_info;
+       struct rte_eth_conf *conf = &dev->data->dev_conf;
+       enum rte_eth_rx_mq_mode mq_mode = conf->rxmode.mq_mode;
+       uint16_t nb_rx_q = dev->data->nb_rx_queues;
+       uint16_t nb_tx_q = dev->data->nb_tx_queues;
+       struct rte_eth_rss_conf rss_conf;
+       uint16_t mtu;
+       int ret;
+
+       /*
+        * Hardware does not support where the number of rx and tx queues is
+        * not equal in hip08.
+        */
+       if (nb_rx_q != nb_tx_q) {
+               hns3_err(hw,
+                        "nb_rx_queues(%u) not equal with nb_tx_queues(%u)! "
+                        "Hardware does not support this configuration!",
+                        nb_rx_q, nb_tx_q);
+               return -EINVAL;
+       }
+
+       if (conf->link_speeds & ETH_LINK_SPEED_FIXED) {
+               hns3_err(hw, "setting link speed/duplex not supported");
+               return -EINVAL;
+       }
+
+       hw->adapter_state = HNS3_NIC_CONFIGURING;
+
+       /* When RSS is not configured, redirect the packet queue 0 */
+       if ((uint32_t)mq_mode & ETH_MQ_RX_RSS_FLAG) {
+               rss_conf = conf->rx_adv_conf.rss_conf;
+               if (rss_conf.rss_key == NULL) {
+                       rss_conf.rss_key = rss_cfg->key;
+                       rss_conf.rss_key_len = HNS3_RSS_KEY_SIZE;
+               }
+
+               ret = hns3_dev_rss_hash_update(dev, &rss_conf);
+               if (ret)
+                       goto cfg_err;
+       }
+
+       /*
+        * If jumbo frames are enabled, MTU needs to be refreshed
+        * according to the maximum RX packet length.
+        */
+       if (conf->rxmode.offloads & DEV_RX_OFFLOAD_JUMBO_FRAME) {
+               /*
+                * Security of max_rx_pkt_len is guaranteed in dpdk frame.
+                * Maximum value of max_rx_pkt_len is HNS3_MAX_FRAME_LEN, so it
+                * can safely assign to "uint16_t" type variable.
+                */
+               mtu = (uint16_t)HNS3_PKTLEN_TO_MTU(conf->rxmode.max_rx_pkt_len);
+               ret = hns3vf_dev_mtu_set(dev, mtu);
+               if (ret)
+                       goto cfg_err;
+               dev->data->mtu = mtu;
+       }
+
+       ret = hns3vf_dev_configure_vlan(dev);
+       if (ret)
+               goto cfg_err;
+
+       hw->adapter_state = HNS3_NIC_CONFIGURED;
+       return 0;
+
+cfg_err:
+       hw->adapter_state = HNS3_NIC_INITIALIZED;
+       return ret;
+}
+
+static int
+hns3vf_config_mtu(struct hns3_hw *hw, uint16_t mtu)
+{
+       int ret;
+
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_SET_MTU, 0, (const uint8_t *)&mtu,
+                               sizeof(mtu), true, NULL, 0);
+       if (ret)
+               hns3_err(hw, "Failed to set mtu (%u) for vf: %d", mtu, ret);
+
+       return ret;
+}
+
+static int
+hns3vf_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
+{
+       struct hns3_hw *hw = HNS3_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+       uint32_t frame_size = mtu + HNS3_ETH_OVERHEAD;
+       int ret;
+
+       if (dev->data->dev_started) {
+               hns3_err(hw, "Failed to set mtu, port %u must be stopped "
+                        "before configuration", dev->data->port_id);
+               return -EBUSY;
+       }
+
+       rte_spinlock_lock(&hw->lock);
+       ret = hns3vf_config_mtu(hw, mtu);
+       if (ret) {
+               rte_spinlock_unlock(&hw->lock);
+               return ret;
+       }
+       if (frame_size > RTE_ETHER_MAX_LEN)
+               dev->data->dev_conf.rxmode.offloads |=
+                                               DEV_RX_OFFLOAD_JUMBO_FRAME;
+       else
+               dev->data->dev_conf.rxmode.offloads &=
+                                               ~DEV_RX_OFFLOAD_JUMBO_FRAME;
+       dev->data->dev_conf.rxmode.max_rx_pkt_len = frame_size;
+       rte_spinlock_unlock(&hw->lock);
+
+       return 0;
+}
+
+static int
+hns3vf_dev_infos_get(struct rte_eth_dev *eth_dev, struct rte_eth_dev_info *info)
+{
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+
+       info->max_rx_queues = hw->tqps_num;
+       info->max_tx_queues = hw->tqps_num;
+       info->max_rx_pktlen = HNS3_MAX_FRAME_LEN; /* CRC included */
+       info->min_rx_bufsize = hw->rx_buf_len;
+       info->max_mac_addrs = HNS3_VF_UC_MACADDR_NUM;
+       info->max_mtu = info->max_rx_pktlen - HNS3_ETH_OVERHEAD;
+
+       info->rx_offload_capa = (DEV_RX_OFFLOAD_IPV4_CKSUM |
+                                DEV_RX_OFFLOAD_UDP_CKSUM |
+                                DEV_RX_OFFLOAD_TCP_CKSUM |
+                                DEV_RX_OFFLOAD_SCTP_CKSUM |
+                                DEV_RX_OFFLOAD_OUTER_IPV4_CKSUM |
+                                DEV_RX_OFFLOAD_OUTER_UDP_CKSUM |
+                                DEV_RX_OFFLOAD_KEEP_CRC |
+                                DEV_RX_OFFLOAD_SCATTER |
+                                DEV_RX_OFFLOAD_VLAN_STRIP |
+                                DEV_RX_OFFLOAD_QINQ_STRIP |
+                                DEV_RX_OFFLOAD_VLAN_FILTER |
+                                DEV_RX_OFFLOAD_JUMBO_FRAME);
+       info->tx_queue_offload_capa = DEV_TX_OFFLOAD_MBUF_FAST_FREE;
+       info->tx_offload_capa = (DEV_TX_OFFLOAD_OUTER_IPV4_CKSUM |
+                                DEV_TX_OFFLOAD_IPV4_CKSUM |
+                                DEV_TX_OFFLOAD_TCP_CKSUM |
+                                DEV_TX_OFFLOAD_UDP_CKSUM |
+                                DEV_TX_OFFLOAD_SCTP_CKSUM |
+                                DEV_TX_OFFLOAD_VLAN_INSERT |
+                                DEV_TX_OFFLOAD_QINQ_INSERT |
+                                DEV_TX_OFFLOAD_MULTI_SEGS |
+                                info->tx_queue_offload_capa);
+
+       info->vmdq_queue_num = 0;
+
+       info->reta_size = HNS3_RSS_IND_TBL_SIZE;
+       info->hash_key_size = HNS3_RSS_KEY_SIZE;
+       info->flow_type_rss_offloads = HNS3_ETH_RSS_SUPPORT;
+
+       return 0;
+}
+
+static void
+hns3vf_clear_event_cause(struct hns3_hw *hw, uint32_t regclr)
+{
+       hns3_write_dev(hw, HNS3_VECTOR0_CMDQ_SRC_REG, regclr);
+}
+
+static void
+hns3vf_disable_irq0(struct hns3_hw *hw)
+{
+       hns3_write_dev(hw, HNS3_MISC_VECTOR_REG_BASE, 0);
+}
+
+static void
+hns3vf_enable_irq0(struct hns3_hw *hw)
+{
+       hns3_write_dev(hw, HNS3_MISC_VECTOR_REG_BASE, 1);
+}
+
+static enum hns3vf_evt_cause
+hns3vf_check_event_cause(struct hns3_adapter *hns, uint32_t *clearval)
+{
+       struct hns3_hw *hw = &hns->hw;
+       enum hns3vf_evt_cause ret;
+       uint32_t cmdq_stat_reg;
+       uint32_t val;
+
+       /* Fetch the events from their corresponding regs */
+       cmdq_stat_reg = hns3_read_dev(hw, HNS3_VECTOR0_CMDQ_STAT_REG);
+
+       /* Check for vector0 mailbox(=CMDQ RX) event source */
+       if (BIT(HNS3_VECTOR0_RX_CMDQ_INT_B) & cmdq_stat_reg) {
+               val = cmdq_stat_reg & ~BIT(HNS3_VECTOR0_RX_CMDQ_INT_B);
+               ret = HNS3VF_VECTOR0_EVENT_MBX;
+               goto out;
+       }
+
+       val = 0;
+       ret = HNS3VF_VECTOR0_EVENT_OTHER;
+out:
+       if (clearval)
+               *clearval = val;
+       return ret;
+}
+
+static void
+hns3vf_interrupt_handler(void *param)
+{
+       struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+       struct hns3_adapter *hns = dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+       enum hns3vf_evt_cause event_cause;
+       uint32_t clearval;
+
+       /* Disable interrupt */
+       hns3vf_disable_irq0(hw);
+
+       /* Read out interrupt causes */
+       event_cause = hns3vf_check_event_cause(hns, &clearval);
+
+       switch (event_cause) {
+       case HNS3VF_VECTOR0_EVENT_MBX:
+               hns3_dev_handle_mbx_msg(hw);
+               break;
+       default:
+               break;
+       }
+
+       /* Clear interrupt causes */
+       hns3vf_clear_event_cause(hw, clearval);
+
+       /* Enable interrupt */
+       hns3vf_enable_irq0(hw);
+}
+
+static int
+hns3vf_check_tqp_info(struct hns3_hw *hw)
+{
+       uint16_t tqps_num;
+
+       tqps_num = hw->tqps_num;
+       if (tqps_num > HNS3_MAX_TQP_NUM_PER_FUNC || tqps_num == 0) {
+               PMD_INIT_LOG(ERR, "Get invalid tqps_num(%u) from PF. valid "
+                                 "range: 1~%d",
+                            tqps_num, HNS3_MAX_TQP_NUM_PER_FUNC);
+               return -EINVAL;
+       }
+
+       if (hw->rx_buf_len == 0)
+               hw->rx_buf_len = HNS3_DEFAULT_RX_BUF_LEN;
+       hw->alloc_rss_size = RTE_MIN(hw->rss_size_max, hw->tqps_num);
+
+       return 0;
+}
+
+static int
+hns3vf_get_queue_info(struct hns3_hw *hw)
+{
+#define HNS3VF_TQPS_RSS_INFO_LEN       6
+       uint8_t resp_msg[HNS3VF_TQPS_RSS_INFO_LEN];
+       int ret;
+
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_GET_QINFO, 0, NULL, 0, true,
+                               resp_msg, HNS3VF_TQPS_RSS_INFO_LEN);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to get tqp info from PF: %d", ret);
+               return ret;
+       }
+
+       memcpy(&hw->tqps_num, &resp_msg[0], sizeof(uint16_t));
+       memcpy(&hw->rss_size_max, &resp_msg[2], sizeof(uint16_t));
+       memcpy(&hw->rx_buf_len, &resp_msg[4], sizeof(uint16_t));
+
+       return hns3vf_check_tqp_info(hw);
+}
+
+static int
+hns3vf_get_queue_depth(struct hns3_hw *hw)
+{
+#define HNS3VF_TQPS_DEPTH_INFO_LEN     4
+       uint8_t resp_msg[HNS3VF_TQPS_DEPTH_INFO_LEN];
+       int ret;
+
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_GET_QDEPTH, 0, NULL, 0, true,
+                               resp_msg, HNS3VF_TQPS_DEPTH_INFO_LEN);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to get tqp depth info from PF: %d",
+                            ret);
+               return ret;
+       }
+
+       memcpy(&hw->num_tx_desc, &resp_msg[0], sizeof(uint16_t));
+       memcpy(&hw->num_rx_desc, &resp_msg[2], sizeof(uint16_t));
+
+       return 0;
+}
+
+static int
+hns3vf_get_tc_info(struct hns3_hw *hw)
+{
+       uint8_t resp_msg;
+       int ret;
+
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_GET_TCINFO, 0, NULL, 0,
+                               true, &resp_msg, sizeof(resp_msg));
+       if (ret) {
+               hns3_err(hw, "VF request to get TC info from PF failed %d",
+                        ret);
+               return ret;
+       }
+
+       hw->hw_tc_map = resp_msg;
+
+       return 0;
+}
+
+static int
+hns3vf_get_configuration(struct hns3_hw *hw)
+{
+       int ret;
+
+       hw->mac.media_type = HNS3_MEDIA_TYPE_NONE;
+
+       /* Get queue configuration from PF */
+       ret = hns3vf_get_queue_info(hw);
+       if (ret)
+               return ret;
+
+       /* Get queue depth info from PF */
+       ret = hns3vf_get_queue_depth(hw);
+       if (ret)
+               return ret;
+
+       /* Get tc configuration from PF */
+       return hns3vf_get_tc_info(hw);
+}
+
+static void
+hns3vf_set_tc_info(struct hns3_adapter *hns)
+{
+       struct hns3_hw *hw = &hns->hw;
+       uint16_t nb_rx_q = hw->data->nb_rx_queues;
+       uint16_t new_tqps;
+       uint8_t i;
+
+       hw->num_tc = 0;
+       for (i = 0; i < HNS3_MAX_TC_NUM; i++)
+               if (hw->hw_tc_map & BIT(i))
+                       hw->num_tc++;
+
+       new_tqps = RTE_MIN(hw->tqps_num, nb_rx_q);
+       hw->alloc_rss_size = RTE_MIN(hw->rss_size_max, new_tqps / hw->num_tc);
+       hw->alloc_tqps = hw->alloc_rss_size * hw->num_tc;
+
+       hns3_tc_queue_mapping_cfg(hw);
+}
+
+static void
+hns3vf_request_link_info(struct hns3_hw *hw)
+{
+       uint8_t resp_msg;
+       int ret;
+
+       if (rte_atomic16_read(&hw->reset.resetting))
+               return;
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_GET_LINK_STATUS, 0, NULL, 0, false,
+                               &resp_msg, sizeof(resp_msg));
+       if (ret)
+               hns3_err(hw, "Failed to fetch link status from PF: %d", ret);
+}
+
+static int
+hns3vf_vlan_filter_configure(struct hns3_adapter *hns, uint16_t vlan_id, int on)
+{
+#define HNS3VF_VLAN_MBX_MSG_LEN 5
+       struct hns3_hw *hw = &hns->hw;
+       uint8_t msg_data[HNS3VF_VLAN_MBX_MSG_LEN];
+       uint16_t proto = htons(RTE_ETHER_TYPE_VLAN);
+       uint8_t is_kill = on ? 0 : 1;
+
+       msg_data[0] = is_kill;
+       memcpy(&msg_data[1], &vlan_id, sizeof(vlan_id));
+       memcpy(&msg_data[3], &proto, sizeof(proto));
+
+       return hns3_send_mbx_msg(hw, HNS3_MBX_SET_VLAN, HNS3_MBX_VLAN_FILTER,
+                                msg_data, HNS3VF_VLAN_MBX_MSG_LEN, true, NULL,
+                                0);
+}
+
+static int
+hns3vf_vlan_filter_set(struct rte_eth_dev *dev, uint16_t vlan_id, int on)
+{
+       struct hns3_adapter *hns = dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+       int ret;
+
+       rte_spinlock_lock(&hw->lock);
+       ret = hns3vf_vlan_filter_configure(hns, vlan_id, on);
+       rte_spinlock_unlock(&hw->lock);
+       if (ret)
+               hns3_err(hw, "vf set vlan id failed, vlan_id =%u, ret =%d",
+                        vlan_id, ret);
+
+       return ret;
+}
+
+static int
+hns3vf_en_hw_strip_rxvtag(struct hns3_hw *hw, bool enable)
+{
+       uint8_t msg_data;
+       int ret;
+
+       msg_data = enable ? 1 : 0;
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_SET_VLAN, HNS3_MBX_VLAN_RX_OFF_CFG,
+                               &msg_data, sizeof(msg_data), false, NULL, 0);
+       if (ret)
+               hns3_err(hw, "vf enable strip failed, ret =%d", ret);
+
+       return ret;
+}
+
+static int
+hns3vf_vlan_offload_set(struct rte_eth_dev *dev, int mask)
+{
+       struct hns3_hw *hw = HNS3_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+       struct rte_eth_conf *dev_conf = &dev->data->dev_conf;
+       unsigned int tmp_mask;
+
+       tmp_mask = (unsigned int)mask;
+       /* Vlan stripping setting */
+       if (tmp_mask & ETH_VLAN_STRIP_MASK) {
+               rte_spinlock_lock(&hw->lock);
+               /* Enable or disable VLAN stripping */
+               if (dev_conf->rxmode.offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
+                       hns3vf_en_hw_strip_rxvtag(hw, true);
+               else
+                       hns3vf_en_hw_strip_rxvtag(hw, false);
+               rte_spinlock_unlock(&hw->lock);
+       }
+
+       return 0;
+}
+
+static int
+hns3vf_dev_configure_vlan(struct rte_eth_dev *dev)
+{
+       struct hns3_adapter *hns = dev->data->dev_private;
+       struct rte_eth_dev_data *data = dev->data;
+       struct hns3_hw *hw = &hns->hw;
+       int ret;
+
+       if (data->dev_conf.txmode.hw_vlan_reject_tagged ||
+           data->dev_conf.txmode.hw_vlan_reject_untagged ||
+           data->dev_conf.txmode.hw_vlan_insert_pvid) {
+               hns3_warn(hw, "hw_vlan_reject_tagged, hw_vlan_reject_untagged "
+                             "or hw_vlan_insert_pvid is not support!");
+       }
+
+       /* Apply vlan offload setting */
+       ret = hns3vf_vlan_offload_set(dev, ETH_VLAN_STRIP_MASK);
+       if (ret)
+               hns3_err(hw, "dev config vlan offload failed, ret =%d", ret);
+
+       return ret;
+}
+
+static int
+hns3vf_set_alive(struct hns3_hw *hw, bool alive)
+{
+       uint8_t msg_data;
+
+       msg_data = alive ? 1 : 0;
+       return hns3_send_mbx_msg(hw, HNS3_MBX_SET_ALIVE, 0, &msg_data,
+                                sizeof(msg_data), false, NULL, 0);
+}
+
+static void
+hns3vf_keep_alive_handler(void *param)
+{
+       struct rte_eth_dev *eth_dev = (struct rte_eth_dev *)param;
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+       uint8_t respmsg;
+       int ret;
+
+       ret = hns3_send_mbx_msg(hw, HNS3_MBX_KEEP_ALIVE, 0, NULL, 0,
+                               false, &respmsg, sizeof(uint8_t));
+       if (ret)
+               hns3_err(hw, "VF sends keeping alive cmd failed(=%d)",
+                        ret);
+
+       rte_eal_alarm_set(HNS3VF_KEEP_ALIVE_INTERVAL, hns3vf_keep_alive_handler,
+                         eth_dev);
+}
+
+static void
+hns3vf_service_handler(void *param)
+{
+       struct rte_eth_dev *eth_dev = (struct rte_eth_dev *)param;
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+
+       hns3vf_request_link_info(hw);
+
+       rte_eal_alarm_set(HNS3VF_SERVICE_INTERVAL, hns3vf_service_handler,
+                         eth_dev);
+}
+
+static int
+hns3vf_init_hardware(struct hns3_adapter *hns)
+{
+       struct hns3_hw *hw = &hns->hw;
+       uint16_t mtu = hw->data->mtu;
+       int ret;
+
+       ret = hns3vf_set_promisc_mode(hw, true);
+       if (ret)
+               return ret;
+
+       ret = hns3vf_config_mtu(hw, mtu);
+       if (ret)
+               goto err_init_hardware;
+
+       ret = hns3vf_vlan_filter_configure(hns, 0, 1);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to initialize VLAN config: %d", ret);
+               goto err_init_hardware;
+       }
+
+       ret = hns3_config_gro(hw, false);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to config gro: %d", ret);
+               goto err_init_hardware;
+       }
+
+       ret = hns3vf_set_alive(hw, true);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to VF send alive to PF: %d", ret);
+               goto err_init_hardware;
+       }
+
+       hns3vf_request_link_info(hw);
+       return 0;
+
+err_init_hardware:
+       (void)hns3vf_set_promisc_mode(hw, false);
+       return ret;
+}
+
+static int
+hns3vf_init_vf(struct rte_eth_dev *eth_dev)
+{
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+       int ret;
+
+       PMD_INIT_FUNC_TRACE();
+
+       /* Get hardware io base address from pcie BAR2 IO space */
+       hw->io_base = pci_dev->mem_resource[2].addr;
+
+       /* Firmware command queue initialize */
+       ret = hns3_cmd_init_queue(hw);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to init cmd queue: %d", ret);
+               goto err_cmd_init_queue;
+       }
+
+       /* Firmware command initialize */
+       ret = hns3_cmd_init(hw);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to init cmd: %d", ret);
+               goto err_cmd_init;
+       }
+
+       rte_spinlock_init(&hw->mbx_resp.lock);
+
+       hns3vf_clear_event_cause(hw, 0);
+
+       ret = rte_intr_callback_register(&pci_dev->intr_handle,
+                                        hns3vf_interrupt_handler, eth_dev);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to register intr: %d", ret);
+               goto err_intr_callback_register;
+       }
+
+       /* Enable interrupt */
+       rte_intr_enable(&pci_dev->intr_handle);
+       hns3vf_enable_irq0(hw);
+
+       /* Get configuration from PF */
+       ret = hns3vf_get_configuration(hw);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to fetch configuration: %d", ret);
+               goto err_get_config;
+       }
+
+       rte_eth_random_addr(hw->mac.mac_addr); /* Generate a random mac addr */
+
+       ret = hns3vf_init_hardware(hns);
+       if (ret)
+               goto err_get_config;
+
+       hns3_set_default_rss_args(hw);
+
+       return 0;
+
+err_get_config:
+       hns3vf_disable_irq0(hw);
+       rte_intr_disable(&pci_dev->intr_handle);
+
+err_intr_callback_register:
+       hns3_cmd_uninit(hw);
+
+err_cmd_init:
+       hns3_cmd_destroy_queue(hw);
+
+err_cmd_init_queue:
+       hw->io_base = NULL;
+
+       return ret;
+}
+
+static void
+hns3vf_uninit_vf(struct rte_eth_dev *eth_dev)
+{
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+
+       PMD_INIT_FUNC_TRACE();
+
+       hns3_rss_uninit(hns);
+       (void)hns3vf_set_alive(hw, false);
+       (void)hns3vf_set_promisc_mode(hw, false);
+       hns3vf_disable_irq0(hw);
+       rte_intr_disable(&pci_dev->intr_handle);
+       hns3_cmd_uninit(hw);
+       hns3_cmd_destroy_queue(hw);
+       hw->io_base = NULL;
+}
+
+static int
+hns3vf_do_stop(struct hns3_adapter *hns)
+{
+       struct hns3_hw *hw = &hns->hw;
+
+       hw->mac.link_status = ETH_LINK_DOWN;
+
+       hns3vf_configure_mac_addr(hns, true);
+
+       return 0;
+}
+
+static void
+hns3vf_dev_stop(struct rte_eth_dev *eth_dev)
+{
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+
+       PMD_INIT_FUNC_TRACE();
+
+       hw->adapter_state = HNS3_NIC_STOPPING;
+
+       rte_spinlock_lock(&hw->lock);
+       hns3vf_do_stop(hns);
+       hw->adapter_state = HNS3_NIC_CONFIGURED;
+       rte_spinlock_unlock(&hw->lock);
+}
+
+static void
+hns3vf_dev_close(struct rte_eth_dev *eth_dev)
+{
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+
+       if (hw->adapter_state == HNS3_NIC_STARTED)
+               hns3vf_dev_stop(eth_dev);
+
+       hw->adapter_state = HNS3_NIC_CLOSING;
+       rte_eal_alarm_cancel(hns3vf_keep_alive_handler, eth_dev);
+       rte_eal_alarm_cancel(hns3vf_service_handler, eth_dev);
+       hns3vf_configure_all_mc_mac_addr(hns, true);
+       hns3vf_uninit_vf(eth_dev);
+       rte_free(eth_dev->process_private);
+       eth_dev->process_private = NULL;
+       hw->adapter_state = HNS3_NIC_CLOSED;
+       hns3_warn(hw, "Close port %d finished", hw->data->port_id);
+}
+
+static int
+hns3vf_dev_link_update(struct rte_eth_dev *eth_dev,
+                      __rte_unused int wait_to_complete)
+{
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+       struct hns3_mac *mac = &hw->mac;
+       struct rte_eth_link new_link;
+
+       hns3vf_request_link_info(hw);
+
+       memset(&new_link, 0, sizeof(new_link));
+       switch (mac->link_speed) {
+       case ETH_SPEED_NUM_10M:
+       case ETH_SPEED_NUM_100M:
+       case ETH_SPEED_NUM_1G:
+       case ETH_SPEED_NUM_10G:
+       case ETH_SPEED_NUM_25G:
+       case ETH_SPEED_NUM_40G:
+       case ETH_SPEED_NUM_50G:
+       case ETH_SPEED_NUM_100G:
+               new_link.link_speed = mac->link_speed;
+               break;
+       default:
+               new_link.link_speed = ETH_SPEED_NUM_100M;
+               break;
+       }
+
+       new_link.link_duplex = mac->link_duplex;
+       new_link.link_status = mac->link_status ? ETH_LINK_UP : ETH_LINK_DOWN;
+       new_link.link_autoneg =
+           !(eth_dev->data->dev_conf.link_speeds & ETH_LINK_SPEED_FIXED);
+
+       return rte_eth_linkstatus_set(eth_dev, &new_link);
+}
+
+static int
+hns3vf_do_start(struct hns3_adapter *hns, __rte_unused bool reset_queue)
+{
+       hns3vf_set_tc_info(hns);
+
+       return 0;
+}
+
+static int
+hns3vf_dev_start(struct rte_eth_dev *eth_dev)
+{
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+       int ret;
+
+       PMD_INIT_FUNC_TRACE();
+       rte_spinlock_lock(&hw->lock);
+       hw->adapter_state = HNS3_NIC_STARTING;
+       ret = hns3vf_do_start(hns, true);
+       if (ret) {
+               hw->adapter_state = HNS3_NIC_CONFIGURED;
+               rte_spinlock_unlock(&hw->lock);
+               return ret;
+       }
+       hw->adapter_state = HNS3_NIC_STARTED;
+       rte_spinlock_unlock(&hw->lock);
+       return 0;
+}
+
+static const struct eth_dev_ops hns3vf_eth_dev_ops = {
+       .dev_start          = hns3vf_dev_start,
+       .dev_stop           = hns3vf_dev_stop,
+       .dev_close          = hns3vf_dev_close,
+       .mtu_set            = hns3vf_dev_mtu_set,
+       .dev_infos_get      = hns3vf_dev_infos_get,
+       .dev_configure      = hns3vf_dev_configure,
+       .mac_addr_add       = hns3vf_add_mac_addr,
+       .mac_addr_remove    = hns3vf_remove_mac_addr,
+       .mac_addr_set       = hns3vf_set_default_mac_addr,
+       .set_mc_addr_list   = hns3vf_set_mc_mac_addr_list,
+       .link_update        = hns3vf_dev_link_update,
+       .rss_hash_update    = hns3_dev_rss_hash_update,
+       .rss_hash_conf_get  = hns3_dev_rss_hash_conf_get,
+       .reta_update        = hns3_dev_rss_reta_update,
+       .reta_query         = hns3_dev_rss_reta_query,
+       .filter_ctrl        = hns3_dev_filter_ctrl,
+       .vlan_filter_set    = hns3vf_vlan_filter_set,
+       .vlan_offload_set   = hns3vf_vlan_offload_set,
+};
+
+static int
+hns3vf_dev_init(struct rte_eth_dev *eth_dev)
+{
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+       int ret;
+
+       PMD_INIT_FUNC_TRACE();
+
+       eth_dev->process_private = (struct hns3_process_private *)
+           rte_zmalloc_socket("hns3_filter_list",
+                              sizeof(struct hns3_process_private),
+                              RTE_CACHE_LINE_SIZE, eth_dev->device->numa_node);
+       if (eth_dev->process_private == NULL) {
+               PMD_INIT_LOG(ERR, "Failed to alloc memory for process private");
+               return -ENOMEM;
+       }
+
+       /* initialize flow filter lists */
+       hns3_filterlist_init(eth_dev);
+
+       eth_dev->dev_ops = &hns3vf_eth_dev_ops;
+       if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+               return 0;
+
+       hw->adapter_state = HNS3_NIC_UNINITIALIZED;
+       hns->is_vf = true;
+       hw->data = eth_dev->data;
+
+       ret = hns3vf_init_vf(eth_dev);
+       if (ret) {
+               PMD_INIT_LOG(ERR, "Failed to init vf: %d", ret);
+               goto err_init_vf;
+       }
+
+       /* Allocate memory for storing MAC addresses */
+       eth_dev->data->mac_addrs = rte_zmalloc("hns3vf-mac",
+                                              sizeof(struct rte_ether_addr) *
+                                              HNS3_VF_UC_MACADDR_NUM, 0);
+       if (eth_dev->data->mac_addrs == NULL) {
+               PMD_INIT_LOG(ERR, "Failed to allocate %zx bytes needed "
+                            "to store MAC addresses",
+                            sizeof(struct rte_ether_addr) *
+                            HNS3_VF_UC_MACADDR_NUM);
+               ret = -ENOMEM;
+               goto err_rte_zmalloc;
+       }
+
+       rte_ether_addr_copy((struct rte_ether_addr *)hw->mac.mac_addr,
+                           &eth_dev->data->mac_addrs[0]);
+       hw->adapter_state = HNS3_NIC_INITIALIZED;
+       /*
+        * Pass the information to the rte_eth_dev_close() that it should also
+        * release the private port resources.
+        */
+       eth_dev->data->dev_flags |= RTE_ETH_DEV_CLOSE_REMOVE;
+
+       rte_eal_alarm_set(HNS3VF_KEEP_ALIVE_INTERVAL, hns3vf_keep_alive_handler,
+                         eth_dev);
+       rte_eal_alarm_set(HNS3VF_SERVICE_INTERVAL, hns3vf_service_handler,
+                         eth_dev);
+       return 0;
+
+err_rte_zmalloc:
+       hns3vf_uninit_vf(eth_dev);
+
+err_init_vf:
+       eth_dev->dev_ops = NULL;
+       eth_dev->rx_pkt_burst = NULL;
+       eth_dev->tx_pkt_burst = NULL;
+       eth_dev->tx_pkt_prepare = NULL;
+       rte_free(eth_dev->process_private);
+       eth_dev->process_private = NULL;
+
+       return ret;
+}
+
+static int
+hns3vf_dev_uninit(struct rte_eth_dev *eth_dev)
+{
+       struct hns3_adapter *hns = eth_dev->data->dev_private;
+       struct hns3_hw *hw = &hns->hw;
+
+       PMD_INIT_FUNC_TRACE();
+
+       if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+               return -EPERM;
+
+       eth_dev->dev_ops = NULL;
+       eth_dev->rx_pkt_burst = NULL;
+       eth_dev->tx_pkt_burst = NULL;
+       eth_dev->tx_pkt_prepare = NULL;
+
+       if (hw->adapter_state < HNS3_NIC_CLOSING)
+               hns3vf_dev_close(eth_dev);
+
+       hw->adapter_state = HNS3_NIC_REMOVED;
+       return 0;
+}
+
+static int
+eth_hns3vf_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
+                    struct rte_pci_device *pci_dev)
+{
+       return rte_eth_dev_pci_generic_probe(pci_dev,
+                                            sizeof(struct hns3_adapter),
+                                            hns3vf_dev_init);
+}
+
+static int
+eth_hns3vf_pci_remove(struct rte_pci_device *pci_dev)
+{
+       return rte_eth_dev_pci_generic_remove(pci_dev, hns3vf_dev_uninit);
+}
+
+static const struct rte_pci_id pci_id_hns3vf_map[] = {
+       { RTE_PCI_DEVICE(PCI_VENDOR_ID_HUAWEI, HNS3_DEV_ID_100G_VF) },
+       { RTE_PCI_DEVICE(PCI_VENDOR_ID_HUAWEI, HNS3_DEV_ID_100G_RDMA_PFC_VF) },
+       { .vendor_id = 0, /* sentinel */ },
+};
+
+static struct rte_pci_driver rte_hns3vf_pmd = {
+       .id_table = pci_id_hns3vf_map,
+       .drv_flags = RTE_PCI_DRV_NEED_MAPPING,
+       .probe = eth_hns3vf_pci_probe,
+       .remove = eth_hns3vf_pci_remove,
+};
+
+RTE_PMD_REGISTER_PCI(net_hns3_vf, rte_hns3vf_pmd);
+RTE_PMD_REGISTER_PCI_TABLE(net_hns3_vf, pci_id_hns3vf_map);
+RTE_PMD_REGISTER_KMOD_DEP(net_hns3_vf, "* igb_uio | vfio-pci");
index 622f0a7..4a70992 100644 (file)
@@ -16,6 +16,7 @@ endif
 sources = files('hns3_cmd.c',
        'hns3_dcb.c',
        'hns3_ethdev.c',
+       'hns3_ethdev_vf.c',
        'hns3_fdir.c',
        'hns3_flow.c',
        'hns3_mbx.c',