ethdev: make driver-only headers private
[dpdk.git] / drivers / net / txgbe / txgbe_ethdev.c
index 544db8b..23f9d17 100644 (file)
@@ -7,9 +7,11 @@
 #include <stdint.h>
 #include <string.h>
 #include <rte_common.h>
-#include <rte_ethdev_pci.h>
+#include <ethdev_pci.h>
 
 #include <rte_interrupts.h>
+#include <rte_log.h>
+#include <rte_debug.h>
 #include <rte_pci.h>
 #include <rte_memory.h>
 #include <rte_eal.h>
 #include "base/txgbe.h"
 #include "txgbe_ethdev.h"
 #include "txgbe_rxtx.h"
+#include "txgbe_regs_group.h"
+
+static const struct reg_info txgbe_regs_general[] = {
+       {TXGBE_RST, 1, 1, "TXGBE_RST"},
+       {TXGBE_STAT, 1, 1, "TXGBE_STAT"},
+       {TXGBE_PORTCTL, 1, 1, "TXGBE_PORTCTL"},
+       {TXGBE_SDP, 1, 1, "TXGBE_SDP"},
+       {TXGBE_SDPCTL, 1, 1, "TXGBE_SDPCTL"},
+       {TXGBE_LEDCTL, 1, 1, "TXGBE_LEDCTL"},
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_nvm[] = {
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_interrupt[] = {
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_fctl_others[] = {
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_rxdma[] = {
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_rx[] = {
+       {0, 0, 0, ""}
+};
+
+static struct reg_info txgbe_regs_tx[] = {
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_wakeup[] = {
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_dcb[] = {
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_mac[] = {
+       {0, 0, 0, ""}
+};
+
+static const struct reg_info txgbe_regs_diagnostic[] = {
+       {0, 0, 0, ""},
+};
 
+/* PF registers */
+static const struct reg_info *txgbe_regs_others[] = {
+                               txgbe_regs_general,
+                               txgbe_regs_nvm,
+                               txgbe_regs_interrupt,
+                               txgbe_regs_fctl_others,
+                               txgbe_regs_rxdma,
+                               txgbe_regs_rx,
+                               txgbe_regs_tx,
+                               txgbe_regs_wakeup,
+                               txgbe_regs_dcb,
+                               txgbe_regs_mac,
+                               txgbe_regs_diagnostic,
+                               NULL};
+
+static int txgbe_fdir_filter_init(struct rte_eth_dev *eth_dev);
+static int txgbe_fdir_filter_uninit(struct rte_eth_dev *eth_dev);
+static int txgbe_l2_tn_filter_init(struct rte_eth_dev *eth_dev);
+static int txgbe_l2_tn_filter_uninit(struct rte_eth_dev *eth_dev);
+static int  txgbe_dev_set_link_up(struct rte_eth_dev *dev);
+static int  txgbe_dev_set_link_down(struct rte_eth_dev *dev);
 static int txgbe_dev_close(struct rte_eth_dev *dev);
+static int txgbe_dev_link_update(struct rte_eth_dev *dev,
+                               int wait_to_complete);
+static int txgbe_dev_stats_reset(struct rte_eth_dev *dev);
+static void txgbe_vlan_hw_strip_enable(struct rte_eth_dev *dev, uint16_t queue);
+static void txgbe_vlan_hw_strip_disable(struct rte_eth_dev *dev,
+                                       uint16_t queue);
 
 static void txgbe_dev_link_status_print(struct rte_eth_dev *dev);
 static int txgbe_dev_lsc_interrupt_setup(struct rte_eth_dev *dev, uint8_t on);
@@ -33,6 +113,27 @@ static void txgbe_dev_interrupt_handler(void *param);
 static void txgbe_dev_interrupt_delayed_handler(void *param);
 static void txgbe_configure_msix(struct rte_eth_dev *dev);
 
+static int txgbe_filter_restore(struct rte_eth_dev *dev);
+static void txgbe_l2_tunnel_conf(struct rte_eth_dev *dev);
+
+#define TXGBE_SET_HWSTRIP(h, q) do {\
+               uint32_t idx = (q) / (sizeof((h)->bitmap[0]) * NBBY); \
+               uint32_t bit = (q) % (sizeof((h)->bitmap[0]) * NBBY); \
+               (h)->bitmap[idx] |= 1 << bit;\
+       } while (0)
+
+#define TXGBE_CLEAR_HWSTRIP(h, q) do {\
+               uint32_t idx = (q) / (sizeof((h)->bitmap[0]) * NBBY); \
+               uint32_t bit = (q) % (sizeof((h)->bitmap[0]) * NBBY); \
+               (h)->bitmap[idx] &= ~(1 << bit);\
+       } while (0)
+
+#define TXGBE_GET_HWSTRIP(h, q, r) do {\
+               uint32_t idx = (q) / (sizeof((h)->bitmap[0]) * NBBY); \
+               uint32_t bit = (q) % (sizeof((h)->bitmap[0]) * NBBY); \
+               (r) = (h)->bitmap[idx] >> bit & 1;\
+       } while (0)
+
 /*
  * The set of PCI devices this driver supports
  */
@@ -58,6 +159,144 @@ static const struct rte_eth_desc_lim tx_desc_lim = {
 
 static const struct eth_dev_ops txgbe_eth_dev_ops;
 
+#define HW_XSTAT(m) {#m, offsetof(struct txgbe_hw_stats, m)}
+#define HW_XSTAT_NAME(m, n) {n, offsetof(struct txgbe_hw_stats, m)}
+static const struct rte_txgbe_xstats_name_off rte_txgbe_stats_strings[] = {
+       /* MNG RxTx */
+       HW_XSTAT(mng_bmc2host_packets),
+       HW_XSTAT(mng_host2bmc_packets),
+       /* Basic RxTx */
+       HW_XSTAT(rx_packets),
+       HW_XSTAT(tx_packets),
+       HW_XSTAT(rx_bytes),
+       HW_XSTAT(tx_bytes),
+       HW_XSTAT(rx_total_bytes),
+       HW_XSTAT(rx_total_packets),
+       HW_XSTAT(tx_total_packets),
+       HW_XSTAT(rx_total_missed_packets),
+       HW_XSTAT(rx_broadcast_packets),
+       HW_XSTAT(rx_multicast_packets),
+       HW_XSTAT(rx_management_packets),
+       HW_XSTAT(tx_management_packets),
+       HW_XSTAT(rx_management_dropped),
+
+       /* Basic Error */
+       HW_XSTAT(rx_crc_errors),
+       HW_XSTAT(rx_illegal_byte_errors),
+       HW_XSTAT(rx_error_bytes),
+       HW_XSTAT(rx_mac_short_packet_dropped),
+       HW_XSTAT(rx_length_errors),
+       HW_XSTAT(rx_undersize_errors),
+       HW_XSTAT(rx_fragment_errors),
+       HW_XSTAT(rx_oversize_errors),
+       HW_XSTAT(rx_jabber_errors),
+       HW_XSTAT(rx_l3_l4_xsum_error),
+       HW_XSTAT(mac_local_errors),
+       HW_XSTAT(mac_remote_errors),
+
+       /* Flow Director */
+       HW_XSTAT(flow_director_added_filters),
+       HW_XSTAT(flow_director_removed_filters),
+       HW_XSTAT(flow_director_filter_add_errors),
+       HW_XSTAT(flow_director_filter_remove_errors),
+       HW_XSTAT(flow_director_matched_filters),
+       HW_XSTAT(flow_director_missed_filters),
+
+       /* FCoE */
+       HW_XSTAT(rx_fcoe_crc_errors),
+       HW_XSTAT(rx_fcoe_mbuf_allocation_errors),
+       HW_XSTAT(rx_fcoe_dropped),
+       HW_XSTAT(rx_fcoe_packets),
+       HW_XSTAT(tx_fcoe_packets),
+       HW_XSTAT(rx_fcoe_bytes),
+       HW_XSTAT(tx_fcoe_bytes),
+       HW_XSTAT(rx_fcoe_no_ddp),
+       HW_XSTAT(rx_fcoe_no_ddp_ext_buff),
+
+       /* MACSEC */
+       HW_XSTAT(tx_macsec_pkts_untagged),
+       HW_XSTAT(tx_macsec_pkts_encrypted),
+       HW_XSTAT(tx_macsec_pkts_protected),
+       HW_XSTAT(tx_macsec_octets_encrypted),
+       HW_XSTAT(tx_macsec_octets_protected),
+       HW_XSTAT(rx_macsec_pkts_untagged),
+       HW_XSTAT(rx_macsec_pkts_badtag),
+       HW_XSTAT(rx_macsec_pkts_nosci),
+       HW_XSTAT(rx_macsec_pkts_unknownsci),
+       HW_XSTAT(rx_macsec_octets_decrypted),
+       HW_XSTAT(rx_macsec_octets_validated),
+       HW_XSTAT(rx_macsec_sc_pkts_unchecked),
+       HW_XSTAT(rx_macsec_sc_pkts_delayed),
+       HW_XSTAT(rx_macsec_sc_pkts_late),
+       HW_XSTAT(rx_macsec_sa_pkts_ok),
+       HW_XSTAT(rx_macsec_sa_pkts_invalid),
+       HW_XSTAT(rx_macsec_sa_pkts_notvalid),
+       HW_XSTAT(rx_macsec_sa_pkts_unusedsa),
+       HW_XSTAT(rx_macsec_sa_pkts_notusingsa),
+
+       /* MAC RxTx */
+       HW_XSTAT(rx_size_64_packets),
+       HW_XSTAT(rx_size_65_to_127_packets),
+       HW_XSTAT(rx_size_128_to_255_packets),
+       HW_XSTAT(rx_size_256_to_511_packets),
+       HW_XSTAT(rx_size_512_to_1023_packets),
+       HW_XSTAT(rx_size_1024_to_max_packets),
+       HW_XSTAT(tx_size_64_packets),
+       HW_XSTAT(tx_size_65_to_127_packets),
+       HW_XSTAT(tx_size_128_to_255_packets),
+       HW_XSTAT(tx_size_256_to_511_packets),
+       HW_XSTAT(tx_size_512_to_1023_packets),
+       HW_XSTAT(tx_size_1024_to_max_packets),
+
+       /* Flow Control */
+       HW_XSTAT(tx_xon_packets),
+       HW_XSTAT(rx_xon_packets),
+       HW_XSTAT(tx_xoff_packets),
+       HW_XSTAT(rx_xoff_packets),
+
+       HW_XSTAT_NAME(tx_xon_packets, "tx_flow_control_xon_packets"),
+       HW_XSTAT_NAME(rx_xon_packets, "rx_flow_control_xon_packets"),
+       HW_XSTAT_NAME(tx_xoff_packets, "tx_flow_control_xoff_packets"),
+       HW_XSTAT_NAME(rx_xoff_packets, "rx_flow_control_xoff_packets"),
+};
+
+#define TXGBE_NB_HW_STATS (sizeof(rte_txgbe_stats_strings) / \
+                          sizeof(rte_txgbe_stats_strings[0]))
+
+/* Per-priority statistics */
+#define UP_XSTAT(m) {#m, offsetof(struct txgbe_hw_stats, up[0].m)}
+static const struct rte_txgbe_xstats_name_off rte_txgbe_up_strings[] = {
+       UP_XSTAT(rx_up_packets),
+       UP_XSTAT(tx_up_packets),
+       UP_XSTAT(rx_up_bytes),
+       UP_XSTAT(tx_up_bytes),
+       UP_XSTAT(rx_up_drop_packets),
+
+       UP_XSTAT(tx_up_xon_packets),
+       UP_XSTAT(rx_up_xon_packets),
+       UP_XSTAT(tx_up_xoff_packets),
+       UP_XSTAT(rx_up_xoff_packets),
+       UP_XSTAT(rx_up_dropped),
+       UP_XSTAT(rx_up_mbuf_alloc_errors),
+       UP_XSTAT(tx_up_xon2off_packets),
+};
+
+#define TXGBE_NB_UP_STATS (sizeof(rte_txgbe_up_strings) / \
+                          sizeof(rte_txgbe_up_strings[0]))
+
+/* Per-queue statistics */
+#define QP_XSTAT(m) {#m, offsetof(struct txgbe_hw_stats, qp[0].m)}
+static const struct rte_txgbe_xstats_name_off rte_txgbe_qp_strings[] = {
+       QP_XSTAT(rx_qp_packets),
+       QP_XSTAT(tx_qp_packets),
+       QP_XSTAT(rx_qp_bytes),
+       QP_XSTAT(tx_qp_bytes),
+       QP_XSTAT(rx_qp_mc_packets),
+};
+
+#define TXGBE_NB_QP_STATS (sizeof(rte_txgbe_qp_strings) / \
+                          sizeof(rte_txgbe_qp_strings[0]))
+
 static inline int
 txgbe_is_sfp(struct txgbe_hw *hw)
 {
@@ -74,6 +313,25 @@ txgbe_is_sfp(struct txgbe_hw *hw)
        }
 }
 
+static inline int32_t
+txgbe_pf_reset_hw(struct txgbe_hw *hw)
+{
+       uint32_t ctrl_ext;
+       int32_t status;
+
+       status = hw->mac.reset_hw(hw);
+
+       ctrl_ext = rd32(hw, TXGBE_PORTCTL);
+       /* Set PF Reset Done bit so PF/VF Mail Ops can work */
+       ctrl_ext |= TXGBE_PORTCTL_RSTDONE;
+       wr32(hw, TXGBE_PORTCTL, ctrl_ext);
+       txgbe_flush(hw);
+
+       if (status == TXGBE_ERR_SFP_NOT_PRESENT)
+               status = 0;
+       return status;
+}
+
 static inline void
 txgbe_enable_intr(struct rte_eth_dev *dev)
 {
@@ -97,19 +355,170 @@ txgbe_disable_intr(struct txgbe_hw *hw)
        txgbe_flush(hw);
 }
 
+static int
+txgbe_dev_queue_stats_mapping_set(struct rte_eth_dev *eth_dev,
+                                 uint16_t queue_id,
+                                 uint8_t stat_idx,
+                                 uint8_t is_rx)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(eth_dev);
+       struct txgbe_stat_mappings *stat_mappings =
+               TXGBE_DEV_STAT_MAPPINGS(eth_dev);
+       uint32_t qsmr_mask = 0;
+       uint32_t clearing_mask = QMAP_FIELD_RESERVED_BITS_MASK;
+       uint32_t q_map;
+       uint8_t n, offset;
+
+       if (hw->mac.type != txgbe_mac_raptor)
+               return -ENOSYS;
+
+       if (stat_idx & !QMAP_FIELD_RESERVED_BITS_MASK)
+               return -EIO;
+
+       PMD_INIT_LOG(DEBUG, "Setting port %d, %s queue_id %d to stat index %d",
+                    (int)(eth_dev->data->port_id), is_rx ? "RX" : "TX",
+                    queue_id, stat_idx);
+
+       n = (uint8_t)(queue_id / NB_QMAP_FIELDS_PER_QSM_REG);
+       if (n >= TXGBE_NB_STAT_MAPPING) {
+               PMD_INIT_LOG(ERR, "Nb of stat mapping registers exceeded");
+               return -EIO;
+       }
+       offset = (uint8_t)(queue_id % NB_QMAP_FIELDS_PER_QSM_REG);
+
+       /* Now clear any previous stat_idx set */
+       clearing_mask <<= (QSM_REG_NB_BITS_PER_QMAP_FIELD * offset);
+       if (!is_rx)
+               stat_mappings->tqsm[n] &= ~clearing_mask;
+       else
+               stat_mappings->rqsm[n] &= ~clearing_mask;
+
+       q_map = (uint32_t)stat_idx;
+       q_map &= QMAP_FIELD_RESERVED_BITS_MASK;
+       qsmr_mask = q_map << (QSM_REG_NB_BITS_PER_QMAP_FIELD * offset);
+       if (!is_rx)
+               stat_mappings->tqsm[n] |= qsmr_mask;
+       else
+               stat_mappings->rqsm[n] |= qsmr_mask;
+
+       PMD_INIT_LOG(DEBUG, "Set port %d, %s queue_id %d to stat index %d",
+                    (int)(eth_dev->data->port_id), is_rx ? "RX" : "TX",
+                    queue_id, stat_idx);
+       PMD_INIT_LOG(DEBUG, "%s[%d] = 0x%08x", is_rx ? "RQSMR" : "TQSM", n,
+                    is_rx ? stat_mappings->rqsm[n] : stat_mappings->tqsm[n]);
+       return 0;
+}
+
+static void
+txgbe_dcb_init(struct txgbe_hw *hw, struct txgbe_dcb_config *dcb_config)
+{
+       int i;
+       u8 bwgp;
+       struct txgbe_dcb_tc_config *tc;
+
+       UNREFERENCED_PARAMETER(hw);
+
+       dcb_config->num_tcs.pg_tcs = TXGBE_DCB_TC_MAX;
+       dcb_config->num_tcs.pfc_tcs = TXGBE_DCB_TC_MAX;
+       bwgp = (u8)(100 / TXGBE_DCB_TC_MAX);
+       for (i = 0; i < TXGBE_DCB_TC_MAX; i++) {
+               tc = &dcb_config->tc_config[i];
+               tc->path[TXGBE_DCB_TX_CONFIG].bwg_id = i;
+               tc->path[TXGBE_DCB_TX_CONFIG].bwg_percent = bwgp + (i & 1);
+               tc->path[TXGBE_DCB_RX_CONFIG].bwg_id = i;
+               tc->path[TXGBE_DCB_RX_CONFIG].bwg_percent = bwgp + (i & 1);
+               tc->pfc = txgbe_dcb_pfc_disabled;
+       }
+
+       /* Initialize default user to priority mapping, UPx->TC0 */
+       tc = &dcb_config->tc_config[0];
+       tc->path[TXGBE_DCB_TX_CONFIG].up_to_tc_bitmap = 0xFF;
+       tc->path[TXGBE_DCB_RX_CONFIG].up_to_tc_bitmap = 0xFF;
+       for (i = 0; i < TXGBE_DCB_BWG_MAX; i++) {
+               dcb_config->bw_percentage[i][TXGBE_DCB_TX_CONFIG] = 100;
+               dcb_config->bw_percentage[i][TXGBE_DCB_RX_CONFIG] = 100;
+       }
+       dcb_config->rx_pba_cfg = txgbe_dcb_pba_equal;
+       dcb_config->pfc_mode_enable = false;
+       dcb_config->vt_mode = true;
+       dcb_config->round_robin_enable = false;
+       /* support all DCB capabilities */
+       dcb_config->support.capabilities = 0xFF;
+}
+
+/*
+ * Ensure that all locks are released before first NVM or PHY access
+ */
+static void
+txgbe_swfw_lock_reset(struct txgbe_hw *hw)
+{
+       uint16_t mask;
+
+       /*
+        * These ones are more tricky since they are common to all ports; but
+        * swfw_sync retries last long enough (1s) to be almost sure that if
+        * lock can not be taken it is due to an improper lock of the
+        * semaphore.
+        */
+       mask = TXGBE_MNGSEM_SWPHY |
+              TXGBE_MNGSEM_SWMBX |
+              TXGBE_MNGSEM_SWFLASH;
+       if (hw->mac.acquire_swfw_sync(hw, mask) < 0)
+               PMD_DRV_LOG(DEBUG, "SWFW common locks released");
+
+       hw->mac.release_swfw_sync(hw, mask);
+}
+
 static int
 eth_txgbe_dev_init(struct rte_eth_dev *eth_dev, void *init_params __rte_unused)
 {
        struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
        struct txgbe_hw *hw = TXGBE_DEV_HW(eth_dev);
+       struct txgbe_vfta *shadow_vfta = TXGBE_DEV_VFTA(eth_dev);
+       struct txgbe_hwstrip *hwstrip = TXGBE_DEV_HWSTRIP(eth_dev);
+       struct txgbe_dcb_config *dcb_config = TXGBE_DEV_DCB_CONFIG(eth_dev);
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(eth_dev);
+       struct txgbe_bw_conf *bw_conf = TXGBE_DEV_BW_CONF(eth_dev);
        struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
        const struct rte_memzone *mz;
+       uint32_t ctrl_ext;
        uint16_t csum;
-       int err;
+       int err, i, ret;
 
        PMD_INIT_FUNC_TRACE();
 
        eth_dev->dev_ops = &txgbe_eth_dev_ops;
+       eth_dev->rx_queue_count       = txgbe_dev_rx_queue_count;
+       eth_dev->rx_descriptor_status = txgbe_dev_rx_descriptor_status;
+       eth_dev->tx_descriptor_status = txgbe_dev_tx_descriptor_status;
+       eth_dev->rx_pkt_burst = &txgbe_recv_pkts;
+       eth_dev->tx_pkt_burst = &txgbe_xmit_pkts;
+       eth_dev->tx_pkt_prepare = &txgbe_prep_pkts;
+
+       /*
+        * For secondary processes, we don't initialise any further as primary
+        * has already done this work. Only check we don't need a different
+        * RX and TX function.
+        */
+       if (rte_eal_process_type() != RTE_PROC_PRIMARY) {
+               struct txgbe_tx_queue *txq;
+               /* TX queue function in primary, set by last queue initialized
+                * Tx queue may not initialized by primary process
+                */
+               if (eth_dev->data->tx_queues) {
+                       uint16_t nb_tx_queues = eth_dev->data->nb_tx_queues;
+                       txq = eth_dev->data->tx_queues[nb_tx_queues - 1];
+                       txgbe_set_tx_function(eth_dev, txq);
+               } else {
+                       /* Use default TX function if we get here */
+                       PMD_INIT_LOG(NOTICE, "No TX queues configured yet. "
+                                    "Using default TX function.");
+               }
+
+               txgbe_set_rx_function(eth_dev);
+
+               return 0;
+       }
 
        rte_eth_copy_pci_info(eth_dev, pci_dev);
 
@@ -135,6 +544,29 @@ eth_txgbe_dev_init(struct rte_eth_dev *eth_dev, void *init_params __rte_unused)
                return -EIO;
        }
 
+       /* Unlock any pending hardware semaphore */
+       txgbe_swfw_lock_reset(hw);
+
+#ifdef RTE_LIB_SECURITY
+       /* Initialize security_ctx only for primary process*/
+       if (txgbe_ipsec_ctx_create(eth_dev))
+               return -ENOMEM;
+#endif
+
+       /* Initialize DCB configuration*/
+       memset(dcb_config, 0, sizeof(struct txgbe_dcb_config));
+       txgbe_dcb_init(hw, dcb_config);
+
+       /* Get Hardware Flow Control setting */
+       hw->fc.requested_mode = txgbe_fc_full;
+       hw->fc.current_mode = txgbe_fc_full;
+       hw->fc.pause_time = TXGBE_FC_PAUSE_TIME;
+       for (i = 0; i < TXGBE_DCB_TC_MAX; i++) {
+               hw->fc.low_water[i] = TXGBE_FC_XON_LOTH;
+               hw->fc.high_water[i] = TXGBE_FC_XOFF_HITH;
+       }
+       hw->fc.send_xon = 1;
+
        err = hw->rom.init_params(hw);
        if (err != 0) {
                PMD_INIT_LOG(ERR, "The EEPROM init failed: %d", err);
@@ -183,6 +615,9 @@ eth_txgbe_dev_init(struct rte_eth_dev *eth_dev, void *init_params __rte_unused)
                return -EIO;
        }
 
+       /* Reset the hw statistics */
+       txgbe_dev_stats_reset(eth_dev);
+
        /* disable interrupt */
        txgbe_disable_intr(hw);
 
@@ -211,6 +646,30 @@ eth_txgbe_dev_init(struct rte_eth_dev *eth_dev, void *init_params __rte_unused)
                return -ENOMEM;
        }
 
+       /* initialize the vfta */
+       memset(shadow_vfta, 0, sizeof(*shadow_vfta));
+
+       /* initialize the hw strip bitmap*/
+       memset(hwstrip, 0, sizeof(*hwstrip));
+
+       /* initialize PF if max_vfs not zero */
+       ret = txgbe_pf_host_init(eth_dev);
+       if (ret) {
+               rte_free(eth_dev->data->mac_addrs);
+               eth_dev->data->mac_addrs = NULL;
+               rte_free(eth_dev->data->hash_mac_addrs);
+               eth_dev->data->hash_mac_addrs = NULL;
+               return ret;
+       }
+
+       ctrl_ext = rd32(hw, TXGBE_PORTCTL);
+       /* let hardware know driver is loaded */
+       ctrl_ext |= TXGBE_PORTCTL_DRVLOAD;
+       /* Set PF Reset Done bit so PF/VF Mail Ops can work */
+       ctrl_ext |= TXGBE_PORTCTL_RSTDONE;
+       wr32(hw, TXGBE_PORTCTL, ctrl_ext);
+       txgbe_flush(hw);
+
        if (txgbe_is_sfp(hw) && hw->phy.sfp_type != txgbe_sfp_type_not_present)
                PMD_INIT_LOG(DEBUG, "MAC: %d, PHY: %d, SFP+: %d",
                             (int)hw->mac.type, (int)hw->phy.type,
@@ -232,6 +691,28 @@ eth_txgbe_dev_init(struct rte_eth_dev *eth_dev, void *init_params __rte_unused)
        /* enable support intr */
        txgbe_enable_intr(eth_dev);
 
+       /* initialize filter info */
+       memset(filter_info, 0,
+              sizeof(struct txgbe_filter_info));
+
+       /* initialize 5tuple filter list */
+       TAILQ_INIT(&filter_info->fivetuple_list);
+
+       /* initialize flow director filter list & hash */
+       txgbe_fdir_filter_init(eth_dev);
+
+       /* initialize l2 tunnel filter list & hash */
+       txgbe_l2_tn_filter_init(eth_dev);
+
+       /* initialize flow filter lists */
+       txgbe_filterlist_init();
+
+       /* initialize bandwidth configuration info */
+       memset(bw_conf, 0, sizeof(struct txgbe_bw_conf));
+
+       /* initialize Traffic Manager configuration */
+       txgbe_tm_conf_init(eth_dev);
+
        return 0;
 }
 
@@ -248,6 +729,135 @@ eth_txgbe_dev_uninit(struct rte_eth_dev *eth_dev)
        return 0;
 }
 
+static int txgbe_ntuple_filter_uninit(struct rte_eth_dev *eth_dev)
+{
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(eth_dev);
+       struct txgbe_5tuple_filter *p_5tuple;
+
+       while ((p_5tuple = TAILQ_FIRST(&filter_info->fivetuple_list))) {
+               TAILQ_REMOVE(&filter_info->fivetuple_list,
+                            p_5tuple,
+                            entries);
+               rte_free(p_5tuple);
+       }
+       memset(filter_info->fivetuple_mask, 0,
+              sizeof(uint32_t) * TXGBE_5TUPLE_ARRAY_SIZE);
+
+       return 0;
+}
+
+static int txgbe_fdir_filter_uninit(struct rte_eth_dev *eth_dev)
+{
+       struct txgbe_hw_fdir_info *fdir_info = TXGBE_DEV_FDIR(eth_dev);
+       struct txgbe_fdir_filter *fdir_filter;
+
+       if (fdir_info->hash_map)
+               rte_free(fdir_info->hash_map);
+       if (fdir_info->hash_handle)
+               rte_hash_free(fdir_info->hash_handle);
+
+       while ((fdir_filter = TAILQ_FIRST(&fdir_info->fdir_list))) {
+               TAILQ_REMOVE(&fdir_info->fdir_list,
+                            fdir_filter,
+                            entries);
+               rte_free(fdir_filter);
+       }
+
+       return 0;
+}
+
+static int txgbe_l2_tn_filter_uninit(struct rte_eth_dev *eth_dev)
+{
+       struct txgbe_l2_tn_info *l2_tn_info = TXGBE_DEV_L2_TN(eth_dev);
+       struct txgbe_l2_tn_filter *l2_tn_filter;
+
+       if (l2_tn_info->hash_map)
+               rte_free(l2_tn_info->hash_map);
+       if (l2_tn_info->hash_handle)
+               rte_hash_free(l2_tn_info->hash_handle);
+
+       while ((l2_tn_filter = TAILQ_FIRST(&l2_tn_info->l2_tn_list))) {
+               TAILQ_REMOVE(&l2_tn_info->l2_tn_list,
+                            l2_tn_filter,
+                            entries);
+               rte_free(l2_tn_filter);
+       }
+
+       return 0;
+}
+
+static int txgbe_fdir_filter_init(struct rte_eth_dev *eth_dev)
+{
+       struct txgbe_hw_fdir_info *fdir_info = TXGBE_DEV_FDIR(eth_dev);
+       char fdir_hash_name[RTE_HASH_NAMESIZE];
+       struct rte_hash_parameters fdir_hash_params = {
+               .name = fdir_hash_name,
+               .entries = TXGBE_MAX_FDIR_FILTER_NUM,
+               .key_len = sizeof(struct txgbe_atr_input),
+               .hash_func = rte_hash_crc,
+               .hash_func_init_val = 0,
+               .socket_id = rte_socket_id(),
+       };
+
+       TAILQ_INIT(&fdir_info->fdir_list);
+       snprintf(fdir_hash_name, RTE_HASH_NAMESIZE,
+                "fdir_%s", TDEV_NAME(eth_dev));
+       fdir_info->hash_handle = rte_hash_create(&fdir_hash_params);
+       if (!fdir_info->hash_handle) {
+               PMD_INIT_LOG(ERR, "Failed to create fdir hash table!");
+               return -EINVAL;
+       }
+       fdir_info->hash_map = rte_zmalloc("txgbe",
+                                         sizeof(struct txgbe_fdir_filter *) *
+                                         TXGBE_MAX_FDIR_FILTER_NUM,
+                                         0);
+       if (!fdir_info->hash_map) {
+               PMD_INIT_LOG(ERR,
+                            "Failed to allocate memory for fdir hash map!");
+               return -ENOMEM;
+       }
+       fdir_info->mask_added = FALSE;
+
+       return 0;
+}
+
+static int txgbe_l2_tn_filter_init(struct rte_eth_dev *eth_dev)
+{
+       struct txgbe_l2_tn_info *l2_tn_info = TXGBE_DEV_L2_TN(eth_dev);
+       char l2_tn_hash_name[RTE_HASH_NAMESIZE];
+       struct rte_hash_parameters l2_tn_hash_params = {
+               .name = l2_tn_hash_name,
+               .entries = TXGBE_MAX_L2_TN_FILTER_NUM,
+               .key_len = sizeof(struct txgbe_l2_tn_key),
+               .hash_func = rte_hash_crc,
+               .hash_func_init_val = 0,
+               .socket_id = rte_socket_id(),
+       };
+
+       TAILQ_INIT(&l2_tn_info->l2_tn_list);
+       snprintf(l2_tn_hash_name, RTE_HASH_NAMESIZE,
+                "l2_tn_%s", TDEV_NAME(eth_dev));
+       l2_tn_info->hash_handle = rte_hash_create(&l2_tn_hash_params);
+       if (!l2_tn_info->hash_handle) {
+               PMD_INIT_LOG(ERR, "Failed to create L2 TN hash table!");
+               return -EINVAL;
+       }
+       l2_tn_info->hash_map = rte_zmalloc("txgbe",
+                                  sizeof(struct txgbe_l2_tn_filter *) *
+                                  TXGBE_MAX_L2_TN_FILTER_NUM,
+                                  0);
+       if (!l2_tn_info->hash_map) {
+               PMD_INIT_LOG(ERR,
+                       "Failed to allocate memory for L2 TN hash map!");
+               return -ENOMEM;
+       }
+       l2_tn_info->e_tag_en = FALSE;
+       l2_tn_info->e_tag_fwd_en = FALSE;
+       l2_tn_info->e_tag_ether_type = RTE_ETHER_TYPE_ETAG;
+
+       return 0;
+}
+
 static int
 eth_txgbe_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
                struct rte_pci_device *pci_dev)
@@ -299,514 +909,4324 @@ static struct rte_pci_driver rte_txgbe_pmd = {
        .remove = eth_txgbe_pci_remove,
 };
 
+static int
+txgbe_vlan_filter_set(struct rte_eth_dev *dev, uint16_t vlan_id, int on)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_vfta *shadow_vfta = TXGBE_DEV_VFTA(dev);
+       uint32_t vfta;
+       uint32_t vid_idx;
+       uint32_t vid_bit;
+
+       vid_idx = (uint32_t)((vlan_id >> 5) & 0x7F);
+       vid_bit = (uint32_t)(1 << (vlan_id & 0x1F));
+       vfta = rd32(hw, TXGBE_VLANTBL(vid_idx));
+       if (on)
+               vfta |= vid_bit;
+       else
+               vfta &= ~vid_bit;
+       wr32(hw, TXGBE_VLANTBL(vid_idx), vfta);
+
+       /* update local VFTA copy */
+       shadow_vfta->vfta[vid_idx] = vfta;
+
+       return 0;
+}
 
 static void
-txgbe_dev_phy_intr_setup(struct rte_eth_dev *dev)
+txgbe_vlan_strip_queue_set(struct rte_eth_dev *dev, uint16_t queue, int on)
 {
        struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
-       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
-       uint32_t gpie;
+       struct txgbe_rx_queue *rxq;
+       bool restart;
+       uint32_t rxcfg, rxbal, rxbah;
 
-       gpie = rd32(hw, TXGBE_GPIOINTEN);
-       gpie |= TXGBE_GPIOBIT_6;
-       wr32(hw, TXGBE_GPIOINTEN, gpie);
-       intr->mask_misc |= TXGBE_ICRMISC_GPIO;
+       if (on)
+               txgbe_vlan_hw_strip_enable(dev, queue);
+       else
+               txgbe_vlan_hw_strip_disable(dev, queue);
+
+       rxq = dev->data->rx_queues[queue];
+       rxbal = rd32(hw, TXGBE_RXBAL(rxq->reg_idx));
+       rxbah = rd32(hw, TXGBE_RXBAH(rxq->reg_idx));
+       rxcfg = rd32(hw, TXGBE_RXCFG(rxq->reg_idx));
+       if (rxq->offloads & DEV_RX_OFFLOAD_VLAN_STRIP) {
+               restart = (rxcfg & TXGBE_RXCFG_ENA) &&
+                       !(rxcfg & TXGBE_RXCFG_VLAN);
+               rxcfg |= TXGBE_RXCFG_VLAN;
+       } else {
+               restart = (rxcfg & TXGBE_RXCFG_ENA) &&
+                       (rxcfg & TXGBE_RXCFG_VLAN);
+               rxcfg &= ~TXGBE_RXCFG_VLAN;
+       }
+       rxcfg &= ~TXGBE_RXCFG_ENA;
+
+       if (restart) {
+               /* set vlan strip for ring */
+               txgbe_dev_rx_queue_stop(dev, queue);
+               wr32(hw, TXGBE_RXBAL(rxq->reg_idx), rxbal);
+               wr32(hw, TXGBE_RXBAH(rxq->reg_idx), rxbah);
+               wr32(hw, TXGBE_RXCFG(rxq->reg_idx), rxcfg);
+               txgbe_dev_rx_queue_start(dev, queue);
+       }
 }
 
-/*
- * Reset and stop device.
- */
 static int
-txgbe_dev_close(struct rte_eth_dev *dev)
+txgbe_vlan_tpid_set(struct rte_eth_dev *dev,
+                   enum rte_vlan_type vlan_type,
+                   uint16_t tpid)
 {
-       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
-       struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
-       int retries = 0;
-       int ret;
-
-       PMD_INIT_FUNC_TRACE();
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       int ret = 0;
+       uint32_t portctrl, vlan_ext, qinq;
+
+       portctrl = rd32(hw, TXGBE_PORTCTL);
+
+       vlan_ext = (portctrl & TXGBE_PORTCTL_VLANEXT);
+       qinq = vlan_ext && (portctrl & TXGBE_PORTCTL_QINQ);
+       switch (vlan_type) {
+       case ETH_VLAN_TYPE_INNER:
+               if (vlan_ext) {
+                       wr32m(hw, TXGBE_VLANCTL,
+                               TXGBE_VLANCTL_TPID_MASK,
+                               TXGBE_VLANCTL_TPID(tpid));
+                       wr32m(hw, TXGBE_DMATXCTRL,
+                               TXGBE_DMATXCTRL_TPID_MASK,
+                               TXGBE_DMATXCTRL_TPID(tpid));
+               } else {
+                       ret = -ENOTSUP;
+                       PMD_DRV_LOG(ERR, "Inner type is not supported"
+                                   " by single VLAN");
+               }
 
-       /* disable uio intr before callback unregister */
-       rte_intr_disable(intr_handle);
+               if (qinq) {
+                       wr32m(hw, TXGBE_TAGTPID(0),
+                               TXGBE_TAGTPID_LSB_MASK,
+                               TXGBE_TAGTPID_LSB(tpid));
+               }
+               break;
+       case ETH_VLAN_TYPE_OUTER:
+               if (vlan_ext) {
+                       /* Only the high 16-bits is valid */
+                       wr32m(hw, TXGBE_EXTAG,
+                               TXGBE_EXTAG_VLAN_MASK,
+                               TXGBE_EXTAG_VLAN(tpid));
+               } else {
+                       wr32m(hw, TXGBE_VLANCTL,
+                               TXGBE_VLANCTL_TPID_MASK,
+                               TXGBE_VLANCTL_TPID(tpid));
+                       wr32m(hw, TXGBE_DMATXCTRL,
+                               TXGBE_DMATXCTRL_TPID_MASK,
+                               TXGBE_DMATXCTRL_TPID(tpid));
+               }
 
-       do {
-               ret = rte_intr_callback_unregister(intr_handle,
-                               txgbe_dev_interrupt_handler, dev);
-               if (ret >= 0 || ret == -ENOENT) {
-                       break;
-               } else if (ret != -EAGAIN) {
-                       PMD_INIT_LOG(ERR,
-                               "intr callback unregister failed: %d",
-                               ret);
+               if (qinq) {
+                       wr32m(hw, TXGBE_TAGTPID(0),
+                               TXGBE_TAGTPID_MSB_MASK,
+                               TXGBE_TAGTPID_MSB(tpid));
                }
-               rte_delay_ms(100);
-       } while (retries++ < (10 + TXGBE_LINK_UP_TIME));
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "Unsupported VLAN type %d", vlan_type);
+               return -EINVAL;
+       }
 
-       /* cancel the delay handler before remove dev */
-       rte_eal_alarm_cancel(txgbe_dev_interrupt_delayed_handler, dev);
+       return ret;
+}
 
-       rte_free(dev->data->mac_addrs);
-       dev->data->mac_addrs = NULL;
+void
+txgbe_vlan_hw_filter_disable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t vlnctrl;
 
-       rte_free(dev->data->hash_mac_addrs);
-       dev->data->hash_mac_addrs = NULL;
+       PMD_INIT_FUNC_TRACE();
 
-       return 0;
+       /* Filter Table Disable */
+       vlnctrl = rd32(hw, TXGBE_VLANCTL);
+       vlnctrl &= ~TXGBE_VLANCTL_VFE;
+       wr32(hw, TXGBE_VLANCTL, vlnctrl);
 }
 
-static int
-txgbe_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
+void
+txgbe_vlan_hw_filter_enable(struct rte_eth_dev *dev)
 {
-       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
        struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_vfta *shadow_vfta = TXGBE_DEV_VFTA(dev);
+       uint32_t vlnctrl;
+       uint16_t i;
 
-       dev_info->max_rx_queues = (uint16_t)hw->mac.max_rx_queues;
-       dev_info->max_tx_queues = (uint16_t)hw->mac.max_tx_queues;
-       dev_info->min_rx_bufsize = 1024;
-       dev_info->max_rx_pktlen = 15872;
-       dev_info->max_mac_addrs = hw->mac.num_rar_entries;
-       dev_info->max_hash_mac_addrs = TXGBE_VMDQ_NUM_UC_MAC;
-       dev_info->max_vfs = pci_dev->max_vfs;
-       dev_info->max_vmdq_pools = ETH_64_POOLS;
-       dev_info->vmdq_queue_num = dev_info->max_rx_queues;
-       dev_info->rx_queue_offload_capa = txgbe_get_rx_queue_offloads(dev);
-       dev_info->rx_offload_capa = (txgbe_get_rx_port_offloads(dev) |
-                                    dev_info->rx_queue_offload_capa);
-       dev_info->tx_queue_offload_capa = txgbe_get_tx_queue_offloads(dev);
-       dev_info->tx_offload_capa = txgbe_get_tx_port_offloads(dev);
+       PMD_INIT_FUNC_TRACE();
 
-       dev_info->default_rxconf = (struct rte_eth_rxconf) {
-               .rx_thresh = {
-                       .pthresh = TXGBE_DEFAULT_RX_PTHRESH,
-                       .hthresh = TXGBE_DEFAULT_RX_HTHRESH,
-                       .wthresh = TXGBE_DEFAULT_RX_WTHRESH,
-               },
-               .rx_free_thresh = TXGBE_DEFAULT_RX_FREE_THRESH,
-               .rx_drop_en = 0,
-               .offloads = 0,
-       };
+       /* Filter Table Enable */
+       vlnctrl = rd32(hw, TXGBE_VLANCTL);
+       vlnctrl &= ~TXGBE_VLANCTL_CFIENA;
+       vlnctrl |= TXGBE_VLANCTL_VFE;
+       wr32(hw, TXGBE_VLANCTL, vlnctrl);
 
-       dev_info->default_txconf = (struct rte_eth_txconf) {
-               .tx_thresh = {
-                       .pthresh = TXGBE_DEFAULT_TX_PTHRESH,
-                       .hthresh = TXGBE_DEFAULT_TX_HTHRESH,
-                       .wthresh = TXGBE_DEFAULT_TX_WTHRESH,
-               },
-               .tx_free_thresh = TXGBE_DEFAULT_TX_FREE_THRESH,
-               .offloads = 0,
-       };
+       /* write whatever is in local vfta copy */
+       for (i = 0; i < TXGBE_VFTA_SIZE; i++)
+               wr32(hw, TXGBE_VLANTBL(i), shadow_vfta->vfta[i]);
+}
 
-       dev_info->rx_desc_lim = rx_desc_lim;
-       dev_info->tx_desc_lim = tx_desc_lim;
+void
+txgbe_vlan_hw_strip_bitmap_set(struct rte_eth_dev *dev, uint16_t queue, bool on)
+{
+       struct txgbe_hwstrip *hwstrip = TXGBE_DEV_HWSTRIP(dev);
+       struct txgbe_rx_queue *rxq;
 
-       dev_info->hash_key_size = TXGBE_HKEY_MAX_INDEX * sizeof(uint32_t);
-       dev_info->reta_size = ETH_RSS_RETA_SIZE_128;
-       dev_info->flow_type_rss_offloads = TXGBE_RSS_OFFLOAD_ALL;
+       if (queue >= TXGBE_MAX_RX_QUEUE_NUM)
+               return;
 
-       dev_info->speed_capa = ETH_LINK_SPEED_1G | ETH_LINK_SPEED_10G;
-       dev_info->speed_capa |= ETH_LINK_SPEED_100M;
+       if (on)
+               TXGBE_SET_HWSTRIP(hwstrip, queue);
+       else
+               TXGBE_CLEAR_HWSTRIP(hwstrip, queue);
 
-       /* Driver-preferred Rx/Tx parameters */
-       dev_info->default_rxportconf.burst_size = 32;
-       dev_info->default_txportconf.burst_size = 32;
-       dev_info->default_rxportconf.nb_queues = 1;
-       dev_info->default_txportconf.nb_queues = 1;
-       dev_info->default_rxportconf.ring_size = 256;
-       dev_info->default_txportconf.ring_size = 256;
+       if (queue >= dev->data->nb_rx_queues)
+               return;
 
-       return 0;
+       rxq = dev->data->rx_queues[queue];
+
+       if (on) {
+               rxq->vlan_flags = PKT_RX_VLAN | PKT_RX_VLAN_STRIPPED;
+               rxq->offloads |= DEV_RX_OFFLOAD_VLAN_STRIP;
+       } else {
+               rxq->vlan_flags = PKT_RX_VLAN;
+               rxq->offloads &= ~DEV_RX_OFFLOAD_VLAN_STRIP;
+       }
+}
+
+static void
+txgbe_vlan_hw_strip_disable(struct rte_eth_dev *dev, uint16_t queue)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t ctrl;
+
+       PMD_INIT_FUNC_TRACE();
+
+       ctrl = rd32(hw, TXGBE_RXCFG(queue));
+       ctrl &= ~TXGBE_RXCFG_VLAN;
+       wr32(hw, TXGBE_RXCFG(queue), ctrl);
+
+       /* record those setting for HW strip per queue */
+       txgbe_vlan_hw_strip_bitmap_set(dev, queue, 0);
+}
+
+static void
+txgbe_vlan_hw_strip_enable(struct rte_eth_dev *dev, uint16_t queue)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t ctrl;
+
+       PMD_INIT_FUNC_TRACE();
+
+       ctrl = rd32(hw, TXGBE_RXCFG(queue));
+       ctrl |= TXGBE_RXCFG_VLAN;
+       wr32(hw, TXGBE_RXCFG(queue), ctrl);
+
+       /* record those setting for HW strip per queue */
+       txgbe_vlan_hw_strip_bitmap_set(dev, queue, 1);
+}
+
+static void
+txgbe_vlan_hw_extend_disable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t ctrl;
+
+       PMD_INIT_FUNC_TRACE();
+
+       ctrl = rd32(hw, TXGBE_PORTCTL);
+       ctrl &= ~TXGBE_PORTCTL_VLANEXT;
+       ctrl &= ~TXGBE_PORTCTL_QINQ;
+       wr32(hw, TXGBE_PORTCTL, ctrl);
+}
+
+static void
+txgbe_vlan_hw_extend_enable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct rte_eth_rxmode *rxmode = &dev->data->dev_conf.rxmode;
+       struct rte_eth_txmode *txmode = &dev->data->dev_conf.txmode;
+       uint32_t ctrl;
+
+       PMD_INIT_FUNC_TRACE();
+
+       ctrl  = rd32(hw, TXGBE_PORTCTL);
+       ctrl |= TXGBE_PORTCTL_VLANEXT;
+       if (rxmode->offloads & DEV_RX_OFFLOAD_QINQ_STRIP ||
+           txmode->offloads & DEV_TX_OFFLOAD_QINQ_INSERT)
+               ctrl |= TXGBE_PORTCTL_QINQ;
+       wr32(hw, TXGBE_PORTCTL, ctrl);
+}
+
+void
+txgbe_vlan_hw_strip_config(struct rte_eth_dev *dev)
+{
+       struct txgbe_rx_queue *rxq;
+       uint16_t i;
+
+       PMD_INIT_FUNC_TRACE();
+
+       for (i = 0; i < dev->data->nb_rx_queues; i++) {
+               rxq = dev->data->rx_queues[i];
+
+               if (rxq->offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
+                       txgbe_vlan_strip_queue_set(dev, i, 1);
+               else
+                       txgbe_vlan_strip_queue_set(dev, i, 0);
+       }
+}
+
+void
+txgbe_config_vlan_strip_on_all_queues(struct rte_eth_dev *dev, int mask)
+{
+       uint16_t i;
+       struct rte_eth_rxmode *rxmode;
+       struct txgbe_rx_queue *rxq;
+
+       if (mask & ETH_VLAN_STRIP_MASK) {
+               rxmode = &dev->data->dev_conf.rxmode;
+               if (rxmode->offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
+                       for (i = 0; i < dev->data->nb_rx_queues; i++) {
+                               rxq = dev->data->rx_queues[i];
+                               rxq->offloads |= DEV_RX_OFFLOAD_VLAN_STRIP;
+                       }
+               else
+                       for (i = 0; i < dev->data->nb_rx_queues; i++) {
+                               rxq = dev->data->rx_queues[i];
+                               rxq->offloads &= ~DEV_RX_OFFLOAD_VLAN_STRIP;
+                       }
+       }
 }
 
 static int
-txgbe_dev_link_update(struct rte_eth_dev *dev, int wait_to_complete)
+txgbe_vlan_offload_config(struct rte_eth_dev *dev, int mask)
 {
-       RTE_SET_USED(dev);
-       RTE_SET_USED(wait_to_complete);
+       struct rte_eth_rxmode *rxmode;
+       rxmode = &dev->data->dev_conf.rxmode;
+
+       if (mask & ETH_VLAN_STRIP_MASK)
+               txgbe_vlan_hw_strip_config(dev);
+
+       if (mask & ETH_VLAN_FILTER_MASK) {
+               if (rxmode->offloads & DEV_RX_OFFLOAD_VLAN_FILTER)
+                       txgbe_vlan_hw_filter_enable(dev);
+               else
+                       txgbe_vlan_hw_filter_disable(dev);
+       }
+
+       if (mask & ETH_VLAN_EXTEND_MASK) {
+               if (rxmode->offloads & DEV_RX_OFFLOAD_VLAN_EXTEND)
+                       txgbe_vlan_hw_extend_enable(dev);
+               else
+                       txgbe_vlan_hw_extend_disable(dev);
+       }
+
        return 0;
 }
 
-/**
- * It clears the interrupt causes and enables the interrupt.
- * It will be called once only during nic initialized.
- *
- * @param dev
- *  Pointer to struct rte_eth_dev.
- * @param on
- *  Enable or Disable.
- *
- * @return
- *  - On success, zero.
- *  - On failure, a negative value.
- */
 static int
-txgbe_dev_lsc_interrupt_setup(struct rte_eth_dev *dev, uint8_t on)
+txgbe_vlan_offload_set(struct rte_eth_dev *dev, int mask)
 {
-       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       txgbe_config_vlan_strip_on_all_queues(dev, mask);
 
-       txgbe_dev_link_status_print(dev);
-       if (on)
-               intr->mask_misc |= TXGBE_ICRMISC_LSC;
-       else
-               intr->mask_misc &= ~TXGBE_ICRMISC_LSC;
+       txgbe_vlan_offload_config(dev, mask);
 
        return 0;
 }
 
-/**
- * It clears the interrupt causes and enables the interrupt.
- * It will be called once only during nic initialized.
- *
- * @param dev
- *  Pointer to struct rte_eth_dev.
- *
- * @return
- *  - On success, zero.
- *  - On failure, a negative value.
- */
+static void
+txgbe_vmdq_vlan_hw_filter_enable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       /* VLNCTL: enable vlan filtering and allow all vlan tags through */
+       uint32_t vlanctrl = rd32(hw, TXGBE_VLANCTL);
+
+       vlanctrl |= TXGBE_VLANCTL_VFE; /* enable vlan filters */
+       wr32(hw, TXGBE_VLANCTL, vlanctrl);
+}
+
 static int
-txgbe_dev_rxq_interrupt_setup(struct rte_eth_dev *dev)
+txgbe_check_vf_rss_rxq_num(struct rte_eth_dev *dev, uint16_t nb_rx_q)
 {
-       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
 
-       intr->mask[0] |= TXGBE_ICR_MASK;
-       intr->mask[1] |= TXGBE_ICR_MASK;
+       switch (nb_rx_q) {
+       case 1:
+       case 2:
+               RTE_ETH_DEV_SRIOV(dev).active = ETH_64_POOLS;
+               break;
+       case 4:
+               RTE_ETH_DEV_SRIOV(dev).active = ETH_32_POOLS;
+               break;
+       default:
+               return -EINVAL;
+       }
 
+       RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool =
+               TXGBE_MAX_RX_QUEUE_NUM / RTE_ETH_DEV_SRIOV(dev).active;
+       RTE_ETH_DEV_SRIOV(dev).def_pool_q_idx =
+               pci_dev->max_vfs * RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool;
        return 0;
 }
 
-/**
- * It clears the interrupt causes and enables the interrupt.
- * It will be called once only during nic initialized.
- *
- * @param dev
- *  Pointer to struct rte_eth_dev.
- *
- * @return
- *  - On success, zero.
- *  - On failure, a negative value.
- */
 static int
-txgbe_dev_macsec_interrupt_setup(struct rte_eth_dev *dev)
+txgbe_check_mq_mode(struct rte_eth_dev *dev)
 {
-       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       struct rte_eth_conf *dev_conf = &dev->data->dev_conf;
+       uint16_t nb_rx_q = dev->data->nb_rx_queues;
+       uint16_t nb_tx_q = dev->data->nb_tx_queues;
+
+       if (RTE_ETH_DEV_SRIOV(dev).active != 0) {
+               /* check multi-queue mode */
+               switch (dev_conf->rxmode.mq_mode) {
+               case ETH_MQ_RX_VMDQ_DCB:
+                       PMD_INIT_LOG(INFO, "ETH_MQ_RX_VMDQ_DCB mode supported in SRIOV");
+                       break;
+               case ETH_MQ_RX_VMDQ_DCB_RSS:
+                       /* DCB/RSS VMDQ in SRIOV mode, not implement yet */
+                       PMD_INIT_LOG(ERR, "SRIOV active,"
+                                       " unsupported mq_mode rx %d.",
+                                       dev_conf->rxmode.mq_mode);
+                       return -EINVAL;
+               case ETH_MQ_RX_RSS:
+               case ETH_MQ_RX_VMDQ_RSS:
+                       dev->data->dev_conf.rxmode.mq_mode = ETH_MQ_RX_VMDQ_RSS;
+                       if (nb_rx_q <= RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool)
+                               if (txgbe_check_vf_rss_rxq_num(dev, nb_rx_q)) {
+                                       PMD_INIT_LOG(ERR, "SRIOV is active,"
+                                               " invalid queue number"
+                                               " for VMDQ RSS, allowed"
+                                               " value are 1, 2 or 4.");
+                                       return -EINVAL;
+                               }
+                       break;
+               case ETH_MQ_RX_VMDQ_ONLY:
+               case ETH_MQ_RX_NONE:
+                       /* if nothing mq mode configure, use default scheme */
+                       dev->data->dev_conf.rxmode.mq_mode =
+                               ETH_MQ_RX_VMDQ_ONLY;
+                       break;
+               default: /* ETH_MQ_RX_DCB, ETH_MQ_RX_DCB_RSS or ETH_MQ_TX_DCB*/
+                       /* SRIOV only works in VMDq enable mode */
+                       PMD_INIT_LOG(ERR, "SRIOV is active,"
+                                       " wrong mq_mode rx %d.",
+                                       dev_conf->rxmode.mq_mode);
+                       return -EINVAL;
+               }
 
-       intr->mask_misc |= TXGBE_ICRMISC_LNKSEC;
+               switch (dev_conf->txmode.mq_mode) {
+               case ETH_MQ_TX_VMDQ_DCB:
+                       PMD_INIT_LOG(INFO, "ETH_MQ_TX_VMDQ_DCB mode supported in SRIOV");
+                       dev->data->dev_conf.txmode.mq_mode = ETH_MQ_TX_VMDQ_DCB;
+                       break;
+               default: /* ETH_MQ_TX_VMDQ_ONLY or ETH_MQ_TX_NONE */
+                       dev->data->dev_conf.txmode.mq_mode =
+                               ETH_MQ_TX_VMDQ_ONLY;
+                       break;
+               }
+
+               /* check valid queue number */
+               if ((nb_rx_q > RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool) ||
+                   (nb_tx_q > RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool)) {
+                       PMD_INIT_LOG(ERR, "SRIOV is active,"
+                                       " nb_rx_q=%d nb_tx_q=%d queue number"
+                                       " must be less than or equal to %d.",
+                                       nb_rx_q, nb_tx_q,
+                                       RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool);
+                       return -EINVAL;
+               }
+       } else {
+               if (dev_conf->rxmode.mq_mode == ETH_MQ_RX_VMDQ_DCB_RSS) {
+                       PMD_INIT_LOG(ERR, "VMDQ+DCB+RSS mq_mode is"
+                                         " not supported.");
+                       return -EINVAL;
+               }
+               /* check configuration for vmdb+dcb mode */
+               if (dev_conf->rxmode.mq_mode == ETH_MQ_RX_VMDQ_DCB) {
+                       const struct rte_eth_vmdq_dcb_conf *conf;
+
+                       if (nb_rx_q != TXGBE_VMDQ_DCB_NB_QUEUES) {
+                               PMD_INIT_LOG(ERR, "VMDQ+DCB, nb_rx_q != %d.",
+                                               TXGBE_VMDQ_DCB_NB_QUEUES);
+                               return -EINVAL;
+                       }
+                       conf = &dev_conf->rx_adv_conf.vmdq_dcb_conf;
+                       if (!(conf->nb_queue_pools == ETH_16_POOLS ||
+                              conf->nb_queue_pools == ETH_32_POOLS)) {
+                               PMD_INIT_LOG(ERR, "VMDQ+DCB selected,"
+                                               " nb_queue_pools must be %d or %d.",
+                                               ETH_16_POOLS, ETH_32_POOLS);
+                               return -EINVAL;
+                       }
+               }
+               if (dev_conf->txmode.mq_mode == ETH_MQ_TX_VMDQ_DCB) {
+                       const struct rte_eth_vmdq_dcb_tx_conf *conf;
+
+                       if (nb_tx_q != TXGBE_VMDQ_DCB_NB_QUEUES) {
+                               PMD_INIT_LOG(ERR, "VMDQ+DCB, nb_tx_q != %d",
+                                                TXGBE_VMDQ_DCB_NB_QUEUES);
+                               return -EINVAL;
+                       }
+                       conf = &dev_conf->tx_adv_conf.vmdq_dcb_tx_conf;
+                       if (!(conf->nb_queue_pools == ETH_16_POOLS ||
+                              conf->nb_queue_pools == ETH_32_POOLS)) {
+                               PMD_INIT_LOG(ERR, "VMDQ+DCB selected,"
+                                               " nb_queue_pools != %d and"
+                                               " nb_queue_pools != %d.",
+                                               ETH_16_POOLS, ETH_32_POOLS);
+                               return -EINVAL;
+                       }
+               }
+
+               /* For DCB mode check our configuration before we go further */
+               if (dev_conf->rxmode.mq_mode == ETH_MQ_RX_DCB) {
+                       const struct rte_eth_dcb_rx_conf *conf;
+
+                       conf = &dev_conf->rx_adv_conf.dcb_rx_conf;
+                       if (!(conf->nb_tcs == ETH_4_TCS ||
+                              conf->nb_tcs == ETH_8_TCS)) {
+                               PMD_INIT_LOG(ERR, "DCB selected, nb_tcs != %d"
+                                               " and nb_tcs != %d.",
+                                               ETH_4_TCS, ETH_8_TCS);
+                               return -EINVAL;
+                       }
+               }
 
+               if (dev_conf->txmode.mq_mode == ETH_MQ_TX_DCB) {
+                       const struct rte_eth_dcb_tx_conf *conf;
+
+                       conf = &dev_conf->tx_adv_conf.dcb_tx_conf;
+                       if (!(conf->nb_tcs == ETH_4_TCS ||
+                              conf->nb_tcs == ETH_8_TCS)) {
+                               PMD_INIT_LOG(ERR, "DCB selected, nb_tcs != %d"
+                                               " and nb_tcs != %d.",
+                                               ETH_4_TCS, ETH_8_TCS);
+                               return -EINVAL;
+                       }
+               }
+       }
        return 0;
 }
 
-/*
- * It reads ICR and sets flag (TXGBE_ICRMISC_LSC) for the link_update.
- *
- * @param dev
- *  Pointer to struct rte_eth_dev.
- *
- * @return
- *  - On success, zero.
- *  - On failure, a negative value.
- */
 static int
-txgbe_dev_interrupt_get_status(struct rte_eth_dev *dev)
+txgbe_dev_configure(struct rte_eth_dev *dev)
 {
-       uint32_t eicr;
-       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
        struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+       int ret;
 
-       /* clear all cause mask */
-       txgbe_disable_intr(hw);
-
-       /* read-on-clear nic registers here */
-       eicr = ((u32 *)hw->isb_mem)[TXGBE_ISB_MISC];
-       PMD_DRV_LOG(DEBUG, "eicr %x", eicr);
-
-       intr->flags = 0;
+       PMD_INIT_FUNC_TRACE();
 
-       /* set flag for async link update */
-       if (eicr & TXGBE_ICRMISC_LSC)
-               intr->flags |= TXGBE_FLAG_NEED_LINK_UPDATE;
+       if (dev->data->dev_conf.rxmode.mq_mode & ETH_MQ_RX_RSS_FLAG)
+               dev->data->dev_conf.rxmode.offloads |= DEV_RX_OFFLOAD_RSS_HASH;
 
-       if (eicr & TXGBE_ICRMISC_VFMBX)
-               intr->flags |= TXGBE_FLAG_MAILBOX;
+       /* multiple queue mode checking */
+       ret  = txgbe_check_mq_mode(dev);
+       if (ret != 0) {
+               PMD_DRV_LOG(ERR, "txgbe_check_mq_mode fails with %d.",
+                           ret);
+               return ret;
+       }
 
-       if (eicr & TXGBE_ICRMISC_LNKSEC)
-               intr->flags |= TXGBE_FLAG_MACSEC;
+       /* set flag to update link status after init */
+       intr->flags |= TXGBE_FLAG_NEED_LINK_UPDATE;
 
-       if (eicr & TXGBE_ICRMISC_GPIO)
-               intr->flags |= TXGBE_FLAG_PHY_INTERRUPT;
+       /*
+        * Initialize to TRUE. If any of Rx queues doesn't meet the bulk
+        * allocation Rx preconditions we will reset it.
+        */
+       adapter->rx_bulk_alloc_allowed = true;
 
        return 0;
 }
 
-/**
- * It gets and then prints the link status.
- *
- * @param dev
- *  Pointer to struct rte_eth_dev.
- *
- * @return
- *  - On success, zero.
- *  - On failure, a negative value.
- */
 static void
-txgbe_dev_link_status_print(struct rte_eth_dev *dev)
+txgbe_dev_phy_intr_setup(struct rte_eth_dev *dev)
 {
-       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       uint32_t gpie;
+
+       gpie = rd32(hw, TXGBE_GPIOINTEN);
+       gpie |= TXGBE_GPIOBIT_6;
+       wr32(hw, TXGBE_GPIOINTEN, gpie);
+       intr->mask_misc |= TXGBE_ICRMISC_GPIO;
+}
+
+int
+txgbe_set_vf_rate_limit(struct rte_eth_dev *dev, uint16_t vf,
+                       uint16_t tx_rate, uint64_t q_msk)
+{
+       struct txgbe_hw *hw;
+       struct txgbe_vf_info *vfinfo;
        struct rte_eth_link link;
+       uint8_t  nb_q_per_pool;
+       uint32_t queue_stride;
+       uint32_t queue_idx, idx = 0, vf_idx;
+       uint32_t queue_end;
+       uint16_t total_rate = 0;
+       struct rte_pci_device *pci_dev;
+       int ret;
 
-       rte_eth_linkstatus_get(dev, &link);
+       pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+       ret = rte_eth_link_get_nowait(dev->data->port_id, &link);
+       if (ret < 0)
+               return ret;
 
-       if (link.link_status) {
-               PMD_INIT_LOG(INFO, "Port %d: Link Up - speed %u Mbps - %s",
-                                       (int)(dev->data->port_id),
-                                       (unsigned int)link.link_speed,
-                       link.link_duplex == ETH_LINK_FULL_DUPLEX ?
-                                       "full-duplex" : "half-duplex");
+       if (vf >= pci_dev->max_vfs)
+               return -EINVAL;
+
+       if (tx_rate > link.link_speed)
+               return -EINVAL;
+
+       if (q_msk == 0)
+               return 0;
+
+       hw = TXGBE_DEV_HW(dev);
+       vfinfo = *(TXGBE_DEV_VFDATA(dev));
+       nb_q_per_pool = RTE_ETH_DEV_SRIOV(dev).nb_q_per_pool;
+       queue_stride = TXGBE_MAX_RX_QUEUE_NUM / RTE_ETH_DEV_SRIOV(dev).active;
+       queue_idx = vf * queue_stride;
+       queue_end = queue_idx + nb_q_per_pool - 1;
+       if (queue_end >= hw->mac.max_tx_queues)
+               return -EINVAL;
+
+       if (vfinfo) {
+               for (vf_idx = 0; vf_idx < pci_dev->max_vfs; vf_idx++) {
+                       if (vf_idx == vf)
+                               continue;
+                       for (idx = 0; idx < RTE_DIM(vfinfo[vf_idx].tx_rate);
+                               idx++)
+                               total_rate += vfinfo[vf_idx].tx_rate[idx];
+               }
        } else {
-               PMD_INIT_LOG(INFO, " Port %d: Link Down",
-                               (int)(dev->data->port_id));
+               return -EINVAL;
        }
-       PMD_INIT_LOG(DEBUG, "PCI Address: " PCI_PRI_FMT,
-                               pci_dev->addr.domain,
-                               pci_dev->addr.bus,
-                               pci_dev->addr.devid,
-                               pci_dev->addr.function);
+
+       /* Store tx_rate for this vf. */
+       for (idx = 0; idx < nb_q_per_pool; idx++) {
+               if (((uint64_t)0x1 << idx) & q_msk) {
+                       if (vfinfo[vf].tx_rate[idx] != tx_rate)
+                               vfinfo[vf].tx_rate[idx] = tx_rate;
+                       total_rate += tx_rate;
+               }
+       }
+
+       if (total_rate > dev->data->dev_link.link_speed) {
+               /* Reset stored TX rate of the VF if it causes exceed
+                * link speed.
+                */
+               memset(vfinfo[vf].tx_rate, 0, sizeof(vfinfo[vf].tx_rate));
+               return -EINVAL;
+       }
+
+       /* Set ARBTXRATE of each queue/pool for vf X  */
+       for (; queue_idx <= queue_end; queue_idx++) {
+               if (0x1 & q_msk)
+                       txgbe_set_queue_rate_limit(dev, queue_idx, tx_rate);
+               q_msk = q_msk >> 1;
+       }
+
+       return 0;
 }
 
 /*
- * It executes link_update after knowing an interrupt occurred.
- *
- * @param dev
- *  Pointer to struct rte_eth_dev.
- *
- * @return
- *  - On success, zero.
- *  - On failure, a negative value.
+ * Configure device link speed and setup link.
+ * It returns 0 on success.
  */
 static int
-txgbe_dev_interrupt_action(struct rte_eth_dev *dev,
-                          struct rte_intr_handle *intr_handle)
+txgbe_dev_start(struct rte_eth_dev *dev)
 {
-       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
-       int64_t timeout;
        struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_hw_stats *hw_stats = TXGBE_DEV_STATS(dev);
+       struct txgbe_vf_info *vfinfo = *TXGBE_DEV_VFDATA(dev);
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+       struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
+       uint32_t intr_vector = 0;
+       int err;
+       bool link_up = false, negotiate = 0;
+       uint32_t speed = 0;
+       uint32_t allowed_speeds = 0;
+       int mask = 0;
+       int status;
+       uint16_t vf, idx;
+       uint32_t *link_speeds;
+       struct txgbe_tm_conf *tm_conf = TXGBE_DEV_TM_CONF(dev);
 
-       PMD_DRV_LOG(DEBUG, "intr action type %d", intr->flags);
-
-       if (intr->flags & TXGBE_FLAG_MAILBOX)
-               intr->flags &= ~TXGBE_FLAG_MAILBOX;
+       PMD_INIT_FUNC_TRACE();
 
-       if (intr->flags & TXGBE_FLAG_PHY_INTERRUPT) {
-               hw->phy.handle_lasi(hw);
-               intr->flags &= ~TXGBE_FLAG_PHY_INTERRUPT;
+       /* TXGBE devices don't support:
+        *    - half duplex (checked afterwards for valid speeds)
+        *    - fixed speed: TODO implement
+        */
+       if (dev->data->dev_conf.link_speeds & ETH_LINK_SPEED_FIXED) {
+               PMD_INIT_LOG(ERR,
+               "Invalid link_speeds for port %u, fix speed not supported",
+                               dev->data->port_id);
+               return -EINVAL;
        }
 
-       if (intr->flags & TXGBE_FLAG_NEED_LINK_UPDATE) {
-               struct rte_eth_link link;
+       /* Stop the link setup handler before resetting the HW. */
+       rte_eal_alarm_cancel(txgbe_dev_setup_link_alarm_handler, dev);
 
-               /*get the link status before link update, for predicting later*/
-               rte_eth_linkstatus_get(dev, &link);
+       /* disable uio/vfio intr/eventfd mapping */
+       rte_intr_disable(intr_handle);
 
-               txgbe_dev_link_update(dev, 0);
+       /* stop adapter */
+       hw->adapter_stopped = 0;
+       txgbe_stop_hw(hw);
 
-               /* likely to up */
-               if (!link.link_status)
-                       /* handle it 1 sec later, wait it being stable */
-                       timeout = TXGBE_LINK_UP_CHECK_TIMEOUT;
-               /* likely to down */
-               else
-                       /* handle it 4 sec later, wait it being stable */
-                       timeout = TXGBE_LINK_DOWN_CHECK_TIMEOUT;
+       /* reinitialize adapter
+        * this calls reset and start
+        */
+       hw->nb_rx_queues = dev->data->nb_rx_queues;
+       hw->nb_tx_queues = dev->data->nb_tx_queues;
+       status = txgbe_pf_reset_hw(hw);
+       if (status != 0)
+               return -1;
+       hw->mac.start_hw(hw);
+       hw->mac.get_link_status = true;
+
+       /* configure PF module if SRIOV enabled */
+       txgbe_pf_host_configure(dev);
+
+       txgbe_dev_phy_intr_setup(dev);
+
+       /* check and configure queue intr-vector mapping */
+       if ((rte_intr_cap_multiple(intr_handle) ||
+            !RTE_ETH_DEV_SRIOV(dev).active) &&
+           dev->data->dev_conf.intr_conf.rxq != 0) {
+               intr_vector = dev->data->nb_rx_queues;
+               if (rte_intr_efd_enable(intr_handle, intr_vector))
+                       return -1;
+       }
 
-               txgbe_dev_link_status_print(dev);
-               if (rte_eal_alarm_set(timeout * 1000,
-                                     txgbe_dev_interrupt_delayed_handler,
-                                     (void *)dev) < 0) {
-                       PMD_DRV_LOG(ERR, "Error setting alarm");
-               } else {
-                       /* remember original mask */
-                       intr->mask_misc_orig = intr->mask_misc;
-                       /* only disable lsc interrupt */
-                       intr->mask_misc &= ~TXGBE_ICRMISC_LSC;
+       if (rte_intr_dp_is_en(intr_handle) && !intr_handle->intr_vec) {
+               intr_handle->intr_vec =
+                       rte_zmalloc("intr_vec",
+                                   dev->data->nb_rx_queues * sizeof(int), 0);
+               if (intr_handle->intr_vec == NULL) {
+                       PMD_INIT_LOG(ERR, "Failed to allocate %d rx_queues"
+                                    " intr_vec", dev->data->nb_rx_queues);
+                       return -ENOMEM;
                }
        }
 
-       PMD_DRV_LOG(DEBUG, "enable intr immediately");
-       txgbe_enable_intr(dev);
+       /* confiugre msix for sleep until rx interrupt */
+       txgbe_configure_msix(dev);
+
+       /* initialize transmission unit */
+       txgbe_dev_tx_init(dev);
+
+       /* This can fail when allocating mbufs for descriptor rings */
+       err = txgbe_dev_rx_init(dev);
+       if (err) {
+               PMD_INIT_LOG(ERR, "Unable to initialize RX hardware");
+               goto error;
+       }
+
+       mask = ETH_VLAN_STRIP_MASK | ETH_VLAN_FILTER_MASK |
+               ETH_VLAN_EXTEND_MASK;
+       err = txgbe_vlan_offload_config(dev, mask);
+       if (err) {
+               PMD_INIT_LOG(ERR, "Unable to set VLAN offload");
+               goto error;
+       }
+
+       if (dev->data->dev_conf.rxmode.mq_mode == ETH_MQ_RX_VMDQ_ONLY) {
+               /* Enable vlan filtering for VMDq */
+               txgbe_vmdq_vlan_hw_filter_enable(dev);
+       }
+
+       /* Configure DCB hw */
+       txgbe_configure_pb(dev);
+       txgbe_configure_port(dev);
+       txgbe_configure_dcb(dev);
+
+       if (dev->data->dev_conf.fdir_conf.mode != RTE_FDIR_MODE_NONE) {
+               err = txgbe_fdir_configure(dev);
+               if (err)
+                       goto error;
+       }
+
+       /* Restore vf rate limit */
+       if (vfinfo != NULL) {
+               for (vf = 0; vf < pci_dev->max_vfs; vf++)
+                       for (idx = 0; idx < TXGBE_MAX_QUEUE_NUM_PER_VF; idx++)
+                               if (vfinfo[vf].tx_rate[idx] != 0)
+                                       txgbe_set_vf_rate_limit(dev, vf,
+                                               vfinfo[vf].tx_rate[idx],
+                                               1 << idx);
+       }
+
+       err = txgbe_dev_rxtx_start(dev);
+       if (err < 0) {
+               PMD_INIT_LOG(ERR, "Unable to start rxtx queues");
+               goto error;
+       }
+
+       /* Skip link setup if loopback mode is enabled. */
+       if (hw->mac.type == txgbe_mac_raptor &&
+           dev->data->dev_conf.lpbk_mode)
+               goto skip_link_setup;
+
+       if (txgbe_is_sfp(hw) && hw->phy.multispeed_fiber) {
+               err = hw->mac.setup_sfp(hw);
+               if (err)
+                       goto error;
+       }
+
+       if (hw->phy.media_type == txgbe_media_type_copper) {
+               /* Turn on the copper */
+               hw->phy.set_phy_power(hw, true);
+       } else {
+               /* Turn on the laser */
+               hw->mac.enable_tx_laser(hw);
+       }
+
+       err = hw->mac.check_link(hw, &speed, &link_up, 0);
+       if (err)
+               goto error;
+       dev->data->dev_link.link_status = link_up;
+
+       err = hw->mac.get_link_capabilities(hw, &speed, &negotiate);
+       if (err)
+               goto error;
+
+       allowed_speeds = ETH_LINK_SPEED_100M | ETH_LINK_SPEED_1G |
+                       ETH_LINK_SPEED_10G;
+
+       link_speeds = &dev->data->dev_conf.link_speeds;
+       if (*link_speeds & ~allowed_speeds) {
+               PMD_INIT_LOG(ERR, "Invalid link setting");
+               goto error;
+       }
+
+       speed = 0x0;
+       if (*link_speeds == ETH_LINK_SPEED_AUTONEG) {
+               speed = (TXGBE_LINK_SPEED_100M_FULL |
+                        TXGBE_LINK_SPEED_1GB_FULL |
+                        TXGBE_LINK_SPEED_10GB_FULL);
+       } else {
+               if (*link_speeds & ETH_LINK_SPEED_10G)
+                       speed |= TXGBE_LINK_SPEED_10GB_FULL;
+               if (*link_speeds & ETH_LINK_SPEED_5G)
+                       speed |= TXGBE_LINK_SPEED_5GB_FULL;
+               if (*link_speeds & ETH_LINK_SPEED_2_5G)
+                       speed |= TXGBE_LINK_SPEED_2_5GB_FULL;
+               if (*link_speeds & ETH_LINK_SPEED_1G)
+                       speed |= TXGBE_LINK_SPEED_1GB_FULL;
+               if (*link_speeds & ETH_LINK_SPEED_100M)
+                       speed |= TXGBE_LINK_SPEED_100M_FULL;
+       }
+
+       err = hw->mac.setup_link(hw, speed, link_up);
+       if (err)
+               goto error;
+
+skip_link_setup:
+
+       if (rte_intr_allow_others(intr_handle)) {
+               /* check if lsc interrupt is enabled */
+               if (dev->data->dev_conf.intr_conf.lsc != 0)
+                       txgbe_dev_lsc_interrupt_setup(dev, TRUE);
+               else
+                       txgbe_dev_lsc_interrupt_setup(dev, FALSE);
+               txgbe_dev_macsec_interrupt_setup(dev);
+               txgbe_set_ivar_map(hw, -1, 1, TXGBE_MISC_VEC_ID);
+       } else {
+               rte_intr_callback_unregister(intr_handle,
+                                            txgbe_dev_interrupt_handler, dev);
+               if (dev->data->dev_conf.intr_conf.lsc != 0)
+                       PMD_INIT_LOG(INFO, "lsc won't enable because of"
+                                    " no intr multiplex");
+       }
+
+       /* check if rxq interrupt is enabled */
+       if (dev->data->dev_conf.intr_conf.rxq != 0 &&
+           rte_intr_dp_is_en(intr_handle))
+               txgbe_dev_rxq_interrupt_setup(dev);
+
+       /* enable uio/vfio intr/eventfd mapping */
        rte_intr_enable(intr_handle);
 
+       /* resume enabled intr since hw reset */
+       txgbe_enable_intr(dev);
+       txgbe_l2_tunnel_conf(dev);
+       txgbe_filter_restore(dev);
+
+       if (tm_conf->root && !tm_conf->committed)
+               PMD_DRV_LOG(WARNING,
+                           "please call hierarchy_commit() "
+                           "before starting the port");
+
+       /*
+        * Update link status right before return, because it may
+        * start link configuration process in a separate thread.
+        */
+       txgbe_dev_link_update(dev, 0);
+
+       wr32m(hw, TXGBE_LEDCTL, 0xFFFFFFFF, TXGBE_LEDCTL_ORD_MASK);
+
+       txgbe_read_stats_registers(hw, hw_stats);
+       hw->offset_loaded = 1;
+
        return 0;
+
+error:
+       PMD_INIT_LOG(ERR, "failure in dev start: %d", err);
+       txgbe_dev_clear_queues(dev);
+       return -EIO;
 }
 
-/**
- * Interrupt handler which shall be registered for alarm callback for delayed
- * handling specific interrupt to wait for the stable nic state. As the
- * NIC interrupt state is not stable for txgbe after link is just down,
- * it needs to wait 4 seconds to get the stable status.
- *
- * @param handle
- *  Pointer to interrupt handle.
- * @param param
- *  The address of parameter (struct rte_eth_dev *) registered before.
- *
- * @return
- *  void
+/*
+ * Stop device: disable rx and tx functions to allow for reconfiguring.
  */
-static void
-txgbe_dev_interrupt_delayed_handler(void *param)
+static int
+txgbe_dev_stop(struct rte_eth_dev *dev)
 {
-       struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+       struct rte_eth_link link;
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_vf_info *vfinfo = *TXGBE_DEV_VFDATA(dev);
        struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
        struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
-       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
-       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
-       uint32_t eicr;
+       int vf;
+       struct txgbe_tm_conf *tm_conf = TXGBE_DEV_TM_CONF(dev);
+
+       if (hw->adapter_stopped)
+               return 0;
 
+       PMD_INIT_FUNC_TRACE();
+
+       rte_eal_alarm_cancel(txgbe_dev_setup_link_alarm_handler, dev);
+
+       /* disable interrupts */
        txgbe_disable_intr(hw);
 
-       eicr = ((u32 *)hw->isb_mem)[TXGBE_ISB_MISC];
+       /* reset the NIC */
+       txgbe_pf_reset_hw(hw);
+       hw->adapter_stopped = 0;
 
-       if (intr->flags & TXGBE_FLAG_PHY_INTERRUPT) {
-               hw->phy.handle_lasi(hw);
-               intr->flags &= ~TXGBE_FLAG_PHY_INTERRUPT;
-       }
+       /* stop adapter */
+       txgbe_stop_hw(hw);
 
-       if (intr->flags & TXGBE_FLAG_NEED_LINK_UPDATE) {
-               txgbe_dev_link_update(dev, 0);
-               intr->flags &= ~TXGBE_FLAG_NEED_LINK_UPDATE;
-               txgbe_dev_link_status_print(dev);
-               rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC,
-                                             NULL);
+       for (vf = 0; vfinfo != NULL && vf < pci_dev->max_vfs; vf++)
+               vfinfo[vf].clear_to_send = false;
+
+       if (hw->phy.media_type == txgbe_media_type_copper) {
+               /* Turn off the copper */
+               hw->phy.set_phy_power(hw, false);
+       } else {
+               /* Turn off the laser */
+               hw->mac.disable_tx_laser(hw);
        }
 
-       if (intr->flags & TXGBE_FLAG_MACSEC) {
-               rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_MACSEC,
-                                             NULL);
-               intr->flags &= ~TXGBE_FLAG_MACSEC;
+       txgbe_dev_clear_queues(dev);
+
+       /* Clear stored conf */
+       dev->data->scattered_rx = 0;
+       dev->data->lro = 0;
+
+       /* Clear recorded link status */
+       memset(&link, 0, sizeof(link));
+       rte_eth_linkstatus_set(dev, &link);
+
+       if (!rte_intr_allow_others(intr_handle))
+               /* resume to the default handler */
+               rte_intr_callback_register(intr_handle,
+                                          txgbe_dev_interrupt_handler,
+                                          (void *)dev);
+
+       /* Clean datapath event and queue/vec mapping */
+       rte_intr_efd_disable(intr_handle);
+       if (intr_handle->intr_vec != NULL) {
+               rte_free(intr_handle->intr_vec);
+               intr_handle->intr_vec = NULL;
        }
 
-       /* restore original mask */
-       intr->mask_misc = intr->mask_misc_orig;
-       intr->mask_misc_orig = 0;
+       /* reset hierarchy commit */
+       tm_conf->committed = false;
 
-       PMD_DRV_LOG(DEBUG, "enable intr in delayed handler S[%08x]", eicr);
-       txgbe_enable_intr(dev);
-       rte_intr_enable(intr_handle);
+       adapter->rss_reta_updated = 0;
+       wr32m(hw, TXGBE_LEDCTL, 0xFFFFFFFF, TXGBE_LEDCTL_SEL_MASK);
+
+       hw->adapter_stopped = true;
+       dev->data->dev_started = 0;
+
+       return 0;
 }
 
-/**
- * Interrupt handler triggered by NIC  for handling
- * specific interrupt.
- *
- * @param handle
- *  Pointer to interrupt handle.
- * @param param
- *  The address of parameter (struct rte_eth_dev *) registered before.
- *
- * @return
- *  void
+/*
+ * Set device link up: enable tx.
  */
-static void
-txgbe_dev_interrupt_handler(void *param)
+static int
+txgbe_dev_set_link_up(struct rte_eth_dev *dev)
 {
-       struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
 
-       txgbe_dev_interrupt_get_status(dev);
-       txgbe_dev_interrupt_action(dev, dev->intr_handle);
+       if (hw->phy.media_type == txgbe_media_type_copper) {
+               /* Turn on the copper */
+               hw->phy.set_phy_power(hw, true);
+       } else {
+               /* Turn on the laser */
+               hw->mac.enable_tx_laser(hw);
+               txgbe_dev_link_update(dev, 0);
+       }
+
+       return 0;
 }
 
-/**
- * set the IVAR registers, mapping interrupt causes to vectors
- * @param hw
- *  pointer to txgbe_hw struct
- * @direction
- *  0 for Rx, 1 for Tx, -1 for other causes
- * @queue
- *  queue to map the corresponding interrupt to
- * @msix_vector
- *  the vector to map to the corresponding queue
+/*
+ * Set device link down: disable tx.
  */
-void
-txgbe_set_ivar_map(struct txgbe_hw *hw, int8_t direction,
-                  uint8_t queue, uint8_t msix_vector)
+static int
+txgbe_dev_set_link_down(struct rte_eth_dev *dev)
 {
-       uint32_t tmp, idx;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
 
-       if (direction == -1) {
-               /* other causes */
-               msix_vector |= TXGBE_IVARMISC_VLD;
-               idx = 0;
-               tmp = rd32(hw, TXGBE_IVARMISC);
-               tmp &= ~(0xFF << idx);
-               tmp |= (msix_vector << idx);
-               wr32(hw, TXGBE_IVARMISC, tmp);
+       if (hw->phy.media_type == txgbe_media_type_copper) {
+               /* Turn off the copper */
+               hw->phy.set_phy_power(hw, false);
        } else {
-               /* rx or tx causes */
-               /* Workround for ICR lost */
-               idx = ((16 * (queue & 1)) + (8 * direction));
-               tmp = rd32(hw, TXGBE_IVAR(queue >> 1));
-               tmp &= ~(0xFF << idx);
-               tmp |= (msix_vector << idx);
-               wr32(hw, TXGBE_IVAR(queue >> 1), tmp);
+               /* Turn off the laser */
+               hw->mac.disable_tx_laser(hw);
+               txgbe_dev_link_update(dev, 0);
        }
+
+       return 0;
 }
 
-/**
- * Sets up the hardware to properly generate MSI-X interrupts
- * @hw
- *  board private structure
+/*
+ * Reset and stop device.
  */
-static void
-txgbe_configure_msix(struct rte_eth_dev *dev)
+static int
+txgbe_dev_close(struct rte_eth_dev *dev)
 {
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
        struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
        struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
-       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
-       uint32_t queue_id, base = TXGBE_MISC_VEC_ID;
-       uint32_t vec = TXGBE_MISC_VEC_ID;
-       uint32_t gpie;
+       int retries = 0;
+       int ret;
 
-       /* won't configure msix register if no mapping is done
-        * between intr vector and event fd
-        * but if misx has been enabled already, need to configure
-        * auto clean, auto mask and throttling.
-        */
-       gpie = rd32(hw, TXGBE_GPIE);
-       if (!rte_intr_dp_is_en(intr_handle) &&
-           !(gpie & TXGBE_GPIE_MSIX))
-               return;
+       PMD_INIT_FUNC_TRACE();
 
-       if (rte_intr_allow_others(intr_handle)) {
-               base = TXGBE_RX_VEC_START;
-               vec = base;
-       }
+       txgbe_pf_reset_hw(hw);
 
-       /* setup GPIE for MSI-x mode */
-       gpie = rd32(hw, TXGBE_GPIE);
-       gpie |= TXGBE_GPIE_MSIX;
-       wr32(hw, TXGBE_GPIE, gpie);
+       ret = txgbe_dev_stop(dev);
 
-       /* Populate the IVAR table and set the ITR values to the
-        * corresponding register.
-        */
-       if (rte_intr_dp_is_en(intr_handle)) {
-               for (queue_id = 0; queue_id < dev->data->nb_rx_queues;
-                       queue_id++) {
-                       /* by default, 1:1 mapping */
-                       txgbe_set_ivar_map(hw, 0, queue_id, vec);
-                       intr_handle->intr_vec[queue_id] = vec;
-                       if (vec < base + intr_handle->nb_efd - 1)
-                               vec++;
-               }
+       txgbe_dev_free_queues(dev);
 
-               txgbe_set_ivar_map(hw, -1, 1, TXGBE_MISC_VEC_ID);
-       }
-       wr32(hw, TXGBE_ITR(TXGBE_MISC_VEC_ID),
-                       TXGBE_ITR_IVAL_10G(TXGBE_QUEUE_ITR_INTERVAL_DEFAULT)
-                       | TXGBE_ITR_WRDSA);
+       /* reprogram the RAR[0] in case user changed it. */
+       txgbe_set_rar(hw, 0, hw->mac.addr, 0, true);
+
+       /* Unlock any pending hardware semaphore */
+       txgbe_swfw_lock_reset(hw);
+
+       /* disable uio intr before callback unregister */
+       rte_intr_disable(intr_handle);
+
+       do {
+               ret = rte_intr_callback_unregister(intr_handle,
+                               txgbe_dev_interrupt_handler, dev);
+               if (ret >= 0 || ret == -ENOENT) {
+                       break;
+               } else if (ret != -EAGAIN) {
+                       PMD_INIT_LOG(ERR,
+                               "intr callback unregister failed: %d",
+                               ret);
+               }
+               rte_delay_ms(100);
+       } while (retries++ < (10 + TXGBE_LINK_UP_TIME));
+
+       /* cancel the delay handler before remove dev */
+       rte_eal_alarm_cancel(txgbe_dev_interrupt_delayed_handler, dev);
+
+       /* uninitialize PF if max_vfs not zero */
+       txgbe_pf_host_uninit(dev);
+
+       rte_free(dev->data->mac_addrs);
+       dev->data->mac_addrs = NULL;
+
+       rte_free(dev->data->hash_mac_addrs);
+       dev->data->hash_mac_addrs = NULL;
+
+       /* remove all the fdir filters & hash */
+       txgbe_fdir_filter_uninit(dev);
+
+       /* remove all the L2 tunnel filters & hash */
+       txgbe_l2_tn_filter_uninit(dev);
+
+       /* Remove all ntuple filters of the device */
+       txgbe_ntuple_filter_uninit(dev);
+
+       /* clear all the filters list */
+       txgbe_filterlist_flush();
+
+       /* Remove all Traffic Manager configuration */
+       txgbe_tm_conf_uninit(dev);
+
+#ifdef RTE_LIB_SECURITY
+       rte_free(dev->security_ctx);
+#endif
+
+       return ret;
+}
+
+/*
+ * Reset PF device.
+ */
+static int
+txgbe_dev_reset(struct rte_eth_dev *dev)
+{
+       int ret;
+
+       /* When a DPDK PMD PF begin to reset PF port, it should notify all
+        * its VF to make them align with it. The detailed notification
+        * mechanism is PMD specific. As to txgbe PF, it is rather complex.
+        * To avoid unexpected behavior in VF, currently reset of PF with
+        * SR-IOV activation is not supported. It might be supported later.
+        */
+       if (dev->data->sriov.active)
+               return -ENOTSUP;
+
+       ret = eth_txgbe_dev_uninit(dev);
+       if (ret)
+               return ret;
+
+       ret = eth_txgbe_dev_init(dev, NULL);
+
+       return ret;
+}
+
+#define UPDATE_QP_COUNTER_32bit(reg, last_counter, counter)     \
+       {                                                       \
+               uint32_t current_counter = rd32(hw, reg);       \
+               if (current_counter < last_counter)             \
+                       current_counter += 0x100000000LL;       \
+               if (!hw->offset_loaded)                         \
+                       last_counter = current_counter;         \
+               counter = current_counter - last_counter;       \
+               counter &= 0xFFFFFFFFLL;                        \
+       }
+
+#define UPDATE_QP_COUNTER_36bit(reg_lsb, reg_msb, last_counter, counter) \
+       {                                                                \
+               uint64_t current_counter_lsb = rd32(hw, reg_lsb);        \
+               uint64_t current_counter_msb = rd32(hw, reg_msb);        \
+               uint64_t current_counter = (current_counter_msb << 32) | \
+                       current_counter_lsb;                             \
+               if (current_counter < last_counter)                      \
+                       current_counter += 0x1000000000LL;               \
+               if (!hw->offset_loaded)                                  \
+                       last_counter = current_counter;                  \
+               counter = current_counter - last_counter;                \
+               counter &= 0xFFFFFFFFFLL;                                \
+       }
+
+void
+txgbe_read_stats_registers(struct txgbe_hw *hw,
+                          struct txgbe_hw_stats *hw_stats)
+{
+       unsigned int i;
+
+       /* QP Stats */
+       for (i = 0; i < hw->nb_rx_queues; i++) {
+               UPDATE_QP_COUNTER_32bit(TXGBE_QPRXPKT(i),
+                       hw->qp_last[i].rx_qp_packets,
+                       hw_stats->qp[i].rx_qp_packets);
+               UPDATE_QP_COUNTER_36bit(TXGBE_QPRXOCTL(i), TXGBE_QPRXOCTH(i),
+                       hw->qp_last[i].rx_qp_bytes,
+                       hw_stats->qp[i].rx_qp_bytes);
+               UPDATE_QP_COUNTER_32bit(TXGBE_QPRXMPKT(i),
+                       hw->qp_last[i].rx_qp_mc_packets,
+                       hw_stats->qp[i].rx_qp_mc_packets);
+       }
+
+       for (i = 0; i < hw->nb_tx_queues; i++) {
+               UPDATE_QP_COUNTER_32bit(TXGBE_QPTXPKT(i),
+                       hw->qp_last[i].tx_qp_packets,
+                       hw_stats->qp[i].tx_qp_packets);
+               UPDATE_QP_COUNTER_36bit(TXGBE_QPTXOCTL(i), TXGBE_QPTXOCTH(i),
+                       hw->qp_last[i].tx_qp_bytes,
+                       hw_stats->qp[i].tx_qp_bytes);
+       }
+       /* PB Stats */
+       for (i = 0; i < TXGBE_MAX_UP; i++) {
+               hw_stats->up[i].rx_up_xon_packets +=
+                               rd32(hw, TXGBE_PBRXUPXON(i));
+               hw_stats->up[i].rx_up_xoff_packets +=
+                               rd32(hw, TXGBE_PBRXUPXOFF(i));
+               hw_stats->up[i].tx_up_xon_packets +=
+                               rd32(hw, TXGBE_PBTXUPXON(i));
+               hw_stats->up[i].tx_up_xoff_packets +=
+                               rd32(hw, TXGBE_PBTXUPXOFF(i));
+               hw_stats->up[i].tx_up_xon2off_packets +=
+                               rd32(hw, TXGBE_PBTXUPOFF(i));
+               hw_stats->up[i].rx_up_dropped +=
+                               rd32(hw, TXGBE_PBRXMISS(i));
+       }
+       hw_stats->rx_xon_packets += rd32(hw, TXGBE_PBRXLNKXON);
+       hw_stats->rx_xoff_packets += rd32(hw, TXGBE_PBRXLNKXOFF);
+       hw_stats->tx_xon_packets += rd32(hw, TXGBE_PBTXLNKXON);
+       hw_stats->tx_xoff_packets += rd32(hw, TXGBE_PBTXLNKXOFF);
+
+       /* DMA Stats */
+       hw_stats->rx_packets += rd32(hw, TXGBE_DMARXPKT);
+       hw_stats->tx_packets += rd32(hw, TXGBE_DMATXPKT);
+
+       hw_stats->rx_bytes += rd64(hw, TXGBE_DMARXOCTL);
+       hw_stats->tx_bytes += rd64(hw, TXGBE_DMATXOCTL);
+       hw_stats->rx_drop_packets += rd32(hw, TXGBE_PBRXDROP);
+
+       /* MAC Stats */
+       hw_stats->rx_crc_errors += rd64(hw, TXGBE_MACRXERRCRCL);
+       hw_stats->rx_multicast_packets += rd64(hw, TXGBE_MACRXMPKTL);
+       hw_stats->tx_multicast_packets += rd64(hw, TXGBE_MACTXMPKTL);
+
+       hw_stats->rx_total_packets += rd64(hw, TXGBE_MACRXPKTL);
+       hw_stats->tx_total_packets += rd64(hw, TXGBE_MACTXPKTL);
+       hw_stats->rx_total_bytes += rd64(hw, TXGBE_MACRXGBOCTL);
+
+       hw_stats->rx_broadcast_packets += rd64(hw, TXGBE_MACRXOCTL);
+       hw_stats->tx_broadcast_packets += rd32(hw, TXGBE_MACTXOCTL);
+
+       hw_stats->rx_size_64_packets += rd64(hw, TXGBE_MACRX1TO64L);
+       hw_stats->rx_size_65_to_127_packets += rd64(hw, TXGBE_MACRX65TO127L);
+       hw_stats->rx_size_128_to_255_packets += rd64(hw, TXGBE_MACRX128TO255L);
+       hw_stats->rx_size_256_to_511_packets += rd64(hw, TXGBE_MACRX256TO511L);
+       hw_stats->rx_size_512_to_1023_packets +=
+                       rd64(hw, TXGBE_MACRX512TO1023L);
+       hw_stats->rx_size_1024_to_max_packets +=
+                       rd64(hw, TXGBE_MACRX1024TOMAXL);
+       hw_stats->tx_size_64_packets += rd64(hw, TXGBE_MACTX1TO64L);
+       hw_stats->tx_size_65_to_127_packets += rd64(hw, TXGBE_MACTX65TO127L);
+       hw_stats->tx_size_128_to_255_packets += rd64(hw, TXGBE_MACTX128TO255L);
+       hw_stats->tx_size_256_to_511_packets += rd64(hw, TXGBE_MACTX256TO511L);
+       hw_stats->tx_size_512_to_1023_packets +=
+                       rd64(hw, TXGBE_MACTX512TO1023L);
+       hw_stats->tx_size_1024_to_max_packets +=
+                       rd64(hw, TXGBE_MACTX1024TOMAXL);
+
+       hw_stats->rx_undersize_errors += rd64(hw, TXGBE_MACRXERRLENL);
+       hw_stats->rx_oversize_errors += rd32(hw, TXGBE_MACRXOVERSIZE);
+       hw_stats->rx_jabber_errors += rd32(hw, TXGBE_MACRXJABBER);
+
+       /* MNG Stats */
+       hw_stats->mng_bmc2host_packets = rd32(hw, TXGBE_MNGBMC2OS);
+       hw_stats->mng_host2bmc_packets = rd32(hw, TXGBE_MNGOS2BMC);
+       hw_stats->rx_management_packets = rd32(hw, TXGBE_DMARXMNG);
+       hw_stats->tx_management_packets = rd32(hw, TXGBE_DMATXMNG);
+
+       /* FCoE Stats */
+       hw_stats->rx_fcoe_crc_errors += rd32(hw, TXGBE_FCOECRC);
+       hw_stats->rx_fcoe_mbuf_allocation_errors += rd32(hw, TXGBE_FCOELAST);
+       hw_stats->rx_fcoe_dropped += rd32(hw, TXGBE_FCOERPDC);
+       hw_stats->rx_fcoe_packets += rd32(hw, TXGBE_FCOEPRC);
+       hw_stats->tx_fcoe_packets += rd32(hw, TXGBE_FCOEPTC);
+       hw_stats->rx_fcoe_bytes += rd32(hw, TXGBE_FCOEDWRC);
+       hw_stats->tx_fcoe_bytes += rd32(hw, TXGBE_FCOEDWTC);
+
+       /* Flow Director Stats */
+       hw_stats->flow_director_matched_filters += rd32(hw, TXGBE_FDIRMATCH);
+       hw_stats->flow_director_missed_filters += rd32(hw, TXGBE_FDIRMISS);
+       hw_stats->flow_director_added_filters +=
+               TXGBE_FDIRUSED_ADD(rd32(hw, TXGBE_FDIRUSED));
+       hw_stats->flow_director_removed_filters +=
+               TXGBE_FDIRUSED_REM(rd32(hw, TXGBE_FDIRUSED));
+       hw_stats->flow_director_filter_add_errors +=
+               TXGBE_FDIRFAIL_ADD(rd32(hw, TXGBE_FDIRFAIL));
+       hw_stats->flow_director_filter_remove_errors +=
+               TXGBE_FDIRFAIL_REM(rd32(hw, TXGBE_FDIRFAIL));
+
+       /* MACsec Stats */
+       hw_stats->tx_macsec_pkts_untagged += rd32(hw, TXGBE_LSECTX_UTPKT);
+       hw_stats->tx_macsec_pkts_encrypted +=
+                       rd32(hw, TXGBE_LSECTX_ENCPKT);
+       hw_stats->tx_macsec_pkts_protected +=
+                       rd32(hw, TXGBE_LSECTX_PROTPKT);
+       hw_stats->tx_macsec_octets_encrypted +=
+                       rd32(hw, TXGBE_LSECTX_ENCOCT);
+       hw_stats->tx_macsec_octets_protected +=
+                       rd32(hw, TXGBE_LSECTX_PROTOCT);
+       hw_stats->rx_macsec_pkts_untagged += rd32(hw, TXGBE_LSECRX_UTPKT);
+       hw_stats->rx_macsec_pkts_badtag += rd32(hw, TXGBE_LSECRX_BTPKT);
+       hw_stats->rx_macsec_pkts_nosci += rd32(hw, TXGBE_LSECRX_NOSCIPKT);
+       hw_stats->rx_macsec_pkts_unknownsci += rd32(hw, TXGBE_LSECRX_UNSCIPKT);
+       hw_stats->rx_macsec_octets_decrypted += rd32(hw, TXGBE_LSECRX_DECOCT);
+       hw_stats->rx_macsec_octets_validated += rd32(hw, TXGBE_LSECRX_VLDOCT);
+       hw_stats->rx_macsec_sc_pkts_unchecked +=
+                       rd32(hw, TXGBE_LSECRX_UNCHKPKT);
+       hw_stats->rx_macsec_sc_pkts_delayed += rd32(hw, TXGBE_LSECRX_DLYPKT);
+       hw_stats->rx_macsec_sc_pkts_late += rd32(hw, TXGBE_LSECRX_LATEPKT);
+       for (i = 0; i < 2; i++) {
+               hw_stats->rx_macsec_sa_pkts_ok +=
+                       rd32(hw, TXGBE_LSECRX_OKPKT(i));
+               hw_stats->rx_macsec_sa_pkts_invalid +=
+                       rd32(hw, TXGBE_LSECRX_INVPKT(i));
+               hw_stats->rx_macsec_sa_pkts_notvalid +=
+                       rd32(hw, TXGBE_LSECRX_BADPKT(i));
+       }
+       hw_stats->rx_macsec_sa_pkts_unusedsa +=
+                       rd32(hw, TXGBE_LSECRX_INVSAPKT);
+       hw_stats->rx_macsec_sa_pkts_notusingsa +=
+                       rd32(hw, TXGBE_LSECRX_BADSAPKT);
+
+       hw_stats->rx_total_missed_packets = 0;
+       for (i = 0; i < TXGBE_MAX_UP; i++) {
+               hw_stats->rx_total_missed_packets +=
+                       hw_stats->up[i].rx_up_dropped;
+       }
+}
+
+static int
+txgbe_dev_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_hw_stats *hw_stats = TXGBE_DEV_STATS(dev);
+       struct txgbe_stat_mappings *stat_mappings =
+                       TXGBE_DEV_STAT_MAPPINGS(dev);
+       uint32_t i, j;
+
+       txgbe_read_stats_registers(hw, hw_stats);
+
+       if (stats == NULL)
+               return -EINVAL;
+
+       /* Fill out the rte_eth_stats statistics structure */
+       stats->ipackets = hw_stats->rx_packets;
+       stats->ibytes = hw_stats->rx_bytes;
+       stats->opackets = hw_stats->tx_packets;
+       stats->obytes = hw_stats->tx_bytes;
+
+       memset(&stats->q_ipackets, 0, sizeof(stats->q_ipackets));
+       memset(&stats->q_opackets, 0, sizeof(stats->q_opackets));
+       memset(&stats->q_ibytes, 0, sizeof(stats->q_ibytes));
+       memset(&stats->q_obytes, 0, sizeof(stats->q_obytes));
+       memset(&stats->q_errors, 0, sizeof(stats->q_errors));
+       for (i = 0; i < TXGBE_MAX_QP; i++) {
+               uint32_t n = i / NB_QMAP_FIELDS_PER_QSM_REG;
+               uint32_t offset = (i % NB_QMAP_FIELDS_PER_QSM_REG) * 8;
+               uint32_t q_map;
+
+               q_map = (stat_mappings->rqsm[n] >> offset)
+                               & QMAP_FIELD_RESERVED_BITS_MASK;
+               j = (q_map < RTE_ETHDEV_QUEUE_STAT_CNTRS
+                    ? q_map : q_map % RTE_ETHDEV_QUEUE_STAT_CNTRS);
+               stats->q_ipackets[j] += hw_stats->qp[i].rx_qp_packets;
+               stats->q_ibytes[j] += hw_stats->qp[i].rx_qp_bytes;
+
+               q_map = (stat_mappings->tqsm[n] >> offset)
+                               & QMAP_FIELD_RESERVED_BITS_MASK;
+               j = (q_map < RTE_ETHDEV_QUEUE_STAT_CNTRS
+                    ? q_map : q_map % RTE_ETHDEV_QUEUE_STAT_CNTRS);
+               stats->q_opackets[j] += hw_stats->qp[i].tx_qp_packets;
+               stats->q_obytes[j] += hw_stats->qp[i].tx_qp_bytes;
+       }
+
+       /* Rx Errors */
+       stats->imissed  = hw_stats->rx_total_missed_packets;
+       stats->ierrors  = hw_stats->rx_crc_errors +
+                         hw_stats->rx_mac_short_packet_dropped +
+                         hw_stats->rx_length_errors +
+                         hw_stats->rx_undersize_errors +
+                         hw_stats->rx_oversize_errors +
+                         hw_stats->rx_drop_packets +
+                         hw_stats->rx_illegal_byte_errors +
+                         hw_stats->rx_error_bytes +
+                         hw_stats->rx_fragment_errors +
+                         hw_stats->rx_fcoe_crc_errors +
+                         hw_stats->rx_fcoe_mbuf_allocation_errors;
+
+       /* Tx Errors */
+       stats->oerrors  = 0;
+       return 0;
+}
+
+static int
+txgbe_dev_stats_reset(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_hw_stats *hw_stats = TXGBE_DEV_STATS(dev);
+
+       /* HW registers are cleared on read */
+       hw->offset_loaded = 0;
+       txgbe_dev_stats_get(dev, NULL);
+       hw->offset_loaded = 1;
+
+       /* Reset software totals */
+       memset(hw_stats, 0, sizeof(*hw_stats));
+
+       return 0;
+}
+
+/* This function calculates the number of xstats based on the current config */
+static unsigned
+txgbe_xstats_calc_num(struct rte_eth_dev *dev)
+{
+       int nb_queues = max(dev->data->nb_rx_queues, dev->data->nb_tx_queues);
+       return TXGBE_NB_HW_STATS +
+              TXGBE_NB_UP_STATS * TXGBE_MAX_UP +
+              TXGBE_NB_QP_STATS * nb_queues;
+}
+
+static inline int
+txgbe_get_name_by_id(uint32_t id, char *name, uint32_t size)
+{
+       int nb, st;
+
+       /* Extended stats from txgbe_hw_stats */
+       if (id < TXGBE_NB_HW_STATS) {
+               snprintf(name, size, "[hw]%s",
+                       rte_txgbe_stats_strings[id].name);
+               return 0;
+       }
+       id -= TXGBE_NB_HW_STATS;
+
+       /* Priority Stats */
+       if (id < TXGBE_NB_UP_STATS * TXGBE_MAX_UP) {
+               nb = id / TXGBE_NB_UP_STATS;
+               st = id % TXGBE_NB_UP_STATS;
+               snprintf(name, size, "[p%u]%s", nb,
+                       rte_txgbe_up_strings[st].name);
+               return 0;
+       }
+       id -= TXGBE_NB_UP_STATS * TXGBE_MAX_UP;
+
+       /* Queue Stats */
+       if (id < TXGBE_NB_QP_STATS * TXGBE_MAX_QP) {
+               nb = id / TXGBE_NB_QP_STATS;
+               st = id % TXGBE_NB_QP_STATS;
+               snprintf(name, size, "[q%u]%s", nb,
+                       rte_txgbe_qp_strings[st].name);
+               return 0;
+       }
+       id -= TXGBE_NB_QP_STATS * TXGBE_MAX_QP;
+
+       return -(int)(id + 1);
+}
+
+static inline int
+txgbe_get_offset_by_id(uint32_t id, uint32_t *offset)
+{
+       int nb, st;
+
+       /* Extended stats from txgbe_hw_stats */
+       if (id < TXGBE_NB_HW_STATS) {
+               *offset = rte_txgbe_stats_strings[id].offset;
+               return 0;
+       }
+       id -= TXGBE_NB_HW_STATS;
+
+       /* Priority Stats */
+       if (id < TXGBE_NB_UP_STATS * TXGBE_MAX_UP) {
+               nb = id / TXGBE_NB_UP_STATS;
+               st = id % TXGBE_NB_UP_STATS;
+               *offset = rte_txgbe_up_strings[st].offset +
+                       nb * (TXGBE_NB_UP_STATS * sizeof(uint64_t));
+               return 0;
+       }
+       id -= TXGBE_NB_UP_STATS * TXGBE_MAX_UP;
+
+       /* Queue Stats */
+       if (id < TXGBE_NB_QP_STATS * TXGBE_MAX_QP) {
+               nb = id / TXGBE_NB_QP_STATS;
+               st = id % TXGBE_NB_QP_STATS;
+               *offset = rte_txgbe_qp_strings[st].offset +
+                       nb * (TXGBE_NB_QP_STATS * sizeof(uint64_t));
+               return 0;
+       }
+
+       return -1;
+}
+
+static int txgbe_dev_xstats_get_names(struct rte_eth_dev *dev,
+       struct rte_eth_xstat_name *xstats_names, unsigned int limit)
+{
+       unsigned int i, count;
+
+       count = txgbe_xstats_calc_num(dev);
+       if (xstats_names == NULL)
+               return count;
+
+       /* Note: limit >= cnt_stats checked upstream
+        * in rte_eth_xstats_names()
+        */
+       limit = min(limit, count);
+
+       /* Extended stats from txgbe_hw_stats */
+       for (i = 0; i < limit; i++) {
+               if (txgbe_get_name_by_id(i, xstats_names[i].name,
+                       sizeof(xstats_names[i].name))) {
+                       PMD_INIT_LOG(WARNING, "id value %d isn't valid", i);
+                       break;
+               }
+       }
+
+       return i;
+}
+
+static int txgbe_dev_xstats_get_names_by_id(struct rte_eth_dev *dev,
+       struct rte_eth_xstat_name *xstats_names,
+       const uint64_t *ids,
+       unsigned int limit)
+{
+       unsigned int i;
+
+       if (ids == NULL)
+               return txgbe_dev_xstats_get_names(dev, xstats_names, limit);
+
+       for (i = 0; i < limit; i++) {
+               if (txgbe_get_name_by_id(ids[i], xstats_names[i].name,
+                               sizeof(xstats_names[i].name))) {
+                       PMD_INIT_LOG(WARNING, "id value %d isn't valid", i);
+                       return -1;
+               }
+       }
+
+       return i;
+}
+
+static int
+txgbe_dev_xstats_get(struct rte_eth_dev *dev, struct rte_eth_xstat *xstats,
+                                        unsigned int limit)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_hw_stats *hw_stats = TXGBE_DEV_STATS(dev);
+       unsigned int i, count;
+
+       txgbe_read_stats_registers(hw, hw_stats);
+
+       /* If this is a reset xstats is NULL, and we have cleared the
+        * registers by reading them.
+        */
+       count = txgbe_xstats_calc_num(dev);
+       if (xstats == NULL)
+               return count;
+
+       limit = min(limit, txgbe_xstats_calc_num(dev));
+
+       /* Extended stats from txgbe_hw_stats */
+       for (i = 0; i < limit; i++) {
+               uint32_t offset = 0;
+
+               if (txgbe_get_offset_by_id(i, &offset)) {
+                       PMD_INIT_LOG(WARNING, "id value %d isn't valid", i);
+                       break;
+               }
+               xstats[i].value = *(uint64_t *)(((char *)hw_stats) + offset);
+               xstats[i].id = i;
+       }
+
+       return i;
+}
+
+static int
+txgbe_dev_xstats_get_(struct rte_eth_dev *dev, uint64_t *values,
+                                        unsigned int limit)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_hw_stats *hw_stats = TXGBE_DEV_STATS(dev);
+       unsigned int i, count;
+
+       txgbe_read_stats_registers(hw, hw_stats);
+
+       /* If this is a reset xstats is NULL, and we have cleared the
+        * registers by reading them.
+        */
+       count = txgbe_xstats_calc_num(dev);
+       if (values == NULL)
+               return count;
+
+       limit = min(limit, txgbe_xstats_calc_num(dev));
+
+       /* Extended stats from txgbe_hw_stats */
+       for (i = 0; i < limit; i++) {
+               uint32_t offset;
+
+               if (txgbe_get_offset_by_id(i, &offset)) {
+                       PMD_INIT_LOG(WARNING, "id value %d isn't valid", i);
+                       break;
+               }
+               values[i] = *(uint64_t *)(((char *)hw_stats) + offset);
+       }
+
+       return i;
+}
+
+static int
+txgbe_dev_xstats_get_by_id(struct rte_eth_dev *dev, const uint64_t *ids,
+               uint64_t *values, unsigned int limit)
+{
+       struct txgbe_hw_stats *hw_stats = TXGBE_DEV_STATS(dev);
+       unsigned int i;
+
+       if (ids == NULL)
+               return txgbe_dev_xstats_get_(dev, values, limit);
+
+       for (i = 0; i < limit; i++) {
+               uint32_t offset;
+
+               if (txgbe_get_offset_by_id(ids[i], &offset)) {
+                       PMD_INIT_LOG(WARNING, "id value %d isn't valid", i);
+                       break;
+               }
+               values[i] = *(uint64_t *)(((char *)hw_stats) + offset);
+       }
+
+       return i;
+}
+
+static int
+txgbe_dev_xstats_reset(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_hw_stats *hw_stats = TXGBE_DEV_STATS(dev);
+
+       /* HW registers are cleared on read */
+       hw->offset_loaded = 0;
+       txgbe_read_stats_registers(hw, hw_stats);
+       hw->offset_loaded = 1;
+
+       /* Reset software totals */
+       memset(hw_stats, 0, sizeof(*hw_stats));
+
+       return 0;
+}
+
+static int
+txgbe_fw_version_get(struct rte_eth_dev *dev, char *fw_version, size_t fw_size)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       u16 eeprom_verh, eeprom_verl;
+       u32 etrack_id;
+       int ret;
+
+       hw->rom.readw_sw(hw, TXGBE_EEPROM_VERSION_H, &eeprom_verh);
+       hw->rom.readw_sw(hw, TXGBE_EEPROM_VERSION_L, &eeprom_verl);
+
+       etrack_id = (eeprom_verh << 16) | eeprom_verl;
+       ret = snprintf(fw_version, fw_size, "0x%08x", etrack_id);
+
+       ret += 1; /* add the size of '\0' */
+       if (fw_size < (u32)ret)
+               return ret;
+       else
+               return 0;
+}
+
+static int
+txgbe_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
+{
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       dev_info->max_rx_queues = (uint16_t)hw->mac.max_rx_queues;
+       dev_info->max_tx_queues = (uint16_t)hw->mac.max_tx_queues;
+       dev_info->min_rx_bufsize = 1024;
+       dev_info->max_rx_pktlen = 15872;
+       dev_info->max_mac_addrs = hw->mac.num_rar_entries;
+       dev_info->max_hash_mac_addrs = TXGBE_VMDQ_NUM_UC_MAC;
+       dev_info->max_vfs = pci_dev->max_vfs;
+       dev_info->max_vmdq_pools = ETH_64_POOLS;
+       dev_info->vmdq_queue_num = dev_info->max_rx_queues;
+       dev_info->rx_queue_offload_capa = txgbe_get_rx_queue_offloads(dev);
+       dev_info->rx_offload_capa = (txgbe_get_rx_port_offloads(dev) |
+                                    dev_info->rx_queue_offload_capa);
+       dev_info->tx_queue_offload_capa = txgbe_get_tx_queue_offloads(dev);
+       dev_info->tx_offload_capa = txgbe_get_tx_port_offloads(dev);
+
+       dev_info->default_rxconf = (struct rte_eth_rxconf) {
+               .rx_thresh = {
+                       .pthresh = TXGBE_DEFAULT_RX_PTHRESH,
+                       .hthresh = TXGBE_DEFAULT_RX_HTHRESH,
+                       .wthresh = TXGBE_DEFAULT_RX_WTHRESH,
+               },
+               .rx_free_thresh = TXGBE_DEFAULT_RX_FREE_THRESH,
+               .rx_drop_en = 0,
+               .offloads = 0,
+       };
+
+       dev_info->default_txconf = (struct rte_eth_txconf) {
+               .tx_thresh = {
+                       .pthresh = TXGBE_DEFAULT_TX_PTHRESH,
+                       .hthresh = TXGBE_DEFAULT_TX_HTHRESH,
+                       .wthresh = TXGBE_DEFAULT_TX_WTHRESH,
+               },
+               .tx_free_thresh = TXGBE_DEFAULT_TX_FREE_THRESH,
+               .offloads = 0,
+       };
+
+       dev_info->rx_desc_lim = rx_desc_lim;
+       dev_info->tx_desc_lim = tx_desc_lim;
+
+       dev_info->hash_key_size = TXGBE_HKEY_MAX_INDEX * sizeof(uint32_t);
+       dev_info->reta_size = ETH_RSS_RETA_SIZE_128;
+       dev_info->flow_type_rss_offloads = TXGBE_RSS_OFFLOAD_ALL;
+
+       dev_info->speed_capa = ETH_LINK_SPEED_1G | ETH_LINK_SPEED_10G;
+       dev_info->speed_capa |= ETH_LINK_SPEED_100M;
+
+       /* Driver-preferred Rx/Tx parameters */
+       dev_info->default_rxportconf.burst_size = 32;
+       dev_info->default_txportconf.burst_size = 32;
+       dev_info->default_rxportconf.nb_queues = 1;
+       dev_info->default_txportconf.nb_queues = 1;
+       dev_info->default_rxportconf.ring_size = 256;
+       dev_info->default_txportconf.ring_size = 256;
+
+       return 0;
+}
+
+const uint32_t *
+txgbe_dev_supported_ptypes_get(struct rte_eth_dev *dev)
+{
+       if (dev->rx_pkt_burst == txgbe_recv_pkts ||
+           dev->rx_pkt_burst == txgbe_recv_pkts_lro_single_alloc ||
+           dev->rx_pkt_burst == txgbe_recv_pkts_lro_bulk_alloc ||
+           dev->rx_pkt_burst == txgbe_recv_pkts_bulk_alloc)
+               return txgbe_get_supported_ptypes();
+
+       return NULL;
+}
+
+void
+txgbe_dev_setup_link_alarm_handler(void *param)
+{
+       struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       u32 speed;
+       bool autoneg = false;
+
+       speed = hw->phy.autoneg_advertised;
+       if (!speed)
+               hw->mac.get_link_capabilities(hw, &speed, &autoneg);
+
+       hw->mac.setup_link(hw, speed, true);
+
+       intr->flags &= ~TXGBE_FLAG_NEED_LINK_CONFIG;
+}
+
+/* return 0 means link status changed, -1 means not changed */
+int
+txgbe_dev_link_update_share(struct rte_eth_dev *dev,
+                           int wait_to_complete)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct rte_eth_link link;
+       u32 link_speed = TXGBE_LINK_SPEED_UNKNOWN;
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       bool link_up;
+       int err;
+       int wait = 1;
+
+       memset(&link, 0, sizeof(link));
+       link.link_status = ETH_LINK_DOWN;
+       link.link_speed = ETH_SPEED_NUM_NONE;
+       link.link_duplex = ETH_LINK_HALF_DUPLEX;
+       link.link_autoneg = ETH_LINK_AUTONEG;
+
+       hw->mac.get_link_status = true;
+
+       if (intr->flags & TXGBE_FLAG_NEED_LINK_CONFIG)
+               return rte_eth_linkstatus_set(dev, &link);
+
+       /* check if it needs to wait to complete, if lsc interrupt is enabled */
+       if (wait_to_complete == 0 || dev->data->dev_conf.intr_conf.lsc != 0)
+               wait = 0;
+
+       err = hw->mac.check_link(hw, &link_speed, &link_up, wait);
+
+       if (err != 0) {
+               link.link_speed = ETH_SPEED_NUM_100M;
+               link.link_duplex = ETH_LINK_FULL_DUPLEX;
+               return rte_eth_linkstatus_set(dev, &link);
+       }
+
+       if (link_up == 0) {
+               if (hw->phy.media_type == txgbe_media_type_fiber) {
+                       intr->flags |= TXGBE_FLAG_NEED_LINK_CONFIG;
+                       rte_eal_alarm_set(10,
+                               txgbe_dev_setup_link_alarm_handler, dev);
+               }
+               return rte_eth_linkstatus_set(dev, &link);
+       }
+
+       intr->flags &= ~TXGBE_FLAG_NEED_LINK_CONFIG;
+       link.link_status = ETH_LINK_UP;
+       link.link_duplex = ETH_LINK_FULL_DUPLEX;
+
+       switch (link_speed) {
+       default:
+       case TXGBE_LINK_SPEED_UNKNOWN:
+               link.link_duplex = ETH_LINK_FULL_DUPLEX;
+               link.link_speed = ETH_SPEED_NUM_100M;
+               break;
+
+       case TXGBE_LINK_SPEED_100M_FULL:
+               link.link_speed = ETH_SPEED_NUM_100M;
+               break;
+
+       case TXGBE_LINK_SPEED_1GB_FULL:
+               link.link_speed = ETH_SPEED_NUM_1G;
+               break;
+
+       case TXGBE_LINK_SPEED_2_5GB_FULL:
+               link.link_speed = ETH_SPEED_NUM_2_5G;
+               break;
+
+       case TXGBE_LINK_SPEED_5GB_FULL:
+               link.link_speed = ETH_SPEED_NUM_5G;
+               break;
+
+       case TXGBE_LINK_SPEED_10GB_FULL:
+               link.link_speed = ETH_SPEED_NUM_10G;
+               break;
+       }
+
+       return rte_eth_linkstatus_set(dev, &link);
+}
+
+static int
+txgbe_dev_link_update(struct rte_eth_dev *dev, int wait_to_complete)
+{
+       return txgbe_dev_link_update_share(dev, wait_to_complete);
+}
+
+static int
+txgbe_dev_promiscuous_enable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t fctrl;
+
+       fctrl = rd32(hw, TXGBE_PSRCTL);
+       fctrl |= (TXGBE_PSRCTL_UCP | TXGBE_PSRCTL_MCP);
+       wr32(hw, TXGBE_PSRCTL, fctrl);
+
+       return 0;
+}
+
+static int
+txgbe_dev_promiscuous_disable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t fctrl;
+
+       fctrl = rd32(hw, TXGBE_PSRCTL);
+       fctrl &= (~TXGBE_PSRCTL_UCP);
+       if (dev->data->all_multicast == 1)
+               fctrl |= TXGBE_PSRCTL_MCP;
+       else
+               fctrl &= (~TXGBE_PSRCTL_MCP);
+       wr32(hw, TXGBE_PSRCTL, fctrl);
+
+       return 0;
+}
+
+static int
+txgbe_dev_allmulticast_enable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t fctrl;
+
+       fctrl = rd32(hw, TXGBE_PSRCTL);
+       fctrl |= TXGBE_PSRCTL_MCP;
+       wr32(hw, TXGBE_PSRCTL, fctrl);
+
+       return 0;
+}
+
+static int
+txgbe_dev_allmulticast_disable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t fctrl;
+
+       if (dev->data->promiscuous == 1)
+               return 0; /* must remain in all_multicast mode */
+
+       fctrl = rd32(hw, TXGBE_PSRCTL);
+       fctrl &= (~TXGBE_PSRCTL_MCP);
+       wr32(hw, TXGBE_PSRCTL, fctrl);
+
+       return 0;
+}
+
+/**
+ * It clears the interrupt causes and enables the interrupt.
+ * It will be called once only during nic initialized.
+ *
+ * @param dev
+ *  Pointer to struct rte_eth_dev.
+ * @param on
+ *  Enable or Disable.
+ *
+ * @return
+ *  - On success, zero.
+ *  - On failure, a negative value.
+ */
+static int
+txgbe_dev_lsc_interrupt_setup(struct rte_eth_dev *dev, uint8_t on)
+{
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+
+       txgbe_dev_link_status_print(dev);
+       if (on)
+               intr->mask_misc |= TXGBE_ICRMISC_LSC;
+       else
+               intr->mask_misc &= ~TXGBE_ICRMISC_LSC;
+
+       return 0;
+}
+
+/**
+ * It clears the interrupt causes and enables the interrupt.
+ * It will be called once only during nic initialized.
+ *
+ * @param dev
+ *  Pointer to struct rte_eth_dev.
+ *
+ * @return
+ *  - On success, zero.
+ *  - On failure, a negative value.
+ */
+static int
+txgbe_dev_rxq_interrupt_setup(struct rte_eth_dev *dev)
+{
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+
+       intr->mask[0] |= TXGBE_ICR_MASK;
+       intr->mask[1] |= TXGBE_ICR_MASK;
+
+       return 0;
+}
+
+/**
+ * It clears the interrupt causes and enables the interrupt.
+ * It will be called once only during nic initialized.
+ *
+ * @param dev
+ *  Pointer to struct rte_eth_dev.
+ *
+ * @return
+ *  - On success, zero.
+ *  - On failure, a negative value.
+ */
+static int
+txgbe_dev_macsec_interrupt_setup(struct rte_eth_dev *dev)
+{
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+
+       intr->mask_misc |= TXGBE_ICRMISC_LNKSEC;
+
+       return 0;
+}
+
+/*
+ * It reads ICR and sets flag (TXGBE_ICRMISC_LSC) for the link_update.
+ *
+ * @param dev
+ *  Pointer to struct rte_eth_dev.
+ *
+ * @return
+ *  - On success, zero.
+ *  - On failure, a negative value.
+ */
+static int
+txgbe_dev_interrupt_get_status(struct rte_eth_dev *dev)
+{
+       uint32_t eicr;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+
+       /* clear all cause mask */
+       txgbe_disable_intr(hw);
+
+       /* read-on-clear nic registers here */
+       eicr = ((u32 *)hw->isb_mem)[TXGBE_ISB_MISC];
+       PMD_DRV_LOG(DEBUG, "eicr %x", eicr);
+
+       intr->flags = 0;
+
+       /* set flag for async link update */
+       if (eicr & TXGBE_ICRMISC_LSC)
+               intr->flags |= TXGBE_FLAG_NEED_LINK_UPDATE;
+
+       if (eicr & TXGBE_ICRMISC_VFMBX)
+               intr->flags |= TXGBE_FLAG_MAILBOX;
+
+       if (eicr & TXGBE_ICRMISC_LNKSEC)
+               intr->flags |= TXGBE_FLAG_MACSEC;
+
+       if (eicr & TXGBE_ICRMISC_GPIO)
+               intr->flags |= TXGBE_FLAG_PHY_INTERRUPT;
+
+       return 0;
+}
+
+/**
+ * It gets and then prints the link status.
+ *
+ * @param dev
+ *  Pointer to struct rte_eth_dev.
+ *
+ * @return
+ *  - On success, zero.
+ *  - On failure, a negative value.
+ */
+static void
+txgbe_dev_link_status_print(struct rte_eth_dev *dev)
+{
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+       struct rte_eth_link link;
+
+       rte_eth_linkstatus_get(dev, &link);
+
+       if (link.link_status) {
+               PMD_INIT_LOG(INFO, "Port %d: Link Up - speed %u Mbps - %s",
+                                       (int)(dev->data->port_id),
+                                       (unsigned int)link.link_speed,
+                       link.link_duplex == ETH_LINK_FULL_DUPLEX ?
+                                       "full-duplex" : "half-duplex");
+       } else {
+               PMD_INIT_LOG(INFO, " Port %d: Link Down",
+                               (int)(dev->data->port_id));
+       }
+       PMD_INIT_LOG(DEBUG, "PCI Address: " PCI_PRI_FMT,
+                               pci_dev->addr.domain,
+                               pci_dev->addr.bus,
+                               pci_dev->addr.devid,
+                               pci_dev->addr.function);
+}
+
+/*
+ * It executes link_update after knowing an interrupt occurred.
+ *
+ * @param dev
+ *  Pointer to struct rte_eth_dev.
+ *
+ * @return
+ *  - On success, zero.
+ *  - On failure, a negative value.
+ */
+static int
+txgbe_dev_interrupt_action(struct rte_eth_dev *dev,
+                          struct rte_intr_handle *intr_handle)
+{
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       int64_t timeout;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       PMD_DRV_LOG(DEBUG, "intr action type %d", intr->flags);
+
+       if (intr->flags & TXGBE_FLAG_MAILBOX) {
+               txgbe_pf_mbx_process(dev);
+               intr->flags &= ~TXGBE_FLAG_MAILBOX;
+       }
+
+       if (intr->flags & TXGBE_FLAG_PHY_INTERRUPT) {
+               hw->phy.handle_lasi(hw);
+               intr->flags &= ~TXGBE_FLAG_PHY_INTERRUPT;
+       }
+
+       if (intr->flags & TXGBE_FLAG_NEED_LINK_UPDATE) {
+               struct rte_eth_link link;
+
+               /*get the link status before link update, for predicting later*/
+               rte_eth_linkstatus_get(dev, &link);
+
+               txgbe_dev_link_update(dev, 0);
+
+               /* likely to up */
+               if (!link.link_status)
+                       /* handle it 1 sec later, wait it being stable */
+                       timeout = TXGBE_LINK_UP_CHECK_TIMEOUT;
+               /* likely to down */
+               else
+                       /* handle it 4 sec later, wait it being stable */
+                       timeout = TXGBE_LINK_DOWN_CHECK_TIMEOUT;
+
+               txgbe_dev_link_status_print(dev);
+               if (rte_eal_alarm_set(timeout * 1000,
+                                     txgbe_dev_interrupt_delayed_handler,
+                                     (void *)dev) < 0) {
+                       PMD_DRV_LOG(ERR, "Error setting alarm");
+               } else {
+                       /* remember original mask */
+                       intr->mask_misc_orig = intr->mask_misc;
+                       /* only disable lsc interrupt */
+                       intr->mask_misc &= ~TXGBE_ICRMISC_LSC;
+               }
+       }
+
+       PMD_DRV_LOG(DEBUG, "enable intr immediately");
+       txgbe_enable_intr(dev);
+       rte_intr_enable(intr_handle);
+
+       return 0;
+}
+
+/**
+ * Interrupt handler which shall be registered for alarm callback for delayed
+ * handling specific interrupt to wait for the stable nic state. As the
+ * NIC interrupt state is not stable for txgbe after link is just down,
+ * it needs to wait 4 seconds to get the stable status.
+ *
+ * @param handle
+ *  Pointer to interrupt handle.
+ * @param param
+ *  The address of parameter (struct rte_eth_dev *) registered before.
+ *
+ * @return
+ *  void
+ */
+static void
+txgbe_dev_interrupt_delayed_handler(void *param)
+{
+       struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+       struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
+       struct txgbe_interrupt *intr = TXGBE_DEV_INTR(dev);
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t eicr;
+
+       txgbe_disable_intr(hw);
+
+       eicr = ((u32 *)hw->isb_mem)[TXGBE_ISB_MISC];
+       if (eicr & TXGBE_ICRMISC_VFMBX)
+               txgbe_pf_mbx_process(dev);
+
+       if (intr->flags & TXGBE_FLAG_PHY_INTERRUPT) {
+               hw->phy.handle_lasi(hw);
+               intr->flags &= ~TXGBE_FLAG_PHY_INTERRUPT;
+       }
+
+       if (intr->flags & TXGBE_FLAG_NEED_LINK_UPDATE) {
+               txgbe_dev_link_update(dev, 0);
+               intr->flags &= ~TXGBE_FLAG_NEED_LINK_UPDATE;
+               txgbe_dev_link_status_print(dev);
+               rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_INTR_LSC,
+                                             NULL);
+       }
+
+       if (intr->flags & TXGBE_FLAG_MACSEC) {
+               rte_eth_dev_callback_process(dev, RTE_ETH_EVENT_MACSEC,
+                                             NULL);
+               intr->flags &= ~TXGBE_FLAG_MACSEC;
+       }
+
+       /* restore original mask */
+       intr->mask_misc = intr->mask_misc_orig;
+       intr->mask_misc_orig = 0;
+
+       PMD_DRV_LOG(DEBUG, "enable intr in delayed handler S[%08x]", eicr);
+       txgbe_enable_intr(dev);
+       rte_intr_enable(intr_handle);
+}
+
+/**
+ * Interrupt handler triggered by NIC  for handling
+ * specific interrupt.
+ *
+ * @param handle
+ *  Pointer to interrupt handle.
+ * @param param
+ *  The address of parameter (struct rte_eth_dev *) registered before.
+ *
+ * @return
+ *  void
+ */
+static void
+txgbe_dev_interrupt_handler(void *param)
+{
+       struct rte_eth_dev *dev = (struct rte_eth_dev *)param;
+
+       txgbe_dev_interrupt_get_status(dev);
+       txgbe_dev_interrupt_action(dev, dev->intr_handle);
+}
+
+static int
+txgbe_dev_led_on(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw;
+
+       hw = TXGBE_DEV_HW(dev);
+       return txgbe_led_on(hw, 4) == 0 ? 0 : -ENOTSUP;
+}
+
+static int
+txgbe_dev_led_off(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw;
+
+       hw = TXGBE_DEV_HW(dev);
+       return txgbe_led_off(hw, 4) == 0 ? 0 : -ENOTSUP;
+}
+
+static int
+txgbe_flow_ctrl_get(struct rte_eth_dev *dev, struct rte_eth_fc_conf *fc_conf)
+{
+       struct txgbe_hw *hw;
+       uint32_t mflcn_reg;
+       uint32_t fccfg_reg;
+       int rx_pause;
+       int tx_pause;
+
+       hw = TXGBE_DEV_HW(dev);
+
+       fc_conf->pause_time = hw->fc.pause_time;
+       fc_conf->high_water = hw->fc.high_water[0];
+       fc_conf->low_water = hw->fc.low_water[0];
+       fc_conf->send_xon = hw->fc.send_xon;
+       fc_conf->autoneg = !hw->fc.disable_fc_autoneg;
+
+       /*
+        * Return rx_pause status according to actual setting of
+        * RXFCCFG register.
+        */
+       mflcn_reg = rd32(hw, TXGBE_RXFCCFG);
+       if (mflcn_reg & (TXGBE_RXFCCFG_FC | TXGBE_RXFCCFG_PFC))
+               rx_pause = 1;
+       else
+               rx_pause = 0;
+
+       /*
+        * Return tx_pause status according to actual setting of
+        * TXFCCFG register.
+        */
+       fccfg_reg = rd32(hw, TXGBE_TXFCCFG);
+       if (fccfg_reg & (TXGBE_TXFCCFG_FC | TXGBE_TXFCCFG_PFC))
+               tx_pause = 1;
+       else
+               tx_pause = 0;
+
+       if (rx_pause && tx_pause)
+               fc_conf->mode = RTE_FC_FULL;
+       else if (rx_pause)
+               fc_conf->mode = RTE_FC_RX_PAUSE;
+       else if (tx_pause)
+               fc_conf->mode = RTE_FC_TX_PAUSE;
+       else
+               fc_conf->mode = RTE_FC_NONE;
+
+       return 0;
+}
+
+static int
+txgbe_flow_ctrl_set(struct rte_eth_dev *dev, struct rte_eth_fc_conf *fc_conf)
+{
+       struct txgbe_hw *hw;
+       int err;
+       uint32_t rx_buf_size;
+       uint32_t max_high_water;
+       enum txgbe_fc_mode rte_fcmode_2_txgbe_fcmode[] = {
+               txgbe_fc_none,
+               txgbe_fc_rx_pause,
+               txgbe_fc_tx_pause,
+               txgbe_fc_full
+       };
+
+       PMD_INIT_FUNC_TRACE();
+
+       hw = TXGBE_DEV_HW(dev);
+       rx_buf_size = rd32(hw, TXGBE_PBRXSIZE(0));
+       PMD_INIT_LOG(DEBUG, "Rx packet buffer size = 0x%x", rx_buf_size);
+
+       /*
+        * At least reserve one Ethernet frame for watermark
+        * high_water/low_water in kilo bytes for txgbe
+        */
+       max_high_water = (rx_buf_size - RTE_ETHER_MAX_LEN) >> 10;
+       if (fc_conf->high_water > max_high_water ||
+           fc_conf->high_water < fc_conf->low_water) {
+               PMD_INIT_LOG(ERR, "Invalid high/low water setup value in KB");
+               PMD_INIT_LOG(ERR, "High_water must <= 0x%x", max_high_water);
+               return -EINVAL;
+       }
+
+       hw->fc.requested_mode = rte_fcmode_2_txgbe_fcmode[fc_conf->mode];
+       hw->fc.pause_time     = fc_conf->pause_time;
+       hw->fc.high_water[0]  = fc_conf->high_water;
+       hw->fc.low_water[0]   = fc_conf->low_water;
+       hw->fc.send_xon       = fc_conf->send_xon;
+       hw->fc.disable_fc_autoneg = !fc_conf->autoneg;
+
+       err = txgbe_fc_enable(hw);
+
+       /* Not negotiated is not an error case */
+       if (err == 0 || err == TXGBE_ERR_FC_NOT_NEGOTIATED) {
+               wr32m(hw, TXGBE_MACRXFLT, TXGBE_MACRXFLT_CTL_MASK,
+                     (fc_conf->mac_ctrl_frame_fwd
+                      ? TXGBE_MACRXFLT_CTL_NOPS : TXGBE_MACRXFLT_CTL_DROP));
+               txgbe_flush(hw);
+
+               return 0;
+       }
+
+       PMD_INIT_LOG(ERR, "txgbe_fc_enable = 0x%x", err);
+       return -EIO;
+}
+
+static int
+txgbe_priority_flow_ctrl_set(struct rte_eth_dev *dev,
+               struct rte_eth_pfc_conf *pfc_conf)
+{
+       int err;
+       uint32_t rx_buf_size;
+       uint32_t max_high_water;
+       uint8_t tc_num;
+       uint8_t  map[TXGBE_DCB_UP_MAX] = { 0 };
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_dcb_config *dcb_config = TXGBE_DEV_DCB_CONFIG(dev);
+
+       enum txgbe_fc_mode rte_fcmode_2_txgbe_fcmode[] = {
+               txgbe_fc_none,
+               txgbe_fc_rx_pause,
+               txgbe_fc_tx_pause,
+               txgbe_fc_full
+       };
+
+       PMD_INIT_FUNC_TRACE();
+
+       txgbe_dcb_unpack_map_cee(dcb_config, TXGBE_DCB_RX_CONFIG, map);
+       tc_num = map[pfc_conf->priority];
+       rx_buf_size = rd32(hw, TXGBE_PBRXSIZE(tc_num));
+       PMD_INIT_LOG(DEBUG, "Rx packet buffer size = 0x%x", rx_buf_size);
+       /*
+        * At least reserve one Ethernet frame for watermark
+        * high_water/low_water in kilo bytes for txgbe
+        */
+       max_high_water = (rx_buf_size - RTE_ETHER_MAX_LEN) >> 10;
+       if (pfc_conf->fc.high_water > max_high_water ||
+           pfc_conf->fc.high_water <= pfc_conf->fc.low_water) {
+               PMD_INIT_LOG(ERR, "Invalid high/low water setup value in KB");
+               PMD_INIT_LOG(ERR, "High_water must <= 0x%x", max_high_water);
+               return -EINVAL;
+       }
+
+       hw->fc.requested_mode = rte_fcmode_2_txgbe_fcmode[pfc_conf->fc.mode];
+       hw->fc.pause_time = pfc_conf->fc.pause_time;
+       hw->fc.send_xon = pfc_conf->fc.send_xon;
+       hw->fc.low_water[tc_num] =  pfc_conf->fc.low_water;
+       hw->fc.high_water[tc_num] = pfc_conf->fc.high_water;
+
+       err = txgbe_dcb_pfc_enable(hw, tc_num);
+
+       /* Not negotiated is not an error case */
+       if (err == 0 || err == TXGBE_ERR_FC_NOT_NEGOTIATED)
+               return 0;
+
+       PMD_INIT_LOG(ERR, "txgbe_dcb_pfc_enable = 0x%x", err);
+       return -EIO;
+}
+
+int
+txgbe_dev_rss_reta_update(struct rte_eth_dev *dev,
+                         struct rte_eth_rss_reta_entry64 *reta_conf,
+                         uint16_t reta_size)
+{
+       uint8_t i, j, mask;
+       uint32_t reta;
+       uint16_t idx, shift;
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       PMD_INIT_FUNC_TRACE();
+
+       if (!txgbe_rss_update_sp(hw->mac.type)) {
+               PMD_DRV_LOG(ERR, "RSS reta update is not supported on this "
+                       "NIC.");
+               return -ENOTSUP;
+       }
+
+       if (reta_size != ETH_RSS_RETA_SIZE_128) {
+               PMD_DRV_LOG(ERR, "The size of hash lookup table configured "
+                       "(%d) doesn't match the number hardware can supported "
+                       "(%d)", reta_size, ETH_RSS_RETA_SIZE_128);
+               return -EINVAL;
+       }
+
+       for (i = 0; i < reta_size; i += 4) {
+               idx = i / RTE_RETA_GROUP_SIZE;
+               shift = i % RTE_RETA_GROUP_SIZE;
+               mask = (uint8_t)RS64(reta_conf[idx].mask, shift, 0xF);
+               if (!mask)
+                       continue;
+
+               reta = rd32a(hw, TXGBE_REG_RSSTBL, i >> 2);
+               for (j = 0; j < 4; j++) {
+                       if (RS8(mask, j, 0x1)) {
+                               reta  &= ~(MS32(8 * j, 0xFF));
+                               reta |= LS32(reta_conf[idx].reta[shift + j],
+                                               8 * j, 0xFF);
+                       }
+               }
+               wr32a(hw, TXGBE_REG_RSSTBL, i >> 2, reta);
+       }
+       adapter->rss_reta_updated = 1;
+
+       return 0;
+}
+
+int
+txgbe_dev_rss_reta_query(struct rte_eth_dev *dev,
+                        struct rte_eth_rss_reta_entry64 *reta_conf,
+                        uint16_t reta_size)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint8_t i, j, mask;
+       uint32_t reta;
+       uint16_t idx, shift;
+
+       PMD_INIT_FUNC_TRACE();
+
+       if (reta_size != ETH_RSS_RETA_SIZE_128) {
+               PMD_DRV_LOG(ERR, "The size of hash lookup table configured "
+                       "(%d) doesn't match the number hardware can supported "
+                       "(%d)", reta_size, ETH_RSS_RETA_SIZE_128);
+               return -EINVAL;
+       }
+
+       for (i = 0; i < reta_size; i += 4) {
+               idx = i / RTE_RETA_GROUP_SIZE;
+               shift = i % RTE_RETA_GROUP_SIZE;
+               mask = (uint8_t)RS64(reta_conf[idx].mask, shift, 0xF);
+               if (!mask)
+                       continue;
+
+               reta = rd32a(hw, TXGBE_REG_RSSTBL, i >> 2);
+               for (j = 0; j < 4; j++) {
+                       if (RS8(mask, j, 0x1))
+                               reta_conf[idx].reta[shift + j] =
+                                       (uint16_t)RS32(reta, 8 * j, 0xFF);
+               }
+       }
+
+       return 0;
+}
+
+static int
+txgbe_add_rar(struct rte_eth_dev *dev, struct rte_ether_addr *mac_addr,
+                               uint32_t index, uint32_t pool)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t enable_addr = 1;
+
+       return txgbe_set_rar(hw, index, mac_addr->addr_bytes,
+                            pool, enable_addr);
+}
+
+static void
+txgbe_remove_rar(struct rte_eth_dev *dev, uint32_t index)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       txgbe_clear_rar(hw, index);
+}
+
+static int
+txgbe_set_default_mac_addr(struct rte_eth_dev *dev, struct rte_ether_addr *addr)
+{
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+
+       txgbe_remove_rar(dev, 0);
+       txgbe_add_rar(dev, addr, 0, pci_dev->max_vfs);
+
+       return 0;
+}
+
+static int
+txgbe_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct rte_eth_dev_info dev_info;
+       uint32_t frame_size = mtu + RTE_ETHER_HDR_LEN + RTE_ETHER_CRC_LEN;
+       struct rte_eth_dev_data *dev_data = dev->data;
+       int ret;
+
+       ret = txgbe_dev_info_get(dev, &dev_info);
+       if (ret != 0)
+               return ret;
+
+       /* check that mtu is within the allowed range */
+       if (mtu < RTE_ETHER_MIN_MTU || frame_size > dev_info.max_rx_pktlen)
+               return -EINVAL;
+
+       /* If device is started, refuse mtu that requires the support of
+        * scattered packets when this feature has not been enabled before.
+        */
+       if (dev_data->dev_started && !dev_data->scattered_rx &&
+           (frame_size + 2 * TXGBE_VLAN_TAG_SIZE >
+            dev->data->min_rx_buf_size - RTE_PKTMBUF_HEADROOM)) {
+               PMD_INIT_LOG(ERR, "Stop port first.");
+               return -EINVAL;
+       }
+
+       /* update max frame size */
+       dev->data->dev_conf.rxmode.max_rx_pkt_len = frame_size;
+
+       if (hw->mode)
+               wr32m(hw, TXGBE_FRMSZ, TXGBE_FRMSZ_MAX_MASK,
+                       TXGBE_FRAME_SIZE_MAX);
+       else
+               wr32m(hw, TXGBE_FRMSZ, TXGBE_FRMSZ_MAX_MASK,
+                       TXGBE_FRMSZ_MAX(frame_size));
+
+       return 0;
+}
+
+static uint32_t
+txgbe_uta_vector(struct txgbe_hw *hw, struct rte_ether_addr *uc_addr)
+{
+       uint32_t vector = 0;
+
+       switch (hw->mac.mc_filter_type) {
+       case 0:   /* use bits [47:36] of the address */
+               vector = ((uc_addr->addr_bytes[4] >> 4) |
+                       (((uint16_t)uc_addr->addr_bytes[5]) << 4));
+               break;
+       case 1:   /* use bits [46:35] of the address */
+               vector = ((uc_addr->addr_bytes[4] >> 3) |
+                       (((uint16_t)uc_addr->addr_bytes[5]) << 5));
+               break;
+       case 2:   /* use bits [45:34] of the address */
+               vector = ((uc_addr->addr_bytes[4] >> 2) |
+                       (((uint16_t)uc_addr->addr_bytes[5]) << 6));
+               break;
+       case 3:   /* use bits [43:32] of the address */
+               vector = ((uc_addr->addr_bytes[4]) |
+                       (((uint16_t)uc_addr->addr_bytes[5]) << 8));
+               break;
+       default:  /* Invalid mc_filter_type */
+               break;
+       }
+
+       /* vector can only be 12-bits or boundary will be exceeded */
+       vector &= 0xFFF;
+       return vector;
+}
+
+static int
+txgbe_uc_hash_table_set(struct rte_eth_dev *dev,
+                       struct rte_ether_addr *mac_addr, uint8_t on)
+{
+       uint32_t vector;
+       uint32_t uta_idx;
+       uint32_t reg_val;
+       uint32_t uta_mask;
+       uint32_t psrctl;
+
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_uta_info *uta_info = TXGBE_DEV_UTA_INFO(dev);
+
+       /* The UTA table only exists on pf hardware */
+       if (hw->mac.type < txgbe_mac_raptor)
+               return -ENOTSUP;
+
+       vector = txgbe_uta_vector(hw, mac_addr);
+       uta_idx = (vector >> 5) & 0x7F;
+       uta_mask = 0x1UL << (vector & 0x1F);
+
+       if (!!on == !!(uta_info->uta_shadow[uta_idx] & uta_mask))
+               return 0;
+
+       reg_val = rd32(hw, TXGBE_UCADDRTBL(uta_idx));
+       if (on) {
+               uta_info->uta_in_use++;
+               reg_val |= uta_mask;
+               uta_info->uta_shadow[uta_idx] |= uta_mask;
+       } else {
+               uta_info->uta_in_use--;
+               reg_val &= ~uta_mask;
+               uta_info->uta_shadow[uta_idx] &= ~uta_mask;
+       }
+
+       wr32(hw, TXGBE_UCADDRTBL(uta_idx), reg_val);
+
+       psrctl = rd32(hw, TXGBE_PSRCTL);
+       if (uta_info->uta_in_use > 0)
+               psrctl |= TXGBE_PSRCTL_UCHFENA;
+       else
+               psrctl &= ~TXGBE_PSRCTL_UCHFENA;
+
+       psrctl &= ~TXGBE_PSRCTL_ADHF12_MASK;
+       psrctl |= TXGBE_PSRCTL_ADHF12(hw->mac.mc_filter_type);
+       wr32(hw, TXGBE_PSRCTL, psrctl);
+
+       return 0;
+}
+
+static int
+txgbe_uc_all_hash_table_set(struct rte_eth_dev *dev, uint8_t on)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_uta_info *uta_info = TXGBE_DEV_UTA_INFO(dev);
+       uint32_t psrctl;
+       int i;
+
+       /* The UTA table only exists on pf hardware */
+       if (hw->mac.type < txgbe_mac_raptor)
+               return -ENOTSUP;
+
+       if (on) {
+               for (i = 0; i < ETH_VMDQ_NUM_UC_HASH_ARRAY; i++) {
+                       uta_info->uta_shadow[i] = ~0;
+                       wr32(hw, TXGBE_UCADDRTBL(i), ~0);
+               }
+       } else {
+               for (i = 0; i < ETH_VMDQ_NUM_UC_HASH_ARRAY; i++) {
+                       uta_info->uta_shadow[i] = 0;
+                       wr32(hw, TXGBE_UCADDRTBL(i), 0);
+               }
+       }
+
+       psrctl = rd32(hw, TXGBE_PSRCTL);
+       if (on)
+               psrctl |= TXGBE_PSRCTL_UCHFENA;
+       else
+               psrctl &= ~TXGBE_PSRCTL_UCHFENA;
+
+       psrctl &= ~TXGBE_PSRCTL_ADHF12_MASK;
+       psrctl |= TXGBE_PSRCTL_ADHF12(hw->mac.mc_filter_type);
+       wr32(hw, TXGBE_PSRCTL, psrctl);
+
+       return 0;
+}
+
+uint32_t
+txgbe_convert_vm_rx_mask_to_val(uint16_t rx_mask, uint32_t orig_val)
+{
+       uint32_t new_val = orig_val;
+
+       if (rx_mask & ETH_VMDQ_ACCEPT_UNTAG)
+               new_val |= TXGBE_POOLETHCTL_UTA;
+       if (rx_mask & ETH_VMDQ_ACCEPT_HASH_MC)
+               new_val |= TXGBE_POOLETHCTL_MCHA;
+       if (rx_mask & ETH_VMDQ_ACCEPT_HASH_UC)
+               new_val |= TXGBE_POOLETHCTL_UCHA;
+       if (rx_mask & ETH_VMDQ_ACCEPT_BROADCAST)
+               new_val |= TXGBE_POOLETHCTL_BCA;
+       if (rx_mask & ETH_VMDQ_ACCEPT_MULTICAST)
+               new_val |= TXGBE_POOLETHCTL_MCP;
+
+       return new_val;
+}
+
+static int
+txgbe_dev_rx_queue_intr_enable(struct rte_eth_dev *dev, uint16_t queue_id)
+{
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+       struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
+       uint32_t mask;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       if (queue_id < 32) {
+               mask = rd32(hw, TXGBE_IMS(0));
+               mask &= (1 << queue_id);
+               wr32(hw, TXGBE_IMS(0), mask);
+       } else if (queue_id < 64) {
+               mask = rd32(hw, TXGBE_IMS(1));
+               mask &= (1 << (queue_id - 32));
+               wr32(hw, TXGBE_IMS(1), mask);
+       }
+       rte_intr_enable(intr_handle);
+
+       return 0;
+}
+
+static int
+txgbe_dev_rx_queue_intr_disable(struct rte_eth_dev *dev, uint16_t queue_id)
+{
+       uint32_t mask;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       if (queue_id < 32) {
+               mask = rd32(hw, TXGBE_IMS(0));
+               mask &= ~(1 << queue_id);
+               wr32(hw, TXGBE_IMS(0), mask);
+       } else if (queue_id < 64) {
+               mask = rd32(hw, TXGBE_IMS(1));
+               mask &= ~(1 << (queue_id - 32));
+               wr32(hw, TXGBE_IMS(1), mask);
+       }
+
+       return 0;
+}
+
+/**
+ * set the IVAR registers, mapping interrupt causes to vectors
+ * @param hw
+ *  pointer to txgbe_hw struct
+ * @direction
+ *  0 for Rx, 1 for Tx, -1 for other causes
+ * @queue
+ *  queue to map the corresponding interrupt to
+ * @msix_vector
+ *  the vector to map to the corresponding queue
+ */
+void
+txgbe_set_ivar_map(struct txgbe_hw *hw, int8_t direction,
+                  uint8_t queue, uint8_t msix_vector)
+{
+       uint32_t tmp, idx;
+
+       if (direction == -1) {
+               /* other causes */
+               msix_vector |= TXGBE_IVARMISC_VLD;
+               idx = 0;
+               tmp = rd32(hw, TXGBE_IVARMISC);
+               tmp &= ~(0xFF << idx);
+               tmp |= (msix_vector << idx);
+               wr32(hw, TXGBE_IVARMISC, tmp);
+       } else {
+               /* rx or tx causes */
+               /* Workround for ICR lost */
+               idx = ((16 * (queue & 1)) + (8 * direction));
+               tmp = rd32(hw, TXGBE_IVAR(queue >> 1));
+               tmp &= ~(0xFF << idx);
+               tmp |= (msix_vector << idx);
+               wr32(hw, TXGBE_IVAR(queue >> 1), tmp);
+       }
+}
+
+/**
+ * Sets up the hardware to properly generate MSI-X interrupts
+ * @hw
+ *  board private structure
+ */
+static void
+txgbe_configure_msix(struct rte_eth_dev *dev)
+{
+       struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(dev);
+       struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t queue_id, base = TXGBE_MISC_VEC_ID;
+       uint32_t vec = TXGBE_MISC_VEC_ID;
+       uint32_t gpie;
+
+       /* won't configure msix register if no mapping is done
+        * between intr vector and event fd
+        * but if misx has been enabled already, need to configure
+        * auto clean, auto mask and throttling.
+        */
+       gpie = rd32(hw, TXGBE_GPIE);
+       if (!rte_intr_dp_is_en(intr_handle) &&
+           !(gpie & TXGBE_GPIE_MSIX))
+               return;
+
+       if (rte_intr_allow_others(intr_handle)) {
+               base = TXGBE_RX_VEC_START;
+               vec = base;
+       }
+
+       /* setup GPIE for MSI-x mode */
+       gpie = rd32(hw, TXGBE_GPIE);
+       gpie |= TXGBE_GPIE_MSIX;
+       wr32(hw, TXGBE_GPIE, gpie);
+
+       /* Populate the IVAR table and set the ITR values to the
+        * corresponding register.
+        */
+       if (rte_intr_dp_is_en(intr_handle)) {
+               for (queue_id = 0; queue_id < dev->data->nb_rx_queues;
+                       queue_id++) {
+                       /* by default, 1:1 mapping */
+                       txgbe_set_ivar_map(hw, 0, queue_id, vec);
+                       intr_handle->intr_vec[queue_id] = vec;
+                       if (vec < base + intr_handle->nb_efd - 1)
+                               vec++;
+               }
+
+               txgbe_set_ivar_map(hw, -1, 1, TXGBE_MISC_VEC_ID);
+       }
+       wr32(hw, TXGBE_ITR(TXGBE_MISC_VEC_ID),
+                       TXGBE_ITR_IVAL_10G(TXGBE_QUEUE_ITR_INTERVAL_DEFAULT)
+                       | TXGBE_ITR_WRDSA);
+}
+
+int
+txgbe_set_queue_rate_limit(struct rte_eth_dev *dev,
+                          uint16_t queue_idx, uint16_t tx_rate)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t bcnrc_val;
+
+       if (queue_idx >= hw->mac.max_tx_queues)
+               return -EINVAL;
+
+       if (tx_rate != 0) {
+               bcnrc_val = TXGBE_ARBTXRATE_MAX(tx_rate);
+               bcnrc_val |= TXGBE_ARBTXRATE_MIN(tx_rate / 2);
+       } else {
+               bcnrc_val = 0;
+       }
+
+       /*
+        * Set global transmit compensation time to the MMW_SIZE in ARBTXMMW
+        * register. MMW_SIZE=0x014 if 9728-byte jumbo is supported.
+        */
+       wr32(hw, TXGBE_ARBTXMMW, 0x14);
+
+       /* Set ARBTXRATE of queue X */
+       wr32(hw, TXGBE_ARBPOOLIDX, queue_idx);
+       wr32(hw, TXGBE_ARBTXRATE, bcnrc_val);
+       txgbe_flush(hw);
+
+       return 0;
+}
+
+int
+txgbe_syn_filter_set(struct rte_eth_dev *dev,
+                       struct rte_eth_syn_filter *filter,
+                       bool add)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       uint32_t syn_info;
+       uint32_t synqf;
+
+       if (filter->queue >= TXGBE_MAX_RX_QUEUE_NUM)
+               return -EINVAL;
+
+       syn_info = filter_info->syn_info;
+
+       if (add) {
+               if (syn_info & TXGBE_SYNCLS_ENA)
+                       return -EINVAL;
+               synqf = (uint32_t)TXGBE_SYNCLS_QPID(filter->queue);
+               synqf |= TXGBE_SYNCLS_ENA;
+
+               if (filter->hig_pri)
+                       synqf |= TXGBE_SYNCLS_HIPRIO;
+               else
+                       synqf &= ~TXGBE_SYNCLS_HIPRIO;
+       } else {
+               synqf = rd32(hw, TXGBE_SYNCLS);
+               if (!(syn_info & TXGBE_SYNCLS_ENA))
+                       return -ENOENT;
+               synqf &= ~(TXGBE_SYNCLS_QPID_MASK | TXGBE_SYNCLS_ENA);
+       }
+
+       filter_info->syn_info = synqf;
+       wr32(hw, TXGBE_SYNCLS, synqf);
+       txgbe_flush(hw);
+       return 0;
+}
+
+static inline enum txgbe_5tuple_protocol
+convert_protocol_type(uint8_t protocol_value)
+{
+       if (protocol_value == IPPROTO_TCP)
+               return TXGBE_5TF_PROT_TCP;
+       else if (protocol_value == IPPROTO_UDP)
+               return TXGBE_5TF_PROT_UDP;
+       else if (protocol_value == IPPROTO_SCTP)
+               return TXGBE_5TF_PROT_SCTP;
+       else
+               return TXGBE_5TF_PROT_NONE;
+}
+
+/* inject a 5-tuple filter to HW */
+static inline void
+txgbe_inject_5tuple_filter(struct rte_eth_dev *dev,
+                          struct txgbe_5tuple_filter *filter)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       int i;
+       uint32_t ftqf, sdpqf;
+       uint32_t l34timir = 0;
+       uint32_t mask = TXGBE_5TFCTL0_MASK;
+
+       i = filter->index;
+       sdpqf = TXGBE_5TFPORT_DST(be_to_le16(filter->filter_info.dst_port));
+       sdpqf |= TXGBE_5TFPORT_SRC(be_to_le16(filter->filter_info.src_port));
+
+       ftqf = TXGBE_5TFCTL0_PROTO(filter->filter_info.proto);
+       ftqf |= TXGBE_5TFCTL0_PRI(filter->filter_info.priority);
+       if (filter->filter_info.src_ip_mask == 0) /* 0 means compare. */
+               mask &= ~TXGBE_5TFCTL0_MSADDR;
+       if (filter->filter_info.dst_ip_mask == 0)
+               mask &= ~TXGBE_5TFCTL0_MDADDR;
+       if (filter->filter_info.src_port_mask == 0)
+               mask &= ~TXGBE_5TFCTL0_MSPORT;
+       if (filter->filter_info.dst_port_mask == 0)
+               mask &= ~TXGBE_5TFCTL0_MDPORT;
+       if (filter->filter_info.proto_mask == 0)
+               mask &= ~TXGBE_5TFCTL0_MPROTO;
+       ftqf |= mask;
+       ftqf |= TXGBE_5TFCTL0_MPOOL;
+       ftqf |= TXGBE_5TFCTL0_ENA;
+
+       wr32(hw, TXGBE_5TFDADDR(i), be_to_le32(filter->filter_info.dst_ip));
+       wr32(hw, TXGBE_5TFSADDR(i), be_to_le32(filter->filter_info.src_ip));
+       wr32(hw, TXGBE_5TFPORT(i), sdpqf);
+       wr32(hw, TXGBE_5TFCTL0(i), ftqf);
+
+       l34timir |= TXGBE_5TFCTL1_QP(filter->queue);
+       wr32(hw, TXGBE_5TFCTL1(i), l34timir);
+}
+
+/*
+ * add a 5tuple filter
+ *
+ * @param
+ * dev: Pointer to struct rte_eth_dev.
+ * index: the index the filter allocates.
+ * filter: pointer to the filter that will be added.
+ * rx_queue: the queue id the filter assigned to.
+ *
+ * @return
+ *    - On success, zero.
+ *    - On failure, a negative value.
+ */
+static int
+txgbe_add_5tuple_filter(struct rte_eth_dev *dev,
+                       struct txgbe_5tuple_filter *filter)
+{
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       int i, idx, shift;
+
+       /*
+        * look for an unused 5tuple filter index,
+        * and insert the filter to list.
+        */
+       for (i = 0; i < TXGBE_MAX_FTQF_FILTERS; i++) {
+               idx = i / (sizeof(uint32_t) * NBBY);
+               shift = i % (sizeof(uint32_t) * NBBY);
+               if (!(filter_info->fivetuple_mask[idx] & (1 << shift))) {
+                       filter_info->fivetuple_mask[idx] |= 1 << shift;
+                       filter->index = i;
+                       TAILQ_INSERT_TAIL(&filter_info->fivetuple_list,
+                                         filter,
+                                         entries);
+                       break;
+               }
+       }
+       if (i >= TXGBE_MAX_FTQF_FILTERS) {
+               PMD_DRV_LOG(ERR, "5tuple filters are full.");
+               return -ENOSYS;
+       }
+
+       txgbe_inject_5tuple_filter(dev, filter);
+
+       return 0;
+}
+
+/*
+ * remove a 5tuple filter
+ *
+ * @param
+ * dev: Pointer to struct rte_eth_dev.
+ * filter: the pointer of the filter will be removed.
+ */
+static void
+txgbe_remove_5tuple_filter(struct rte_eth_dev *dev,
+                       struct txgbe_5tuple_filter *filter)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       uint16_t index = filter->index;
+
+       filter_info->fivetuple_mask[index / (sizeof(uint32_t) * NBBY)] &=
+                               ~(1 << (index % (sizeof(uint32_t) * NBBY)));
+       TAILQ_REMOVE(&filter_info->fivetuple_list, filter, entries);
+       rte_free(filter);
+
+       wr32(hw, TXGBE_5TFDADDR(index), 0);
+       wr32(hw, TXGBE_5TFSADDR(index), 0);
+       wr32(hw, TXGBE_5TFPORT(index), 0);
+       wr32(hw, TXGBE_5TFCTL0(index), 0);
+       wr32(hw, TXGBE_5TFCTL1(index), 0);
+}
+
+static inline struct txgbe_5tuple_filter *
+txgbe_5tuple_filter_lookup(struct txgbe_5tuple_filter_list *filter_list,
+                       struct txgbe_5tuple_filter_info *key)
+{
+       struct txgbe_5tuple_filter *it;
+
+       TAILQ_FOREACH(it, filter_list, entries) {
+               if (memcmp(key, &it->filter_info,
+                       sizeof(struct txgbe_5tuple_filter_info)) == 0) {
+                       return it;
+               }
+       }
+       return NULL;
+}
+
+/* translate elements in struct rte_eth_ntuple_filter
+ * to struct txgbe_5tuple_filter_info
+ */
+static inline int
+ntuple_filter_to_5tuple(struct rte_eth_ntuple_filter *filter,
+                       struct txgbe_5tuple_filter_info *filter_info)
+{
+       if (filter->queue >= TXGBE_MAX_RX_QUEUE_NUM ||
+               filter->priority > TXGBE_5TUPLE_MAX_PRI ||
+               filter->priority < TXGBE_5TUPLE_MIN_PRI)
+               return -EINVAL;
+
+       switch (filter->dst_ip_mask) {
+       case UINT32_MAX:
+               filter_info->dst_ip_mask = 0;
+               filter_info->dst_ip = filter->dst_ip;
+               break;
+       case 0:
+               filter_info->dst_ip_mask = 1;
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "invalid dst_ip mask.");
+               return -EINVAL;
+       }
+
+       switch (filter->src_ip_mask) {
+       case UINT32_MAX:
+               filter_info->src_ip_mask = 0;
+               filter_info->src_ip = filter->src_ip;
+               break;
+       case 0:
+               filter_info->src_ip_mask = 1;
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "invalid src_ip mask.");
+               return -EINVAL;
+       }
+
+       switch (filter->dst_port_mask) {
+       case UINT16_MAX:
+               filter_info->dst_port_mask = 0;
+               filter_info->dst_port = filter->dst_port;
+               break;
+       case 0:
+               filter_info->dst_port_mask = 1;
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "invalid dst_port mask.");
+               return -EINVAL;
+       }
+
+       switch (filter->src_port_mask) {
+       case UINT16_MAX:
+               filter_info->src_port_mask = 0;
+               filter_info->src_port = filter->src_port;
+               break;
+       case 0:
+               filter_info->src_port_mask = 1;
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "invalid src_port mask.");
+               return -EINVAL;
+       }
+
+       switch (filter->proto_mask) {
+       case UINT8_MAX:
+               filter_info->proto_mask = 0;
+               filter_info->proto =
+                       convert_protocol_type(filter->proto);
+               break;
+       case 0:
+               filter_info->proto_mask = 1;
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "invalid protocol mask.");
+               return -EINVAL;
+       }
+
+       filter_info->priority = (uint8_t)filter->priority;
+       return 0;
+}
+
+/*
+ * add or delete a ntuple filter
+ *
+ * @param
+ * dev: Pointer to struct rte_eth_dev.
+ * ntuple_filter: Pointer to struct rte_eth_ntuple_filter
+ * add: if true, add filter, if false, remove filter
+ *
+ * @return
+ *    - On success, zero.
+ *    - On failure, a negative value.
+ */
+int
+txgbe_add_del_ntuple_filter(struct rte_eth_dev *dev,
+                       struct rte_eth_ntuple_filter *ntuple_filter,
+                       bool add)
+{
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       struct txgbe_5tuple_filter_info filter_5tuple;
+       struct txgbe_5tuple_filter *filter;
+       int ret;
+
+       if (ntuple_filter->flags != RTE_5TUPLE_FLAGS) {
+               PMD_DRV_LOG(ERR, "only 5tuple is supported.");
+               return -EINVAL;
+       }
+
+       memset(&filter_5tuple, 0, sizeof(struct txgbe_5tuple_filter_info));
+       ret = ntuple_filter_to_5tuple(ntuple_filter, &filter_5tuple);
+       if (ret < 0)
+               return ret;
+
+       filter = txgbe_5tuple_filter_lookup(&filter_info->fivetuple_list,
+                                        &filter_5tuple);
+       if (filter != NULL && add) {
+               PMD_DRV_LOG(ERR, "filter exists.");
+               return -EEXIST;
+       }
+       if (filter == NULL && !add) {
+               PMD_DRV_LOG(ERR, "filter doesn't exist.");
+               return -ENOENT;
+       }
+
+       if (add) {
+               filter = rte_zmalloc("txgbe_5tuple_filter",
+                               sizeof(struct txgbe_5tuple_filter), 0);
+               if (filter == NULL)
+                       return -ENOMEM;
+               rte_memcpy(&filter->filter_info,
+                                &filter_5tuple,
+                                sizeof(struct txgbe_5tuple_filter_info));
+               filter->queue = ntuple_filter->queue;
+               ret = txgbe_add_5tuple_filter(dev, filter);
+               if (ret < 0) {
+                       rte_free(filter);
+                       return ret;
+               }
+       } else {
+               txgbe_remove_5tuple_filter(dev, filter);
+       }
+
+       return 0;
+}
+
+int
+txgbe_add_del_ethertype_filter(struct rte_eth_dev *dev,
+                       struct rte_eth_ethertype_filter *filter,
+                       bool add)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       uint32_t etqf = 0;
+       uint32_t etqs = 0;
+       int ret;
+       struct txgbe_ethertype_filter ethertype_filter;
+
+       if (filter->queue >= TXGBE_MAX_RX_QUEUE_NUM)
+               return -EINVAL;
+
+       if (filter->ether_type == RTE_ETHER_TYPE_IPV4 ||
+           filter->ether_type == RTE_ETHER_TYPE_IPV6) {
+               PMD_DRV_LOG(ERR, "unsupported ether_type(0x%04x) in"
+                       " ethertype filter.", filter->ether_type);
+               return -EINVAL;
+       }
+
+       if (filter->flags & RTE_ETHTYPE_FLAGS_MAC) {
+               PMD_DRV_LOG(ERR, "mac compare is unsupported.");
+               return -EINVAL;
+       }
+       if (filter->flags & RTE_ETHTYPE_FLAGS_DROP) {
+               PMD_DRV_LOG(ERR, "drop option is unsupported.");
+               return -EINVAL;
+       }
+
+       ret = txgbe_ethertype_filter_lookup(filter_info, filter->ether_type);
+       if (ret >= 0 && add) {
+               PMD_DRV_LOG(ERR, "ethertype (0x%04x) filter exists.",
+                           filter->ether_type);
+               return -EEXIST;
+       }
+       if (ret < 0 && !add) {
+               PMD_DRV_LOG(ERR, "ethertype (0x%04x) filter doesn't exist.",
+                           filter->ether_type);
+               return -ENOENT;
+       }
+
+       if (add) {
+               etqf = TXGBE_ETFLT_ENA;
+               etqf |= TXGBE_ETFLT_ETID(filter->ether_type);
+               etqs |= TXGBE_ETCLS_QPID(filter->queue);
+               etqs |= TXGBE_ETCLS_QENA;
+
+               ethertype_filter.ethertype = filter->ether_type;
+               ethertype_filter.etqf = etqf;
+               ethertype_filter.etqs = etqs;
+               ethertype_filter.conf = FALSE;
+               ret = txgbe_ethertype_filter_insert(filter_info,
+                                                   &ethertype_filter);
+               if (ret < 0) {
+                       PMD_DRV_LOG(ERR, "ethertype filters are full.");
+                       return -ENOSPC;
+               }
+       } else {
+               ret = txgbe_ethertype_filter_remove(filter_info, (uint8_t)ret);
+               if (ret < 0)
+                       return -ENOSYS;
+       }
+       wr32(hw, TXGBE_ETFLT(ret), etqf);
+       wr32(hw, TXGBE_ETCLS(ret), etqs);
+       txgbe_flush(hw);
+
+       return 0;
+}
+
+static int
+txgbe_dev_filter_ctrl(__rte_unused struct rte_eth_dev *dev,
+                    enum rte_filter_type filter_type,
+                    enum rte_filter_op filter_op,
+                    void *arg)
+{
+       int ret = 0;
+
+       switch (filter_type) {
+       case RTE_ETH_FILTER_GENERIC:
+               if (filter_op != RTE_ETH_FILTER_GET)
+                       return -EINVAL;
+               *(const void **)arg = &txgbe_flow_ops;
+               break;
+       default:
+               PMD_DRV_LOG(WARNING, "Filter type (%d) not supported",
+                                                       filter_type);
+               ret = -EINVAL;
+               break;
+       }
+
+       return ret;
+}
+
+static u8 *
+txgbe_dev_addr_list_itr(__rte_unused struct txgbe_hw *hw,
+                       u8 **mc_addr_ptr, u32 *vmdq)
+{
+       u8 *mc_addr;
+
+       *vmdq = 0;
+       mc_addr = *mc_addr_ptr;
+       *mc_addr_ptr = (mc_addr + sizeof(struct rte_ether_addr));
+       return mc_addr;
+}
+
+int
+txgbe_dev_set_mc_addr_list(struct rte_eth_dev *dev,
+                         struct rte_ether_addr *mc_addr_set,
+                         uint32_t nb_mc_addr)
+{
+       struct txgbe_hw *hw;
+       u8 *mc_addr_list;
+
+       hw = TXGBE_DEV_HW(dev);
+       mc_addr_list = (u8 *)mc_addr_set;
+       return txgbe_update_mc_addr_list(hw, mc_addr_list, nb_mc_addr,
+                                        txgbe_dev_addr_list_itr, TRUE);
+}
+
+static uint64_t
+txgbe_read_systime_cyclecounter(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint64_t systime_cycles;
+
+       systime_cycles = (uint64_t)rd32(hw, TXGBE_TSTIMEL);
+       systime_cycles |= (uint64_t)rd32(hw, TXGBE_TSTIMEH) << 32;
+
+       return systime_cycles;
+}
+
+static uint64_t
+txgbe_read_rx_tstamp_cyclecounter(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint64_t rx_tstamp_cycles;
+
+       /* TSRXSTMPL stores ns and TSRXSTMPH stores seconds. */
+       rx_tstamp_cycles = (uint64_t)rd32(hw, TXGBE_TSRXSTMPL);
+       rx_tstamp_cycles |= (uint64_t)rd32(hw, TXGBE_TSRXSTMPH) << 32;
+
+       return rx_tstamp_cycles;
+}
+
+static uint64_t
+txgbe_read_tx_tstamp_cyclecounter(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint64_t tx_tstamp_cycles;
+
+       /* TSTXSTMPL stores ns and TSTXSTMPH stores seconds. */
+       tx_tstamp_cycles = (uint64_t)rd32(hw, TXGBE_TSTXSTMPL);
+       tx_tstamp_cycles |= (uint64_t)rd32(hw, TXGBE_TSTXSTMPH) << 32;
+
+       return tx_tstamp_cycles;
+}
+
+static void
+txgbe_start_timecounters(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+       struct rte_eth_link link;
+       uint32_t incval = 0;
+       uint32_t shift = 0;
+
+       /* Get current link speed. */
+       txgbe_dev_link_update(dev, 1);
+       rte_eth_linkstatus_get(dev, &link);
+
+       switch (link.link_speed) {
+       case ETH_SPEED_NUM_100M:
+               incval = TXGBE_INCVAL_100;
+               shift = TXGBE_INCVAL_SHIFT_100;
+               break;
+       case ETH_SPEED_NUM_1G:
+               incval = TXGBE_INCVAL_1GB;
+               shift = TXGBE_INCVAL_SHIFT_1GB;
+               break;
+       case ETH_SPEED_NUM_10G:
+       default:
+               incval = TXGBE_INCVAL_10GB;
+               shift = TXGBE_INCVAL_SHIFT_10GB;
+               break;
+       }
+
+       wr32(hw, TXGBE_TSTIMEINC, TXGBE_TSTIMEINC_VP(incval, 2));
+
+       memset(&adapter->systime_tc, 0, sizeof(struct rte_timecounter));
+       memset(&adapter->rx_tstamp_tc, 0, sizeof(struct rte_timecounter));
+       memset(&adapter->tx_tstamp_tc, 0, sizeof(struct rte_timecounter));
+
+       adapter->systime_tc.cc_mask = TXGBE_CYCLECOUNTER_MASK;
+       adapter->systime_tc.cc_shift = shift;
+       adapter->systime_tc.nsec_mask = (1ULL << shift) - 1;
+
+       adapter->rx_tstamp_tc.cc_mask = TXGBE_CYCLECOUNTER_MASK;
+       adapter->rx_tstamp_tc.cc_shift = shift;
+       adapter->rx_tstamp_tc.nsec_mask = (1ULL << shift) - 1;
+
+       adapter->tx_tstamp_tc.cc_mask = TXGBE_CYCLECOUNTER_MASK;
+       adapter->tx_tstamp_tc.cc_shift = shift;
+       adapter->tx_tstamp_tc.nsec_mask = (1ULL << shift) - 1;
+}
+
+static int
+txgbe_timesync_adjust_time(struct rte_eth_dev *dev, int64_t delta)
+{
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+
+       adapter->systime_tc.nsec += delta;
+       adapter->rx_tstamp_tc.nsec += delta;
+       adapter->tx_tstamp_tc.nsec += delta;
+
+       return 0;
+}
+
+static int
+txgbe_timesync_write_time(struct rte_eth_dev *dev, const struct timespec *ts)
+{
+       uint64_t ns;
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+
+       ns = rte_timespec_to_ns(ts);
+       /* Set the timecounters to a new value. */
+       adapter->systime_tc.nsec = ns;
+       adapter->rx_tstamp_tc.nsec = ns;
+       adapter->tx_tstamp_tc.nsec = ns;
+
+       return 0;
+}
+
+static int
+txgbe_timesync_read_time(struct rte_eth_dev *dev, struct timespec *ts)
+{
+       uint64_t ns, systime_cycles;
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+
+       systime_cycles = txgbe_read_systime_cyclecounter(dev);
+       ns = rte_timecounter_update(&adapter->systime_tc, systime_cycles);
+       *ts = rte_ns_to_timespec(ns);
+
+       return 0;
+}
+
+static int
+txgbe_timesync_enable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t tsync_ctl;
+
+       /* Stop the timesync system time. */
+       wr32(hw, TXGBE_TSTIMEINC, 0x0);
+       /* Reset the timesync system time value. */
+       wr32(hw, TXGBE_TSTIMEL, 0x0);
+       wr32(hw, TXGBE_TSTIMEH, 0x0);
+
+       txgbe_start_timecounters(dev);
+
+       /* Enable L2 filtering of IEEE1588/802.1AS Ethernet frame types. */
+       wr32(hw, TXGBE_ETFLT(TXGBE_ETF_ID_1588),
+               RTE_ETHER_TYPE_1588 | TXGBE_ETFLT_ENA | TXGBE_ETFLT_1588);
+
+       /* Enable timestamping of received PTP packets. */
+       tsync_ctl = rd32(hw, TXGBE_TSRXCTL);
+       tsync_ctl |= TXGBE_TSRXCTL_ENA;
+       wr32(hw, TXGBE_TSRXCTL, tsync_ctl);
+
+       /* Enable timestamping of transmitted PTP packets. */
+       tsync_ctl = rd32(hw, TXGBE_TSTXCTL);
+       tsync_ctl |= TXGBE_TSTXCTL_ENA;
+       wr32(hw, TXGBE_TSTXCTL, tsync_ctl);
+
+       txgbe_flush(hw);
+
+       return 0;
+}
+
+static int
+txgbe_timesync_disable(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t tsync_ctl;
+
+       /* Disable timestamping of transmitted PTP packets. */
+       tsync_ctl = rd32(hw, TXGBE_TSTXCTL);
+       tsync_ctl &= ~TXGBE_TSTXCTL_ENA;
+       wr32(hw, TXGBE_TSTXCTL, tsync_ctl);
+
+       /* Disable timestamping of received PTP packets. */
+       tsync_ctl = rd32(hw, TXGBE_TSRXCTL);
+       tsync_ctl &= ~TXGBE_TSRXCTL_ENA;
+       wr32(hw, TXGBE_TSRXCTL, tsync_ctl);
+
+       /* Disable L2 filtering of IEEE1588/802.1AS Ethernet frame types. */
+       wr32(hw, TXGBE_ETFLT(TXGBE_ETF_ID_1588), 0);
+
+       /* Stop incrementating the System Time registers. */
+       wr32(hw, TXGBE_TSTIMEINC, 0);
+
+       return 0;
+}
+
+static int
+txgbe_timesync_read_rx_timestamp(struct rte_eth_dev *dev,
+                                struct timespec *timestamp,
+                                uint32_t flags __rte_unused)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+       uint32_t tsync_rxctl;
+       uint64_t rx_tstamp_cycles;
+       uint64_t ns;
+
+       tsync_rxctl = rd32(hw, TXGBE_TSRXCTL);
+       if ((tsync_rxctl & TXGBE_TSRXCTL_VLD) == 0)
+               return -EINVAL;
+
+       rx_tstamp_cycles = txgbe_read_rx_tstamp_cyclecounter(dev);
+       ns = rte_timecounter_update(&adapter->rx_tstamp_tc, rx_tstamp_cycles);
+       *timestamp = rte_ns_to_timespec(ns);
+
+       return  0;
+}
+
+static int
+txgbe_timesync_read_tx_timestamp(struct rte_eth_dev *dev,
+                                struct timespec *timestamp)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_adapter *adapter = TXGBE_DEV_ADAPTER(dev);
+       uint32_t tsync_txctl;
+       uint64_t tx_tstamp_cycles;
+       uint64_t ns;
+
+       tsync_txctl = rd32(hw, TXGBE_TSTXCTL);
+       if ((tsync_txctl & TXGBE_TSTXCTL_VLD) == 0)
+               return -EINVAL;
+
+       tx_tstamp_cycles = txgbe_read_tx_tstamp_cyclecounter(dev);
+       ns = rte_timecounter_update(&adapter->tx_tstamp_tc, tx_tstamp_cycles);
+       *timestamp = rte_ns_to_timespec(ns);
+
+       return 0;
+}
+
+static int
+txgbe_get_reg_length(struct rte_eth_dev *dev __rte_unused)
+{
+       int count = 0;
+       int g_ind = 0;
+       const struct reg_info *reg_group;
+       const struct reg_info **reg_set = txgbe_regs_others;
+
+       while ((reg_group = reg_set[g_ind++]))
+               count += txgbe_regs_group_count(reg_group);
+
+       return count;
+}
+
+static int
+txgbe_get_regs(struct rte_eth_dev *dev,
+             struct rte_dev_reg_info *regs)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t *data = regs->data;
+       int g_ind = 0;
+       int count = 0;
+       const struct reg_info *reg_group;
+       const struct reg_info **reg_set = txgbe_regs_others;
+
+       if (data == NULL) {
+               regs->length = txgbe_get_reg_length(dev);
+               regs->width = sizeof(uint32_t);
+               return 0;
+       }
+
+       /* Support only full register dump */
+       if (regs->length == 0 ||
+           regs->length == (uint32_t)txgbe_get_reg_length(dev)) {
+               regs->version = hw->mac.type << 24 |
+                               hw->revision_id << 16 |
+                               hw->device_id;
+               while ((reg_group = reg_set[g_ind++]))
+                       count += txgbe_read_regs_group(dev, &data[count],
+                                                     reg_group);
+               return 0;
+       }
+
+       return -ENOTSUP;
+}
+
+static int
+txgbe_get_eeprom_length(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       /* Return unit is byte count */
+       return hw->rom.word_size * 2;
+}
+
+static int
+txgbe_get_eeprom(struct rte_eth_dev *dev,
+               struct rte_dev_eeprom_info *in_eeprom)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_rom_info *eeprom = &hw->rom;
+       uint16_t *data = in_eeprom->data;
+       int first, length;
+
+       first = in_eeprom->offset >> 1;
+       length = in_eeprom->length >> 1;
+       if (first > hw->rom.word_size ||
+           ((first + length) > hw->rom.word_size))
+               return -EINVAL;
+
+       in_eeprom->magic = hw->vendor_id | (hw->device_id << 16);
+
+       return eeprom->readw_buffer(hw, first, length, data);
+}
+
+static int
+txgbe_set_eeprom(struct rte_eth_dev *dev,
+               struct rte_dev_eeprom_info *in_eeprom)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_rom_info *eeprom = &hw->rom;
+       uint16_t *data = in_eeprom->data;
+       int first, length;
+
+       first = in_eeprom->offset >> 1;
+       length = in_eeprom->length >> 1;
+       if (first > hw->rom.word_size ||
+           ((first + length) > hw->rom.word_size))
+               return -EINVAL;
+
+       in_eeprom->magic = hw->vendor_id | (hw->device_id << 16);
+
+       return eeprom->writew_buffer(hw,  first, length, data);
+}
+
+static int
+txgbe_get_module_info(struct rte_eth_dev *dev,
+                     struct rte_eth_dev_module_info *modinfo)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t status;
+       uint8_t sff8472_rev, addr_mode;
+       bool page_swap = false;
+
+       /* Check whether we support SFF-8472 or not */
+       status = hw->phy.read_i2c_eeprom(hw,
+                                            TXGBE_SFF_SFF_8472_COMP,
+                                            &sff8472_rev);
+       if (status != 0)
+               return -EIO;
+
+       /* addressing mode is not supported */
+       status = hw->phy.read_i2c_eeprom(hw,
+                                            TXGBE_SFF_SFF_8472_SWAP,
+                                            &addr_mode);
+       if (status != 0)
+               return -EIO;
+
+       if (addr_mode & TXGBE_SFF_ADDRESSING_MODE) {
+               PMD_DRV_LOG(ERR,
+                           "Address change required to access page 0xA2, "
+                           "but not supported. Please report the module "
+                           "type to the driver maintainers.");
+               page_swap = true;
+       }
+
+       if (sff8472_rev == TXGBE_SFF_SFF_8472_UNSUP || page_swap) {
+               /* We have a SFP, but it does not support SFF-8472 */
+               modinfo->type = RTE_ETH_MODULE_SFF_8079;
+               modinfo->eeprom_len = RTE_ETH_MODULE_SFF_8079_LEN;
+       } else {
+               /* We have a SFP which supports a revision of SFF-8472. */
+               modinfo->type = RTE_ETH_MODULE_SFF_8472;
+               modinfo->eeprom_len = RTE_ETH_MODULE_SFF_8472_LEN;
+       }
+
+       return 0;
+}
+
+static int
+txgbe_get_module_eeprom(struct rte_eth_dev *dev,
+                       struct rte_dev_eeprom_info *info)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t status = TXGBE_ERR_PHY_ADDR_INVALID;
+       uint8_t databyte = 0xFF;
+       uint8_t *data = info->data;
+       uint32_t i = 0;
+
+       if (info->length == 0)
+               return -EINVAL;
+
+       for (i = info->offset; i < info->offset + info->length; i++) {
+               if (i < RTE_ETH_MODULE_SFF_8079_LEN)
+                       status = hw->phy.read_i2c_eeprom(hw, i, &databyte);
+               else
+                       status = hw->phy.read_i2c_sff8472(hw, i, &databyte);
+
+               if (status != 0)
+                       return -EIO;
+
+               data[i - info->offset] = databyte;
+       }
+
+       return 0;
+}
+
+bool
+txgbe_rss_update_sp(enum txgbe_mac_type mac_type)
+{
+       switch (mac_type) {
+       case txgbe_mac_raptor:
+               return 1;
+       default:
+               return 0;
+       }
+}
+
+static int
+txgbe_dev_get_dcb_info(struct rte_eth_dev *dev,
+                       struct rte_eth_dcb_info *dcb_info)
+{
+       struct txgbe_dcb_config *dcb_config = TXGBE_DEV_DCB_CONFIG(dev);
+       struct txgbe_dcb_tc_config *tc;
+       struct rte_eth_dcb_tc_queue_mapping *tc_queue;
+       uint8_t nb_tcs;
+       uint8_t i, j;
+
+       if (dev->data->dev_conf.rxmode.mq_mode & ETH_MQ_RX_DCB_FLAG)
+               dcb_info->nb_tcs = dcb_config->num_tcs.pg_tcs;
+       else
+               dcb_info->nb_tcs = 1;
+
+       tc_queue = &dcb_info->tc_queue;
+       nb_tcs = dcb_info->nb_tcs;
+
+       if (dcb_config->vt_mode) { /* vt is enabled */
+               struct rte_eth_vmdq_dcb_conf *vmdq_rx_conf =
+                               &dev->data->dev_conf.rx_adv_conf.vmdq_dcb_conf;
+               for (i = 0; i < ETH_DCB_NUM_USER_PRIORITIES; i++)
+                       dcb_info->prio_tc[i] = vmdq_rx_conf->dcb_tc[i];
+               if (RTE_ETH_DEV_SRIOV(dev).active > 0) {
+                       for (j = 0; j < nb_tcs; j++) {
+                               tc_queue->tc_rxq[0][j].base = j;
+                               tc_queue->tc_rxq[0][j].nb_queue = 1;
+                               tc_queue->tc_txq[0][j].base = j;
+                               tc_queue->tc_txq[0][j].nb_queue = 1;
+                       }
+               } else {
+                       for (i = 0; i < vmdq_rx_conf->nb_queue_pools; i++) {
+                               for (j = 0; j < nb_tcs; j++) {
+                                       tc_queue->tc_rxq[i][j].base =
+                                               i * nb_tcs + j;
+                                       tc_queue->tc_rxq[i][j].nb_queue = 1;
+                                       tc_queue->tc_txq[i][j].base =
+                                               i * nb_tcs + j;
+                                       tc_queue->tc_txq[i][j].nb_queue = 1;
+                               }
+                       }
+               }
+       } else { /* vt is disabled */
+               struct rte_eth_dcb_rx_conf *rx_conf =
+                               &dev->data->dev_conf.rx_adv_conf.dcb_rx_conf;
+               for (i = 0; i < ETH_DCB_NUM_USER_PRIORITIES; i++)
+                       dcb_info->prio_tc[i] = rx_conf->dcb_tc[i];
+               if (dcb_info->nb_tcs == ETH_4_TCS) {
+                       for (i = 0; i < dcb_info->nb_tcs; i++) {
+                               dcb_info->tc_queue.tc_rxq[0][i].base = i * 32;
+                               dcb_info->tc_queue.tc_rxq[0][i].nb_queue = 16;
+                       }
+                       dcb_info->tc_queue.tc_txq[0][0].base = 0;
+                       dcb_info->tc_queue.tc_txq[0][1].base = 64;
+                       dcb_info->tc_queue.tc_txq[0][2].base = 96;
+                       dcb_info->tc_queue.tc_txq[0][3].base = 112;
+                       dcb_info->tc_queue.tc_txq[0][0].nb_queue = 64;
+                       dcb_info->tc_queue.tc_txq[0][1].nb_queue = 32;
+                       dcb_info->tc_queue.tc_txq[0][2].nb_queue = 16;
+                       dcb_info->tc_queue.tc_txq[0][3].nb_queue = 16;
+               } else if (dcb_info->nb_tcs == ETH_8_TCS) {
+                       for (i = 0; i < dcb_info->nb_tcs; i++) {
+                               dcb_info->tc_queue.tc_rxq[0][i].base = i * 16;
+                               dcb_info->tc_queue.tc_rxq[0][i].nb_queue = 16;
+                       }
+                       dcb_info->tc_queue.tc_txq[0][0].base = 0;
+                       dcb_info->tc_queue.tc_txq[0][1].base = 32;
+                       dcb_info->tc_queue.tc_txq[0][2].base = 64;
+                       dcb_info->tc_queue.tc_txq[0][3].base = 80;
+                       dcb_info->tc_queue.tc_txq[0][4].base = 96;
+                       dcb_info->tc_queue.tc_txq[0][5].base = 104;
+                       dcb_info->tc_queue.tc_txq[0][6].base = 112;
+                       dcb_info->tc_queue.tc_txq[0][7].base = 120;
+                       dcb_info->tc_queue.tc_txq[0][0].nb_queue = 32;
+                       dcb_info->tc_queue.tc_txq[0][1].nb_queue = 32;
+                       dcb_info->tc_queue.tc_txq[0][2].nb_queue = 16;
+                       dcb_info->tc_queue.tc_txq[0][3].nb_queue = 16;
+                       dcb_info->tc_queue.tc_txq[0][4].nb_queue = 8;
+                       dcb_info->tc_queue.tc_txq[0][5].nb_queue = 8;
+                       dcb_info->tc_queue.tc_txq[0][6].nb_queue = 8;
+                       dcb_info->tc_queue.tc_txq[0][7].nb_queue = 8;
+               }
+       }
+       for (i = 0; i < dcb_info->nb_tcs; i++) {
+               tc = &dcb_config->tc_config[i];
+               dcb_info->tc_bws[i] = tc->path[TXGBE_DCB_TX_CONFIG].bwg_percent;
+       }
+       return 0;
+}
+
+/* Update e-tag ether type */
+static int
+txgbe_update_e_tag_eth_type(struct txgbe_hw *hw,
+                           uint16_t ether_type)
+{
+       uint32_t etag_etype;
+
+       etag_etype = rd32(hw, TXGBE_EXTAG);
+       etag_etype &= ~TXGBE_EXTAG_ETAG_MASK;
+       etag_etype |= ether_type;
+       wr32(hw, TXGBE_EXTAG, etag_etype);
+       txgbe_flush(hw);
+
+       return 0;
+}
+
+/* Enable e-tag tunnel */
+static int
+txgbe_e_tag_enable(struct txgbe_hw *hw)
+{
+       uint32_t etag_etype;
+
+       etag_etype = rd32(hw, TXGBE_PORTCTL);
+       etag_etype |= TXGBE_PORTCTL_ETAG;
+       wr32(hw, TXGBE_PORTCTL, etag_etype);
+       txgbe_flush(hw);
+
+       return 0;
+}
+
+static int
+txgbe_e_tag_filter_del(struct rte_eth_dev *dev,
+                      struct txgbe_l2_tunnel_conf  *l2_tunnel)
+{
+       int ret = 0;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t i, rar_entries;
+       uint32_t rar_low, rar_high;
+
+       rar_entries = hw->mac.num_rar_entries;
+
+       for (i = 1; i < rar_entries; i++) {
+               wr32(hw, TXGBE_ETHADDRIDX, i);
+               rar_high = rd32(hw, TXGBE_ETHADDRH);
+               rar_low  = rd32(hw, TXGBE_ETHADDRL);
+               if ((rar_high & TXGBE_ETHADDRH_VLD) &&
+                   (rar_high & TXGBE_ETHADDRH_ETAG) &&
+                   (TXGBE_ETHADDRL_ETAG(rar_low) ==
+                    l2_tunnel->tunnel_id)) {
+                       wr32(hw, TXGBE_ETHADDRL, 0);
+                       wr32(hw, TXGBE_ETHADDRH, 0);
+
+                       txgbe_clear_vmdq(hw, i, BIT_MASK32);
+
+                       return ret;
+               }
+       }
+
+       return ret;
+}
+
+static int
+txgbe_e_tag_filter_add(struct rte_eth_dev *dev,
+                      struct txgbe_l2_tunnel_conf *l2_tunnel)
+{
+       int ret = 0;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       uint32_t i, rar_entries;
+       uint32_t rar_low, rar_high;
+
+       /* One entry for one tunnel. Try to remove potential existing entry. */
+       txgbe_e_tag_filter_del(dev, l2_tunnel);
+
+       rar_entries = hw->mac.num_rar_entries;
+
+       for (i = 1; i < rar_entries; i++) {
+               wr32(hw, TXGBE_ETHADDRIDX, i);
+               rar_high = rd32(hw, TXGBE_ETHADDRH);
+               if (rar_high & TXGBE_ETHADDRH_VLD) {
+                       continue;
+               } else {
+                       txgbe_set_vmdq(hw, i, l2_tunnel->pool);
+                       rar_high = TXGBE_ETHADDRH_VLD | TXGBE_ETHADDRH_ETAG;
+                       rar_low = l2_tunnel->tunnel_id;
+
+                       wr32(hw, TXGBE_ETHADDRL, rar_low);
+                       wr32(hw, TXGBE_ETHADDRH, rar_high);
+
+                       return ret;
+               }
+       }
+
+       PMD_INIT_LOG(NOTICE, "The table of E-tag forwarding rule is full."
+                    " Please remove a rule before adding a new one.");
+       return -EINVAL;
+}
+
+static inline struct txgbe_l2_tn_filter *
+txgbe_l2_tn_filter_lookup(struct txgbe_l2_tn_info *l2_tn_info,
+                         struct txgbe_l2_tn_key *key)
+{
+       int ret;
+
+       ret = rte_hash_lookup(l2_tn_info->hash_handle, (const void *)key);
+       if (ret < 0)
+               return NULL;
+
+       return l2_tn_info->hash_map[ret];
+}
+
+static inline int
+txgbe_insert_l2_tn_filter(struct txgbe_l2_tn_info *l2_tn_info,
+                         struct txgbe_l2_tn_filter *l2_tn_filter)
+{
+       int ret;
+
+       ret = rte_hash_add_key(l2_tn_info->hash_handle,
+                              &l2_tn_filter->key);
+
+       if (ret < 0) {
+               PMD_DRV_LOG(ERR,
+                           "Failed to insert L2 tunnel filter"
+                           " to hash table %d!",
+                           ret);
+               return ret;
+       }
+
+       l2_tn_info->hash_map[ret] = l2_tn_filter;
+
+       TAILQ_INSERT_TAIL(&l2_tn_info->l2_tn_list, l2_tn_filter, entries);
+
+       return 0;
+}
+
+static inline int
+txgbe_remove_l2_tn_filter(struct txgbe_l2_tn_info *l2_tn_info,
+                         struct txgbe_l2_tn_key *key)
+{
+       int ret;
+       struct txgbe_l2_tn_filter *l2_tn_filter;
+
+       ret = rte_hash_del_key(l2_tn_info->hash_handle, key);
+
+       if (ret < 0) {
+               PMD_DRV_LOG(ERR,
+                           "No such L2 tunnel filter to delete %d!",
+                           ret);
+               return ret;
+       }
+
+       l2_tn_filter = l2_tn_info->hash_map[ret];
+       l2_tn_info->hash_map[ret] = NULL;
+
+       TAILQ_REMOVE(&l2_tn_info->l2_tn_list, l2_tn_filter, entries);
+       rte_free(l2_tn_filter);
+
+       return 0;
+}
+
+/* Add l2 tunnel filter */
+int
+txgbe_dev_l2_tunnel_filter_add(struct rte_eth_dev *dev,
+                              struct txgbe_l2_tunnel_conf *l2_tunnel,
+                              bool restore)
+{
+       int ret;
+       struct txgbe_l2_tn_info *l2_tn_info = TXGBE_DEV_L2_TN(dev);
+       struct txgbe_l2_tn_key key;
+       struct txgbe_l2_tn_filter *node;
+
+       if (!restore) {
+               key.l2_tn_type = l2_tunnel->l2_tunnel_type;
+               key.tn_id = l2_tunnel->tunnel_id;
+
+               node = txgbe_l2_tn_filter_lookup(l2_tn_info, &key);
+
+               if (node) {
+                       PMD_DRV_LOG(ERR,
+                                   "The L2 tunnel filter already exists!");
+                       return -EINVAL;
+               }
+
+               node = rte_zmalloc("txgbe_l2_tn",
+                                  sizeof(struct txgbe_l2_tn_filter),
+                                  0);
+               if (!node)
+                       return -ENOMEM;
+
+               rte_memcpy(&node->key,
+                                &key,
+                                sizeof(struct txgbe_l2_tn_key));
+               node->pool = l2_tunnel->pool;
+               ret = txgbe_insert_l2_tn_filter(l2_tn_info, node);
+               if (ret < 0) {
+                       rte_free(node);
+                       return ret;
+               }
+       }
+
+       switch (l2_tunnel->l2_tunnel_type) {
+       case RTE_L2_TUNNEL_TYPE_E_TAG:
+               ret = txgbe_e_tag_filter_add(dev, l2_tunnel);
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "Invalid tunnel type");
+               ret = -EINVAL;
+               break;
+       }
+
+       if (!restore && ret < 0)
+               (void)txgbe_remove_l2_tn_filter(l2_tn_info, &key);
+
+       return ret;
+}
+
+/* Delete l2 tunnel filter */
+int
+txgbe_dev_l2_tunnel_filter_del(struct rte_eth_dev *dev,
+                              struct txgbe_l2_tunnel_conf *l2_tunnel)
+{
+       int ret;
+       struct txgbe_l2_tn_info *l2_tn_info = TXGBE_DEV_L2_TN(dev);
+       struct txgbe_l2_tn_key key;
+
+       key.l2_tn_type = l2_tunnel->l2_tunnel_type;
+       key.tn_id = l2_tunnel->tunnel_id;
+       ret = txgbe_remove_l2_tn_filter(l2_tn_info, &key);
+       if (ret < 0)
+               return ret;
+
+       switch (l2_tunnel->l2_tunnel_type) {
+       case RTE_L2_TUNNEL_TYPE_E_TAG:
+               ret = txgbe_e_tag_filter_del(dev, l2_tunnel);
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "Invalid tunnel type");
+               ret = -EINVAL;
+               break;
+       }
+
+       return ret;
+}
+
+static int
+txgbe_e_tag_forwarding_en_dis(struct rte_eth_dev *dev, bool en)
+{
+       int ret = 0;
+       uint32_t ctrl;
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       ctrl = rd32(hw, TXGBE_POOLCTL);
+       ctrl &= ~TXGBE_POOLCTL_MODE_MASK;
+       if (en)
+               ctrl |= TXGBE_PSRPOOL_MODE_ETAG;
+       wr32(hw, TXGBE_POOLCTL, ctrl);
+
+       return ret;
+}
+
+/* Add UDP tunneling port */
+static int
+txgbe_dev_udp_tunnel_port_add(struct rte_eth_dev *dev,
+                             struct rte_eth_udp_tunnel *udp_tunnel)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       int ret = 0;
+
+       if (udp_tunnel == NULL)
+               return -EINVAL;
+
+       switch (udp_tunnel->prot_type) {
+       case RTE_TUNNEL_TYPE_VXLAN:
+               if (udp_tunnel->udp_port == 0) {
+                       PMD_DRV_LOG(ERR, "Add VxLAN port 0 is not allowed.");
+                       ret = -EINVAL;
+                       break;
+               }
+               wr32(hw, TXGBE_VXLANPORT, udp_tunnel->udp_port);
+               wr32(hw, TXGBE_VXLANPORTGPE, udp_tunnel->udp_port);
+               break;
+       case RTE_TUNNEL_TYPE_GENEVE:
+               if (udp_tunnel->udp_port == 0) {
+                       PMD_DRV_LOG(ERR, "Add Geneve port 0 is not allowed.");
+                       ret = -EINVAL;
+                       break;
+               }
+               wr32(hw, TXGBE_GENEVEPORT, udp_tunnel->udp_port);
+               break;
+       case RTE_TUNNEL_TYPE_TEREDO:
+               if (udp_tunnel->udp_port == 0) {
+                       PMD_DRV_LOG(ERR, "Add Teredo port 0 is not allowed.");
+                       ret = -EINVAL;
+                       break;
+               }
+               wr32(hw, TXGBE_TEREDOPORT, udp_tunnel->udp_port);
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "Invalid tunnel type");
+               ret = -EINVAL;
+               break;
+       }
+
+       txgbe_flush(hw);
+
+       return ret;
+}
+
+/* Remove UDP tunneling port */
+static int
+txgbe_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
+                             struct rte_eth_udp_tunnel *udp_tunnel)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       int ret = 0;
+       uint16_t cur_port;
+
+       if (udp_tunnel == NULL)
+               return -EINVAL;
+
+       switch (udp_tunnel->prot_type) {
+       case RTE_TUNNEL_TYPE_VXLAN:
+               cur_port = (uint16_t)rd32(hw, TXGBE_VXLANPORT);
+               if (cur_port != udp_tunnel->udp_port) {
+                       PMD_DRV_LOG(ERR, "Port %u does not exist.",
+                                       udp_tunnel->udp_port);
+                       ret = -EINVAL;
+                       break;
+               }
+               wr32(hw, TXGBE_VXLANPORT, 0);
+               wr32(hw, TXGBE_VXLANPORTGPE, 0);
+               break;
+       case RTE_TUNNEL_TYPE_GENEVE:
+               cur_port = (uint16_t)rd32(hw, TXGBE_GENEVEPORT);
+               if (cur_port != udp_tunnel->udp_port) {
+                       PMD_DRV_LOG(ERR, "Port %u does not exist.",
+                                       udp_tunnel->udp_port);
+                       ret = -EINVAL;
+                       break;
+               }
+               wr32(hw, TXGBE_GENEVEPORT, 0);
+               break;
+       case RTE_TUNNEL_TYPE_TEREDO:
+               cur_port = (uint16_t)rd32(hw, TXGBE_TEREDOPORT);
+               if (cur_port != udp_tunnel->udp_port) {
+                       PMD_DRV_LOG(ERR, "Port %u does not exist.",
+                                       udp_tunnel->udp_port);
+                       ret = -EINVAL;
+                       break;
+               }
+               wr32(hw, TXGBE_TEREDOPORT, 0);
+               break;
+       default:
+               PMD_DRV_LOG(ERR, "Invalid tunnel type");
+               ret = -EINVAL;
+               break;
+       }
+
+       txgbe_flush(hw);
+
+       return ret;
+}
+
+/* restore n-tuple filter */
+static inline void
+txgbe_ntuple_filter_restore(struct rte_eth_dev *dev)
+{
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       struct txgbe_5tuple_filter *node;
+
+       TAILQ_FOREACH(node, &filter_info->fivetuple_list, entries) {
+               txgbe_inject_5tuple_filter(dev, node);
+       }
+}
+
+/* restore ethernet type filter */
+static inline void
+txgbe_ethertype_filter_restore(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       int i;
+
+       for (i = 0; i < TXGBE_ETF_ID_MAX; i++) {
+               if (filter_info->ethertype_mask & (1 << i)) {
+                       wr32(hw, TXGBE_ETFLT(i),
+                                       filter_info->ethertype_filters[i].etqf);
+                       wr32(hw, TXGBE_ETCLS(i),
+                                       filter_info->ethertype_filters[i].etqs);
+                       txgbe_flush(hw);
+               }
+       }
+}
+
+/* restore SYN filter */
+static inline void
+txgbe_syn_filter_restore(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       uint32_t synqf;
+
+       synqf = filter_info->syn_info;
+
+       if (synqf & TXGBE_SYNCLS_ENA) {
+               wr32(hw, TXGBE_SYNCLS, synqf);
+               txgbe_flush(hw);
+       }
+}
+
+/* restore L2 tunnel filter */
+static inline void
+txgbe_l2_tn_filter_restore(struct rte_eth_dev *dev)
+{
+       struct txgbe_l2_tn_info *l2_tn_info = TXGBE_DEV_L2_TN(dev);
+       struct txgbe_l2_tn_filter *node;
+       struct txgbe_l2_tunnel_conf l2_tn_conf;
+
+       TAILQ_FOREACH(node, &l2_tn_info->l2_tn_list, entries) {
+               l2_tn_conf.l2_tunnel_type = node->key.l2_tn_type;
+               l2_tn_conf.tunnel_id      = node->key.tn_id;
+               l2_tn_conf.pool           = node->pool;
+               (void)txgbe_dev_l2_tunnel_filter_add(dev, &l2_tn_conf, TRUE);
+       }
+}
+
+/* restore rss filter */
+static inline void
+txgbe_rss_filter_restore(struct rte_eth_dev *dev)
+{
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+
+       if (filter_info->rss_info.conf.queue_num)
+               txgbe_config_rss_filter(dev,
+                       &filter_info->rss_info, TRUE);
+}
+
+static int
+txgbe_filter_restore(struct rte_eth_dev *dev)
+{
+       txgbe_ntuple_filter_restore(dev);
+       txgbe_ethertype_filter_restore(dev);
+       txgbe_syn_filter_restore(dev);
+       txgbe_fdir_filter_restore(dev);
+       txgbe_l2_tn_filter_restore(dev);
+       txgbe_rss_filter_restore(dev);
+
+       return 0;
+}
+
+static void
+txgbe_l2_tunnel_conf(struct rte_eth_dev *dev)
+{
+       struct txgbe_l2_tn_info *l2_tn_info = TXGBE_DEV_L2_TN(dev);
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+
+       if (l2_tn_info->e_tag_en)
+               (void)txgbe_e_tag_enable(hw);
+
+       if (l2_tn_info->e_tag_fwd_en)
+               (void)txgbe_e_tag_forwarding_en_dis(dev, 1);
+
+       (void)txgbe_update_e_tag_eth_type(hw, l2_tn_info->e_tag_ether_type);
+}
+
+/* remove all the n-tuple filters */
+void
+txgbe_clear_all_ntuple_filter(struct rte_eth_dev *dev)
+{
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       struct txgbe_5tuple_filter *p_5tuple;
+
+       while ((p_5tuple = TAILQ_FIRST(&filter_info->fivetuple_list)))
+               txgbe_remove_5tuple_filter(dev, p_5tuple);
+}
+
+/* remove all the ether type filters */
+void
+txgbe_clear_all_ethertype_filter(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+       int i;
+
+       for (i = 0; i < TXGBE_ETF_ID_MAX; i++) {
+               if (filter_info->ethertype_mask & (1 << i) &&
+                   !filter_info->ethertype_filters[i].conf) {
+                       (void)txgbe_ethertype_filter_remove(filter_info,
+                                                           (uint8_t)i);
+                       wr32(hw, TXGBE_ETFLT(i), 0);
+                       wr32(hw, TXGBE_ETCLS(i), 0);
+                       txgbe_flush(hw);
+               }
+       }
+}
+
+/* remove the SYN filter */
+void
+txgbe_clear_syn_filter(struct rte_eth_dev *dev)
+{
+       struct txgbe_hw *hw = TXGBE_DEV_HW(dev);
+       struct txgbe_filter_info *filter_info = TXGBE_DEV_FILTER(dev);
+
+       if (filter_info->syn_info & TXGBE_SYNCLS_ENA) {
+               filter_info->syn_info = 0;
+
+               wr32(hw, TXGBE_SYNCLS, 0);
+               txgbe_flush(hw);
+       }
+}
+
+/* remove all the L2 tunnel filters */
+int
+txgbe_clear_all_l2_tn_filter(struct rte_eth_dev *dev)
+{
+       struct txgbe_l2_tn_info *l2_tn_info = TXGBE_DEV_L2_TN(dev);
+       struct txgbe_l2_tn_filter *l2_tn_filter;
+       struct txgbe_l2_tunnel_conf l2_tn_conf;
+       int ret = 0;
+
+       while ((l2_tn_filter = TAILQ_FIRST(&l2_tn_info->l2_tn_list))) {
+               l2_tn_conf.l2_tunnel_type = l2_tn_filter->key.l2_tn_type;
+               l2_tn_conf.tunnel_id      = l2_tn_filter->key.tn_id;
+               l2_tn_conf.pool           = l2_tn_filter->pool;
+               ret = txgbe_dev_l2_tunnel_filter_del(dev, &l2_tn_conf);
+               if (ret < 0)
+                       return ret;
+       }
+
+       return 0;
 }
 
 static const struct eth_dev_ops txgbe_eth_dev_ops = {
+       .dev_configure              = txgbe_dev_configure,
        .dev_infos_get              = txgbe_dev_info_get,
+       .dev_start                  = txgbe_dev_start,
+       .dev_stop                   = txgbe_dev_stop,
+       .dev_set_link_up            = txgbe_dev_set_link_up,
+       .dev_set_link_down          = txgbe_dev_set_link_down,
+       .dev_close                  = txgbe_dev_close,
+       .dev_reset                  = txgbe_dev_reset,
+       .promiscuous_enable         = txgbe_dev_promiscuous_enable,
+       .promiscuous_disable        = txgbe_dev_promiscuous_disable,
+       .allmulticast_enable        = txgbe_dev_allmulticast_enable,
+       .allmulticast_disable       = txgbe_dev_allmulticast_disable,
+       .link_update                = txgbe_dev_link_update,
+       .stats_get                  = txgbe_dev_stats_get,
+       .xstats_get                 = txgbe_dev_xstats_get,
+       .xstats_get_by_id           = txgbe_dev_xstats_get_by_id,
+       .stats_reset                = txgbe_dev_stats_reset,
+       .xstats_reset               = txgbe_dev_xstats_reset,
+       .xstats_get_names           = txgbe_dev_xstats_get_names,
+       .xstats_get_names_by_id     = txgbe_dev_xstats_get_names_by_id,
+       .queue_stats_mapping_set    = txgbe_dev_queue_stats_mapping_set,
+       .fw_version_get             = txgbe_fw_version_get,
+       .dev_supported_ptypes_get   = txgbe_dev_supported_ptypes_get,
+       .mtu_set                    = txgbe_dev_mtu_set,
+       .vlan_filter_set            = txgbe_vlan_filter_set,
+       .vlan_tpid_set              = txgbe_vlan_tpid_set,
+       .vlan_offload_set           = txgbe_vlan_offload_set,
+       .vlan_strip_queue_set       = txgbe_vlan_strip_queue_set,
+       .rx_queue_start             = txgbe_dev_rx_queue_start,
+       .rx_queue_stop              = txgbe_dev_rx_queue_stop,
+       .tx_queue_start             = txgbe_dev_tx_queue_start,
+       .tx_queue_stop              = txgbe_dev_tx_queue_stop,
+       .rx_queue_setup             = txgbe_dev_rx_queue_setup,
+       .rx_queue_intr_enable       = txgbe_dev_rx_queue_intr_enable,
+       .rx_queue_intr_disable      = txgbe_dev_rx_queue_intr_disable,
+       .rx_queue_release           = txgbe_dev_rx_queue_release,
+       .tx_queue_setup             = txgbe_dev_tx_queue_setup,
+       .tx_queue_release           = txgbe_dev_tx_queue_release,
+       .dev_led_on                 = txgbe_dev_led_on,
+       .dev_led_off                = txgbe_dev_led_off,
+       .flow_ctrl_get              = txgbe_flow_ctrl_get,
+       .flow_ctrl_set              = txgbe_flow_ctrl_set,
+       .priority_flow_ctrl_set     = txgbe_priority_flow_ctrl_set,
+       .mac_addr_add               = txgbe_add_rar,
+       .mac_addr_remove            = txgbe_remove_rar,
+       .mac_addr_set               = txgbe_set_default_mac_addr,
+       .uc_hash_table_set          = txgbe_uc_hash_table_set,
+       .uc_all_hash_table_set      = txgbe_uc_all_hash_table_set,
+       .set_queue_rate_limit       = txgbe_set_queue_rate_limit,
+       .reta_update                = txgbe_dev_rss_reta_update,
+       .reta_query                 = txgbe_dev_rss_reta_query,
+       .rss_hash_update            = txgbe_dev_rss_hash_update,
+       .rss_hash_conf_get          = txgbe_dev_rss_hash_conf_get,
+       .filter_ctrl                = txgbe_dev_filter_ctrl,
+       .set_mc_addr_list           = txgbe_dev_set_mc_addr_list,
+       .rxq_info_get               = txgbe_rxq_info_get,
+       .txq_info_get               = txgbe_txq_info_get,
+       .timesync_enable            = txgbe_timesync_enable,
+       .timesync_disable           = txgbe_timesync_disable,
+       .timesync_read_rx_timestamp = txgbe_timesync_read_rx_timestamp,
+       .timesync_read_tx_timestamp = txgbe_timesync_read_tx_timestamp,
+       .get_reg                    = txgbe_get_regs,
+       .get_eeprom_length          = txgbe_get_eeprom_length,
+       .get_eeprom                 = txgbe_get_eeprom,
+       .set_eeprom                 = txgbe_set_eeprom,
+       .get_module_info            = txgbe_get_module_info,
+       .get_module_eeprom          = txgbe_get_module_eeprom,
+       .get_dcb_info               = txgbe_dev_get_dcb_info,
+       .timesync_adjust_time       = txgbe_timesync_adjust_time,
+       .timesync_read_time         = txgbe_timesync_read_time,
+       .timesync_write_time        = txgbe_timesync_write_time,
+       .udp_tunnel_port_add        = txgbe_dev_udp_tunnel_port_add,
+       .udp_tunnel_port_del        = txgbe_dev_udp_tunnel_port_del,
+       .tm_ops_get                 = txgbe_tm_ops_get,
+       .tx_done_cleanup            = txgbe_dev_tx_done_cleanup,
 };
 
 RTE_PMD_REGISTER_PCI(net_txgbe, rte_txgbe_pmd);