net/bnxt: cleanup VF representor device operations
[dpdk.git] / drivers / net / bnxt / bnxt_reps.c
index 4561115..6fa9a30 100644 (file)
 #include "bnxt.h"
 #include "bnxt_ring.h"
 #include "bnxt_reps.h"
+#include "bnxt_rxq.h"
+#include "bnxt_rxr.h"
+#include "bnxt_txq.h"
+#include "bnxt_txr.h"
+#include "bnxt_hwrm.h"
 #include "hsi_struct_def_dpdk.h"
+#include "bnxt_tf_common.h"
+#include "ulp_port_db.h"
+#include "ulp_flow_db.h"
 
 static const struct eth_dev_ops bnxt_vf_rep_dev_ops = {
        .dev_infos_get = bnxt_vf_rep_dev_info_get_op,
        .dev_configure = bnxt_vf_rep_dev_configure_op,
        .dev_start = bnxt_vf_rep_dev_start_op,
        .rx_queue_setup = bnxt_vf_rep_rx_queue_setup_op,
+       .rx_queue_release = bnxt_vf_rep_rx_queue_release_op,
        .tx_queue_setup = bnxt_vf_rep_tx_queue_setup_op,
+       .tx_queue_release = bnxt_vf_rep_tx_queue_release_op,
        .link_update = bnxt_vf_rep_link_update_op,
        .dev_close = bnxt_vf_rep_dev_close_op,
-       .dev_stop = bnxt_vf_rep_dev_stop_op
+       .dev_stop = bnxt_vf_rep_dev_stop_op,
+       .stats_get = bnxt_vf_rep_stats_get_op,
+       .stats_reset = bnxt_vf_rep_stats_reset_op,
+       .filter_ctrl = bnxt_filter_ctrl_op
 };
 
-static uint16_t
-bnxt_vf_rep_rx_burst(__rte_unused void *rx_queue,
-                    __rte_unused struct rte_mbuf **rx_pkts,
-                    __rte_unused uint16_t nb_pkts)
+uint16_t
+bnxt_vfr_recv(uint16_t port_id, uint16_t queue_id, struct rte_mbuf *mbuf)
 {
+       struct bnxt_sw_rx_bd *prod_rx_buf;
+       struct bnxt_rx_ring_info *rep_rxr;
+       struct bnxt_rx_queue *rep_rxq;
+       struct rte_eth_dev *vfr_eth_dev;
+       struct bnxt_vf_representor *vfr_bp;
+       uint16_t mask;
+       uint8_t que;
+
+       vfr_eth_dev = &rte_eth_devices[port_id];
+       if (!vfr_eth_dev)
+               return 1;
+       vfr_bp = vfr_eth_dev->data->dev_private;
+       /* If rxq_id happens to be > max rep_queue, use rxq0 */
+       que = queue_id < BNXT_MAX_VF_REP_RINGS ? queue_id : 0;
+       rep_rxq = vfr_bp->rx_queues[que];
+       rep_rxr = rep_rxq->rx_ring;
+       mask = rep_rxr->rx_ring_struct->ring_mask;
+
+       /* Put this mbuf on the RxQ of the Representor */
+       prod_rx_buf =
+               &rep_rxr->rx_buf_ring[rep_rxr->rx_prod++ & mask];
+       if (!prod_rx_buf->mbuf) {
+               prod_rx_buf->mbuf = mbuf;
+               vfr_bp->rx_bytes[que] += mbuf->pkt_len;
+               vfr_bp->rx_pkts[que]++;
+       } else {
+               vfr_bp->rx_drop_bytes[que] += mbuf->pkt_len;
+               vfr_bp->rx_drop_pkts[que]++;
+               rte_free(mbuf); /* Representor Rx ring full, drop pkt */
+       }
+
        return 0;
 }
 
 static uint16_t
-bnxt_vf_rep_tx_burst(__rte_unused void *tx_queue,
-                    __rte_unused struct rte_mbuf **tx_pkts,
+bnxt_vf_rep_rx_burst(void *rx_queue,
+                    struct rte_mbuf **rx_pkts,
+                    uint16_t nb_pkts)
+{
+       struct bnxt_rx_queue *rxq = rx_queue;
+       struct bnxt_sw_rx_bd *cons_rx_buf;
+       struct bnxt_rx_ring_info *rxr;
+       uint16_t nb_rx_pkts = 0;
+       uint16_t mask, i;
+
+       if (!rxq)
+               return 0;
+
+       rxr = rxq->rx_ring;
+       mask = rxr->rx_ring_struct->ring_mask;
+       for (i = 0; i < nb_pkts; i++) {
+               cons_rx_buf = &rxr->rx_buf_ring[rxr->rx_cons & mask];
+               if (!cons_rx_buf->mbuf)
+                       return nb_rx_pkts;
+               rx_pkts[nb_rx_pkts] = cons_rx_buf->mbuf;
+               rx_pkts[nb_rx_pkts]->port = rxq->port_id;
+               cons_rx_buf->mbuf = NULL;
+               nb_rx_pkts++;
+               rxr->rx_cons++;
+       }
+
+       return nb_rx_pkts;
+}
+
+static uint16_t
+bnxt_vf_rep_tx_burst(void *tx_queue,
+                    struct rte_mbuf **tx_pkts,
                     __rte_unused uint16_t nb_pkts)
 {
-       return 0;
+       struct bnxt_vf_rep_tx_queue *vfr_txq = tx_queue;
+       struct bnxt_tx_queue *ptxq;
+       struct bnxt *parent;
+       struct  bnxt_vf_representor *vf_rep_bp;
+       int qid;
+       int rc;
+       int i;
+
+       if (!vfr_txq)
+               return 0;
+
+       qid = vfr_txq->txq->queue_id;
+       vf_rep_bp = vfr_txq->bp;
+       parent = vf_rep_bp->parent_dev->data->dev_private;
+       pthread_mutex_lock(&parent->rep_info->vfr_lock);
+       ptxq = parent->tx_queues[qid];
+
+       ptxq->vfr_tx_cfa_action = vf_rep_bp->vfr_tx_cfa_action;
+
+       for (i = 0; i < nb_pkts; i++) {
+               vf_rep_bp->tx_bytes[qid] += tx_pkts[i]->pkt_len;
+               vf_rep_bp->tx_pkts[qid]++;
+       }
+
+       rc = bnxt_xmit_pkts(ptxq, tx_pkts, nb_pkts);
+       ptxq->vfr_tx_cfa_action = 0;
+       pthread_mutex_unlock(&parent->rep_info->vfr_lock);
+
+       return rc;
 }
 
 int bnxt_vf_representor_init(struct rte_eth_dev *eth_dev, void *params)
@@ -42,10 +142,11 @@ int bnxt_vf_representor_init(struct rte_eth_dev *eth_dev, void *params)
                                 (struct bnxt_vf_representor *)params;
        struct rte_eth_link *link;
        struct bnxt *parent_bp;
+       int rc = 0;
 
        vf_rep_bp->vf_id = rep_params->vf_id;
        vf_rep_bp->switch_domain_id = rep_params->switch_domain_id;
-       vf_rep_bp->parent_priv = rep_params->parent_priv;
+       vf_rep_bp->parent_dev = rep_params->parent_dev;
 
        eth_dev->data->dev_flags |= RTE_ETH_DEV_REPRESENTOR;
        eth_dev->data->representor_id = rep_params->vf_id;
@@ -63,7 +164,7 @@ int bnxt_vf_representor_init(struct rte_eth_dev *eth_dev, void *params)
        eth_dev->rx_pkt_burst = bnxt_vf_rep_rx_burst;
        eth_dev->tx_pkt_burst = bnxt_vf_rep_tx_burst;
        /* Link state. Inherited from PF or trusted VF */
-       parent_bp = vf_rep_bp->parent_priv;
+       parent_bp = vf_rep_bp->parent_dev->data->dev_private;
        link = &parent_bp->eth_dev->data->dev_link;
 
        eth_dev->data->dev_link.link_speed = link->link_speed;
@@ -71,6 +172,17 @@ int bnxt_vf_representor_init(struct rte_eth_dev *eth_dev, void *params)
        eth_dev->data->dev_link.link_status = link->link_status;
        eth_dev->data->dev_link.link_autoneg = link->link_autoneg;
 
+       vf_rep_bp->fw_fid = rep_params->vf_id + parent_bp->first_vf_id;
+       PMD_DRV_LOG(INFO, "vf_rep->fw_fid = %d\n", vf_rep_bp->fw_fid);
+       rc = bnxt_hwrm_get_dflt_vnic_svif(parent_bp, vf_rep_bp->fw_fid,
+                                         &vf_rep_bp->dflt_vnic_id,
+                                         &vf_rep_bp->svif);
+       if (rc)
+               PMD_DRV_LOG(ERR, "Failed to get default vnic id of VF\n");
+       else
+               PMD_DRV_LOG(INFO, "vf_rep->dflt_vnic_id = %d\n",
+                           vf_rep_bp->dflt_vnic_id);
+
        PMD_DRV_LOG(INFO, "calling bnxt_print_link_info\n");
        bnxt_print_link_info(eth_dev);
 
@@ -94,16 +206,18 @@ int bnxt_vf_representor_uninit(struct rte_eth_dev *eth_dev)
        uint16_t vf_id;
 
        eth_dev->data->mac_addrs = NULL;
+       eth_dev->dev_ops = NULL;
+
+       parent_bp = rep->parent_dev->data->dev_private;
+       if (!parent_bp)
+               return 0;
 
-       parent_bp = rep->parent_priv;
-       if (parent_bp) {
-               parent_bp->num_reps--;
-               vf_id = rep->vf_id;
+       parent_bp->num_reps--;
+       vf_id = rep->vf_id;
+       if (parent_bp->rep_info)
                memset(&parent_bp->rep_info[vf_id], 0,
                       sizeof(parent_bp->rep_info[vf_id]));
                /* mark that this representor has been freed */
-       }
-       eth_dev->dev_ops = NULL;
        return 0;
 }
 
@@ -115,7 +229,10 @@ int bnxt_vf_rep_link_update_op(struct rte_eth_dev *eth_dev, int wait_to_compl)
        struct rte_eth_link *link;
        int rc;
 
-       parent_bp = rep->parent_priv;
+       parent_bp = rep->parent_dev->data->dev_private;
+       if (!parent_bp)
+               return 0;
+
        rc = bnxt_link_update_op(parent_bp->eth_dev, wait_to_compl);
 
        /* Link state. Inherited from PF or trusted VF */
@@ -130,15 +247,206 @@ int bnxt_vf_rep_link_update_op(struct rte_eth_dev *eth_dev, int wait_to_compl)
        return rc;
 }
 
+static int bnxt_tf_vfr_alloc(struct rte_eth_dev *vfr_ethdev)
+{
+       int rc;
+       struct bnxt_vf_representor *vfr = vfr_ethdev->data->dev_private;
+       struct rte_eth_dev *parent_dev = vfr->parent_dev;
+       struct bnxt *parent_bp = parent_dev->data->dev_private;
+       uint16_t vfr_port_id = vfr_ethdev->data->port_id;
+       struct ulp_tlv_param param_list[] = {
+               {
+                       .type = BNXT_ULP_DF_PARAM_TYPE_DEV_PORT_ID,
+                       .length = 2,
+                       .value = {(vfr_port_id >> 8) & 0xff, vfr_port_id & 0xff}
+               },
+               {
+                       .type = BNXT_ULP_DF_PARAM_TYPE_LAST,
+                       .length = 0,
+                       .value = {0}
+               }
+       };
+
+       ulp_port_db_dev_port_intf_update(parent_bp->ulp_ctx, vfr_ethdev);
+
+       rc = ulp_default_flow_create(parent_dev, param_list,
+                                    BNXT_ULP_DF_TPL_VFREP_TO_VF,
+                                    &vfr->rep2vf_flow_id);
+       if (rc) {
+               BNXT_TF_DBG(DEBUG,
+                           "Default flow rule creation for VFR->VF failed!\n");
+               goto err;
+       }
+
+       BNXT_TF_DBG(DEBUG, "*** Default flow rule created for VFR->VF! ***\n");
+       BNXT_TF_DBG(DEBUG, "rep2vf_flow_id = %d\n", vfr->rep2vf_flow_id);
+       rc = ulp_default_flow_db_cfa_action_get(parent_bp->ulp_ctx,
+                                               vfr->rep2vf_flow_id,
+                                               &vfr->vfr_tx_cfa_action);
+       if (rc) {
+               BNXT_TF_DBG(DEBUG,
+                           "Failed to get action_ptr for VFR->VF dflt rule\n");
+               goto rep2vf_free;
+       }
+       BNXT_TF_DBG(DEBUG, "tx_cfa_action = %d\n", vfr->vfr_tx_cfa_action);
+       rc = ulp_default_flow_create(parent_dev, param_list,
+                                    BNXT_ULP_DF_TPL_VF_TO_VFREP,
+                                    &vfr->vf2rep_flow_id);
+       if (rc) {
+               BNXT_TF_DBG(DEBUG,
+                           "Default flow rule creation for VF->VFR failed!\n");
+               goto rep2vf_free;
+       }
+
+       BNXT_TF_DBG(DEBUG, "*** Default flow rule created for VF->VFR! ***\n");
+       BNXT_TF_DBG(DEBUG, "vfr2rep_flow_id = %d\n", vfr->vf2rep_flow_id);
+
+       rc = bnxt_hwrm_cfa_vfr_alloc(parent_bp, vfr->vf_id);
+       if (rc)
+               goto vf2rep_free;
+
+       return 0;
+
+vf2rep_free:
+       ulp_default_flow_destroy(vfr->parent_dev, vfr->vf2rep_flow_id);
+rep2vf_free:
+       ulp_default_flow_destroy(vfr->parent_dev, vfr->rep2vf_flow_id);
+err:
+       return -EIO;
+}
+
+static int bnxt_vfr_alloc(struct rte_eth_dev *vfr_ethdev)
+{
+       int rc = 0;
+       struct bnxt_vf_representor *vfr = vfr_ethdev->data->dev_private;
+
+       if (!vfr || !vfr->parent_dev) {
+               PMD_DRV_LOG(ERR,
+                           "No memory allocated for representor\n");
+               return -ENOMEM;
+       }
+
+       /* Check if representor has been already allocated in FW */
+       if (vfr->vfr_tx_cfa_action)
+               return 0;
+
+       /*
+        * Alloc VF rep rules in CFA after default VNIC is created.
+        * Otherwise the FW will create the VF-rep rules with
+        * default drop action.
+        */
+       rc = bnxt_tf_vfr_alloc(vfr_ethdev);
+       if (!rc)
+               PMD_DRV_LOG(DEBUG, "allocated representor %d in FW\n",
+                           vfr->vf_id);
+       else
+               PMD_DRV_LOG(ERR,
+                           "Failed to alloc representor %d in FW\n",
+                           vfr->vf_id);
+
+       return rc;
+}
+
+static void bnxt_vf_rep_free_rx_mbufs(struct bnxt_vf_representor *rep_bp)
+{
+       struct bnxt_rx_queue *rxq;
+       unsigned int i;
+
+       for (i = 0; i < rep_bp->rx_nr_rings; i++) {
+               rxq = rep_bp->rx_queues[i];
+               bnxt_rx_queue_release_mbufs(rxq);
+       }
+}
+
 int bnxt_vf_rep_dev_start_op(struct rte_eth_dev *eth_dev)
 {
-       bnxt_vf_rep_link_update_op(eth_dev, 1);
+       struct bnxt_vf_representor *rep_bp = eth_dev->data->dev_private;
+       int rc;
+
+       rc = bnxt_vfr_alloc(eth_dev);
+
+       if (!rc) {
+               eth_dev->rx_pkt_burst = &bnxt_vf_rep_rx_burst;
+               eth_dev->tx_pkt_burst = &bnxt_vf_rep_tx_burst;
+
+               bnxt_vf_rep_link_update_op(eth_dev, 1);
+       } else {
+               eth_dev->data->dev_link.link_status = 0;
+               bnxt_vf_rep_free_rx_mbufs(rep_bp);
+       }
+
+       return rc;
+}
+
+static int bnxt_tf_vfr_free(struct bnxt_vf_representor *vfr)
+{
+       int rc = 0;
 
+       rc = ulp_default_flow_destroy(vfr->parent_dev,
+                                     vfr->rep2vf_flow_id);
+       if (rc)
+               PMD_DRV_LOG(ERR,
+                           "default flow destroy failed rep2vf flowid: %d\n",
+                           vfr->rep2vf_flow_id);
+       rc = ulp_default_flow_destroy(vfr->parent_dev,
+                                     vfr->vf2rep_flow_id);
+       if (rc)
+               PMD_DRV_LOG(ERR,
+                           "default flow destroy failed vf2rep flowid: %d\n",
+                           vfr->vf2rep_flow_id);
        return 0;
 }
 
-void bnxt_vf_rep_dev_stop_op(__rte_unused struct rte_eth_dev *eth_dev)
+static int bnxt_vfr_free(struct bnxt_vf_representor *vfr)
 {
+       int rc = 0;
+       struct bnxt *parent_bp;
+
+       if (!vfr || !vfr->parent_dev) {
+               PMD_DRV_LOG(ERR,
+                           "No memory allocated for representor\n");
+               return -ENOMEM;
+       }
+
+       parent_bp = vfr->parent_dev->data->dev_private;
+       if (!parent_bp)
+               return 0;
+
+       /* Check if representor has been already freed in FW */
+       if (!vfr->vfr_tx_cfa_action)
+               return 0;
+
+       rc = bnxt_tf_vfr_free(vfr);
+       if (rc) {
+               PMD_DRV_LOG(ERR,
+                           "Failed to free representor %d in FW\n",
+                           vfr->vf_id);
+               return rc;
+       }
+
+       PMD_DRV_LOG(DEBUG, "freed representor %d in FW\n",
+                   vfr->vf_id);
+       vfr->vfr_tx_cfa_action = 0;
+
+       rc = bnxt_hwrm_cfa_vfr_free(parent_bp, vfr->vf_id);
+
+       return rc;
+}
+
+void bnxt_vf_rep_dev_stop_op(struct rte_eth_dev *eth_dev)
+{
+       struct bnxt_vf_representor *vfr_bp = eth_dev->data->dev_private;
+
+       /* Avoid crashes as we are about to free queues */
+       eth_dev->rx_pkt_burst = &bnxt_dummy_recv_pkts;
+       eth_dev->tx_pkt_burst = &bnxt_dummy_xmit_pkts;
+
+       bnxt_vfr_free(vfr_bp);
+
+       if (eth_dev->data->dev_started)
+               eth_dev->data->dev_link.link_status = 0;
+
+       bnxt_vf_rep_free_rx_mbufs(vfr_bp);
 }
 
 void bnxt_vf_rep_dev_close_op(struct rte_eth_dev *eth_dev)
@@ -151,12 +459,11 @@ int bnxt_vf_rep_dev_info_get_op(struct rte_eth_dev *eth_dev,
 {
        struct bnxt_vf_representor *rep_bp = eth_dev->data->dev_private;
        struct bnxt *parent_bp;
-       uint16_t max_vnics, i, j, vpool, vrxq;
        unsigned int max_rx_rings;
        int rc = 0;
 
        /* MAC Specifics */
-       parent_bp = rep_bp->parent_priv;
+       parent_bp = rep_bp->parent_dev->data->dev_private;
        if (!parent_bp) {
                PMD_DRV_LOG(ERR, "Rep parent NULL!\n");
                return rc;
@@ -171,7 +478,6 @@ int bnxt_vf_rep_dev_info_get_op(struct rte_eth_dev *eth_dev,
        dev_info->max_tx_queues = max_rx_rings;
        dev_info->reta_size = bnxt_rss_hash_tbl_size(parent_bp);
        dev_info->hash_key_size = 40;
-       max_vnics = parent_bp->max_vnics;
 
        /* MTU specifics */
        dev_info->min_mtu = RTE_ETHER_MIN_MTU;
@@ -187,94 +493,230 @@ int bnxt_vf_rep_dev_info_get_op(struct rte_eth_dev *eth_dev,
        dev_info->tx_offload_capa = BNXT_DEV_TX_OFFLOAD_SUPPORT;
        dev_info->flow_type_rss_offloads = BNXT_ETH_RSS_SUPPORT;
 
-       /* *INDENT-OFF* */
-       dev_info->default_rxconf = (struct rte_eth_rxconf) {
-               .rx_thresh = {
-                       .pthresh = 8,
-                       .hthresh = 8,
-                       .wthresh = 0,
-               },
-               .rx_free_thresh = 32,
-               /* If no descriptors available, pkts are dropped by default */
-               .rx_drop_en = 1,
-       };
+       return 0;
+}
 
-       dev_info->default_txconf = (struct rte_eth_txconf) {
-               .tx_thresh = {
-                       .pthresh = 32,
-                       .hthresh = 0,
-                       .wthresh = 0,
-               },
-               .tx_free_thresh = 32,
-               .tx_rs_thresh = 32,
-       };
-       eth_dev->data->dev_conf.intr_conf.lsc = 1;
+int bnxt_vf_rep_dev_configure_op(__rte_unused struct rte_eth_dev *eth_dev)
+{
+       struct bnxt_vf_representor *rep_bp = eth_dev->data->dev_private;
 
-       eth_dev->data->dev_conf.intr_conf.rxq = 1;
-       dev_info->rx_desc_lim.nb_min = BNXT_MIN_RING_DESC;
-       dev_info->rx_desc_lim.nb_max = BNXT_MAX_RX_RING_DESC;
-       dev_info->tx_desc_lim.nb_min = BNXT_MIN_RING_DESC;
-       dev_info->tx_desc_lim.nb_max = BNXT_MAX_TX_RING_DESC;
+       PMD_DRV_LOG(DEBUG, "Representor dev_configure_op\n");
+       rep_bp->rx_queues = (void *)eth_dev->data->rx_queues;
+       rep_bp->tx_nr_rings = eth_dev->data->nb_tx_queues;
+       rep_bp->rx_nr_rings = eth_dev->data->nb_rx_queues;
 
-       /* *INDENT-ON* */
+       return 0;
+}
 
-       /*
-        * TODO: default_rxconf, default_txconf, rx_desc_lim, and tx_desc_lim
-        *       need further investigation.
-        */
+int bnxt_vf_rep_rx_queue_setup_op(struct rte_eth_dev *eth_dev,
+                         uint16_t queue_idx,
+                         uint16_t nb_desc,
+                         unsigned int socket_id,
+                         __rte_unused const struct rte_eth_rxconf *rx_conf,
+                         __rte_unused struct rte_mempool *mp)
+{
+       struct bnxt_vf_representor *rep_bp = eth_dev->data->dev_private;
+       struct bnxt *parent_bp = rep_bp->parent_dev->data->dev_private;
+       struct bnxt_rx_queue *parent_rxq;
+       struct bnxt_rx_queue *rxq;
+       struct bnxt_sw_rx_bd *buf_ring;
+       int rc = 0;
 
-       /* VMDq resources */
-       vpool = 64; /* ETH_64_POOLS */
-       vrxq = 128; /* ETH_VMDQ_DCB_NUM_QUEUES */
-       for (i = 0; i < 4; vpool >>= 1, i++) {
-               if (max_vnics > vpool) {
-                       for (j = 0; j < 5; vrxq >>= 1, j++) {
-                               if (dev_info->max_rx_queues > vrxq) {
-                                       if (vpool > vrxq)
-                                               vpool = vrxq;
-                                       goto found;
-                               }
-                       }
-                       /* Not enough resources to support VMDq */
-                       break;
-               }
+       if (queue_idx >= BNXT_MAX_VF_REP_RINGS) {
+               PMD_DRV_LOG(ERR,
+                           "Cannot create Rx ring %d. %d rings available\n",
+                           queue_idx, BNXT_MAX_VF_REP_RINGS);
+               return -EINVAL;
+       }
+
+       if (!nb_desc || nb_desc > MAX_RX_DESC_CNT) {
+               PMD_DRV_LOG(ERR, "nb_desc %d is invalid\n", nb_desc);
+               return -EINVAL;
+       }
+
+       parent_rxq = parent_bp->rx_queues[queue_idx];
+       if (!parent_rxq) {
+               PMD_DRV_LOG(ERR, "Parent RxQ has not been configured yet\n");
+               return -EINVAL;
+       }
+
+       if (nb_desc != parent_rxq->nb_rx_desc) {
+               PMD_DRV_LOG(ERR, "nb_desc %d do not match parent rxq", nb_desc);
+               return -EINVAL;
        }
-       /* Not enough resources to support VMDq */
-       vpool = 0;
-       vrxq = 0;
-found:
-       dev_info->max_vmdq_pools = vpool;
-       dev_info->vmdq_queue_num = vrxq;
 
-       dev_info->vmdq_pool_base = 0;
-       dev_info->vmdq_queue_base = 0;
+       if (eth_dev->data->rx_queues) {
+               rxq = eth_dev->data->rx_queues[queue_idx];
+               if (rxq)
+                       bnxt_rx_queue_release_op(rxq);
+       }
+
+       rxq = rte_zmalloc_socket("bnxt_vfr_rx_queue",
+                                sizeof(struct bnxt_rx_queue),
+                                RTE_CACHE_LINE_SIZE, socket_id);
+       if (!rxq) {
+               PMD_DRV_LOG(ERR, "bnxt_vfr_rx_queue allocation failed!\n");
+               return -ENOMEM;
+       }
+
+       rxq->nb_rx_desc = nb_desc;
+
+       rc = bnxt_init_rx_ring_struct(rxq, socket_id);
+       if (rc)
+               goto out;
+
+       buf_ring = rte_zmalloc_socket("bnxt_rx_vfr_buf_ring",
+                                     sizeof(struct bnxt_sw_rx_bd) *
+                                     rxq->rx_ring->rx_ring_struct->ring_size,
+                                     RTE_CACHE_LINE_SIZE, socket_id);
+       if (!buf_ring) {
+               PMD_DRV_LOG(ERR, "bnxt_rx_vfr_buf_ring allocation failed!\n");
+               rc = -ENOMEM;
+               goto out;
+       }
+
+       rxq->rx_ring->rx_buf_ring = buf_ring;
+       rxq->queue_id = queue_idx;
+       rxq->port_id = eth_dev->data->port_id;
+       eth_dev->data->rx_queues[queue_idx] = rxq;
 
        return 0;
+
+out:
+       if (rxq)
+               bnxt_rx_queue_release_op(rxq);
+
+       return rc;
 }
 
-int bnxt_vf_rep_dev_configure_op(__rte_unused struct rte_eth_dev *eth_dev)
+void bnxt_vf_rep_rx_queue_release_op(void *rx_queue)
 {
-       PMD_DRV_LOG(DEBUG, "Representor dev_configure_op\n");
+       struct bnxt_rx_queue *rxq = (struct bnxt_rx_queue *)rx_queue;
+
+       if (!rxq)
+               return;
+
+       bnxt_rx_queue_release_mbufs(rxq);
+
+       bnxt_free_ring(rxq->rx_ring->rx_ring_struct);
+       bnxt_free_ring(rxq->rx_ring->ag_ring_struct);
+       bnxt_free_ring(rxq->cp_ring->cp_ring_struct);
+
+       rte_free(rxq);
+}
+
+int bnxt_vf_rep_tx_queue_setup_op(struct rte_eth_dev *eth_dev,
+                         uint16_t queue_idx,
+                         uint16_t nb_desc,
+                         unsigned int socket_id,
+                         __rte_unused const struct rte_eth_txconf *tx_conf)
+{
+       struct bnxt_vf_representor *rep_bp = eth_dev->data->dev_private;
+       struct bnxt *parent_bp = rep_bp->parent_dev->data->dev_private;
+       struct bnxt_tx_queue *parent_txq, *txq;
+       struct bnxt_vf_rep_tx_queue *vfr_txq;
+
+       if (queue_idx >= BNXT_MAX_VF_REP_RINGS) {
+               PMD_DRV_LOG(ERR,
+                           "Cannot create Tx rings %d. %d rings available\n",
+                           queue_idx, BNXT_MAX_VF_REP_RINGS);
+               return -EINVAL;
+       }
+
+       if (!nb_desc || nb_desc > MAX_TX_DESC_CNT) {
+               PMD_DRV_LOG(ERR, "nb_desc %d is invalid", nb_desc);
+               return -EINVAL;
+       }
+
+       parent_txq = parent_bp->tx_queues[queue_idx];
+       if (!parent_txq) {
+               PMD_DRV_LOG(ERR, "Parent TxQ has not been configured yet\n");
+               return -EINVAL;
+       }
+
+       if (nb_desc != parent_txq->nb_tx_desc) {
+               PMD_DRV_LOG(ERR, "nb_desc %d do not match parent txq", nb_desc);
+               return -EINVAL;
+       }
+
+       if (eth_dev->data->tx_queues) {
+               vfr_txq = eth_dev->data->tx_queues[queue_idx];
+               bnxt_vf_rep_tx_queue_release_op(vfr_txq);
+               vfr_txq = NULL;
+       }
+
+       vfr_txq = rte_zmalloc_socket("bnxt_vfr_tx_queue",
+                                    sizeof(struct bnxt_vf_rep_tx_queue),
+                                    RTE_CACHE_LINE_SIZE, socket_id);
+       if (!vfr_txq) {
+               PMD_DRV_LOG(ERR, "bnxt_vfr_tx_queue allocation failed!");
+               return -ENOMEM;
+       }
+       txq = rte_zmalloc_socket("bnxt_tx_queue",
+                                sizeof(struct bnxt_tx_queue),
+                                RTE_CACHE_LINE_SIZE, socket_id);
+       if (!txq) {
+               PMD_DRV_LOG(ERR, "bnxt_tx_queue allocation failed!");
+               rte_free(vfr_txq);
+               return -ENOMEM;
+       }
+
+       txq->nb_tx_desc = nb_desc;
+       txq->queue_id = queue_idx;
+       txq->port_id = eth_dev->data->port_id;
+       vfr_txq->txq = txq;
+       vfr_txq->bp = rep_bp;
+       eth_dev->data->tx_queues[queue_idx] = vfr_txq;
+
        return 0;
 }
 
-int bnxt_vf_rep_rx_queue_setup_op(__rte_unused struct rte_eth_dev *eth_dev,
-                                 __rte_unused uint16_t queue_idx,
-                                 __rte_unused uint16_t nb_desc,
-                                 __rte_unused unsigned int socket_id,
-                                 __rte_unused const struct rte_eth_rxconf *
-                                 rx_conf,
-                                 __rte_unused struct rte_mempool *mp)
+void bnxt_vf_rep_tx_queue_release_op(void *tx_queue)
 {
+       struct bnxt_vf_rep_tx_queue *vfr_txq = tx_queue;
+
+       if (!vfr_txq)
+               return;
+
+       rte_free(vfr_txq->txq);
+       rte_free(vfr_txq);
+}
+
+int bnxt_vf_rep_stats_get_op(struct rte_eth_dev *eth_dev,
+                            struct rte_eth_stats *stats)
+{
+       struct bnxt_vf_representor *rep_bp = eth_dev->data->dev_private;
+       int i;
+
+       memset(stats, 0, sizeof(*stats));
+       for (i = 0; i < BNXT_MAX_VF_REP_RINGS; i++) {
+               stats->obytes += rep_bp->tx_bytes[i];
+               stats->opackets += rep_bp->tx_pkts[i];
+               stats->ibytes += rep_bp->rx_bytes[i];
+               stats->ipackets += rep_bp->rx_pkts[i];
+               stats->imissed += rep_bp->rx_drop_pkts[i];
+
+               stats->q_ipackets[i] = rep_bp->rx_pkts[i];
+               stats->q_ibytes[i] = rep_bp->rx_bytes[i];
+               stats->q_opackets[i] = rep_bp->tx_pkts[i];
+               stats->q_obytes[i] = rep_bp->tx_bytes[i];
+               stats->q_errors[i] = rep_bp->rx_drop_pkts[i];
+       }
+
        return 0;
 }
 
-int bnxt_vf_rep_tx_queue_setup_op(__rte_unused struct rte_eth_dev *eth_dev,
-                                 __rte_unused uint16_t queue_idx,
-                                 __rte_unused uint16_t nb_desc,
-                                 __rte_unused unsigned int socket_id,
-                                 __rte_unused const struct rte_eth_txconf *
-                                 tx_conf)
+int bnxt_vf_rep_stats_reset_op(struct rte_eth_dev *eth_dev)
 {
+       struct bnxt_vf_representor *rep_bp = eth_dev->data->dev_private;
+       int i;
+
+       for (i = 0; i < BNXT_MAX_VF_REP_RINGS; i++) {
+               rep_bp->tx_pkts[i] = 0;
+               rep_bp->tx_bytes[i] = 0;
+               rep_bp->rx_pkts[i] = 0;
+               rep_bp->rx_bytes[i] = 0;
+               rep_bp->rx_drop_pkts[i] = 0;
+       }
        return 0;
 }