net: add macro for MAC address print
[dpdk.git] / drivers / net / virtio / virtio_ethdev.c
index 4f2fa0d..9ca8bae 100644 (file)
@@ -8,46 +8,41 @@
 #include <errno.h>
 #include <unistd.h>
 
-#include <rte_ethdev_driver.h>
-#include <rte_ethdev_pci.h>
+#include <ethdev_driver.h>
 #include <rte_memcpy.h>
 #include <rte_string_fns.h>
 #include <rte_memzone.h>
 #include <rte_malloc.h>
 #include <rte_branch_prediction.h>
-#include <rte_pci.h>
-#include <rte_bus_pci.h>
 #include <rte_ether.h>
 #include <rte_ip.h>
 #include <rte_arp.h>
 #include <rte_common.h>
 #include <rte_errno.h>
 #include <rte_cpuflags.h>
-
+#include <rte_vect.h>
 #include <rte_memory.h>
+#include <rte_eal_paging.h>
 #include <rte_eal.h>
 #include <rte_dev.h>
 #include <rte_cycles.h>
 #include <rte_kvargs.h>
 
 #include "virtio_ethdev.h"
-#include "virtio_pci.h"
+#include "virtio.h"
 #include "virtio_logs.h"
 #include "virtqueue.h"
 #include "virtio_rxtx.h"
 #include "virtio_user/virtio_user_dev.h"
 
-static int eth_virtio_dev_uninit(struct rte_eth_dev *eth_dev);
 static int  virtio_dev_configure(struct rte_eth_dev *dev);
 static int  virtio_dev_start(struct rte_eth_dev *dev);
-static void virtio_dev_stop(struct rte_eth_dev *dev);
 static int virtio_dev_promiscuous_enable(struct rte_eth_dev *dev);
 static int virtio_dev_promiscuous_disable(struct rte_eth_dev *dev);
 static int virtio_dev_allmulticast_enable(struct rte_eth_dev *dev);
 static int virtio_dev_allmulticast_disable(struct rte_eth_dev *dev);
 static uint32_t virtio_dev_speed_capa_get(uint32_t speed);
 static int virtio_dev_devargs_parse(struct rte_devargs *devargs,
-       int *vdpa,
        uint32_t *speed,
        int *vectorized);
 static int virtio_dev_info_get(struct rte_eth_dev *dev,
@@ -88,15 +83,6 @@ static int virtio_dev_queue_stats_mapping_set(
 static void virtio_notify_peers(struct rte_eth_dev *dev);
 static void virtio_ack_link_announce(struct rte_eth_dev *dev);
 
-/*
- * The set of PCI devices this driver supports
- */
-static const struct rte_pci_id pci_id_virtio_map[] = {
-       { RTE_PCI_DEVICE(VIRTIO_PCI_VENDORID, VIRTIO_PCI_LEGACY_DEVICEID_NET) },
-       { RTE_PCI_DEVICE(VIRTIO_PCI_VENDORID, VIRTIO_PCI_MODERN_DEVICEID_NET) },
-       { .vendor_id = 0, /* sentinel */ },
-};
-
 struct rte_virtio_xstats_name_off {
        char name[RTE_ETH_XSTATS_NAME_SIZE];
        unsigned offset;
@@ -147,7 +133,7 @@ virtio_send_command_packed(struct virtnet_ctl *cvq,
                           struct virtio_pmd_ctrl *ctrl,
                           int *dlen, int pkt_num)
 {
-       struct virtqueue *vq = cvq->vq;
+       struct virtqueue *vq = virtnet_cq_to_vq(cvq);
        int head;
        struct vring_packed_desc *desc = vq->vq_packed.ring.desc;
        struct virtio_pmd_ctrl *result;
@@ -202,18 +188,18 @@ virtio_send_command_packed(struct virtnet_ctl *cvq,
                vq->vq_packed.cached_flags ^= VRING_PACKED_DESC_F_AVAIL_USED;
        }
 
-       virtio_wmb(vq->hw->weak_barriers);
-       desc[head].flags = VRING_DESC_F_NEXT | flags;
+       virtqueue_store_flags_packed(&desc[head], VRING_DESC_F_NEXT | flags,
+                       vq->hw->weak_barriers);
 
        virtio_wmb(vq->hw->weak_barriers);
        virtqueue_notify(vq);
 
-       /* wait for used descriptors in virtqueue */
+       /* wait for used desc in virtqueue
+        * desc_is_used has a load-acquire or rte_io_rmb inside
+        */
        while (!desc_is_used(&desc[head], vq))
                usleep(100);
 
-       virtio_rmb(vq->hw->weak_barriers);
-
        /* now get used descriptors */
        vq->vq_free_cnt += nb_descs;
        vq->vq_used_cons_idx += nb_descs;
@@ -243,7 +229,7 @@ virtio_send_command_split(struct virtnet_ctl *cvq,
                          int *dlen, int pkt_num)
 {
        struct virtio_pmd_ctrl *result;
-       struct virtqueue *vq = cvq->vq;
+       struct virtqueue *vq = virtnet_cq_to_vq(cvq);
        uint32_t head, i;
        int k, sum = 0;
 
@@ -330,13 +316,13 @@ virtio_send_command(struct virtnet_ctl *cvq, struct virtio_pmd_ctrl *ctrl,
 
        ctrl->status = status;
 
-       if (!cvq || !cvq->vq) {
+       if (!cvq) {
                PMD_INIT_LOG(ERR, "Control queue is not supported.");
                return -1;
        }
 
        rte_spinlock_lock(&cvq->lock);
-       vq = cvq->vq;
+       vq = virtnet_cq_to_vq(cvq);
 
        PMD_INIT_LOG(DEBUG, "vq->vq_desc_head_idx = %d, status = %d, "
                "vq->hw->cvq = %p vq = %p",
@@ -350,7 +336,7 @@ virtio_send_command(struct virtnet_ctl *cvq, struct virtio_pmd_ctrl *ctrl,
        memcpy(cvq->virtio_net_hdr_mz->addr, ctrl,
                sizeof(struct virtio_pmd_ctrl));
 
-       if (vtpci_packed_queue(vq->hw))
+       if (virtio_with_packed_queue(vq->hw))
                result = virtio_send_command_packed(cvq, ctrl, dlen, pkt_num);
        else
                result = virtio_send_command_split(cvq, ctrl, dlen, pkt_num);
@@ -394,7 +380,7 @@ virtio_get_nr_vq(struct virtio_hw *hw)
 {
        uint16_t nr_vq = hw->max_queue_pairs * 2;
 
-       if (vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_VQ))
+       if (virtio_with_feature(hw, VIRTIO_NET_F_CTRL_VQ))
                nr_vq += 1;
 
        return nr_vq;
@@ -416,14 +402,14 @@ virtio_init_vring(struct virtqueue *vq)
        vq->vq_desc_tail_idx = (uint16_t)(vq->vq_nentries - 1);
        vq->vq_free_cnt = vq->vq_nentries;
        memset(vq->vq_descx, 0, sizeof(struct vq_desc_extra) * vq->vq_nentries);
-       if (vtpci_packed_queue(vq->hw)) {
+       if (virtio_with_packed_queue(vq->hw)) {
                vring_init_packed(&vq->vq_packed.ring, ring_mem,
-                                 VIRTIO_PCI_VRING_ALIGN, size);
+                                 VIRTIO_VRING_ALIGN, size);
                vring_desc_init_packed(vq, size);
        } else {
                struct vring *vr = &vq->vq_split.ring;
 
-               vring_init_split(vr, ring_mem, VIRTIO_PCI_VRING_ALIGN, size);
+               vring_init_split(vr, ring_mem, VIRTIO_VRING_ALIGN, size);
                vring_desc_init_split(vr->desc, size);
        }
        /*
@@ -433,7 +419,7 @@ virtio_init_vring(struct virtqueue *vq)
 }
 
 static int
-virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
+virtio_init_queue(struct rte_eth_dev *dev, uint16_t queue_idx)
 {
        char vq_name[VIRTQUEUE_MAX_NAME_SZ];
        char vq_hdr_name[VIRTQUEUE_MAX_NAME_SZ];
@@ -446,31 +432,32 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
        struct virtqueue *vq;
        size_t sz_hdr_mz = 0;
        void *sw_ring = NULL;
-       int queue_type = virtio_get_queue_type(hw, vtpci_queue_idx);
+       int queue_type = virtio_get_queue_type(hw, queue_idx);
        int ret;
        int numa_node = dev->device->numa_node;
+       struct rte_mbuf *fake_mbuf = NULL;
 
        PMD_INIT_LOG(INFO, "setting up queue: %u on NUMA node %d",
-                       vtpci_queue_idx, numa_node);
+                       queue_idx, numa_node);
 
        /*
         * Read the virtqueue size from the Queue Size field
         * Always power of 2 and if 0 virtqueue does not exist
         */
-       vq_size = VTPCI_OPS(hw)->get_queue_num(hw, vtpci_queue_idx);
+       vq_size = VIRTIO_OPS(hw)->get_queue_num(hw, queue_idx);
        PMD_INIT_LOG(DEBUG, "vq_size: %u", vq_size);
        if (vq_size == 0) {
                PMD_INIT_LOG(ERR, "virtqueue does not exist");
                return -EINVAL;
        }
 
-       if (!vtpci_packed_queue(hw) && !rte_is_power_of_2(vq_size)) {
+       if (!virtio_with_packed_queue(hw) && !rte_is_power_of_2(vq_size)) {
                PMD_INIT_LOG(ERR, "split virtqueue size is not power of 2");
                return -EINVAL;
        }
 
        snprintf(vq_name, sizeof(vq_name), "port%d_vq%d",
-                dev->data->port_id, vtpci_queue_idx);
+                dev->data->port_id, queue_idx);
 
        size = RTE_ALIGN_CEIL(sizeof(*vq) +
                                vq_size * sizeof(struct vq_desc_extra),
@@ -483,7 +470,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
                sz_hdr_mz = vq_size * sizeof(struct virtio_tx_region);
        } else if (queue_type == VTNET_CQ) {
                /* Allocate a page for control vq command, data and status */
-               sz_hdr_mz = PAGE_SIZE;
+               sz_hdr_mz = rte_mem_page_size();
        }
 
        vq = rte_zmalloc_socket(vq_name, size, RTE_CACHE_LINE_SIZE,
@@ -492,12 +479,12 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
                PMD_INIT_LOG(ERR, "can not allocate vq");
                return -ENOMEM;
        }
-       hw->vqs[vtpci_queue_idx] = vq;
+       hw->vqs[queue_idx] = vq;
 
        vq->hw = hw;
-       vq->vq_queue_index = vtpci_queue_idx;
+       vq->vq_queue_index = queue_idx;
        vq->vq_nentries = vq_size;
-       if (vtpci_packed_queue(hw)) {
+       if (virtio_with_packed_queue(hw)) {
                vq->vq_packed.used_wrap_counter = 1;
                vq->vq_packed.cached_flags = VRING_PACKED_DESC_F_AVAIL;
                vq->vq_packed.event_flags_shadow = 0;
@@ -508,20 +495,20 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
        /*
         * Reserve a memzone for vring elements
         */
-       size = vring_size(hw, vq_size, VIRTIO_PCI_VRING_ALIGN);
-       vq->vq_ring_size = RTE_ALIGN_CEIL(size, VIRTIO_PCI_VRING_ALIGN);
+       size = vring_size(hw, vq_size, VIRTIO_VRING_ALIGN);
+       vq->vq_ring_size = RTE_ALIGN_CEIL(size, VIRTIO_VRING_ALIGN);
        PMD_INIT_LOG(DEBUG, "vring_size: %d, rounded_vring_size: %d",
                     size, vq->vq_ring_size);
 
        mz = rte_memzone_reserve_aligned(vq_name, vq->vq_ring_size,
                        numa_node, RTE_MEMZONE_IOVA_CONTIG,
-                       VIRTIO_PCI_VRING_ALIGN);
+                       VIRTIO_VRING_ALIGN);
        if (mz == NULL) {
                if (rte_errno == EEXIST)
                        mz = rte_memzone_lookup(vq_name);
                if (mz == NULL) {
                        ret = -ENOMEM;
-                       goto fail_q_alloc;
+                       goto free_vq;
                }
        }
 
@@ -538,7 +525,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
 
        if (sz_hdr_mz) {
                snprintf(vq_hdr_name, sizeof(vq_hdr_name), "port%d_vq%d_hdr",
-                        dev->data->port_id, vtpci_queue_idx);
+                        dev->data->port_id, queue_idx);
                hdr_mz = rte_memzone_reserve_aligned(vq_hdr_name, sz_hdr_mz,
                                numa_node, RTE_MEMZONE_IOVA_CONTIG,
                                RTE_CACHE_LINE_SIZE);
@@ -547,7 +534,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
                                hdr_mz = rte_memzone_lookup(vq_hdr_name);
                        if (hdr_mz == NULL) {
                                ret = -ENOMEM;
-                               goto fail_q_alloc;
+                               goto free_mz;
                        }
                }
        }
@@ -561,47 +548,38 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
                if (!sw_ring) {
                        PMD_INIT_LOG(ERR, "can not allocate RX soft ring");
                        ret = -ENOMEM;
-                       goto fail_q_alloc;
+                       goto free_hdr_mz;
+               }
+
+               fake_mbuf = rte_zmalloc_socket("sw_ring", sizeof(*fake_mbuf),
+                               RTE_CACHE_LINE_SIZE, numa_node);
+               if (!fake_mbuf) {
+                       PMD_INIT_LOG(ERR, "can not allocate fake mbuf");
+                       ret = -ENOMEM;
+                       goto free_sw_ring;
                }
 
                vq->sw_ring = sw_ring;
                rxvq = &vq->rxq;
-               rxvq->vq = vq;
                rxvq->port_id = dev->data->port_id;
                rxvq->mz = mz;
+               rxvq->fake_mbuf = fake_mbuf;
        } else if (queue_type == VTNET_TQ) {
                txvq = &vq->txq;
-               txvq->vq = vq;
                txvq->port_id = dev->data->port_id;
                txvq->mz = mz;
                txvq->virtio_net_hdr_mz = hdr_mz;
                txvq->virtio_net_hdr_mem = hdr_mz->iova;
        } else if (queue_type == VTNET_CQ) {
                cvq = &vq->cq;
-               cvq->vq = vq;
                cvq->mz = mz;
                cvq->virtio_net_hdr_mz = hdr_mz;
                cvq->virtio_net_hdr_mem = hdr_mz->iova;
-               memset(cvq->virtio_net_hdr_mz->addr, 0, PAGE_SIZE);
+               memset(cvq->virtio_net_hdr_mz->addr, 0, rte_mem_page_size());
 
                hw->cvq = cvq;
        }
 
-       /* For virtio_user case (that is when hw->virtio_user_dev is not NULL),
-        * we use virtual address. And we need properly set _offset_, please see
-        * VIRTIO_MBUF_DATA_DMA_ADDR in virtqueue.h for more information.
-        */
-       if (!hw->virtio_user_dev)
-               vq->offset = offsetof(struct rte_mbuf, buf_iova);
-       else {
-               vq->vq_ring_mem = (uintptr_t)mz->addr;
-               vq->offset = offsetof(struct rte_mbuf, buf_addr);
-               if (queue_type == VTNET_TQ)
-                       txvq->virtio_net_hdr_mem = (uintptr_t)hdr_mz->addr;
-               else if (queue_type == VTNET_CQ)
-                       cvq->virtio_net_hdr_mem = (uintptr_t)hdr_mz->addr;
-       }
-
        if (queue_type == VTNET_TQ) {
                struct virtio_tx_region *txr;
                unsigned int i;
@@ -609,10 +587,9 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
                txr = hdr_mz->addr;
                memset(txr, 0, vq_size * sizeof(*txr));
                for (i = 0; i < vq_size; i++) {
-                       struct vring_desc *start_dp = txr[i].tx_indir;
-
                        /* first indirect descriptor is always the tx header */
-                       if (!vtpci_packed_queue(hw)) {
+                       if (!virtio_with_packed_queue(hw)) {
+                               struct vring_desc *start_dp = txr[i].tx_indir;
                                vring_desc_init_split(start_dp,
                                                      RTE_DIM(txr[i].tx_indir));
                                start_dp->addr = txvq->virtio_net_hdr_mem
@@ -621,21 +598,38 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx)
                                                   tx_hdr);
                                start_dp->len = hw->vtnet_hdr_size;
                                start_dp->flags = VRING_DESC_F_NEXT;
+                       } else {
+                               struct vring_packed_desc *start_dp =
+                                       txr[i].tx_packed_indir;
+                               vring_desc_init_indirect_packed(start_dp,
+                                     RTE_DIM(txr[i].tx_packed_indir));
+                               start_dp->addr = txvq->virtio_net_hdr_mem
+                                       + i * sizeof(*txr)
+                                       + offsetof(struct virtio_tx_region,
+                                                  tx_hdr);
+                               start_dp->len = hw->vtnet_hdr_size;
                        }
                }
        }
 
-       if (VTPCI_OPS(hw)->setup_queue(hw, vq) < 0) {
+       if (VIRTIO_OPS(hw)->setup_queue(hw, vq) < 0) {
                PMD_INIT_LOG(ERR, "setup_queue failed");
-               return -EINVAL;
+               ret = -EINVAL;
+               goto clean_vq;
        }
 
        return 0;
 
-fail_q_alloc:
+clean_vq:
+       hw->cvq = NULL;
+       rte_free(fake_mbuf);
+free_sw_ring:
        rte_free(sw_ring);
+free_hdr_mz:
        rte_memzone_free(hdr_mz);
+free_mz:
        rte_memzone_free(mz);
+free_vq:
        rte_free(vq);
 
        return ret;
@@ -659,6 +653,7 @@ virtio_free_queues(struct virtio_hw *hw)
 
                queue_type = virtio_get_queue_type(hw, i);
                if (queue_type == VTNET_RQ) {
+                       rte_free(vq->rxq.fake_mbuf);
                        rte_free(vq->sw_ring);
                        rte_memzone_free(vq->rxq.mz);
                } else if (queue_type == VTNET_TQ) {
@@ -704,21 +699,23 @@ virtio_alloc_queues(struct rte_eth_dev *dev)
 
 static void virtio_queues_unbind_intr(struct rte_eth_dev *dev);
 
-static void
+int
 virtio_dev_close(struct rte_eth_dev *dev)
 {
        struct virtio_hw *hw = dev->data->dev_private;
        struct rte_intr_conf *intr_conf = &dev->data->dev_conf.intr_conf;
 
        PMD_INIT_LOG(DEBUG, "virtio_dev_close");
+       if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+               return 0;
 
        if (!hw->opened)
-               return;
-       hw->opened = false;
+               return 0;
+       hw->opened = 0;
 
        /* reset the NIC */
        if (dev->data->dev_flags & RTE_ETH_DEV_INTR_LSC)
-               VTPCI_OPS(hw)->set_config_irq(hw, VIRTIO_MSI_NO_VECTOR);
+               VIRTIO_OPS(hw)->set_config_irq(hw, VIRTIO_MSI_NO_VECTOR);
        if (intr_conf->rxq)
                virtio_queues_unbind_intr(dev);
 
@@ -729,20 +726,11 @@ virtio_dev_close(struct rte_eth_dev *dev)
                dev->intr_handle->intr_vec = NULL;
        }
 
-       vtpci_reset(hw);
+       virtio_reset(hw);
        virtio_dev_free_mbufs(dev);
        virtio_free_queues(hw);
 
-#ifdef RTE_VIRTIO_USER
-       if (hw->virtio_user_dev)
-               virtio_user_dev_uninit(hw->virtio_user_dev);
-       else
-#endif
-       if (dev->device) {
-               rte_pci_unmap_device(RTE_ETH_DEV_TO_PCI(dev));
-               if (!hw->modern)
-                       rte_pci_ioport_unmap(VTPCI_IO(hw));
-       }
+       return VIRTIO_OPS(hw)->dev_close(hw);
 }
 
 static int
@@ -753,7 +741,7 @@ virtio_dev_promiscuous_enable(struct rte_eth_dev *dev)
        int dlen[1];
        int ret;
 
-       if (!vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_RX)) {
+       if (!virtio_with_feature(hw, VIRTIO_NET_F_CTRL_RX)) {
                PMD_INIT_LOG(INFO, "host does not support rx control");
                return -ENOTSUP;
        }
@@ -780,7 +768,7 @@ virtio_dev_promiscuous_disable(struct rte_eth_dev *dev)
        int dlen[1];
        int ret;
 
-       if (!vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_RX)) {
+       if (!virtio_with_feature(hw, VIRTIO_NET_F_CTRL_RX)) {
                PMD_INIT_LOG(INFO, "host does not support rx control");
                return -ENOTSUP;
        }
@@ -807,7 +795,7 @@ virtio_dev_allmulticast_enable(struct rte_eth_dev *dev)
        int dlen[1];
        int ret;
 
-       if (!vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_RX)) {
+       if (!virtio_with_feature(hw, VIRTIO_NET_F_CTRL_RX)) {
                PMD_INIT_LOG(INFO, "host does not support rx control");
                return -ENOTSUP;
        }
@@ -834,7 +822,7 @@ virtio_dev_allmulticast_disable(struct rte_eth_dev *dev)
        int dlen[1];
        int ret;
 
-       if (!vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_RX)) {
+       if (!virtio_with_feature(hw, VIRTIO_NET_F_CTRL_RX)) {
                PMD_INIT_LOG(INFO, "host does not support rx control");
                return -ENOTSUP;
        }
@@ -853,6 +841,56 @@ virtio_dev_allmulticast_disable(struct rte_eth_dev *dev)
        return 0;
 }
 
+uint16_t
+virtio_rx_mem_pool_buf_size(struct rte_mempool *mp)
+{
+       return rte_pktmbuf_data_room_size(mp) - RTE_PKTMBUF_HEADROOM;
+}
+
+bool
+virtio_rx_check_scatter(uint16_t max_rx_pkt_len, uint16_t rx_buf_size,
+                       bool rx_scatter_enabled, const char **error)
+{
+       if (!rx_scatter_enabled && max_rx_pkt_len > rx_buf_size) {
+               *error = "Rx scatter is disabled and RxQ mbuf pool object size is too small";
+               return false;
+       }
+
+       return true;
+}
+
+static bool
+virtio_check_scatter_on_all_rx_queues(struct rte_eth_dev *dev,
+                                     uint16_t frame_size)
+{
+       struct virtio_hw *hw = dev->data->dev_private;
+       struct virtnet_rx *rxvq;
+       struct virtqueue *vq;
+       unsigned int qidx;
+       uint16_t buf_size;
+       const char *error;
+
+       if (hw->vqs == NULL)
+               return true;
+
+       for (qidx = 0; (vq = hw->vqs[2 * qidx + VTNET_SQ_RQ_QUEUE_IDX]) != NULL;
+            qidx++) {
+               rxvq = &vq->rxq;
+               if (rxvq->mpool == NULL)
+                       continue;
+               buf_size = virtio_rx_mem_pool_buf_size(rxvq->mpool);
+
+               if (!virtio_rx_check_scatter(frame_size, buf_size,
+                                            hw->rx_ol_scatter, &error)) {
+                       PMD_INIT_LOG(ERR, "MTU check for RxQ %u failed: %s",
+                                    qidx, error);
+                       return false;
+               }
+       }
+
+       return true;
+}
+
 #define VLAN_TAG_LEN           4    /* 802.3ac tag (not DMA'd) */
 static int
 virtio_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
@@ -870,6 +908,15 @@ virtio_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
                        RTE_ETHER_MIN_MTU, max_frame_size - ether_hdr_len);
                return -EINVAL;
        }
+
+       if (!virtio_check_scatter_on_all_rx_queues(dev, frame_size)) {
+               PMD_INIT_LOG(ERR, "MTU vs Rx scatter and Rx buffers check failed");
+               return -EINVAL;
+       }
+
+       hw->max_rx_pkt_len = frame_size;
+       dev->data->dev_conf.rxmode.max_rx_pkt_len = hw->max_rx_pkt_len;
+
        return 0;
 }
 
@@ -878,7 +925,7 @@ virtio_dev_rx_queue_intr_enable(struct rte_eth_dev *dev, uint16_t queue_id)
 {
        struct virtio_hw *hw = dev->data->dev_private;
        struct virtnet_rx *rxvq = dev->data->rx_queues[queue_id];
-       struct virtqueue *vq = rxvq->vq;
+       struct virtqueue *vq = virtnet_rxq_to_vq(rxvq);
 
        virtqueue_enable_intr(vq);
        virtio_mb(hw->weak_barriers);
@@ -889,7 +936,7 @@ static int
 virtio_dev_rx_queue_intr_disable(struct rte_eth_dev *dev, uint16_t queue_id)
 {
        struct virtnet_rx *rxvq = dev->data->rx_queues[queue_id];
-       struct virtqueue *vq = rxvq->vq;
+       struct virtqueue *vq = virtnet_rxq_to_vq(rxvq);
 
        virtqueue_disable_intr(vq);
        return 0;
@@ -1120,7 +1167,7 @@ virtio_dev_stats_reset(struct rte_eth_dev *dev)
 static void
 virtio_set_hwaddr(struct virtio_hw *hw)
 {
-       vtpci_write_dev_config(hw,
+       virtio_write_dev_config(hw,
                        offsetof(struct virtio_net_config, mac),
                        &hw->mac_addr, RTE_ETHER_ADDR_LEN);
 }
@@ -1128,8 +1175,8 @@ virtio_set_hwaddr(struct virtio_hw *hw)
 static void
 virtio_get_hwaddr(struct virtio_hw *hw)
 {
-       if (vtpci_with_feature(hw, VIRTIO_NET_F_MAC)) {
-               vtpci_read_dev_config(hw,
+       if (virtio_with_feature(hw, VIRTIO_NET_F_MAC)) {
+               virtio_read_dev_config(hw,
                        offsetof(struct virtio_net_config, mac),
                        &hw->mac_addr, RTE_ETHER_ADDR_LEN);
        } else {
@@ -1146,7 +1193,7 @@ virtio_mac_table_set(struct virtio_hw *hw,
        struct virtio_pmd_ctrl ctrl;
        int err, len[2];
 
-       if (!vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
+       if (!virtio_with_feature(hw, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
                PMD_DRV_LOG(INFO, "host does not support mac table");
                return -1;
        }
@@ -1241,7 +1288,7 @@ virtio_mac_addr_set(struct rte_eth_dev *dev, struct rte_ether_addr *mac_addr)
        memcpy(hw->mac_addr, mac_addr, RTE_ETHER_ADDR_LEN);
 
        /* Use atomic update if available */
-       if (vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
+       if (virtio_with_feature(hw, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
                struct virtio_pmd_ctrl ctrl;
                int len = RTE_ETHER_ADDR_LEN;
 
@@ -1252,7 +1299,7 @@ virtio_mac_addr_set(struct rte_eth_dev *dev, struct rte_ether_addr *mac_addr)
                return virtio_send_command(hw->cvq, &ctrl, &len, 1);
        }
 
-       if (!vtpci_with_feature(hw, VIRTIO_NET_F_MAC))
+       if (!virtio_with_feature(hw, VIRTIO_NET_F_MAC))
                return -ENOTSUP;
 
        virtio_set_hwaddr(hw);
@@ -1266,7 +1313,7 @@ virtio_vlan_filter_set(struct rte_eth_dev *dev, uint16_t vlan_id, int on)
        struct virtio_pmd_ctrl ctrl;
        int len;
 
-       if (!vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_VLAN))
+       if (!virtio_with_feature(hw, VIRTIO_NET_F_CTRL_VLAN))
                return -ENOTSUP;
 
        ctrl.hdr.class = VIRTIO_NET_CTRL_VLAN;
@@ -1285,8 +1332,8 @@ virtio_intr_unmask(struct rte_eth_dev *dev)
        if (rte_intr_ack(dev->intr_handle) < 0)
                return -1;
 
-       if (!hw->virtio_user_dev)
-               hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
+       if (VIRTIO_OPS(hw)->intr_detect)
+               VIRTIO_OPS(hw)->intr_detect(hw);
 
        return 0;
 }
@@ -1299,8 +1346,8 @@ virtio_intr_enable(struct rte_eth_dev *dev)
        if (rte_intr_enable(dev->intr_handle) < 0)
                return -1;
 
-       if (!hw->virtio_user_dev)
-               hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
+       if (VIRTIO_OPS(hw)->intr_detect)
+               VIRTIO_OPS(hw)->intr_detect(hw);
 
        return 0;
 }
@@ -1313,14 +1360,14 @@ virtio_intr_disable(struct rte_eth_dev *dev)
        if (rte_intr_disable(dev->intr_handle) < 0)
                return -1;
 
-       if (!hw->virtio_user_dev)
-               hw->use_msix = vtpci_msix_detect(RTE_ETH_DEV_TO_PCI(dev));
+       if (VIRTIO_OPS(hw)->intr_detect)
+               VIRTIO_OPS(hw)->intr_detect(hw);
 
        return 0;
 }
 
 static int
-virtio_negotiate_features(struct virtio_hw *hw, uint64_t req_features)
+virtio_ethdev_negotiate_features(struct virtio_hw *hw, uint64_t req_features)
 {
        uint64_t host_features;
 
@@ -1329,7 +1376,7 @@ virtio_negotiate_features(struct virtio_hw *hw, uint64_t req_features)
                req_features);
 
        /* Read device(host) feature bits */
-       host_features = VTPCI_OPS(hw)->get_features(hw);
+       host_features = VIRTIO_OPS(hw)->get_features(hw);
        PMD_INIT_LOG(DEBUG, "host_features before negotiate = %" PRIx64,
                host_features);
 
@@ -1337,7 +1384,7 @@ virtio_negotiate_features(struct virtio_hw *hw, uint64_t req_features)
        if (host_features & req_features & (1ULL << VIRTIO_NET_F_MTU)) {
                struct virtio_net_config config;
 
-               vtpci_read_dev_config(hw,
+               virtio_read_dev_config(hw,
                        offsetof(struct virtio_net_config, mtu),
                        &config.mtu, sizeof(config.mtu));
 
@@ -1350,20 +1397,18 @@ virtio_negotiate_features(struct virtio_hw *hw, uint64_t req_features)
         * guest feature bits.
         */
        hw->guest_features = req_features;
-       hw->guest_features = vtpci_negotiate_features(hw, host_features);
+       hw->guest_features = virtio_negotiate_features(hw, host_features);
        PMD_INIT_LOG(DEBUG, "features after negotiate = %" PRIx64,
                hw->guest_features);
 
-       if (hw->modern) {
-               if (!vtpci_with_feature(hw, VIRTIO_F_VERSION_1)) {
-                       PMD_INIT_LOG(ERR,
-                               "VIRTIO_F_VERSION_1 features is not enabled.");
-                       return -1;
-               }
-               vtpci_set_status(hw, VIRTIO_CONFIG_STATUS_FEATURES_OK);
-               if (!(vtpci_get_status(hw) & VIRTIO_CONFIG_STATUS_FEATURES_OK)) {
-                       PMD_INIT_LOG(ERR,
-                               "failed to set FEATURES_OK status!");
+       if (VIRTIO_OPS(hw)->features_ok(hw) < 0)
+               return -1;
+
+       if (virtio_with_feature(hw, VIRTIO_F_VERSION_1)) {
+               virtio_set_status(hw, VIRTIO_CONFIG_STATUS_FEATURES_OK);
+
+               if (!(virtio_get_status(hw) & VIRTIO_CONFIG_STATUS_FEATURES_OK)) {
+                       PMD_INIT_LOG(ERR, "Failed to set FEATURES_OK status!");
                        return -1;
                }
        }
@@ -1481,20 +1526,20 @@ virtio_interrupt_handler(void *param)
        uint16_t status;
 
        /* Read interrupt status which clears interrupt */
-       isr = vtpci_isr(hw);
+       isr = virtio_get_isr(hw);
        PMD_DRV_LOG(INFO, "interrupt status = %#x", isr);
 
        if (virtio_intr_unmask(dev) < 0)
                PMD_DRV_LOG(ERR, "interrupt enable failed");
 
-       if (isr & VIRTIO_PCI_ISR_CONFIG) {
+       if (isr & VIRTIO_ISR_CONFIG) {
                if (virtio_dev_link_update(dev, 0) == 0)
-                       _rte_eth_dev_callback_process(dev,
-                                                     RTE_ETH_EVENT_INTR_LSC,
-                                                     NULL);
+                       rte_eth_dev_callback_process(dev,
+                                                    RTE_ETH_EVENT_INTR_LSC,
+                                                    NULL);
 
-               if (vtpci_with_feature(hw, VIRTIO_NET_F_STATUS)) {
-                       vtpci_read_dev_config(hw,
+               if (virtio_with_feature(hw, VIRTIO_NET_F_STATUS)) {
+                       virtio_read_dev_config(hw,
                                offsetof(struct virtio_net_config, status),
                                &status, sizeof(status));
                        if (status & VIRTIO_NET_S_ANNOUNCE) {
@@ -1513,7 +1558,7 @@ set_rxtx_funcs(struct rte_eth_dev *eth_dev)
        struct virtio_hw *hw = eth_dev->data->dev_private;
 
        eth_dev->tx_pkt_prepare = virtio_xmit_pkts_prepare;
-       if (vtpci_packed_queue(hw)) {
+       if (virtio_with_packed_queue(hw)) {
                PMD_INIT_LOG(INFO,
                        "virtio: using packed ring %s Tx path on port %u",
                        hw->use_vec_tx ? "vectorized" : "standard",
@@ -1534,14 +1579,14 @@ set_rxtx_funcs(struct rte_eth_dev *eth_dev)
                }
        }
 
-       if (vtpci_packed_queue(hw)) {
+       if (virtio_with_packed_queue(hw)) {
                if (hw->use_vec_rx) {
                        PMD_INIT_LOG(INFO,
                                "virtio: using packed ring vectorized Rx path on port %u",
                                eth_dev->data->port_id);
                        eth_dev->rx_pkt_burst =
                                &virtio_recv_pkts_packed_vec;
-               } else if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) {
+               } else if (virtio_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) {
                        PMD_INIT_LOG(INFO,
                                "virtio: using packed ring mergeable buffer Rx path on port %u",
                                eth_dev->data->port_id);
@@ -1563,7 +1608,7 @@ set_rxtx_funcs(struct rte_eth_dev *eth_dev)
                                "virtio: using inorder Rx path on port %u",
                                eth_dev->data->port_id);
                        eth_dev->rx_pkt_burst = &virtio_recv_pkts_inorder;
-               } else if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) {
+               } else if (virtio_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) {
                        PMD_INIT_LOG(INFO,
                                "virtio: using mergeable buffer Rx path on port %u",
                                eth_dev->data->port_id);
@@ -1590,7 +1635,7 @@ virtio_queues_bind_intr(struct rte_eth_dev *dev)
        PMD_INIT_LOG(INFO, "queue/interrupt binding");
        for (i = 0; i < dev->data->nb_rx_queues; ++i) {
                dev->intr_handle->intr_vec[i] = i + 1;
-               if (VTPCI_OPS(hw)->set_queue_irq(hw, hw->vqs[i * 2], i + 1) ==
+               if (VIRTIO_OPS(hw)->set_queue_irq(hw, hw->vqs[i * 2], i + 1) ==
                                                 VIRTIO_MSI_NO_VECTOR) {
                        PMD_DRV_LOG(ERR, "failed to set queue vector");
                        return -EBUSY;
@@ -1608,7 +1653,7 @@ virtio_queues_unbind_intr(struct rte_eth_dev *dev)
 
        PMD_INIT_LOG(INFO, "queue/interrupt unbinding");
        for (i = 0; i < dev->data->nb_rx_queues; ++i)
-               VTPCI_OPS(hw)->set_queue_irq(hw,
+               VIRTIO_OPS(hw)->set_queue_irq(hw,
                                             hw->vqs[i * VTNET_CQ],
                                             VIRTIO_MSI_NO_VECTOR);
 }
@@ -1665,7 +1710,6 @@ virtio_configure_intr(struct rte_eth_dev *dev)
 
        return 0;
 }
-#define SPEED_UNKNOWN    0xffffffff
 #define DUPLEX_UNKNOWN   0xff
 /* reset device and renegotiate features if needed */
 static int
@@ -1674,11 +1718,10 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t req_features)
        struct virtio_hw *hw = eth_dev->data->dev_private;
        struct virtio_net_config *config;
        struct virtio_net_config local_config;
-       struct rte_pci_device *pci_dev = NULL;
        int ret;
 
        /* Reset the device although not necessary at startup */
-       vtpci_reset(hw);
+       virtio_reset(hw);
 
        if (hw->vqs) {
                virtio_dev_free_mbufs(eth_dev);
@@ -1686,29 +1729,27 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t req_features)
        }
 
        /* Tell the host we've noticed this device. */
-       vtpci_set_status(hw, VIRTIO_CONFIG_STATUS_ACK);
+       virtio_set_status(hw, VIRTIO_CONFIG_STATUS_ACK);
 
        /* Tell the host we've known how to drive the device. */
-       vtpci_set_status(hw, VIRTIO_CONFIG_STATUS_DRIVER);
-       if (virtio_negotiate_features(hw, req_features) < 0)
+       virtio_set_status(hw, VIRTIO_CONFIG_STATUS_DRIVER);
+       if (virtio_ethdev_negotiate_features(hw, req_features) < 0)
                return -1;
 
-       hw->weak_barriers = !vtpci_with_feature(hw, VIRTIO_F_ORDER_PLATFORM);
-
-       if (!hw->virtio_user_dev)
-               pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
+       hw->weak_barriers = !virtio_with_feature(hw, VIRTIO_F_ORDER_PLATFORM);
 
        /* If host does not support both status and MSI-X then disable LSC */
-       if (vtpci_with_feature(hw, VIRTIO_NET_F_STATUS) &&
-           hw->use_msix != VIRTIO_MSIX_NONE)
+       if (virtio_with_feature(hw, VIRTIO_NET_F_STATUS) && hw->intr_lsc)
                eth_dev->data->dev_flags |= RTE_ETH_DEV_INTR_LSC;
        else
                eth_dev->data->dev_flags &= ~RTE_ETH_DEV_INTR_LSC;
 
+       eth_dev->data->dev_flags |= RTE_ETH_DEV_AUTOFILL_QUEUE_XSTATS;
+
        /* Setting up rx_header size for the device */
-       if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF) ||
-           vtpci_with_feature(hw, VIRTIO_F_VERSION_1) ||
-           vtpci_with_feature(hw, VIRTIO_F_RING_PACKED))
+       if (virtio_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF) ||
+           virtio_with_feature(hw, VIRTIO_F_VERSION_1) ||
+           virtio_with_packed_queue(hw))
                hw->vtnet_hdr_size = sizeof(struct virtio_net_hdr_mrg_rxbuf);
        else
                hw->vtnet_hdr_size = sizeof(struct virtio_net_hdr);
@@ -1718,38 +1759,36 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t req_features)
        rte_ether_addr_copy((struct rte_ether_addr *)hw->mac_addr,
                        &eth_dev->data->mac_addrs[0]);
        PMD_INIT_LOG(DEBUG,
-                    "PORT MAC: %02X:%02X:%02X:%02X:%02X:%02X",
+                    "PORT MAC: " RTE_ETHER_ADDR_PRT_FMT,
                     hw->mac_addr[0], hw->mac_addr[1], hw->mac_addr[2],
                     hw->mac_addr[3], hw->mac_addr[4], hw->mac_addr[5]);
 
-       if (hw->speed == SPEED_UNKNOWN) {
-               if (vtpci_with_feature(hw, VIRTIO_NET_F_SPEED_DUPLEX)) {
+       if (hw->speed == ETH_SPEED_NUM_UNKNOWN) {
+               if (virtio_with_feature(hw, VIRTIO_NET_F_SPEED_DUPLEX)) {
                        config = &local_config;
-                       vtpci_read_dev_config(hw,
+                       virtio_read_dev_config(hw,
                                offsetof(struct virtio_net_config, speed),
                                &config->speed, sizeof(config->speed));
-                       vtpci_read_dev_config(hw,
+                       virtio_read_dev_config(hw,
                                offsetof(struct virtio_net_config, duplex),
                                &config->duplex, sizeof(config->duplex));
                        hw->speed = config->speed;
                        hw->duplex = config->duplex;
                }
        }
-       if (hw->speed == SPEED_UNKNOWN)
-               hw->speed = ETH_SPEED_NUM_10G;
        if (hw->duplex == DUPLEX_UNKNOWN)
                hw->duplex = ETH_LINK_FULL_DUPLEX;
        PMD_INIT_LOG(DEBUG, "link speed = %d, duplex = %d",
                hw->speed, hw->duplex);
-       if (vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_VQ)) {
+       if (virtio_with_feature(hw, VIRTIO_NET_F_CTRL_VQ)) {
                config = &local_config;
 
-               vtpci_read_dev_config(hw,
+               virtio_read_dev_config(hw,
                        offsetof(struct virtio_net_config, mac),
                        &config->mac, sizeof(config->mac));
 
-               if (vtpci_with_feature(hw, VIRTIO_NET_F_STATUS)) {
-                       vtpci_read_dev_config(hw,
+               if (virtio_with_feature(hw, VIRTIO_NET_F_STATUS)) {
+                       virtio_read_dev_config(hw,
                                offsetof(struct virtio_net_config, status),
                                &config->status, sizeof(config->status));
                } else {
@@ -1758,8 +1797,8 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t req_features)
                        config->status = 0;
                }
 
-               if (vtpci_with_feature(hw, VIRTIO_NET_F_MQ)) {
-                       vtpci_read_dev_config(hw,
+               if (virtio_with_feature(hw, VIRTIO_NET_F_MQ)) {
+                       virtio_read_dev_config(hw,
                                offsetof(struct virtio_net_config, max_virtqueue_pairs),
                                &config->max_virtqueue_pairs,
                                sizeof(config->max_virtqueue_pairs));
@@ -1771,8 +1810,8 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t req_features)
 
                hw->max_queue_pairs = config->max_virtqueue_pairs;
 
-               if (vtpci_with_feature(hw, VIRTIO_NET_F_MTU)) {
-                       vtpci_read_dev_config(hw,
+               if (virtio_with_feature(hw, VIRTIO_NET_F_MTU)) {
+                       virtio_read_dev_config(hw,
                                offsetof(struct virtio_net_config, mtu),
                                &config->mtu,
                                sizeof(config->mtu));
@@ -1801,7 +1840,7 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t req_features)
                                config->max_virtqueue_pairs);
                PMD_INIT_LOG(DEBUG, "config->status=%d", config->status);
                PMD_INIT_LOG(DEBUG,
-                               "PORT MAC: %02X:%02X:%02X:%02X:%02X:%02X",
+                               "PORT MAC: " RTE_ETHER_ADDR_PRT_FMT,
                                config->mac[0], config->mac[1],
                                config->mac[2], config->mac[3],
                                config->mac[4], config->mac[5]);
@@ -1824,62 +1863,11 @@ virtio_init_device(struct rte_eth_dev *eth_dev, uint64_t req_features)
                }
        }
 
-       vtpci_reinit_complete(hw);
-
-       if (pci_dev)
-               PMD_INIT_LOG(DEBUG, "port %d vendorID=0x%x deviceID=0x%x",
-                       eth_dev->data->port_id, pci_dev->id.vendor_id,
-                       pci_dev->id.device_id);
+       virtio_reinit_complete(hw);
 
        return 0;
 }
 
-/*
- * Remap the PCI device again (IO port map for legacy device and
- * memory map for modern device), so that the secondary process
- * could have the PCI initiated correctly.
- */
-static int
-virtio_remap_pci(struct rte_pci_device *pci_dev, struct virtio_hw *hw)
-{
-       if (hw->modern) {
-               /*
-                * We don't have to re-parse the PCI config space, since
-                * rte_pci_map_device() makes sure the mapped address
-                * in secondary process would equal to the one mapped in
-                * the primary process: error will be returned if that
-                * requirement is not met.
-                *
-                * That said, we could simply reuse all cap pointers
-                * (such as dev_cfg, common_cfg, etc.) parsed from the
-                * primary process, which is stored in shared memory.
-                */
-               if (rte_pci_map_device(pci_dev)) {
-                       PMD_INIT_LOG(DEBUG, "failed to map pci device!");
-                       return -1;
-               }
-       } else {
-               if (rte_pci_ioport_map(pci_dev, 0, VTPCI_IO(hw)) < 0)
-                       return -1;
-       }
-
-       return 0;
-}
-
-static void
-virtio_set_vtpci_ops(struct virtio_hw *hw)
-{
-#ifdef RTE_VIRTIO_USER
-       if (hw->virtio_user_dev)
-               VTPCI_OPS(hw) = &virtio_user_ops;
-       else
-#endif
-       if (hw->modern)
-               VTPCI_OPS(hw) = &modern_ops;
-       else
-               VTPCI_OPS(hw) = &legacy_ops;
-}
-
 /*
  * This function is based on probe() function in virtio_pci.c
  * It returns 0 on success.
@@ -1888,7 +1876,7 @@ int
 eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
 {
        struct virtio_hw *hw = eth_dev->data->dev_private;
-       uint32_t speed = SPEED_UNKNOWN;
+       uint32_t speed = ETH_SPEED_NUM_UNKNOWN;
        int vectorized = 0;
        int ret;
 
@@ -1905,27 +1893,15 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
        eth_dev->rx_descriptor_done = virtio_dev_rx_queue_done;
 
        if (rte_eal_process_type() == RTE_PROC_SECONDARY) {
-               if (!hw->virtio_user_dev) {
-                       ret = virtio_remap_pci(RTE_ETH_DEV_TO_PCI(eth_dev), hw);
-                       if (ret)
-                               return ret;
-               }
-
-               virtio_set_vtpci_ops(hw);
                set_rxtx_funcs(eth_dev);
-
                return 0;
        }
-       ret = virtio_dev_devargs_parse(eth_dev->device->devargs,
-                NULL, &speed, &vectorized);
+
+       ret = virtio_dev_devargs_parse(eth_dev->device->devargs, &speed, &vectorized);
        if (ret < 0)
                return ret;
        hw->speed = speed;
-       /*
-        * Pass the information to the rte_eth_dev_close() that it should also
-        * release the private port resources.
-        */
-       eth_dev->data->dev_flags |= RTE_ETH_DEV_CLOSE_REMOVE;
+       hw->duplex = DUPLEX_UNKNOWN;
 
        /* Allocate memory for storing MAC addresses */
        eth_dev->data->mac_addrs = rte_zmalloc("virtio",
@@ -1937,16 +1913,6 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
                return -ENOMEM;
        }
 
-       hw->port_id = eth_dev->data->port_id;
-       /* For virtio_user case the hw->virtio_user_dev is populated by
-        * virtio_user_eth_dev_alloc() before eth_virtio_dev_init() is called.
-        */
-       if (!hw->virtio_user_dev) {
-               ret = vtpci_init(RTE_ETH_DEV_TO_PCI(eth_dev), hw);
-               if (ret)
-                       goto err_vtpci_init;
-       }
-
        rte_spinlock_init(&hw->state_lock);
 
        /* reset device and negotiate default features */
@@ -1955,68 +1921,29 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
                goto err_virtio_init;
 
        if (vectorized) {
-               if (!vtpci_packed_queue(hw)) {
+               if (!virtio_with_packed_queue(hw)) {
                        hw->use_vec_rx = 1;
                } else {
-#if !defined(CC_AVX512_SUPPORT)
-                       PMD_DRV_LOG(INFO,
-                               "building environment do not support packed ring vectorized");
-#else
+#if defined(CC_AVX512_SUPPORT) || defined(RTE_ARCH_ARM)
                        hw->use_vec_rx = 1;
                        hw->use_vec_tx = 1;
+#else
+                       PMD_DRV_LOG(INFO,
+                               "building environment do not support packed ring vectorized");
 #endif
                }
        }
 
-       hw->opened = true;
+       hw->opened = 1;
 
        return 0;
 
 err_virtio_init:
-       if (!hw->virtio_user_dev) {
-               rte_pci_unmap_device(RTE_ETH_DEV_TO_PCI(eth_dev));
-               if (!hw->modern)
-                       rte_pci_ioport_unmap(VTPCI_IO(hw));
-       }
-err_vtpci_init:
        rte_free(eth_dev->data->mac_addrs);
        eth_dev->data->mac_addrs = NULL;
        return ret;
 }
 
-static int
-eth_virtio_dev_uninit(struct rte_eth_dev *eth_dev)
-{
-       PMD_INIT_FUNC_TRACE();
-
-       if (rte_eal_process_type() == RTE_PROC_SECONDARY)
-               return 0;
-
-       virtio_dev_stop(eth_dev);
-       virtio_dev_close(eth_dev);
-
-       eth_dev->dev_ops = NULL;
-       eth_dev->tx_pkt_burst = NULL;
-       eth_dev->rx_pkt_burst = NULL;
-
-       PMD_INIT_LOG(DEBUG, "dev_uninit completed");
-
-       return 0;
-}
-
-
-static int vdpa_check_handler(__rte_unused const char *key,
-               const char *value, void *ret_val)
-{
-       if (strcmp(value, "1") == 0)
-               *(int *)ret_val = 1;
-       else
-               *(int *)ret_val = 0;
-
-       return 0;
-}
-
-
 static uint32_t
 virtio_dev_speed_capa_get(uint32_t speed)
 {
@@ -2035,6 +1962,8 @@ virtio_dev_speed_capa_get(uint32_t speed)
                return ETH_LINK_SPEED_56G;
        case ETH_SPEED_NUM_100G:
                return ETH_LINK_SPEED_100G;
+       case ETH_SPEED_NUM_200G:
+               return ETH_LINK_SPEED_200G;
        default:
                return 0;
        }
@@ -2052,10 +1981,8 @@ static int vectorized_check_handler(__rte_unused const char *key,
 }
 
 #define VIRTIO_ARG_SPEED      "speed"
-#define VIRTIO_ARG_VDPA       "vdpa"
 #define VIRTIO_ARG_VECTORIZED "vectorized"
 
-
 static int
 link_speed_handler(const char *key __rte_unused,
                const char *value, void *ret_val)
@@ -2074,8 +2001,7 @@ link_speed_handler(const char *key __rte_unused,
 
 
 static int
-virtio_dev_devargs_parse(struct rte_devargs *devargs, int *vdpa,
-       uint32_t *speed, int *vectorized)
+virtio_dev_devargs_parse(struct rte_devargs *devargs, uint32_t *speed, int *vectorized)
 {
        struct rte_kvargs *kvlist;
        int ret = 0;
@@ -2088,18 +2014,7 @@ virtio_dev_devargs_parse(struct rte_devargs *devargs, int *vdpa,
                PMD_INIT_LOG(ERR, "error when parsing param");
                return 0;
        }
-       if (vdpa && rte_kvargs_count(kvlist, VIRTIO_ARG_VDPA) == 1) {
-               /* vdpa mode selected when there's a key-value pair:
-                * vdpa=1
-                */
-               ret = rte_kvargs_process(kvlist, VIRTIO_ARG_VDPA,
-                               vdpa_check_handler, vdpa);
-               if (ret < 0) {
-                       PMD_INIT_LOG(ERR, "Failed to parse %s",
-                               VIRTIO_ARG_VDPA);
-                       goto exit;
-               }
-       }
+
        if (speed && rte_kvargs_count(kvlist, VIRTIO_ARG_SPEED) == 1) {
                ret = rte_kvargs_process(kvlist,
                                        VIRTIO_ARG_SPEED,
@@ -2128,67 +2043,20 @@ exit:
        return ret;
 }
 
-static int eth_virtio_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
-       struct rte_pci_device *pci_dev)
-{
-       int vdpa = 0;
-       int ret = 0;
-
-       ret = virtio_dev_devargs_parse(pci_dev->device.devargs, &vdpa, NULL,
-               NULL);
-       if (ret < 0) {
-               PMD_INIT_LOG(ERR, "devargs parsing is failed");
-               return ret;
-       }
-       /* virtio pmd skips probe if device needs to work in vdpa mode */
-       if (vdpa == 1)
-               return 1;
-
-       return rte_eth_dev_pci_generic_probe(pci_dev, sizeof(struct virtio_hw),
-               eth_virtio_dev_init);
-}
-
-static int eth_virtio_pci_remove(struct rte_pci_device *pci_dev)
-{
-       int ret;
-
-       ret = rte_eth_dev_pci_generic_remove(pci_dev, eth_virtio_dev_uninit);
-       /* Port has already been released by close. */
-       if (ret == -ENODEV)
-               ret = 0;
-       return ret;
-}
-
-static struct rte_pci_driver rte_virtio_pmd = {
-       .driver = {
-               .name = "net_virtio",
-       },
-       .id_table = pci_id_virtio_map,
-       .drv_flags = 0,
-       .probe = eth_virtio_pci_probe,
-       .remove = eth_virtio_pci_remove,
-};
-
-RTE_INIT(rte_virtio_pmd_init)
-{
-       rte_eal_iopl_init();
-       rte_pci_register(&rte_virtio_pmd);
-}
-
-static bool
+static uint8_t
 rx_offload_enabled(struct virtio_hw *hw)
 {
-       return vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_CSUM) ||
-               vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4) ||
-               vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6);
+       return virtio_with_feature(hw, VIRTIO_NET_F_GUEST_CSUM) ||
+               virtio_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4) ||
+               virtio_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6);
 }
 
-static bool
+static uint8_t
 tx_offload_enabled(struct virtio_hw *hw)
 {
-       return vtpci_with_feature(hw, VIRTIO_NET_F_CSUM) ||
-               vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO4) ||
-               vtpci_with_feature(hw, VIRTIO_NET_F_HOST_TSO6);
+       return virtio_with_feature(hw, VIRTIO_NET_F_CSUM) ||
+               virtio_with_feature(hw, VIRTIO_NET_F_HOST_TSO4) ||
+               virtio_with_feature(hw, VIRTIO_NET_F_HOST_TSO6);
 }
 
 /*
@@ -2234,6 +2102,8 @@ virtio_dev_configure(struct rte_eth_dev *dev)
        if (rxmode->max_rx_pkt_len > hw->max_mtu + ether_hdr_len)
                req_features &= ~(1ULL << VIRTIO_NET_F_MTU);
 
+       hw->max_rx_pkt_len = rxmode->max_rx_pkt_len;
+
        if (rx_offloads & (DEV_RX_OFFLOAD_UDP_CKSUM |
                           DEV_RX_OFFLOAD_TCP_CKSUM))
                req_features |= (1ULL << VIRTIO_NET_F_GUEST_CSUM);
@@ -2261,29 +2131,31 @@ virtio_dev_configure(struct rte_eth_dev *dev)
 
        if ((rx_offloads & (DEV_RX_OFFLOAD_UDP_CKSUM |
                            DEV_RX_OFFLOAD_TCP_CKSUM)) &&
-               !vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_CSUM)) {
+               !virtio_with_feature(hw, VIRTIO_NET_F_GUEST_CSUM)) {
                PMD_DRV_LOG(ERR,
                        "rx checksum not available on this host");
                return -ENOTSUP;
        }
 
        if ((rx_offloads & DEV_RX_OFFLOAD_TCP_LRO) &&
-               (!vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4) ||
-                !vtpci_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6))) {
+               (!virtio_with_feature(hw, VIRTIO_NET_F_GUEST_TSO4) ||
+                !virtio_with_feature(hw, VIRTIO_NET_F_GUEST_TSO6))) {
                PMD_DRV_LOG(ERR,
                        "Large Receive Offload not available on this host");
                return -ENOTSUP;
        }
 
        /* start control queue */
-       if (vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_VQ))
+       if (virtio_with_feature(hw, VIRTIO_NET_F_CTRL_VQ))
                virtio_dev_cq_start(dev);
 
        if (rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
                hw->vlan_strip = 1;
 
-       if ((rx_offloads & DEV_RX_OFFLOAD_VLAN_FILTER)
-           && !vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_VLAN)) {
+       hw->rx_ol_scatter = (rx_offloads & DEV_RX_OFFLOAD_SCATTER);
+
+       if ((rx_offloads & DEV_RX_OFFLOAD_VLAN_FILTER) &&
+                       !virtio_with_feature(hw, VIRTIO_NET_F_CTRL_VLAN)) {
                PMD_DRV_LOG(ERR,
                            "vlan filtering not available on this host");
                return -ENOTSUP;
@@ -2294,18 +2166,30 @@ virtio_dev_configure(struct rte_eth_dev *dev)
 
        if (dev->data->dev_flags & RTE_ETH_DEV_INTR_LSC)
                /* Enable vector (0) for Link State Intrerrupt */
-               if (VTPCI_OPS(hw)->set_config_irq(hw, 0) ==
+               if (VIRTIO_OPS(hw)->set_config_irq(hw, 0) ==
                                VIRTIO_MSI_NO_VECTOR) {
                        PMD_DRV_LOG(ERR, "failed to set config vector");
                        return -EBUSY;
                }
 
-       if (vtpci_packed_queue(hw)) {
+       if (virtio_with_packed_queue(hw)) {
 #if defined(RTE_ARCH_X86_64) && defined(CC_AVX512_SUPPORT)
                if ((hw->use_vec_rx || hw->use_vec_tx) &&
                    (!rte_cpu_get_flag_enabled(RTE_CPUFLAG_AVX512F) ||
-                    !vtpci_with_feature(hw, VIRTIO_F_IN_ORDER) ||
-                    !vtpci_with_feature(hw, VIRTIO_F_VERSION_1))) {
+                    !virtio_with_feature(hw, VIRTIO_F_IN_ORDER) ||
+                    !virtio_with_feature(hw, VIRTIO_F_VERSION_1) ||
+                    rte_vect_get_max_simd_bitwidth() < RTE_VECT_SIMD_512)) {
+                       PMD_DRV_LOG(INFO,
+                               "disabled packed ring vectorized path for requirements not met");
+                       hw->use_vec_rx = 0;
+                       hw->use_vec_tx = 0;
+               }
+#elif defined(RTE_ARCH_ARM)
+               if ((hw->use_vec_rx || hw->use_vec_tx) &&
+                   (!rte_cpu_get_flag_enabled(RTE_CPUFLAG_NEON) ||
+                    !virtio_with_feature(hw, VIRTIO_F_IN_ORDER) ||
+                    !virtio_with_feature(hw, VIRTIO_F_VERSION_1) ||
+                    rte_vect_get_max_simd_bitwidth() < RTE_VECT_SIMD_128)) {
                        PMD_DRV_LOG(INFO,
                                "disabled packed ring vectorized path for requirements not met");
                        hw->use_vec_rx = 0;
@@ -2317,7 +2201,7 @@ virtio_dev_configure(struct rte_eth_dev *dev)
 #endif
 
                if (hw->use_vec_rx) {
-                       if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) {
+                       if (virtio_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) {
                                PMD_DRV_LOG(INFO,
                                        "disabled packed ring vectorized rx for mrg_rxbuf enabled");
                                hw->use_vec_rx = 0;
@@ -2330,21 +2214,21 @@ virtio_dev_configure(struct rte_eth_dev *dev)
                        }
                }
        } else {
-               if (vtpci_with_feature(hw, VIRTIO_F_IN_ORDER)) {
+               if (virtio_with_feature(hw, VIRTIO_F_IN_ORDER)) {
                        hw->use_inorder_tx = 1;
                        hw->use_inorder_rx = 1;
                        hw->use_vec_rx = 0;
                }
 
                if (hw->use_vec_rx) {
-#if defined RTE_ARCH_ARM64 || defined RTE_ARCH_ARM
+#if defined RTE_ARCH_ARM
                        if (!rte_cpu_get_flag_enabled(RTE_CPUFLAG_NEON)) {
                                PMD_DRV_LOG(INFO,
                                        "disabled split ring vectorized path for requirement not met");
                                hw->use_vec_rx = 0;
                        }
 #endif
-                       if (vtpci_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) {
+                       if (virtio_with_feature(hw, VIRTIO_NET_F_MRG_RXBUF)) {
                                PMD_DRV_LOG(INFO,
                                        "disabled split ring vectorized rx for mrg_rxbuf enabled");
                                hw->use_vec_rx = 0;
@@ -2358,6 +2242,12 @@ virtio_dev_configure(struct rte_eth_dev *dev)
                                        "disabled split ring vectorized rx for offloading enabled");
                                hw->use_vec_rx = 0;
                        }
+
+                       if (rte_vect_get_max_simd_bitwidth() < RTE_VECT_SIMD_128) {
+                               PMD_DRV_LOG(INFO,
+                                       "disabled split ring vectorized rx, max SIMD bitwidth too low");
+                               hw->use_vec_rx = 0;
+                       }
                }
        }
 
@@ -2369,8 +2259,7 @@ static int
 virtio_dev_start(struct rte_eth_dev *dev)
 {
        uint16_t nb_queues, i;
-       struct virtnet_rx *rxvq;
-       struct virtnet_tx *txvq __rte_unused;
+       struct virtqueue *vq;
        struct virtio_hw *hw = dev->data->dev_private;
        int ret;
 
@@ -2427,31 +2316,31 @@ virtio_dev_start(struct rte_eth_dev *dev)
        PMD_INIT_LOG(DEBUG, "nb_queues=%d", nb_queues);
 
        for (i = 0; i < dev->data->nb_rx_queues; i++) {
-               rxvq = dev->data->rx_queues[i];
+               vq = virtnet_rxq_to_vq(dev->data->rx_queues[i]);
                /* Flush the old packets */
-               virtqueue_rxvq_flush(rxvq->vq);
-               virtqueue_notify(rxvq->vq);
+               virtqueue_rxvq_flush(vq);
+               virtqueue_notify(vq);
        }
 
        for (i = 0; i < dev->data->nb_tx_queues; i++) {
-               txvq = dev->data->tx_queues[i];
-               virtqueue_notify(txvq->vq);
+               vq = virtnet_txq_to_vq(dev->data->tx_queues[i]);
+               virtqueue_notify(vq);
        }
 
        PMD_INIT_LOG(DEBUG, "Notified backend at initialization");
 
        for (i = 0; i < dev->data->nb_rx_queues; i++) {
-               rxvq = dev->data->rx_queues[i];
-               VIRTQUEUE_DUMP(rxvq->vq);
+               vq = virtnet_rxq_to_vq(dev->data->rx_queues[i]);
+               VIRTQUEUE_DUMP(vq);
        }
 
        for (i = 0; i < dev->data->nb_tx_queues; i++) {
-               txvq = dev->data->tx_queues[i];
-               VIRTQUEUE_DUMP(txvq->vq);
+               vq = virtnet_txq_to_vq(dev->data->tx_queues[i]);
+               VIRTQUEUE_DUMP(vq);
        }
 
        set_rxtx_funcs(dev);
-       hw->started = true;
+       hw->started = 1;
 
        /* Initialize Link state */
        virtio_dev_link_update(dev, 0);
@@ -2507,7 +2396,7 @@ static void virtio_dev_free_mbufs(struct rte_eth_dev *dev)
 /*
  * Stop device: disable interrupt and mark link down
  */
-static void
+int
 virtio_dev_stop(struct rte_eth_dev *dev)
 {
        struct virtio_hw *hw = dev->data->dev_private;
@@ -2515,11 +2404,12 @@ virtio_dev_stop(struct rte_eth_dev *dev)
        struct rte_intr_conf *intr_conf = &dev->data->dev_conf.intr_conf;
 
        PMD_INIT_LOG(DEBUG, "stop");
+       dev->data->dev_started = 0;
 
        rte_spinlock_lock(&hw->state_lock);
        if (!hw->started)
                goto out_unlock;
-       hw->started = false;
+       hw->started = 0;
 
        if (intr_conf->lsc || intr_conf->rxq) {
                virtio_intr_disable(dev);
@@ -2536,6 +2426,8 @@ virtio_dev_stop(struct rte_eth_dev *dev)
        rte_eth_linkstatus_set(dev, &link);
 out_unlock:
        rte_spinlock_unlock(&hw->state_lock);
+
+       return 0;
 }
 
 static int
@@ -2552,13 +2444,15 @@ virtio_dev_link_update(struct rte_eth_dev *dev, __rte_unused int wait_to_complet
 
        if (!hw->started) {
                link.link_status = ETH_LINK_DOWN;
-       } else if (vtpci_with_feature(hw, VIRTIO_NET_F_STATUS)) {
+               link.link_speed = ETH_SPEED_NUM_NONE;
+       } else if (virtio_with_feature(hw, VIRTIO_NET_F_STATUS)) {
                PMD_INIT_LOG(DEBUG, "Get link status from hw");
-               vtpci_read_dev_config(hw,
+               virtio_read_dev_config(hw,
                                offsetof(struct virtio_net_config, status),
                                &status, sizeof(status));
                if ((status & VIRTIO_NET_S_LINK_UP) == 0) {
                        link.link_status = ETH_LINK_DOWN;
+                       link.link_speed = ETH_SPEED_NUM_NONE;
                        PMD_INIT_LOG(DEBUG, "Port %d is down",
                                     dev->data->port_id);
                } else {
@@ -2582,7 +2476,7 @@ virtio_dev_vlan_offload_set(struct rte_eth_dev *dev, int mask)
 
        if (mask & ETH_VLAN_FILTER_MASK) {
                if ((offloads & DEV_RX_OFFLOAD_VLAN_FILTER) &&
-                               !vtpci_with_feature(hw, VIRTIO_NET_F_CTRL_VLAN)) {
+                               !virtio_with_feature(hw, VIRTIO_NET_F_CTRL_VLAN)) {
 
                        PMD_DRV_LOG(NOTICE,
                                "vlan filtering not available on this host");
@@ -2611,10 +2505,13 @@ virtio_dev_info_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
        dev_info->min_rx_bufsize = VIRTIO_MIN_RX_BUFSIZE;
        dev_info->max_rx_pktlen = VIRTIO_MAX_RX_PKTLEN;
        dev_info->max_mac_addrs = VIRTIO_MAX_MAC_ADDRS;
+       dev_info->max_mtu = hw->max_mtu;
 
-       host_features = VTPCI_OPS(hw)->get_features(hw);
+       host_features = VIRTIO_OPS(hw)->get_features(hw);
        dev_info->rx_offload_capa = DEV_RX_OFFLOAD_VLAN_STRIP;
        dev_info->rx_offload_capa |= DEV_RX_OFFLOAD_JUMBO_FRAME;
+       if (host_features & (1ULL << VIRTIO_NET_F_MRG_RXBUF))
+               dev_info->rx_offload_capa |= DEV_RX_OFFLOAD_SCATTER;
        if (host_features & (1ULL << VIRTIO_NET_F_GUEST_CSUM)) {
                dev_info->rx_offload_capa |=
                        DEV_RX_OFFLOAD_TCP_CKSUM |
@@ -2653,8 +2550,5 @@ __rte_unused uint8_t is_rx)
        return 0;
 }
 
-RTE_PMD_EXPORT_NAME(net_virtio, __COUNTER__);
-RTE_PMD_REGISTER_PCI_TABLE(net_virtio, pci_id_virtio_map);
-RTE_PMD_REGISTER_KMOD_DEP(net_virtio, "* igb_uio | uio_pci_generic | vfio-pci");
-RTE_LOG_REGISTER(virtio_logtype_init, pmd.net.virtio.init, NOTICE);
-RTE_LOG_REGISTER(virtio_logtype_driver, pmd.net.virtio.driver, NOTICE);
+RTE_LOG_REGISTER_SUFFIX(virtio_logtype_init, init, NOTICE);
+RTE_LOG_REGISTER_SUFFIX(virtio_logtype_driver, driver, NOTICE);