net: add rte prefix to ether structures
[dpdk.git] / drivers / net / vhost / rte_eth_vhost.c
index cdd8c31..6705e90 100644 (file)
@@ -1,56 +1,36 @@
-/*-
- *   BSD LICENSE
- *
- *   Copyright (c) 2016 IGEL Co., Ltd.
- *   All rights reserved.
- *
- *   Redistribution and use in source and binary forms, with or without
- *   modification, are permitted provided that the following conditions
- *   are met:
- *
- *     * Redistributions of source code must retain the above copyright
- *       notice, this list of conditions and the following disclaimer.
- *     * Redistributions in binary form must reproduce the above copyright
- *       notice, this list of conditions and the following disclaimer in
- *       the documentation and/or other materials provided with the
- *       distribution.
- *     * Neither the name of IGEL Co.,Ltd. nor the names of its
- *       contributors may be used to endorse or promote products derived
- *       from this software without specific prior written permission.
- *
- *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+/* SPDX-License-Identifier: BSD-3-Clause
+ * Copyright(c) 2016 IGEL Co., Ltd.
+ * Copyright(c) 2016-2018 Intel Corporation
  */
 #include <unistd.h>
 #include <pthread.h>
 #include <stdbool.h>
 
 #include <rte_mbuf.h>
-#include <rte_ethdev.h>
+#include <rte_ethdev_driver.h>
+#include <rte_ethdev_vdev.h>
 #include <rte_malloc.h>
 #include <rte_memcpy.h>
-#include <rte_vdev.h>
+#include <rte_bus_vdev.h>
 #include <rte_kvargs.h>
 #include <rte_vhost.h>
 #include <rte_spinlock.h>
 
 #include "rte_eth_vhost.h"
 
+static int vhost_logtype;
+
+#define VHOST_LOG(level, ...) \
+       rte_log(RTE_LOG_ ## level, vhost_logtype, __VA_ARGS__)
+
 enum {VIRTIO_RXQ, VIRTIO_TXQ, VIRTIO_QNUM};
 
 #define ETH_VHOST_IFACE_ARG            "iface"
 #define ETH_VHOST_QUEUES_ARG           "queues"
 #define ETH_VHOST_CLIENT_ARG           "client"
 #define ETH_VHOST_DEQUEUE_ZERO_COPY    "dequeue-zero-copy"
+#define ETH_VHOST_IOMMU_SUPPORT                "iommu-support"
+#define ETH_VHOST_POSTCOPY_SUPPORT     "postcopy-support"
 #define VHOST_MAX_PKT_BURST 32
 
 static const char *valid_arguments[] = {
@@ -58,10 +38,12 @@ static const char *valid_arguments[] = {
        ETH_VHOST_QUEUES_ARG,
        ETH_VHOST_CLIENT_ARG,
        ETH_VHOST_DEQUEUE_ZERO_COPY,
+       ETH_VHOST_IOMMU_SUPPORT,
+       ETH_VHOST_POSTCOPY_SUPPORT,
        NULL
 };
 
-static struct ether_addr base_eth_addr = {
+static struct rte_ether_addr base_eth_addr = {
        .addr_bytes = {
                0x56 /* V */,
                0x48 /* H */,
@@ -104,7 +86,7 @@ struct vhost_queue {
        rte_atomic32_t while_queuing;
        struct pmd_internal *internal;
        struct rte_mempool *mb_pool;
-       uint8_t port;
+       uint16_t port;
        uint16_t virtqueue_id;
        struct vhost_stats stats;
 };
@@ -114,7 +96,9 @@ struct pmd_internal {
        char *dev_name;
        char *iface_name;
        uint16_t max_queues;
+       int vid;
        rte_atomic32_t started;
+       uint8_t vlan_strip;
 };
 
 struct internal_list {
@@ -341,10 +325,10 @@ static inline void
 vhost_count_multicast_broadcast(struct vhost_queue *vq,
                                struct rte_mbuf *mbuf)
 {
-       struct ether_addr *ea = NULL;
+       struct rte_ether_addr *ea = NULL;
        struct vhost_stats *pstats = &vq->stats;
 
-       ea = rte_pktmbuf_mtod(mbuf, struct ether_addr *);
+       ea = rte_pktmbuf_mtod(mbuf, struct rte_ether_addr *);
        if (is_multicast_ether_addr(ea)) {
                if (is_broadcast_ether_addr(ea))
                        pstats->xstats[VHOST_BROADCAST_PKT]++;
@@ -418,6 +402,11 @@ eth_vhost_rx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
 
        for (i = 0; likely(i < nb_rx); i++) {
                bufs[i]->port = r->port;
+               bufs[i]->vlan_tci = 0;
+
+               if (r->internal->vlan_strip)
+                       rte_vlan_strip(bufs[i]);
+
                r->stats.bytes += bufs[i]->pkt_len;
        }
 
@@ -434,7 +423,7 @@ eth_vhost_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
 {
        struct vhost_queue *r = q;
        uint16_t i, nb_tx = 0;
-       uint16_t nb_send = nb_bufs;
+       uint16_t nb_send = 0;
 
        if (unlikely(rte_atomic32_read(&r->allow_queuing) == 0))
                return 0;
@@ -444,6 +433,22 @@ eth_vhost_tx(void *q, struct rte_mbuf **bufs, uint16_t nb_bufs)
        if (unlikely(rte_atomic32_read(&r->allow_queuing) == 0))
                goto out;
 
+       for (i = 0; i < nb_bufs; i++) {
+               struct rte_mbuf *m = bufs[i];
+
+               /* Do VLAN tag insertion */
+               if (m->ol_flags & PKT_TX_VLAN_PKT) {
+                       int error = rte_vlan_insert(&m);
+                       if (unlikely(error)) {
+                               rte_pktmbuf_free(m);
+                               continue;
+                       }
+               }
+
+               bufs[nb_send] = m;
+               ++nb_send;
+       }
+
        /* Enqueue packets to guest RX queue */
        while (nb_send) {
                uint16_t nb_pkts;
@@ -485,6 +490,11 @@ out:
 static int
 eth_dev_configure(struct rte_eth_dev *dev __rte_unused)
 {
+       struct pmd_internal *internal = dev->data->dev_private;
+       const struct rte_eth_rxmode *rxmode = &dev->data->dev_conf.rxmode;
+
+       internal->vlan_strip = !!(rxmode->offloads & DEV_RX_OFFLOAD_VLAN_STRIP);
+
        return 0;
 }
 
@@ -516,6 +526,136 @@ find_internal_resource(char *ifname)
        return list;
 }
 
+static int
+eth_rxq_intr_enable(struct rte_eth_dev *dev, uint16_t qid)
+{
+       struct rte_vhost_vring vring;
+       struct vhost_queue *vq;
+       int ret = 0;
+
+       vq = dev->data->rx_queues[qid];
+       if (!vq) {
+               VHOST_LOG(ERR, "rxq%d is not setup yet\n", qid);
+               return -1;
+       }
+
+       ret = rte_vhost_get_vhost_vring(vq->vid, (qid << 1) + 1, &vring);
+       if (ret < 0) {
+               VHOST_LOG(ERR, "Failed to get rxq%d's vring\n", qid);
+               return ret;
+       }
+       VHOST_LOG(INFO, "Enable interrupt for rxq%d\n", qid);
+       rte_vhost_enable_guest_notification(vq->vid, (qid << 1) + 1, 1);
+       rte_wmb();
+
+       return ret;
+}
+
+static int
+eth_rxq_intr_disable(struct rte_eth_dev *dev, uint16_t qid)
+{
+       struct rte_vhost_vring vring;
+       struct vhost_queue *vq;
+       int ret = 0;
+
+       vq = dev->data->rx_queues[qid];
+       if (!vq) {
+               VHOST_LOG(ERR, "rxq%d is not setup yet\n", qid);
+               return -1;
+       }
+
+       ret = rte_vhost_get_vhost_vring(vq->vid, (qid << 1) + 1, &vring);
+       if (ret < 0) {
+               VHOST_LOG(ERR, "Failed to get rxq%d's vring", qid);
+               return ret;
+       }
+       VHOST_LOG(INFO, "Disable interrupt for rxq%d\n", qid);
+       rte_vhost_enable_guest_notification(vq->vid, (qid << 1) + 1, 0);
+       rte_wmb();
+
+       return 0;
+}
+
+static void
+eth_vhost_uninstall_intr(struct rte_eth_dev *dev)
+{
+       struct rte_intr_handle *intr_handle = dev->intr_handle;
+
+       if (intr_handle) {
+               if (intr_handle->intr_vec)
+                       free(intr_handle->intr_vec);
+               free(intr_handle);
+       }
+
+       dev->intr_handle = NULL;
+}
+
+static int
+eth_vhost_install_intr(struct rte_eth_dev *dev)
+{
+       struct rte_vhost_vring vring;
+       struct vhost_queue *vq;
+       int count = 0;
+       int nb_rxq = dev->data->nb_rx_queues;
+       int i;
+       int ret;
+
+       /* uninstall firstly if we are reconnecting */
+       if (dev->intr_handle)
+               eth_vhost_uninstall_intr(dev);
+
+       dev->intr_handle = malloc(sizeof(*dev->intr_handle));
+       if (!dev->intr_handle) {
+               VHOST_LOG(ERR, "Fail to allocate intr_handle\n");
+               return -ENOMEM;
+       }
+       memset(dev->intr_handle, 0, sizeof(*dev->intr_handle));
+
+       dev->intr_handle->efd_counter_size = sizeof(uint64_t);
+
+       dev->intr_handle->intr_vec =
+               malloc(nb_rxq * sizeof(dev->intr_handle->intr_vec[0]));
+
+       if (!dev->intr_handle->intr_vec) {
+               VHOST_LOG(ERR,
+                       "Failed to allocate memory for interrupt vector\n");
+               free(dev->intr_handle);
+               return -ENOMEM;
+       }
+
+       VHOST_LOG(INFO, "Prepare intr vec\n");
+       for (i = 0; i < nb_rxq; i++) {
+               vq = dev->data->rx_queues[i];
+               if (!vq) {
+                       VHOST_LOG(INFO, "rxq-%d not setup yet, skip!\n", i);
+                       continue;
+               }
+
+               ret = rte_vhost_get_vhost_vring(vq->vid, (i << 1) + 1, &vring);
+               if (ret < 0) {
+                       VHOST_LOG(INFO,
+                               "Failed to get rxq-%d's vring, skip!\n", i);
+                       continue;
+               }
+
+               if (vring.kickfd < 0) {
+                       VHOST_LOG(INFO,
+                               "rxq-%d's kickfd is invalid, skip!\n", i);
+                       continue;
+               }
+               dev->intr_handle->intr_vec[i] = RTE_INTR_VEC_RXTX_OFFSET + i;
+               dev->intr_handle->efds[i] = vring.kickfd;
+               count++;
+               VHOST_LOG(INFO, "Installed intr vec for rxq-%d\n", i);
+       }
+
+       dev->intr_handle->nb_efd = count;
+       dev->intr_handle->max_intr = count + 1;
+       dev->intr_handle->type = RTE_INTR_HANDLE_VDEV;
+
+       return 0;
+}
+
 static void
 update_queuing_status(struct rte_eth_dev *dev)
 {
@@ -524,6 +664,9 @@ update_queuing_status(struct rte_eth_dev *dev)
        unsigned int i;
        int allow_queuing = 1;
 
+       if (!dev->data->rx_queues || !dev->data->tx_queues)
+               return;
+
        if (rte_atomic32_read(&internal->started) == 0 ||
            rte_atomic32_read(&internal->dev_attached) == 0)
                allow_queuing = 0;
@@ -548,13 +691,37 @@ update_queuing_status(struct rte_eth_dev *dev)
        }
 }
 
+static void
+queue_setup(struct rte_eth_dev *eth_dev, struct pmd_internal *internal)
+{
+       struct vhost_queue *vq;
+       int i;
+
+       for (i = 0; i < eth_dev->data->nb_rx_queues; i++) {
+               vq = eth_dev->data->rx_queues[i];
+               if (!vq)
+                       continue;
+               vq->vid = internal->vid;
+               vq->internal = internal;
+               vq->port = eth_dev->data->port_id;
+       }
+       for (i = 0; i < eth_dev->data->nb_tx_queues; i++) {
+               vq = eth_dev->data->tx_queues[i];
+               if (!vq)
+                       continue;
+               vq->vid = internal->vid;
+               vq->internal = internal;
+               vq->port = eth_dev->data->port_id;
+       }
+}
+
 static int
 new_device(int vid)
 {
        struct rte_eth_dev *eth_dev;
        struct internal_list *list;
        struct pmd_internal *internal;
-       struct vhost_queue *vq;
+       struct rte_eth_conf *dev_conf;
        unsigned i;
        char ifname[PATH_MAX];
 #ifdef RTE_LIBRTE_VHOST_NUMA
@@ -564,12 +731,13 @@ new_device(int vid)
        rte_vhost_get_ifname(vid, ifname, sizeof(ifname));
        list = find_internal_resource(ifname);
        if (list == NULL) {
-               RTE_LOG(INFO, PMD, "Invalid device name: %s\n", ifname);
+               VHOST_LOG(INFO, "Invalid device name: %s\n", ifname);
                return -1;
        }
 
        eth_dev = list->eth_dev;
        internal = eth_dev->data->dev_private;
+       dev_conf = &eth_dev->data->dev_conf;
 
 #ifdef RTE_LIBRTE_VHOST_NUMA
        newnode = rte_vhost_get_numa_node(vid);
@@ -577,21 +745,19 @@ new_device(int vid)
                eth_dev->data->numa_node = newnode;
 #endif
 
-       for (i = 0; i < eth_dev->data->nb_rx_queues; i++) {
-               vq = eth_dev->data->rx_queues[i];
-               if (vq == NULL)
-                       continue;
-               vq->vid = vid;
-               vq->internal = internal;
-               vq->port = eth_dev->data->port_id;
-       }
-       for (i = 0; i < eth_dev->data->nb_tx_queues; i++) {
-               vq = eth_dev->data->tx_queues[i];
-               if (vq == NULL)
-                       continue;
-               vq->vid = vid;
-               vq->internal = internal;
-               vq->port = eth_dev->data->port_id;
+       internal->vid = vid;
+       if (rte_atomic32_read(&internal->started) == 1) {
+               queue_setup(eth_dev, internal);
+
+               if (dev_conf->intr_conf.rxq) {
+                       if (eth_vhost_install_intr(eth_dev) < 0) {
+                               VHOST_LOG(INFO,
+                                       "Failed to install interrupt handler.");
+                                       return -1;
+                       }
+               }
+       } else {
+               VHOST_LOG(INFO, "RX/TX queues not exist yet\n");
        }
 
        for (i = 0; i < rte_vhost_get_vring_num(vid); i++)
@@ -604,7 +770,7 @@ new_device(int vid)
        rte_atomic32_set(&internal->dev_attached, 1);
        update_queuing_status(eth_dev);
 
-       RTE_LOG(INFO, PMD, "New connection established\n");
+       VHOST_LOG(INFO, "Vhost device %d created\n", vid);
 
        _rte_eth_dev_callback_process(eth_dev, RTE_ETH_EVENT_INTR_LSC, NULL);
 
@@ -625,7 +791,7 @@ destroy_device(int vid)
        rte_vhost_get_ifname(vid, ifname, sizeof(ifname));
        list = find_internal_resource(ifname);
        if (list == NULL) {
-               RTE_LOG(ERR, PMD, "Invalid interface name: %s\n", ifname);
+               VHOST_LOG(ERR, "Invalid interface name: %s\n", ifname);
                return;
        }
        eth_dev = list->eth_dev;
@@ -636,17 +802,19 @@ destroy_device(int vid)
 
        eth_dev->data->dev_link.link_status = ETH_LINK_DOWN;
 
-       for (i = 0; i < eth_dev->data->nb_rx_queues; i++) {
-               vq = eth_dev->data->rx_queues[i];
-               if (vq == NULL)
-                       continue;
-               vq->vid = -1;
-       }
-       for (i = 0; i < eth_dev->data->nb_tx_queues; i++) {
-               vq = eth_dev->data->tx_queues[i];
-               if (vq == NULL)
-                       continue;
-               vq->vid = -1;
+       if (eth_dev->data->rx_queues && eth_dev->data->tx_queues) {
+               for (i = 0; i < eth_dev->data->nb_rx_queues; i++) {
+                       vq = eth_dev->data->rx_queues[i];
+                       if (!vq)
+                               continue;
+                       vq->vid = -1;
+               }
+               for (i = 0; i < eth_dev->data->nb_tx_queues; i++) {
+                       vq = eth_dev->data->tx_queues[i];
+                       if (!vq)
+                               continue;
+                       vq->vid = -1;
+               }
        }
 
        state = vring_states[eth_dev->data->port_id];
@@ -658,7 +826,8 @@ destroy_device(int vid)
        state->max_vring = 0;
        rte_spinlock_unlock(&state->lock);
 
-       RTE_LOG(INFO, PMD, "Connection closed\n");
+       VHOST_LOG(INFO, "Vhost device %d destroyed\n", vid);
+       eth_vhost_uninstall_intr(eth_dev);
 
        _rte_eth_dev_callback_process(eth_dev, RTE_ETH_EVENT_INTR_LSC, NULL);
 }
@@ -674,7 +843,7 @@ vring_state_changed(int vid, uint16_t vring, int enable)
        rte_vhost_get_ifname(vid, ifname, sizeof(ifname));
        list = find_internal_resource(ifname);
        if (list == NULL) {
-               RTE_LOG(ERR, PMD, "Invalid interface name: %s\n", ifname);
+               VHOST_LOG(ERR, "Invalid interface name: %s\n", ifname);
                return -1;
        }
 
@@ -686,7 +855,7 @@ vring_state_changed(int vid, uint16_t vring, int enable)
        state->max_vring = RTE_MAX(vring, state->max_vring);
        rte_spinlock_unlock(&state->lock);
 
-       RTE_LOG(INFO, PMD, "vring%u is %s\n",
+       VHOST_LOG(INFO, "vring%u is %s\n",
                        vring, enable ? "enabled" : "disabled");
 
        _rte_eth_dev_callback_process(eth_dev, RTE_ETH_EVENT_QUEUE_STATE, NULL);
@@ -701,7 +870,7 @@ static struct vhost_device_ops vhost_ops = {
 };
 
 int
-rte_eth_vhost_get_queue_event(uint8_t port_id,
+rte_eth_vhost_get_queue_event(uint16_t port_id,
                struct rte_eth_vhost_queue_event *event)
 {
        struct rte_vhost_vring_state *state;
@@ -709,13 +878,13 @@ rte_eth_vhost_get_queue_event(uint8_t port_id,
        int idx;
 
        if (port_id >= RTE_MAX_ETHPORTS) {
-               RTE_LOG(ERR, PMD, "Invalid port id\n");
+               VHOST_LOG(ERR, "Invalid port id\n");
                return -1;
        }
 
        state = vring_states[port_id];
        if (!state) {
-               RTE_LOG(ERR, PMD, "Unused port\n");
+               VHOST_LOG(ERR, "Unused port\n");
                return -1;
        }
 
@@ -738,7 +907,7 @@ rte_eth_vhost_get_queue_event(uint8_t port_id,
 }
 
 int
-rte_eth_vhost_get_vid_from_port_id(uint8_t port_id)
+rte_eth_vhost_get_vid_from_port_id(uint16_t port_id)
 {
        struct internal_list *list;
        struct rte_eth_dev *eth_dev;
@@ -767,12 +936,25 @@ rte_eth_vhost_get_vid_from_port_id(uint8_t port_id)
 }
 
 static int
-eth_dev_start(struct rte_eth_dev *dev)
+eth_dev_start(struct rte_eth_dev *eth_dev)
 {
-       struct pmd_internal *internal = dev->data->dev_private;
+       struct pmd_internal *internal = eth_dev->data->dev_private;
+       struct rte_eth_conf *dev_conf = &eth_dev->data->dev_conf;
+
+       queue_setup(eth_dev, internal);
+
+       if (rte_atomic32_read(&internal->dev_attached) == 1) {
+               if (dev_conf->intr_conf.rxq) {
+                       if (eth_vhost_install_intr(eth_dev) < 0) {
+                               VHOST_LOG(INFO,
+                                       "Failed to install interrupt handler.");
+                                       return -1;
+                       }
+               }
+       }
 
        rte_atomic32_set(&internal->started, 1);
-       update_queuing_status(dev);
+       update_queuing_status(eth_dev);
 
        return 0;
 }
@@ -791,11 +973,14 @@ eth_dev_close(struct rte_eth_dev *dev)
 {
        struct pmd_internal *internal;
        struct internal_list *list;
+       unsigned int i;
 
        internal = dev->data->dev_private;
        if (!internal)
                return;
 
+       eth_dev_stop(dev);
+
        rte_vhost_driver_unregister(internal->iface_name);
 
        list = find_internal_resource(internal->iface_name);
@@ -807,9 +992,19 @@ eth_dev_close(struct rte_eth_dev *dev)
        pthread_mutex_unlock(&internal_list_lock);
        rte_free(list);
 
+       if (dev->data->rx_queues)
+               for (i = 0; i < dev->data->nb_rx_queues; i++)
+                       rte_free(dev->data->rx_queues[i]);
+
+       if (dev->data->tx_queues)
+               for (i = 0; i < dev->data->nb_tx_queues; i++)
+                       rte_free(dev->data->tx_queues[i]);
+
        free(internal->dev_name);
        free(internal->iface_name);
        rte_free(internal);
+
+       dev->data->dev_private = NULL;
 }
 
 static int
@@ -824,7 +1019,7 @@ eth_rx_queue_setup(struct rte_eth_dev *dev, uint16_t rx_queue_id,
        vq = rte_zmalloc_socket(NULL, sizeof(struct vhost_queue),
                        RTE_CACHE_LINE_SIZE, socket_id);
        if (vq == NULL) {
-               RTE_LOG(ERR, PMD, "Failed to allocate memory for rx queue\n");
+               VHOST_LOG(ERR, "Failed to allocate memory for rx queue\n");
                return -ENOMEM;
        }
 
@@ -846,7 +1041,7 @@ eth_tx_queue_setup(struct rte_eth_dev *dev, uint16_t tx_queue_id,
        vq = rte_zmalloc_socket(NULL, sizeof(struct vhost_queue),
                        RTE_CACHE_LINE_SIZE, socket_id);
        if (vq == NULL) {
-               RTE_LOG(ERR, PMD, "Failed to allocate memory for tx queue\n");
+               VHOST_LOG(ERR, "Failed to allocate memory for tx queue\n");
                return -ENOMEM;
        }
 
@@ -864,7 +1059,7 @@ eth_dev_info(struct rte_eth_dev *dev,
 
        internal = dev->data->dev_private;
        if (internal == NULL) {
-               RTE_LOG(ERR, PMD, "Invalid device specified\n");
+               VHOST_LOG(ERR, "Invalid device specified\n");
                return;
        }
 
@@ -873,9 +1068,13 @@ eth_dev_info(struct rte_eth_dev *dev,
        dev_info->max_rx_queues = internal->max_queues;
        dev_info->max_tx_queues = internal->max_queues;
        dev_info->min_rx_bufsize = 0;
+
+       dev_info->tx_offload_capa = DEV_TX_OFFLOAD_MULTI_SEGS |
+                               DEV_TX_OFFLOAD_VLAN_INSERT;
+       dev_info->rx_offload_capa = DEV_RX_OFFLOAD_VLAN_STRIP;
 }
 
-static void
+static int
 eth_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
 {
        unsigned i;
@@ -913,6 +1112,8 @@ eth_stats_get(struct rte_eth_dev *dev, struct rte_eth_stats *stats)
        stats->oerrors = tx_missed_total;
        stats->ibytes = rx_total_bytes;
        stats->obytes = tx_total_bytes;
+
+       return 0;
 }
 
 static void
@@ -961,6 +1162,18 @@ eth_link_update(struct rte_eth_dev *dev __rte_unused,
        return 0;
 }
 
+static uint32_t
+eth_rx_queue_count(struct rte_eth_dev *dev, uint16_t rx_queue_id)
+{
+       struct vhost_queue *vq;
+
+       vq = dev->data->rx_queues[rx_queue_id];
+       if (vq == NULL)
+               return 0;
+
+       return rte_vhost_rx_queue_count(vq->vid, vq->virtqueue_id);
+}
+
 static const struct eth_dev_ops ops = {
        .dev_start = eth_dev_start,
        .dev_stop = eth_dev_stop,
@@ -972,53 +1185,48 @@ static const struct eth_dev_ops ops = {
        .rx_queue_release = eth_queue_release,
        .tx_queue_release = eth_queue_release,
        .tx_done_cleanup = eth_tx_done_cleanup,
+       .rx_queue_count = eth_rx_queue_count,
        .link_update = eth_link_update,
        .stats_get = eth_stats_get,
        .stats_reset = eth_stats_reset,
        .xstats_reset = vhost_dev_xstats_reset,
        .xstats_get = vhost_dev_xstats_get,
        .xstats_get_names = vhost_dev_xstats_get_names,
+       .rx_queue_intr_enable = eth_rxq_intr_enable,
+       .rx_queue_intr_disable = eth_rxq_intr_disable,
 };
 
 static struct rte_vdev_driver pmd_vhost_drv;
 
 static int
-eth_dev_vhost_create(const char *name, char *iface_name, int16_t queues,
-                    const unsigned numa_node, uint64_t flags)
+eth_dev_vhost_create(struct rte_vdev_device *dev, char *iface_name,
+       int16_t queues, const unsigned int numa_node, uint64_t flags)
 {
-       struct rte_eth_dev_data *data = NULL;
+       const char *name = rte_vdev_device_name(dev);
+       struct rte_eth_dev_data *data;
        struct pmd_internal *internal = NULL;
        struct rte_eth_dev *eth_dev = NULL;
-       struct ether_addr *eth_addr = NULL;
+       struct rte_ether_addr *eth_addr = NULL;
        struct rte_vhost_vring_state *vring_state = NULL;
        struct internal_list *list = NULL;
 
-       RTE_LOG(INFO, PMD, "Creating VHOST-USER backend on numa socket %u\n",
+       VHOST_LOG(INFO, "Creating VHOST-USER backend on numa socket %u\n",
                numa_node);
 
-       /* now do all data allocation - for eth_dev structure, dummy pci driver
-        * and internal (private) data
-        */
-       data = rte_zmalloc_socket(name, sizeof(*data), 0, numa_node);
-       if (data == NULL)
-               goto error;
-
-       internal = rte_zmalloc_socket(name, sizeof(*internal), 0, numa_node);
-       if (internal == NULL)
-               goto error;
-
        list = rte_zmalloc_socket(name, sizeof(*list), 0, numa_node);
        if (list == NULL)
                goto error;
 
        /* reserve an ethdev entry */
-       eth_dev = rte_eth_dev_allocate(name);
+       eth_dev = rte_eth_vdev_allocate(dev, sizeof(*internal));
        if (eth_dev == NULL)
                goto error;
+       data = eth_dev->data;
 
        eth_addr = rte_zmalloc_socket(name, sizeof(*eth_addr), 0, numa_node);
        if (eth_addr == NULL)
                goto error;
+       data->mac_addrs = eth_addr;
        *eth_addr = base_eth_addr;
        eth_addr->addr_bytes[5] = eth_dev->data->port_id;
 
@@ -1029,10 +1237,10 @@ eth_dev_vhost_create(const char *name, char *iface_name, int16_t queues,
 
        /* now put it all together
         * - store queue data in internal,
-        * - store numa_node info in ethdev data
         * - point eth_dev_data to internals
         * - and point eth_dev structure to new eth_dev_data structure
         */
+       internal = eth_dev->data->dev_private;
        internal->dev_name = strdup(name);
        if (internal->dev_name == NULL)
                goto error;
@@ -1048,26 +1256,14 @@ eth_dev_vhost_create(const char *name, char *iface_name, int16_t queues,
        rte_spinlock_init(&vring_state->lock);
        vring_states[eth_dev->data->port_id] = vring_state;
 
-       data->dev_private = internal;
-       data->port_id = eth_dev->data->port_id;
-       memmove(data->name, eth_dev->data->name, sizeof(data->name));
        data->nb_rx_queues = queues;
        data->nb_tx_queues = queues;
        internal->max_queues = queues;
+       internal->vid = -1;
        data->dev_link = pmd_link;
-       data->mac_addrs = eth_addr;
+       data->dev_flags = RTE_ETH_DEV_INTR_LSC;
 
-       /* We'll replace the 'data' originally allocated by eth_dev. So the
-        * vhost PMD resources won't be shared between multi processes.
-        */
-       eth_dev->data = data;
        eth_dev->dev_ops = &ops;
-       eth_dev->driver = NULL;
-       data->dev_flags =
-               RTE_ETH_DEV_DETACHABLE | RTE_ETH_DEV_INTR_LSC;
-       data->kdrv = RTE_KDRV_NONE;
-       data->drv_name = pmd_vhost_drv.driver.name;
-       data->numa_node = numa_node;
 
        /* finally assign rx and tx ops */
        eth_dev->rx_pkt_burst = eth_vhost_rx;
@@ -1077,28 +1273,27 @@ eth_dev_vhost_create(const char *name, char *iface_name, int16_t queues,
                goto error;
 
        if (rte_vhost_driver_callback_register(iface_name, &vhost_ops) < 0) {
-               RTE_LOG(ERR, PMD, "Can't register callbacks\n");
+               VHOST_LOG(ERR, "Can't register callbacks\n");
                goto error;
        }
 
        if (rte_vhost_driver_start(iface_name) < 0) {
-               RTE_LOG(ERR, PMD, "Failed to start driver for %s\n",
+               VHOST_LOG(ERR, "Failed to start driver for %s\n",
                        iface_name);
                goto error;
        }
 
+       rte_eth_dev_probing_finish(eth_dev);
        return data->port_id;
 
 error:
-       if (internal)
+       if (internal) {
+               free(internal->iface_name);
                free(internal->dev_name);
+       }
        rte_free(vring_state);
-       rte_free(eth_addr);
-       if (eth_dev)
-               rte_eth_dev_release_port(eth_dev);
-       rte_free(internal);
+       rte_eth_dev_release_port(eth_dev);
        rte_free(list);
-       rte_free(data);
 
        return -1;
 }
@@ -1134,7 +1329,6 @@ open_int(const char *key __rte_unused, const char *value, void *extra_args)
 static int
 rte_pmd_vhost_probe(struct rte_vdev_device *dev)
 {
-       const char *name;
        struct rte_kvargs *kvlist = NULL;
        int ret = 0;
        char *iface_name;
@@ -1142,9 +1336,25 @@ rte_pmd_vhost_probe(struct rte_vdev_device *dev)
        uint64_t flags = 0;
        int client_mode = 0;
        int dequeue_zero_copy = 0;
+       int iommu_support = 0;
+       int postcopy_support = 0;
+       struct rte_eth_dev *eth_dev;
+       const char *name = rte_vdev_device_name(dev);
 
-       name = rte_vdev_device_name(dev);
-       RTE_LOG(INFO, PMD, "Initializing pmd_vhost for %s\n", name);
+       VHOST_LOG(INFO, "Initializing pmd_vhost for %s\n", name);
+
+       if (rte_eal_process_type() == RTE_PROC_SECONDARY) {
+               eth_dev = rte_eth_dev_attach_secondary(name);
+               if (!eth_dev) {
+                       VHOST_LOG(ERR, "Failed to probe %s\n", name);
+                       return -1;
+               }
+               /* TODO: request info from primary to set up Rx and Tx */
+               eth_dev->dev_ops = &ops;
+               eth_dev->device = &dev->device;
+               rte_eth_dev_probing_finish(eth_dev);
+               return 0;
+       }
 
        kvlist = rte_kvargs_parse(rte_vdev_device_args(dev), valid_arguments);
        if (kvlist == NULL)
@@ -1189,7 +1399,31 @@ rte_pmd_vhost_probe(struct rte_vdev_device *dev)
                        flags |= RTE_VHOST_USER_DEQUEUE_ZERO_COPY;
        }
 
-       eth_dev_vhost_create(name, iface_name, queues, rte_socket_id(), flags);
+       if (rte_kvargs_count(kvlist, ETH_VHOST_IOMMU_SUPPORT) == 1) {
+               ret = rte_kvargs_process(kvlist, ETH_VHOST_IOMMU_SUPPORT,
+                                        &open_int, &iommu_support);
+               if (ret < 0)
+                       goto out_free;
+
+               if (iommu_support)
+                       flags |= RTE_VHOST_USER_IOMMU_SUPPORT;
+       }
+
+       if (rte_kvargs_count(kvlist, ETH_VHOST_POSTCOPY_SUPPORT) == 1) {
+               ret = rte_kvargs_process(kvlist, ETH_VHOST_POSTCOPY_SUPPORT,
+                                        &open_int, &postcopy_support);
+               if (ret < 0)
+                       goto out_free;
+
+               if (postcopy_support)
+                       flags |= RTE_VHOST_USER_POSTCOPY_SUPPORT;
+       }
+
+       if (dev->device.numa_node == SOCKET_ID_ANY)
+               dev->device.numa_node = rte_socket_id();
+
+       eth_dev_vhost_create(dev, iface_name, queues, dev->device.numa_node,
+               flags);
 
 out_free:
        rte_kvargs_free(kvlist);
@@ -1201,31 +1435,23 @@ rte_pmd_vhost_remove(struct rte_vdev_device *dev)
 {
        const char *name;
        struct rte_eth_dev *eth_dev = NULL;
-       unsigned int i;
 
        name = rte_vdev_device_name(dev);
-       RTE_LOG(INFO, PMD, "Un-Initializing pmd_vhost for %s\n", name);
+       VHOST_LOG(INFO, "Un-Initializing pmd_vhost for %s\n", name);
 
        /* find an ethdev entry */
        eth_dev = rte_eth_dev_allocated(name);
        if (eth_dev == NULL)
                return -ENODEV;
 
-       eth_dev_stop(eth_dev);
+       if (rte_eal_process_type() != RTE_PROC_PRIMARY)
+               return rte_eth_dev_release_port(eth_dev);
 
        eth_dev_close(eth_dev);
 
        rte_free(vring_states[eth_dev->data->port_id]);
        vring_states[eth_dev->data->port_id] = NULL;
 
-       for (i = 0; i < eth_dev->data->nb_rx_queues; i++)
-               rte_free(eth_dev->data->rx_queues[i]);
-       for (i = 0; i < eth_dev->data->nb_tx_queues; i++)
-               rte_free(eth_dev->data->tx_queues[i]);
-
-       rte_free(eth_dev->data->mac_addrs);
-       rte_free(eth_dev->data);
-
        rte_eth_dev_release_port(eth_dev);
 
        return 0;
@@ -1240,4 +1466,15 @@ RTE_PMD_REGISTER_VDEV(net_vhost, pmd_vhost_drv);
 RTE_PMD_REGISTER_ALIAS(net_vhost, eth_vhost);
 RTE_PMD_REGISTER_PARAM_STRING(net_vhost,
        "iface=<ifc> "
-       "queues=<int>");
+       "queues=<int> "
+       "client=<0|1> "
+       "dequeue-zero-copy=<0|1> "
+       "iommu-support=<0|1> "
+       "postcopy-support=<0|1>");
+
+RTE_INIT(vhost_init_log)
+{
+       vhost_logtype = rte_log_register("pmd.net.vhost");
+       if (vhost_logtype >= 0)
+               rte_log_set_level(vhost_logtype, RTE_LOG_NOTICE);
+}