Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / net / xen-netback / interface.c
diff --git a/kernel/drivers/net/xen-netback/interface.c b/kernel/drivers/net/xen-netback/interface.c
new file mode 100644 (file)
index 0000000..1a83e19
--- /dev/null
@@ -0,0 +1,689 @@
+/*
+ * Network-device interface management.
+ *
+ * Copyright (c) 2004-2005, Keir Fraser
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "common.h"
+
+#include <linux/kthread.h>
+#include <linux/ethtool.h>
+#include <linux/rtnetlink.h>
+#include <linux/if_vlan.h>
+#include <linux/vmalloc.h>
+
+#include <xen/events.h>
+#include <asm/xen/hypercall.h>
+#include <xen/balloon.h>
+
+#define XENVIF_QUEUE_LENGTH 32
+#define XENVIF_NAPI_WEIGHT  64
+
+/* Number of bytes allowed on the internal guest Rx queue. */
+#define XENVIF_RX_QUEUE_BYTES (XEN_NETIF_RX_RING_SIZE/2 * PAGE_SIZE)
+
+/* This function is used to set SKBTX_DEV_ZEROCOPY as well as
+ * increasing the inflight counter. We need to increase the inflight
+ * counter because core driver calls into xenvif_zerocopy_callback
+ * which calls xenvif_skb_zerocopy_complete.
+ */
+void xenvif_skb_zerocopy_prepare(struct xenvif_queue *queue,
+                                struct sk_buff *skb)
+{
+       skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
+       atomic_inc(&queue->inflight_packets);
+}
+
+void xenvif_skb_zerocopy_complete(struct xenvif_queue *queue)
+{
+       atomic_dec(&queue->inflight_packets);
+}
+
+int xenvif_schedulable(struct xenvif *vif)
+{
+       return netif_running(vif->dev) &&
+               test_bit(VIF_STATUS_CONNECTED, &vif->status) &&
+               !vif->disabled;
+}
+
+static irqreturn_t xenvif_tx_interrupt(int irq, void *dev_id)
+{
+       struct xenvif_queue *queue = dev_id;
+
+       if (RING_HAS_UNCONSUMED_REQUESTS(&queue->tx))
+               napi_schedule(&queue->napi);
+
+       return IRQ_HANDLED;
+}
+
+static int xenvif_poll(struct napi_struct *napi, int budget)
+{
+       struct xenvif_queue *queue =
+               container_of(napi, struct xenvif_queue, napi);
+       int work_done;
+
+       /* This vif is rogue, we pretend we've there is nothing to do
+        * for this vif to deschedule it from NAPI. But this interface
+        * will be turned off in thread context later.
+        */
+       if (unlikely(queue->vif->disabled)) {
+               napi_complete(napi);
+               return 0;
+       }
+
+       work_done = xenvif_tx_action(queue, budget);
+
+       if (work_done < budget) {
+               napi_complete(napi);
+               xenvif_napi_schedule_or_enable_events(queue);
+       }
+
+       return work_done;
+}
+
+static irqreturn_t xenvif_rx_interrupt(int irq, void *dev_id)
+{
+       struct xenvif_queue *queue = dev_id;
+
+       xenvif_kick_thread(queue);
+
+       return IRQ_HANDLED;
+}
+
+irqreturn_t xenvif_interrupt(int irq, void *dev_id)
+{
+       xenvif_tx_interrupt(irq, dev_id);
+       xenvif_rx_interrupt(irq, dev_id);
+
+       return IRQ_HANDLED;
+}
+
+int xenvif_queue_stopped(struct xenvif_queue *queue)
+{
+       struct net_device *dev = queue->vif->dev;
+       unsigned int id = queue->id;
+       return netif_tx_queue_stopped(netdev_get_tx_queue(dev, id));
+}
+
+void xenvif_wake_queue(struct xenvif_queue *queue)
+{
+       struct net_device *dev = queue->vif->dev;
+       unsigned int id = queue->id;
+       netif_tx_wake_queue(netdev_get_tx_queue(dev, id));
+}
+
+static int xenvif_start_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+       struct xenvif *vif = netdev_priv(dev);
+       struct xenvif_queue *queue = NULL;
+       unsigned int num_queues = vif->num_queues;
+       u16 index;
+       struct xenvif_rx_cb *cb;
+
+       BUG_ON(skb->dev != dev);
+
+       /* Drop the packet if queues are not set up */
+       if (num_queues < 1)
+               goto drop;
+
+       /* Obtain the queue to be used to transmit this packet */
+       index = skb_get_queue_mapping(skb);
+       if (index >= num_queues) {
+               pr_warn_ratelimited("Invalid queue %hu for packet on interface %s\n.",
+                                   index, vif->dev->name);
+               index %= num_queues;
+       }
+       queue = &vif->queues[index];
+
+       /* Drop the packet if queue is not ready */
+       if (queue->task == NULL ||
+           queue->dealloc_task == NULL ||
+           !xenvif_schedulable(vif))
+               goto drop;
+
+       cb = XENVIF_RX_CB(skb);
+       cb->expires = jiffies + vif->drain_timeout;
+
+       xenvif_rx_queue_tail(queue, skb);
+       xenvif_kick_thread(queue);
+
+       return NETDEV_TX_OK;
+
+ drop:
+       vif->dev->stats.tx_dropped++;
+       dev_kfree_skb(skb);
+       return NETDEV_TX_OK;
+}
+
+static struct net_device_stats *xenvif_get_stats(struct net_device *dev)
+{
+       struct xenvif *vif = netdev_priv(dev);
+       struct xenvif_queue *queue = NULL;
+       unsigned int num_queues = vif->num_queues;
+       unsigned long rx_bytes = 0;
+       unsigned long rx_packets = 0;
+       unsigned long tx_bytes = 0;
+       unsigned long tx_packets = 0;
+       unsigned int index;
+
+       if (vif->queues == NULL)
+               goto out;
+
+       /* Aggregate tx and rx stats from each queue */
+       for (index = 0; index < num_queues; ++index) {
+               queue = &vif->queues[index];
+               rx_bytes += queue->stats.rx_bytes;
+               rx_packets += queue->stats.rx_packets;
+               tx_bytes += queue->stats.tx_bytes;
+               tx_packets += queue->stats.tx_packets;
+       }
+
+out:
+       vif->dev->stats.rx_bytes = rx_bytes;
+       vif->dev->stats.rx_packets = rx_packets;
+       vif->dev->stats.tx_bytes = tx_bytes;
+       vif->dev->stats.tx_packets = tx_packets;
+
+       return &vif->dev->stats;
+}
+
+static void xenvif_up(struct xenvif *vif)
+{
+       struct xenvif_queue *queue = NULL;
+       unsigned int num_queues = vif->num_queues;
+       unsigned int queue_index;
+
+       for (queue_index = 0; queue_index < num_queues; ++queue_index) {
+               queue = &vif->queues[queue_index];
+               napi_enable(&queue->napi);
+               enable_irq(queue->tx_irq);
+               if (queue->tx_irq != queue->rx_irq)
+                       enable_irq(queue->rx_irq);
+               xenvif_napi_schedule_or_enable_events(queue);
+       }
+}
+
+static void xenvif_down(struct xenvif *vif)
+{
+       struct xenvif_queue *queue = NULL;
+       unsigned int num_queues = vif->num_queues;
+       unsigned int queue_index;
+
+       for (queue_index = 0; queue_index < num_queues; ++queue_index) {
+               queue = &vif->queues[queue_index];
+               disable_irq(queue->tx_irq);
+               if (queue->tx_irq != queue->rx_irq)
+                       disable_irq(queue->rx_irq);
+               napi_disable(&queue->napi);
+               del_timer_sync(&queue->credit_timeout);
+       }
+}
+
+static int xenvif_open(struct net_device *dev)
+{
+       struct xenvif *vif = netdev_priv(dev);
+       if (test_bit(VIF_STATUS_CONNECTED, &vif->status))
+               xenvif_up(vif);
+       netif_tx_start_all_queues(dev);
+       return 0;
+}
+
+static int xenvif_close(struct net_device *dev)
+{
+       struct xenvif *vif = netdev_priv(dev);
+       if (test_bit(VIF_STATUS_CONNECTED, &vif->status))
+               xenvif_down(vif);
+       netif_tx_stop_all_queues(dev);
+       return 0;
+}
+
+static int xenvif_change_mtu(struct net_device *dev, int mtu)
+{
+       struct xenvif *vif = netdev_priv(dev);
+       int max = vif->can_sg ? 65535 - VLAN_ETH_HLEN : ETH_DATA_LEN;
+
+       if (mtu > max)
+               return -EINVAL;
+       dev->mtu = mtu;
+       return 0;
+}
+
+static netdev_features_t xenvif_fix_features(struct net_device *dev,
+       netdev_features_t features)
+{
+       struct xenvif *vif = netdev_priv(dev);
+
+       if (!vif->can_sg)
+               features &= ~NETIF_F_SG;
+       if (~(vif->gso_mask | vif->gso_prefix_mask) & GSO_BIT(TCPV4))
+               features &= ~NETIF_F_TSO;
+       if (~(vif->gso_mask | vif->gso_prefix_mask) & GSO_BIT(TCPV6))
+               features &= ~NETIF_F_TSO6;
+       if (!vif->ip_csum)
+               features &= ~NETIF_F_IP_CSUM;
+       if (!vif->ipv6_csum)
+               features &= ~NETIF_F_IPV6_CSUM;
+
+       return features;
+}
+
+static const struct xenvif_stat {
+       char name[ETH_GSTRING_LEN];
+       u16 offset;
+} xenvif_stats[] = {
+       {
+               "rx_gso_checksum_fixup",
+               offsetof(struct xenvif_stats, rx_gso_checksum_fixup)
+       },
+       /* If (sent != success + fail), there are probably packets never
+        * freed up properly!
+        */
+       {
+               "tx_zerocopy_sent",
+               offsetof(struct xenvif_stats, tx_zerocopy_sent),
+       },
+       {
+               "tx_zerocopy_success",
+               offsetof(struct xenvif_stats, tx_zerocopy_success),
+       },
+       {
+               "tx_zerocopy_fail",
+               offsetof(struct xenvif_stats, tx_zerocopy_fail)
+       },
+       /* Number of packets exceeding MAX_SKB_FRAG slots. You should use
+        * a guest with the same MAX_SKB_FRAG
+        */
+       {
+               "tx_frag_overflow",
+               offsetof(struct xenvif_stats, tx_frag_overflow)
+       },
+};
+
+static int xenvif_get_sset_count(struct net_device *dev, int string_set)
+{
+       switch (string_set) {
+       case ETH_SS_STATS:
+               return ARRAY_SIZE(xenvif_stats);
+       default:
+               return -EINVAL;
+       }
+}
+
+static void xenvif_get_ethtool_stats(struct net_device *dev,
+                                    struct ethtool_stats *stats, u64 * data)
+{
+       struct xenvif *vif = netdev_priv(dev);
+       unsigned int num_queues = vif->num_queues;
+       int i;
+       unsigned int queue_index;
+
+       for (i = 0; i < ARRAY_SIZE(xenvif_stats); i++) {
+               unsigned long accum = 0;
+               for (queue_index = 0; queue_index < num_queues; ++queue_index) {
+                       void *vif_stats = &vif->queues[queue_index].stats;
+                       accum += *(unsigned long *)(vif_stats + xenvif_stats[i].offset);
+               }
+               data[i] = accum;
+       }
+}
+
+static void xenvif_get_strings(struct net_device *dev, u32 stringset, u8 * data)
+{
+       int i;
+
+       switch (stringset) {
+       case ETH_SS_STATS:
+               for (i = 0; i < ARRAY_SIZE(xenvif_stats); i++)
+                       memcpy(data + i * ETH_GSTRING_LEN,
+                              xenvif_stats[i].name, ETH_GSTRING_LEN);
+               break;
+       }
+}
+
+static const struct ethtool_ops xenvif_ethtool_ops = {
+       .get_link       = ethtool_op_get_link,
+
+       .get_sset_count = xenvif_get_sset_count,
+       .get_ethtool_stats = xenvif_get_ethtool_stats,
+       .get_strings = xenvif_get_strings,
+};
+
+static const struct net_device_ops xenvif_netdev_ops = {
+       .ndo_start_xmit = xenvif_start_xmit,
+       .ndo_get_stats  = xenvif_get_stats,
+       .ndo_open       = xenvif_open,
+       .ndo_stop       = xenvif_close,
+       .ndo_change_mtu = xenvif_change_mtu,
+       .ndo_fix_features = xenvif_fix_features,
+       .ndo_set_mac_address = eth_mac_addr,
+       .ndo_validate_addr   = eth_validate_addr,
+};
+
+struct xenvif *xenvif_alloc(struct device *parent, domid_t domid,
+                           unsigned int handle)
+{
+       int err;
+       struct net_device *dev;
+       struct xenvif *vif;
+       char name[IFNAMSIZ] = {};
+
+       snprintf(name, IFNAMSIZ - 1, "vif%u.%u", domid, handle);
+       /* Allocate a netdev with the max. supported number of queues.
+        * When the guest selects the desired number, it will be updated
+        * via netif_set_real_num_*_queues().
+        */
+       dev = alloc_netdev_mq(sizeof(struct xenvif), name, NET_NAME_UNKNOWN,
+                             ether_setup, xenvif_max_queues);
+       if (dev == NULL) {
+               pr_warn("Could not allocate netdev for %s\n", name);
+               return ERR_PTR(-ENOMEM);
+       }
+
+       SET_NETDEV_DEV(dev, parent);
+
+       vif = netdev_priv(dev);
+
+       vif->domid  = domid;
+       vif->handle = handle;
+       vif->can_sg = 1;
+       vif->ip_csum = 1;
+       vif->dev = dev;
+       vif->disabled = false;
+       vif->drain_timeout = msecs_to_jiffies(rx_drain_timeout_msecs);
+       vif->stall_timeout = msecs_to_jiffies(rx_stall_timeout_msecs);
+
+       /* Start out with no queues. */
+       vif->queues = NULL;
+       vif->num_queues = 0;
+
+       spin_lock_init(&vif->lock);
+
+       dev->netdev_ops = &xenvif_netdev_ops;
+       dev->hw_features = NETIF_F_SG |
+               NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
+               NETIF_F_TSO | NETIF_F_TSO6;
+       dev->features = dev->hw_features | NETIF_F_RXCSUM;
+       dev->ethtool_ops = &xenvif_ethtool_ops;
+
+       dev->tx_queue_len = XENVIF_QUEUE_LENGTH;
+
+       /*
+        * Initialise a dummy MAC address. We choose the numerically
+        * largest non-broadcast address to prevent the address getting
+        * stolen by an Ethernet bridge for STP purposes.
+        * (FE:FF:FF:FF:FF:FF)
+        */
+       eth_broadcast_addr(dev->dev_addr);
+       dev->dev_addr[0] &= ~0x01;
+
+       netif_carrier_off(dev);
+
+       err = register_netdev(dev);
+       if (err) {
+               netdev_warn(dev, "Could not register device: err=%d\n", err);
+               free_netdev(dev);
+               return ERR_PTR(err);
+       }
+
+       netdev_dbg(dev, "Successfully created xenvif\n");
+
+       __module_get(THIS_MODULE);
+
+       return vif;
+}
+
+int xenvif_init_queue(struct xenvif_queue *queue)
+{
+       int err, i;
+
+       queue->credit_bytes = queue->remaining_credit = ~0UL;
+       queue->credit_usec  = 0UL;
+       init_timer(&queue->credit_timeout);
+       queue->credit_timeout.function = xenvif_tx_credit_callback;
+       queue->credit_window_start = get_jiffies_64();
+
+       queue->rx_queue_max = XENVIF_RX_QUEUE_BYTES;
+
+       skb_queue_head_init(&queue->rx_queue);
+       skb_queue_head_init(&queue->tx_queue);
+
+       queue->pending_cons = 0;
+       queue->pending_prod = MAX_PENDING_REQS;
+       for (i = 0; i < MAX_PENDING_REQS; ++i)
+               queue->pending_ring[i] = i;
+
+       spin_lock_init(&queue->callback_lock);
+       spin_lock_init(&queue->response_lock);
+
+       /* If ballooning is disabled, this will consume real memory, so you
+        * better enable it. The long term solution would be to use just a
+        * bunch of valid page descriptors, without dependency on ballooning
+        */
+       err = gnttab_alloc_pages(MAX_PENDING_REQS,
+                                queue->mmap_pages);
+       if (err) {
+               netdev_err(queue->vif->dev, "Could not reserve mmap_pages\n");
+               return -ENOMEM;
+       }
+
+       for (i = 0; i < MAX_PENDING_REQS; i++) {
+               queue->pending_tx_info[i].callback_struct = (struct ubuf_info)
+                       { .callback = xenvif_zerocopy_callback,
+                         .ctx = NULL,
+                         .desc = i };
+               queue->grant_tx_handle[i] = NETBACK_INVALID_HANDLE;
+       }
+
+       return 0;
+}
+
+void xenvif_carrier_on(struct xenvif *vif)
+{
+       rtnl_lock();
+       if (!vif->can_sg && vif->dev->mtu > ETH_DATA_LEN)
+               dev_set_mtu(vif->dev, ETH_DATA_LEN);
+       netdev_update_features(vif->dev);
+       set_bit(VIF_STATUS_CONNECTED, &vif->status);
+       if (netif_running(vif->dev))
+               xenvif_up(vif);
+       rtnl_unlock();
+}
+
+int xenvif_connect(struct xenvif_queue *queue, unsigned long tx_ring_ref,
+                  unsigned long rx_ring_ref, unsigned int tx_evtchn,
+                  unsigned int rx_evtchn)
+{
+       struct task_struct *task;
+       int err = -ENOMEM;
+
+       BUG_ON(queue->tx_irq);
+       BUG_ON(queue->task);
+       BUG_ON(queue->dealloc_task);
+
+       err = xenvif_map_frontend_rings(queue, tx_ring_ref, rx_ring_ref);
+       if (err < 0)
+               goto err;
+
+       init_waitqueue_head(&queue->wq);
+       init_waitqueue_head(&queue->dealloc_wq);
+       atomic_set(&queue->inflight_packets, 0);
+
+       netif_napi_add(queue->vif->dev, &queue->napi, xenvif_poll,
+                       XENVIF_NAPI_WEIGHT);
+
+       if (tx_evtchn == rx_evtchn) {
+               /* feature-split-event-channels == 0 */
+               err = bind_interdomain_evtchn_to_irqhandler(
+                       queue->vif->domid, tx_evtchn, xenvif_interrupt, 0,
+                       queue->name, queue);
+               if (err < 0)
+                       goto err_unmap;
+               queue->tx_irq = queue->rx_irq = err;
+               disable_irq(queue->tx_irq);
+       } else {
+               /* feature-split-event-channels == 1 */
+               snprintf(queue->tx_irq_name, sizeof(queue->tx_irq_name),
+                        "%s-tx", queue->name);
+               err = bind_interdomain_evtchn_to_irqhandler(
+                       queue->vif->domid, tx_evtchn, xenvif_tx_interrupt, 0,
+                       queue->tx_irq_name, queue);
+               if (err < 0)
+                       goto err_unmap;
+               queue->tx_irq = err;
+               disable_irq(queue->tx_irq);
+
+               snprintf(queue->rx_irq_name, sizeof(queue->rx_irq_name),
+                        "%s-rx", queue->name);
+               err = bind_interdomain_evtchn_to_irqhandler(
+                       queue->vif->domid, rx_evtchn, xenvif_rx_interrupt, 0,
+                       queue->rx_irq_name, queue);
+               if (err < 0)
+                       goto err_tx_unbind;
+               queue->rx_irq = err;
+               disable_irq(queue->rx_irq);
+       }
+
+       queue->stalled = true;
+
+       task = kthread_create(xenvif_kthread_guest_rx,
+                             (void *)queue, "%s-guest-rx", queue->name);
+       if (IS_ERR(task)) {
+               pr_warn("Could not allocate kthread for %s\n", queue->name);
+               err = PTR_ERR(task);
+               goto err_rx_unbind;
+       }
+       queue->task = task;
+       get_task_struct(task);
+
+       task = kthread_create(xenvif_dealloc_kthread,
+                             (void *)queue, "%s-dealloc", queue->name);
+       if (IS_ERR(task)) {
+               pr_warn("Could not allocate kthread for %s\n", queue->name);
+               err = PTR_ERR(task);
+               goto err_rx_unbind;
+       }
+       queue->dealloc_task = task;
+
+       wake_up_process(queue->task);
+       wake_up_process(queue->dealloc_task);
+
+       return 0;
+
+err_rx_unbind:
+       unbind_from_irqhandler(queue->rx_irq, queue);
+       queue->rx_irq = 0;
+err_tx_unbind:
+       unbind_from_irqhandler(queue->tx_irq, queue);
+       queue->tx_irq = 0;
+err_unmap:
+       xenvif_unmap_frontend_rings(queue);
+err:
+       module_put(THIS_MODULE);
+       return err;
+}
+
+void xenvif_carrier_off(struct xenvif *vif)
+{
+       struct net_device *dev = vif->dev;
+
+       rtnl_lock();
+       if (test_and_clear_bit(VIF_STATUS_CONNECTED, &vif->status)) {
+               netif_carrier_off(dev); /* discard queued packets */
+               if (netif_running(dev))
+                       xenvif_down(vif);
+       }
+       rtnl_unlock();
+}
+
+void xenvif_disconnect(struct xenvif *vif)
+{
+       struct xenvif_queue *queue = NULL;
+       unsigned int num_queues = vif->num_queues;
+       unsigned int queue_index;
+
+       xenvif_carrier_off(vif);
+
+       for (queue_index = 0; queue_index < num_queues; ++queue_index) {
+               queue = &vif->queues[queue_index];
+
+               netif_napi_del(&queue->napi);
+
+               if (queue->task) {
+                       kthread_stop(queue->task);
+                       put_task_struct(queue->task);
+                       queue->task = NULL;
+               }
+
+               if (queue->dealloc_task) {
+                       kthread_stop(queue->dealloc_task);
+                       queue->dealloc_task = NULL;
+               }
+
+               if (queue->tx_irq) {
+                       if (queue->tx_irq == queue->rx_irq)
+                               unbind_from_irqhandler(queue->tx_irq, queue);
+                       else {
+                               unbind_from_irqhandler(queue->tx_irq, queue);
+                               unbind_from_irqhandler(queue->rx_irq, queue);
+                       }
+                       queue->tx_irq = 0;
+               }
+
+               xenvif_unmap_frontend_rings(queue);
+       }
+}
+
+/* Reverse the relevant parts of xenvif_init_queue().
+ * Used for queue teardown from xenvif_free(), and on the
+ * error handling paths in xenbus.c:connect().
+ */
+void xenvif_deinit_queue(struct xenvif_queue *queue)
+{
+       gnttab_free_pages(MAX_PENDING_REQS, queue->mmap_pages);
+}
+
+void xenvif_free(struct xenvif *vif)
+{
+       struct xenvif_queue *queue = NULL;
+       unsigned int num_queues = vif->num_queues;
+       unsigned int queue_index;
+
+       unregister_netdev(vif->dev);
+
+       for (queue_index = 0; queue_index < num_queues; ++queue_index) {
+               queue = &vif->queues[queue_index];
+               xenvif_deinit_queue(queue);
+       }
+
+       vfree(vif->queues);
+       vif->queues = NULL;
+       vif->num_queues = 0;
+
+       free_netdev(vif->dev);
+
+       module_put(THIS_MODULE);
+}