[mirror_ubuntu-bionic-kernel.git] / drivers / net / virtio_net.c

/* A simple network driver using virtio.
 *
 * Copyright 2007 Rusty Russell <rusty@rustcorp.com.au> IBM Corporation
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */
//#define DEBUG
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/ethtool.h>
#include <linux/module.h>
#include <linux/virtio.h>
#include <linux/virtio_net.h>
#include <linux/scatterlist.h>

static int napi_weight = 128;
module_param(napi_weight, int, 0444);

static int csum = 1, gso = 1;
module_param(csum, bool, 0444);
module_param(gso, bool, 0444);

/* FIXME: MTU in config. */
#define MAX_PACKET_LEN (ETH_HLEN+ETH_DATA_LEN)

struct virtnet_info
{
	struct virtio_device *vdev;
	struct virtqueue *rvq, *svq;
	struct net_device *dev;
	struct napi_struct napi;

	/* The skb we couldn't send because buffers were full. */
	struct sk_buff *last_xmit_skb;

	/* If we need to free in a timer, this is it. */
	struct timer_list xmit_free_timer;

	/* Number of input buffers, and max we've ever had. */
	unsigned int num, max;

	/* For cleaning up after transmission. */
	struct tasklet_struct tasklet;
	bool free_in_tasklet;

	/* Receive & send queues. */
	struct sk_buff_head recv;
	struct sk_buff_head send;
};

static inline struct virtio_net_hdr *skb_vnet_hdr(struct sk_buff *skb)
{
	return (struct virtio_net_hdr *)skb->cb;
}

static inline void vnet_hdr_to_sg(struct scatterlist *sg, struct sk_buff *skb)
{
	sg_init_one(sg, skb_vnet_hdr(skb), sizeof(struct virtio_net_hdr));
}

static void skb_xmit_done(struct virtqueue *svq)
{
	struct virtnet_info *vi = svq->vdev->priv;

	/* Suppress further interrupts. */
	svq->vq_ops->disable_cb(svq);

	/* We were probably waiting for more output buffers. */
	netif_wake_queue(vi->dev);

	/* Make sure we re-xmit last_xmit_skb: if there are no more packets
	 * queued, start_xmit won't be called. */
	tasklet_schedule(&vi->tasklet);
}

static void receive_skb(struct net_device *dev, struct sk_buff *skb,
			unsigned len)
{
	struct virtio_net_hdr *hdr = skb_vnet_hdr(skb);

	if (unlikely(len < sizeof(struct virtio_net_hdr) + ETH_HLEN)) {
		pr_debug("%s: short packet %i\n", dev->name, len);
		dev->stats.rx_length_errors++;
		goto drop;
	}
	len -= sizeof(struct virtio_net_hdr);
	BUG_ON(len > MAX_PACKET_LEN);

	skb_trim(skb, len);

	dev->stats.rx_bytes += skb->len;
	dev->stats.rx_packets++;

	if (hdr->flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
		pr_debug("Needs csum!\n");
		if (!skb_partial_csum_set(skb,hdr->csum_start,hdr->csum_offset))
			goto frame_err;
	}

	skb->protocol = eth_type_trans(skb, dev);
	pr_debug("Receiving skb proto 0x%04x len %i type %i\n",
		 ntohs(skb->protocol), skb->len, skb->pkt_type);

	if (hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
		pr_debug("GSO!\n");
		switch (hdr->gso_type & ~VIRTIO_NET_HDR_GSO_ECN) {
		case VIRTIO_NET_HDR_GSO_TCPV4:
			skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
			break;
		case VIRTIO_NET_HDR_GSO_UDP:
			skb_shinfo(skb)->gso_type = SKB_GSO_UDP;
			break;
		case VIRTIO_NET_HDR_GSO_TCPV6:
			skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
			break;
		default:
			if (net_ratelimit())
				printk(KERN_WARNING "%s: bad gso type %u.\n",
				       dev->name, hdr->gso_type);
			goto frame_err;
		}

		if (hdr->gso_type & VIRTIO_NET_HDR_GSO_ECN)
			skb_shinfo(skb)->gso_type |= SKB_GSO_TCP_ECN;

		skb_shinfo(skb)->gso_size = hdr->gso_size;
		if (skb_shinfo(skb)->gso_size == 0) {
			if (net_ratelimit())
				printk(KERN_WARNING "%s: zero gso size.\n",
				       dev->name);
			goto frame_err;
		}

		/* Header must be checked, and gso_segs computed. */
		skb_shinfo(skb)->gso_type |= SKB_GSO_DODGY;
		skb_shinfo(skb)->gso_segs = 0;
	}

	netif_receive_skb(skb);
	return;

frame_err:
	dev->stats.rx_frame_errors++;
drop:
	dev_kfree_skb(skb);
}

static void try_fill_recv(struct virtnet_info *vi)
{
	struct sk_buff *skb;
	struct scatterlist sg[2+MAX_SKB_FRAGS];
	int num, err;

	sg_init_table(sg, 2+MAX_SKB_FRAGS);
	for (;;) {
		skb = netdev_alloc_skb(vi->dev, MAX_PACKET_LEN);
		if (unlikely(!skb))
			break;

		skb_put(skb, MAX_PACKET_LEN);
		vnet_hdr_to_sg(sg, skb);
		num = skb_to_sgvec(skb, sg+1, 0, skb->len) + 1;
		skb_queue_head(&vi->recv, skb);

		err = vi->rvq->vq_ops->add_buf(vi->rvq, sg, 0, num, skb);
		if (err) {
			skb_unlink(skb, &vi->recv);
			kfree_skb(skb);
			break;
		}
		vi->num++;
	}
	if (unlikely(vi->num > vi->max))
		vi->max = vi->num;
	vi->rvq->vq_ops->kick(vi->rvq);
}

static void skb_recv_done(struct virtqueue *rvq)
{
	struct virtnet_info *vi = rvq->vdev->priv;
	/* Schedule NAPI, Suppress further interrupts if successful. */
	if (netif_rx_schedule_prep(vi->dev, &vi->napi)) {
		rvq->vq_ops->disable_cb(rvq);
		__netif_rx_schedule(vi->dev, &vi->napi);
	}
}

static int virtnet_poll(struct napi_struct *napi, int budget)
{
	struct virtnet_info *vi = container_of(napi, struct virtnet_info, napi);
	struct sk_buff *skb = NULL;
	unsigned int len, received = 0;

again:
	while (received < budget &&
	       (skb = vi->rvq->vq_ops->get_buf(vi->rvq, &len)) != NULL) {
		__skb_unlink(skb, &vi->recv);
		receive_skb(vi->dev, skb, len);
		vi->num--;
		received++;
	}

	/* FIXME: If we oom and completely run out of inbufs, we need
	 * to start a timer trying to fill more. */
	if (vi->num < vi->max / 2)
		try_fill_recv(vi);

	/* Out of packets? */
	if (received < budget) {
		netif_rx_complete(vi->dev, napi);
		if (unlikely(!vi->rvq->vq_ops->enable_cb(vi->rvq))
		    && napi_schedule_prep(napi)) {
			vi->rvq->vq_ops->disable_cb(vi->rvq);
			__netif_rx_schedule(vi->dev, napi);
			goto again;
		}
	}

	return received;
}

static void free_old_xmit_skbs(struct virtnet_info *vi)
{
	struct sk_buff *skb;
	unsigned int len;

	while ((skb = vi->svq->vq_ops->get_buf(vi->svq, &len)) != NULL) {
		pr_debug("Sent skb %p\n", skb);
		__skb_unlink(skb, &vi->send);
		vi->dev->stats.tx_bytes += skb->len;
		vi->dev->stats.tx_packets++;
		kfree_skb(skb);
	}
}

/* If the virtio transport doesn't always notify us when all in-flight packets
 * are consumed, we fall back to using this function on a timer to free them. */
static void xmit_free(unsigned long data)
{
	struct virtnet_info *vi = (void *)data;

	netif_tx_lock(vi->dev);

	free_old_xmit_skbs(vi);

	if (!skb_queue_empty(&vi->send))
		mod_timer(&vi->xmit_free_timer, jiffies + (HZ/10));

	netif_tx_unlock(vi->dev);
}

static int xmit_skb(struct virtnet_info *vi, struct sk_buff *skb)
{
	int num, err;
	struct scatterlist sg[2+MAX_SKB_FRAGS];
	struct virtio_net_hdr *hdr;
	const unsigned char *dest = ((struct ethhdr *)skb->data)->h_dest;

	sg_init_table(sg, 2+MAX_SKB_FRAGS);

	pr_debug("%s: xmit %p " MAC_FMT "\n", vi->dev->name, skb,
		 dest[0], dest[1], dest[2],
		 dest[3], dest[4], dest[5]);

	/* Encode metadata header at front. */
	hdr = skb_vnet_hdr(skb);
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
		hdr->flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
		hdr->csum_start = skb->csum_start - skb_headroom(skb);
		hdr->csum_offset = skb->csum_offset;
	} else {
		hdr->flags = 0;
		hdr->csum_offset = hdr->csum_start = 0;
	}

	if (skb_is_gso(skb)) {
		hdr->hdr_len = skb_transport_header(skb) - skb->data;
		hdr->gso_size = skb_shinfo(skb)->gso_size;
		if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
			hdr->gso_type = VIRTIO_NET_HDR_GSO_TCPV4;
		else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
			hdr->gso_type = VIRTIO_NET_HDR_GSO_TCPV6;
		else if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP)
			hdr->gso_type = VIRTIO_NET_HDR_GSO_UDP;
		else
			BUG();
		if (skb_shinfo(skb)->gso_type & SKB_GSO_TCP_ECN)
			hdr->gso_type |= VIRTIO_NET_HDR_GSO_ECN;
	} else {
		hdr->gso_type = VIRTIO_NET_HDR_GSO_NONE;
		hdr->gso_size = hdr->hdr_len = 0;
	}

	vnet_hdr_to_sg(sg, skb);
	num = skb_to_sgvec(skb, sg+1, 0, skb->len) + 1;

	err = vi->svq->vq_ops->add_buf(vi->svq, sg, num, 0, skb);
	if (!err && !vi->free_in_tasklet)
		mod_timer(&vi->xmit_free_timer, jiffies + (HZ/10));

	return err;
}

static void xmit_tasklet(unsigned long data)
{
	struct virtnet_info *vi = (void *)data;

	netif_tx_lock_bh(vi->dev);
	if (vi->last_xmit_skb && xmit_skb(vi, vi->last_xmit_skb) == 0) {
		vi->svq->vq_ops->kick(vi->svq);
		vi->last_xmit_skb = NULL;
	}
	if (vi->free_in_tasklet)
		free_old_xmit_skbs(vi);
	netif_tx_unlock_bh(vi->dev);
}

static int start_xmit(struct sk_buff *skb, struct net_device *dev)
{
	struct virtnet_info *vi = netdev_priv(dev);

again:
	/* Free up any pending old buffers before queueing new ones. */
	free_old_xmit_skbs(vi);

	/* If we has a buffer left over from last time, send it now. */
	if (unlikely(vi->last_xmit_skb) &&
	    xmit_skb(vi, vi->last_xmit_skb) != 0)
		goto stop_queue;

	vi->last_xmit_skb = NULL;

	/* Put new one in send queue and do transmit */
	if (likely(skb)) {
		__skb_queue_head(&vi->send, skb);
		if (xmit_skb(vi, skb) != 0) {
			vi->last_xmit_skb = skb;
			skb = NULL;
			goto stop_queue;
		}
	}
done:
	vi->svq->vq_ops->kick(vi->svq);
	return NETDEV_TX_OK;

stop_queue:
	pr_debug("%s: virtio not prepared to send\n", dev->name);
	netif_stop_queue(dev);

	/* Activate callback for using skbs: if this returns false it
	 * means some were used in the meantime. */
	if (unlikely(!vi->svq->vq_ops->enable_cb(vi->svq))) {
		vi->svq->vq_ops->disable_cb(vi->svq);
		netif_start_queue(dev);
		goto again;
	}
	if (skb) {
		/* Drop this skb: we only queue one. */
		vi->dev->stats.tx_dropped++;
		kfree_skb(skb);
	}
	goto done;
}

#ifdef CONFIG_NET_POLL_CONTROLLER
static void virtnet_netpoll(struct net_device *dev)
{
	struct virtnet_info *vi = netdev_priv(dev);

	napi_schedule(&vi->napi);
}
#endif

static int virtnet_open(struct net_device *dev)
{
	struct virtnet_info *vi = netdev_priv(dev);

	napi_enable(&vi->napi);

	/* If all buffers were filled by other side before we napi_enabled, we
	 * won't get another interrupt, so process any outstanding packets
	 * now.  virtnet_poll wants re-enable the queue, so we disable here.
	 * We synchronize against interrupts via NAPI_STATE_SCHED */
	if (netif_rx_schedule_prep(dev, &vi->napi)) {
		vi->rvq->vq_ops->disable_cb(vi->rvq);
		__netif_rx_schedule(dev, &vi->napi);
	}
	return 0;
}

static int virtnet_close(struct net_device *dev)
{
	struct virtnet_info *vi = netdev_priv(dev);

	napi_disable(&vi->napi);

	return 0;
}

static int virtnet_set_tx_csum(struct net_device *dev, u32 data)
{
	struct virtnet_info *vi = netdev_priv(dev);
	struct virtio_device *vdev = vi->vdev;

	if (data && !virtio_has_feature(vdev, VIRTIO_NET_F_CSUM))
		return -ENOSYS;

	return ethtool_op_set_tx_hw_csum(dev, data);
}

static struct ethtool_ops virtnet_ethtool_ops = {
	.set_tx_csum = virtnet_set_tx_csum,
	.set_sg = ethtool_op_set_sg,
};

static int virtnet_probe(struct virtio_device *vdev)
{
	int err;
	struct net_device *dev;
	struct virtnet_info *vi;

	/* Allocate ourselves a network device with room for our info */
	dev = alloc_etherdev(sizeof(struct virtnet_info));
	if (!dev)
		return -ENOMEM;

	/* Set up network device as normal. */
	dev->open = virtnet_open;
	dev->stop = virtnet_close;
	dev->hard_start_xmit = start_xmit;
	dev->features = NETIF_F_HIGHDMA;
#ifdef CONFIG_NET_POLL_CONTROLLER
	dev->poll_controller = virtnet_netpoll;
#endif
	SET_ETHTOOL_OPS(dev, &virtnet_ethtool_ops);
	SET_NETDEV_DEV(dev, &vdev->dev);

	/* Do we support "hardware" checksums? */
	if (csum && virtio_has_feature(vdev, VIRTIO_NET_F_CSUM)) {
		/* This opens up the world of extra features. */
		dev->features |= NETIF_F_HW_CSUM|NETIF_F_SG|NETIF_F_FRAGLIST;
		if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_GSO)) {
			dev->features |= NETIF_F_TSO | NETIF_F_UFO
				| NETIF_F_TSO_ECN | NETIF_F_TSO6;
		}
		/* Individual feature bits: what can host handle? */
		if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_HOST_TSO4))
			dev->features |= NETIF_F_TSO;
		if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_HOST_TSO6))
			dev->features |= NETIF_F_TSO6;
		if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_HOST_ECN))
			dev->features |= NETIF_F_TSO_ECN;
		if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_HOST_UFO))
			dev->features |= NETIF_F_UFO;
	}

	/* Configuration may specify what MAC to use.  Otherwise random. */
	if (virtio_has_feature(vdev, VIRTIO_NET_F_MAC)) {
		vdev->config->get(vdev,
				  offsetof(struct virtio_net_config, mac),
				  dev->dev_addr, dev->addr_len);
	} else
		random_ether_addr(dev->dev_addr);

	/* Set up our device-specific information */
	vi = netdev_priv(dev);
	netif_napi_add(dev, &vi->napi, virtnet_poll, napi_weight);
	vi->dev = dev;
	vi->vdev = vdev;
	vdev->priv = vi;

	/* If they give us a callback when all buffers are done, we don't need
	 * the timer. */
	vi->free_in_tasklet = virtio_has_feature(vdev,VIRTIO_F_NOTIFY_ON_EMPTY);

	/* We expect two virtqueues, receive then send. */
	vi->rvq = vdev->config->find_vq(vdev, 0, skb_recv_done);
	if (IS_ERR(vi->rvq)) {
		err = PTR_ERR(vi->rvq);
		goto free;
	}

	vi->svq = vdev->config->find_vq(vdev, 1, skb_xmit_done);
	if (IS_ERR(vi->svq)) {
		err = PTR_ERR(vi->svq);
		goto free_recv;
	}

	/* Initialize our empty receive and send queues. */
	skb_queue_head_init(&vi->recv);
	skb_queue_head_init(&vi->send);

	tasklet_init(&vi->tasklet, xmit_tasklet, (unsigned long)vi);

	if (!vi->free_in_tasklet)
		setup_timer(&vi->xmit_free_timer, xmit_free, (unsigned long)vi);

	err = register_netdev(dev);
	if (err) {
		pr_debug("virtio_net: registering device failed\n");
		goto free_send;
	}

	/* Last of all, set up some receive buffers. */
	try_fill_recv(vi);

	/* If we didn't even get one input buffer, we're useless. */
	if (vi->num == 0) {
		err = -ENOMEM;
		goto unregister;
	}

	pr_debug("virtnet: registered device %s\n", dev->name);
	return 0;

unregister:
	unregister_netdev(dev);
free_send:
	vdev->config->del_vq(vi->svq);
free_recv:
	vdev->config->del_vq(vi->rvq);
free:
	free_netdev(dev);
	return err;
}

static void virtnet_remove(struct virtio_device *vdev)
{
	struct virtnet_info *vi = vdev->priv;
	struct sk_buff *skb;

	/* Stop all the virtqueues. */
	vdev->config->reset(vdev);

	if (!vi->free_in_tasklet)
		del_timer_sync(&vi->xmit_free_timer);

	/* Free our skbs in send and recv queues, if any. */
	while ((skb = __skb_dequeue(&vi->recv)) != NULL) {
		kfree_skb(skb);
		vi->num--;
	}
	__skb_queue_purge(&vi->send);

	BUG_ON(vi->num != 0);

	vdev->config->del_vq(vi->svq);
	vdev->config->del_vq(vi->rvq);
	unregister_netdev(vi->dev);
	free_netdev(vi->dev);
}

static struct virtio_device_id id_table[] = {
	{ VIRTIO_ID_NET, VIRTIO_DEV_ANY_ID },
	{ 0 },
};

static unsigned int features[] = {
	VIRTIO_NET_F_CSUM, VIRTIO_NET_F_GUEST_CSUM,
	VIRTIO_NET_F_GSO, VIRTIO_NET_F_MAC,
	VIRTIO_NET_F_HOST_TSO4, VIRTIO_NET_F_HOST_UFO, VIRTIO_NET_F_HOST_TSO6,
	VIRTIO_NET_F_HOST_ECN, VIRTIO_F_NOTIFY_ON_EMPTY,
};

static struct virtio_driver virtio_net = {
	.feature_table = features,
	.feature_table_size = ARRAY_SIZE(features),
	.driver.name =	KBUILD_MODNAME,
	.driver.owner =	THIS_MODULE,
	.id_table =	id_table,
	.probe =	virtnet_probe,
	.remove =	__devexit_p(virtnet_remove),
};

static int __init init(void)
{
	return register_virtio_driver(&virtio_net);
}

static void __exit fini(void)
{
	unregister_virtio_driver(&virtio_net);
}
module_init(init);
module_exit(fini);

MODULE_DEVICE_TABLE(virtio, id_table);
MODULE_DESCRIPTION("Virtio network driver");
MODULE_LICENSE("GPL");
Commit	Line	Data
296f96fc RR	1	/* A simple network driver using virtio.
	2	*
	3	* Copyright 2007 Rusty Russell <rusty@rustcorp.com.au> IBM Corporation
	4	*
	5	* This program is free software; you can redistribute it and/or modify
	6	* it under the terms of the GNU General Public License as published by
	7	* the Free Software Foundation; either version 2 of the License, or
	8	* (at your option) any later version.
	9	*
	10	* This program is distributed in the hope that it will be useful,
	11	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	12	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	13	* GNU General Public License for more details.
	14	*
	15	* You should have received a copy of the GNU General Public License
	16	* along with this program; if not, write to the Free Software
	17	* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
	18	*/
	19	//#define DEBUG
	20	#include <linux/netdevice.h>
	21	#include <linux/etherdevice.h>
a9ea3fc6	22	#include <linux/ethtool.h>
296f96fc RR	23	#include <linux/module.h>
	24	#include <linux/virtio.h>
	25	#include <linux/virtio_net.h>
	26	#include <linux/scatterlist.h>
	27
6c0cd7c0 DL	28	static int napi_weight = 128;
	29	module_param(napi_weight, int, 0444);
	30
34a48579 RR	31	static int csum = 1, gso = 1;
	32	module_param(csum, bool, 0444);
	33	module_param(gso, bool, 0444);
	34
296f96fc RR	35	/* FIXME: MTU in config. */
	36	#define MAX_PACKET_LEN (ETH_HLEN+ETH_DATA_LEN)
	37
	38	struct virtnet_info
	39	{
	40	struct virtio_device *vdev;
	41	struct virtqueue rvq, svq;
	42	struct net_device *dev;
	43	struct napi_struct napi;
	44
99ffc696 RR	45	/* The skb we couldn't send because buffers were full. */
	46	struct sk_buff *last_xmit_skb;
	47
363f1514	48	/* If we need to free in a timer, this is it. */
14c998f0 MM	49	struct timer_list xmit_free_timer;
14c998f0 MM	50
296f96fc RR	51	/* Number of input buffers, and max we've ever had. */
	52	unsigned int num, max;
	53
11a3a154 RR	54	/* For cleaning up after transmission. */
11a3a154 RR	55	struct tasklet_struct tasklet;
363f1514	56	bool free_in_tasklet;
11a3a154	57
296f96fc RR	58	/* Receive & send queues. */
	59	struct sk_buff_head recv;
	60	struct sk_buff_head send;
	61	};
	62
	63	static inline struct virtio_net_hdr skb_vnet_hdr(struct sk_buff skb)
	64	{
	65	return (struct virtio_net_hdr *)skb->cb;
	66	}
	67
	68	static inline void vnet_hdr_to_sg(struct scatterlist sg, struct sk_buff skb)
	69	{
	70	sg_init_one(sg, skb_vnet_hdr(skb), sizeof(struct virtio_net_hdr));
	71	}
	72
2cb9c6ba	73	static void skb_xmit_done(struct virtqueue *svq)
296f96fc	74	{
2cb9c6ba	75	struct virtnet_info *vi = svq->vdev->priv;
296f96fc	76
2cb9c6ba RR	77	/* Suppress further interrupts. */
2cb9c6ba RR	78	svq->vq_ops->disable_cb(svq);
11a3a154	79
363f1514	80	/* We were probably waiting for more output buffers. */
296f96fc	81	netif_wake_queue(vi->dev);
11a3a154 RR	82
	83	/* Make sure we re-xmit last_xmit_skb: if there are no more packets
	84	* queued, start_xmit won't be called. */
	85	tasklet_schedule(&vi->tasklet);
296f96fc RR	86	}
	87
	88	static void receive_skb(struct net_device dev, struct sk_buff skb,
	89	unsigned len)
	90	{
	91	struct virtio_net_hdr *hdr = skb_vnet_hdr(skb);
	92
	93	if (unlikely(len < sizeof(struct virtio_net_hdr) + ETH_HLEN)) {
	94	pr_debug("%s: short packet %i\n", dev->name, len);
	95	dev->stats.rx_length_errors++;
	96	goto drop;
	97	}
	98	len -= sizeof(struct virtio_net_hdr);
	99	BUG_ON(len > MAX_PACKET_LEN);
	100
	101	skb_trim(skb, len);
23cde76d	102
296f96fc RR	103	dev->stats.rx_bytes += skb->len;
	104	dev->stats.rx_packets++;
	105
	106	if (hdr->flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
	107	pr_debug("Needs csum!\n");
f35d9d8a	108	if (!skb_partial_csum_set(skb,hdr->csum_start,hdr->csum_offset))
296f96fc	109	goto frame_err;
296f96fc RR	110	}
296f96fc RR	111
23cde76d MM	112	skb->protocol = eth_type_trans(skb, dev);
	113	pr_debug("Receiving skb proto 0x%04x len %i type %i\n",
	114	ntohs(skb->protocol), skb->len, skb->pkt_type);
	115
296f96fc RR	116	if (hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
296f96fc RR	117	pr_debug("GSO!\n");
34a48579	118	switch (hdr->gso_type & ~VIRTIO_NET_HDR_GSO_ECN) {
296f96fc RR	119	case VIRTIO_NET_HDR_GSO_TCPV4:
	120	skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
	121	break;
296f96fc RR	122	case VIRTIO_NET_HDR_GSO_UDP:
	123	skb_shinfo(skb)->gso_type = SKB_GSO_UDP;
	124	break;
	125	case VIRTIO_NET_HDR_GSO_TCPV6:
	126	skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
	127	break;
	128	default:
	129	if (net_ratelimit())
	130	printk(KERN_WARNING "%s: bad gso type %u.\n",
	131	dev->name, hdr->gso_type);
	132	goto frame_err;
	133	}
	134
34a48579 RR	135	if (hdr->gso_type & VIRTIO_NET_HDR_GSO_ECN)
	136	skb_shinfo(skb)->gso_type \|= SKB_GSO_TCP_ECN;
	137
296f96fc RR	138	skb_shinfo(skb)->gso_size = hdr->gso_size;
	139	if (skb_shinfo(skb)->gso_size == 0) {
	140	if (net_ratelimit())
	141	printk(KERN_WARNING "%s: zero gso size.\n",
	142	dev->name);
	143	goto frame_err;
	144	}
	145
	146	/* Header must be checked, and gso_segs computed. */
	147	skb_shinfo(skb)->gso_type \|= SKB_GSO_DODGY;
	148	skb_shinfo(skb)->gso_segs = 0;
	149	}
	150
	151	netif_receive_skb(skb);
	152	return;
	153
	154	frame_err:
	155	dev->stats.rx_frame_errors++;
	156	drop:
	157	dev_kfree_skb(skb);
	158	}
	159
	160	static void try_fill_recv(struct virtnet_info *vi)
	161	{
	162	struct sk_buff *skb;
05271685	163	struct scatterlist sg[2+MAX_SKB_FRAGS];
296f96fc RR	164	int num, err;
296f96fc RR	165
05271685	166	sg_init_table(sg, 2+MAX_SKB_FRAGS);
296f96fc RR	167	for (;;) {
	168	skb = netdev_alloc_skb(vi->dev, MAX_PACKET_LEN);
	169	if (unlikely(!skb))
	170	break;
	171
	172	skb_put(skb, MAX_PACKET_LEN);
	173	vnet_hdr_to_sg(sg, skb);
	174	num = skb_to_sgvec(skb, sg+1, 0, skb->len) + 1;
	175	skb_queue_head(&vi->recv, skb);
	176
	177	err = vi->rvq->vq_ops->add_buf(vi->rvq, sg, 0, num, skb);
	178	if (err) {
	179	skb_unlink(skb, &vi->recv);
	180	kfree_skb(skb);
	181	break;
	182	}
	183	vi->num++;
	184	}
	185	if (unlikely(vi->num > vi->max))
	186	vi->max = vi->num;
	187	vi->rvq->vq_ops->kick(vi->rvq);
	188	}
	189
18445c4d	190	static void skb_recv_done(struct virtqueue *rvq)
296f96fc RR	191	{
296f96fc RR	192	struct virtnet_info *vi = rvq->vdev->priv;
18445c4d RR	193	/* Schedule NAPI, Suppress further interrupts if successful. */
	194	if (netif_rx_schedule_prep(vi->dev, &vi->napi)) {
	195	rvq->vq_ops->disable_cb(rvq);
	196	__netif_rx_schedule(vi->dev, &vi->napi);
	197	}
296f96fc RR	198	}
	199
	200	static int virtnet_poll(struct napi_struct *napi, int budget)
	201	{
	202	struct virtnet_info *vi = container_of(napi, struct virtnet_info, napi);
	203	struct sk_buff *skb = NULL;
	204	unsigned int len, received = 0;
	205
	206	again:
	207	while (received < budget &&
	208	(skb = vi->rvq->vq_ops->get_buf(vi->rvq, &len)) != NULL) {
	209	__skb_unlink(skb, &vi->recv);
	210	receive_skb(vi->dev, skb, len);
	211	vi->num--;
	212	received++;
	213	}
	214
	215	/* FIXME: If we oom and completely run out of inbufs, we need
	216	* to start a timer trying to fill more. */
	217	if (vi->num < vi->max / 2)
	218	try_fill_recv(vi);
	219
8329d98e RR	220	/* Out of packets? */
8329d98e RR	221	if (received < budget) {
296f96fc	222	netif_rx_complete(vi->dev, napi);
18445c4d	223	if (unlikely(!vi->rvq->vq_ops->enable_cb(vi->rvq))
4265f161 CB	224	&& napi_schedule_prep(napi)) {
	225	vi->rvq->vq_ops->disable_cb(vi->rvq);
	226	__netif_rx_schedule(vi->dev, napi);
296f96fc	227	goto again;
4265f161	228	}
296f96fc RR	229	}
	230
	231	return received;
	232	}
	233
	234	static void free_old_xmit_skbs(struct virtnet_info *vi)
	235	{
	236	struct sk_buff *skb;
	237	unsigned int len;
	238
	239	while ((skb = vi->svq->vq_ops->get_buf(vi->svq, &len)) != NULL) {
	240	pr_debug("Sent skb %p\n", skb);
	241	__skb_unlink(skb, &vi->send);
655aa31f	242	vi->dev->stats.tx_bytes += skb->len;
296f96fc RR	243	vi->dev->stats.tx_packets++;
	244	kfree_skb(skb);
	245	}
	246	}
	247
363f1514 RR	248	/* If the virtio transport doesn't always notify us when all in-flight packets
363f1514 RR	249	* are consumed, we fall back to using this function on a timer to free them. */
14c998f0 MM	250	static void xmit_free(unsigned long data)
	251	{
	252	struct virtnet_info vi = (void )data;
	253
	254	netif_tx_lock(vi->dev);
	255
	256	free_old_xmit_skbs(vi);
	257
	258	if (!skb_queue_empty(&vi->send))
	259	mod_timer(&vi->xmit_free_timer, jiffies + (HZ/10));
	260
	261	netif_tx_unlock(vi->dev);
	262	}
	263
99ffc696	264	static int xmit_skb(struct virtnet_info vi, struct sk_buff skb)
296f96fc	265	{
14c998f0	266	int num, err;
05271685	267	struct scatterlist sg[2+MAX_SKB_FRAGS];
296f96fc RR	268	struct virtio_net_hdr *hdr;
296f96fc RR	269	const unsigned char dest = ((struct ethhdr )skb->data)->h_dest;
296f96fc	270
05271685	271	sg_init_table(sg, 2+MAX_SKB_FRAGS);
4d125de3	272
99ffc696	273	pr_debug("%s: xmit %p " MAC_FMT "\n", vi->dev->name, skb,
21f644f3 DM	274	dest[0], dest[1], dest[2],
21f644f3 DM	275	dest[3], dest[4], dest[5]);
296f96fc	276
296f96fc RR	277	/* Encode metadata header at front. */
	278	hdr = skb_vnet_hdr(skb);
	279	if (skb->ip_summed == CHECKSUM_PARTIAL) {
	280	hdr->flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
	281	hdr->csum_start = skb->csum_start - skb_headroom(skb);
	282	hdr->csum_offset = skb->csum_offset;
	283	} else {
	284	hdr->flags = 0;
	285	hdr->csum_offset = hdr->csum_start = 0;
	286	}
	287
	288	if (skb_is_gso(skb)) {
50c8ea80	289	hdr->hdr_len = skb_transport_header(skb) - skb->data;
296f96fc	290	hdr->gso_size = skb_shinfo(skb)->gso_size;
34a48579	291	if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
296f96fc RR	292	hdr->gso_type = VIRTIO_NET_HDR_GSO_TCPV4;
	293	else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
	294	hdr->gso_type = VIRTIO_NET_HDR_GSO_TCPV6;
	295	else if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP)
	296	hdr->gso_type = VIRTIO_NET_HDR_GSO_UDP;
	297	else
	298	BUG();
34a48579 RR	299	if (skb_shinfo(skb)->gso_type & SKB_GSO_TCP_ECN)
34a48579 RR	300	hdr->gso_type \|= VIRTIO_NET_HDR_GSO_ECN;
296f96fc RR	301	} else {
296f96fc RR	302	hdr->gso_type = VIRTIO_NET_HDR_GSO_NONE;
50c8ea80	303	hdr->gso_size = hdr->hdr_len = 0;
296f96fc RR	304	}
	305
	306	vnet_hdr_to_sg(sg, skb);
	307	num = skb_to_sgvec(skb, sg+1, 0, skb->len) + 1;
99ffc696	308
14c998f0	309	err = vi->svq->vq_ops->add_buf(vi->svq, sg, num, 0, skb);
363f1514	310	if (!err && !vi->free_in_tasklet)
14c998f0 MM	311	mod_timer(&vi->xmit_free_timer, jiffies + (HZ/10));
	312
	313	return err;
99ffc696 RR	314	}
99ffc696 RR	315
11a3a154 RR	316	static void xmit_tasklet(unsigned long data)
	317	{
	318	struct virtnet_info vi = (void )data;
	319
	320	netif_tx_lock_bh(vi->dev);
	321	if (vi->last_xmit_skb && xmit_skb(vi, vi->last_xmit_skb) == 0) {
	322	vi->svq->vq_ops->kick(vi->svq);
	323	vi->last_xmit_skb = NULL;
	324	}
363f1514 RR	325	if (vi->free_in_tasklet)
363f1514 RR	326	free_old_xmit_skbs(vi);
11a3a154 RR	327	netif_tx_unlock_bh(vi->dev);
	328	}
	329
99ffc696 RR	330	static int start_xmit(struct sk_buff skb, struct net_device dev)
	331	{
	332	struct virtnet_info *vi = netdev_priv(dev);
2cb9c6ba RR	333
	334	again:
	335	/* Free up any pending old buffers before queueing new ones. */
	336	free_old_xmit_skbs(vi);
99ffc696 RR	337
99ffc696 RR	338	/* If we has a buffer left over from last time, send it now. */
9953ca6c MM	339	if (unlikely(vi->last_xmit_skb) &&
	340	xmit_skb(vi, vi->last_xmit_skb) != 0)
	341	goto stop_queue;
	342
	343	vi->last_xmit_skb = NULL;
2cb9c6ba	344
99ffc696	345	/* Put new one in send queue and do transmit */
7eb2e251 RR	346	if (likely(skb)) {
	347	__skb_queue_head(&vi->send, skb);
	348	if (xmit_skb(vi, skb) != 0) {
	349	vi->last_xmit_skb = skb;
	350	skb = NULL;
	351	goto stop_queue;
	352	}
296f96fc	353	}
99ffc696	354	done:
296f96fc	355	vi->svq->vq_ops->kick(vi->svq);
99ffc696 RR	356	return NETDEV_TX_OK;
	357
	358	stop_queue:
	359	pr_debug("%s: virtio not prepared to send\n", dev->name);
	360	netif_stop_queue(dev);
	361
	362	/* Activate callback for using skbs: if this returns false it
	363	* means some were used in the meantime. */
	364	if (unlikely(!vi->svq->vq_ops->enable_cb(vi->svq))) {
	365	vi->svq->vq_ops->disable_cb(vi->svq);
	366	netif_start_queue(dev);
	367	goto again;
	368	}
9953ca6c MM	369	if (skb) {
	370	/* Drop this skb: we only queue one. */
	371	vi->dev->stats.tx_dropped++;
	372	kfree_skb(skb);
	373	}
99ffc696	374	goto done;
296f96fc RR	375	}
296f96fc RR	376
da74e89d AS	377	#ifdef CONFIG_NET_POLL_CONTROLLER
	378	static void virtnet_netpoll(struct net_device *dev)
	379	{
	380	struct virtnet_info *vi = netdev_priv(dev);
	381
	382	napi_schedule(&vi->napi);
	383	}
	384	#endif
	385
296f96fc RR	386	static int virtnet_open(struct net_device *dev)
	387	{
	388	struct virtnet_info *vi = netdev_priv(dev);
	389
296f96fc	390	napi_enable(&vi->napi);
a48bd8f6 RR	391
	392	/* If all buffers were filled by other side before we napi_enabled, we
	393	* won't get another interrupt, so process any outstanding packets
370076d9 CB	394	* now. virtnet_poll wants re-enable the queue, so we disable here.
	395	* We synchronize against interrupts via NAPI_STATE_SCHED */
	396	if (netif_rx_schedule_prep(dev, &vi->napi)) {
	397	vi->rvq->vq_ops->disable_cb(vi->rvq);
	398	__netif_rx_schedule(dev, &vi->napi);
	399	}
296f96fc RR	400	return 0;
	401	}
	402
	403	static int virtnet_close(struct net_device *dev)
	404	{
	405	struct virtnet_info *vi = netdev_priv(dev);
296f96fc RR	406
	407	napi_disable(&vi->napi);
	408
296f96fc RR	409	return 0;
	410	}
	411
a9ea3fc6 HX	412	static int virtnet_set_tx_csum(struct net_device *dev, u32 data)
	413	{
	414	struct virtnet_info *vi = netdev_priv(dev);
	415	struct virtio_device *vdev = vi->vdev;
	416
	417	if (data && !virtio_has_feature(vdev, VIRTIO_NET_F_CSUM))
	418	return -ENOSYS;
	419
	420	return ethtool_op_set_tx_hw_csum(dev, data);
	421	}
	422
	423	static struct ethtool_ops virtnet_ethtool_ops = {
	424	.set_tx_csum = virtnet_set_tx_csum,
	425	.set_sg = ethtool_op_set_sg,
	426	};
	427
296f96fc RR	428	static int virtnet_probe(struct virtio_device *vdev)
	429	{
	430	int err;
296f96fc RR	431	struct net_device *dev;
296f96fc RR	432	struct virtnet_info *vi;
296f96fc RR	433
	434	/* Allocate ourselves a network device with room for our info */
	435	dev = alloc_etherdev(sizeof(struct virtnet_info));
	436	if (!dev)
	437	return -ENOMEM;
	438
	439	/* Set up network device as normal. */
296f96fc RR	440	dev->open = virtnet_open;
	441	dev->stop = virtnet_close;
	442	dev->hard_start_xmit = start_xmit;
	443	dev->features = NETIF_F_HIGHDMA;
da74e89d AS	444	#ifdef CONFIG_NET_POLL_CONTROLLER
	445	dev->poll_controller = virtnet_netpoll;
	446	#endif
a9ea3fc6	447	SET_ETHTOOL_OPS(dev, &virtnet_ethtool_ops);
296f96fc RR	448	SET_NETDEV_DEV(dev, &vdev->dev);
	449
	450	/* Do we support "hardware" checksums? */
c45a6816	451	if (csum && virtio_has_feature(vdev, VIRTIO_NET_F_CSUM)) {
296f96fc RR	452	/* This opens up the world of extra features. */
296f96fc RR	453	dev->features \|= NETIF_F_HW_CSUM\|NETIF_F_SG\|NETIF_F_FRAGLIST;
c45a6816	454	if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_GSO)) {
34a48579 RR	455	dev->features \|= NETIF_F_TSO \| NETIF_F_UFO
	456	\| NETIF_F_TSO_ECN \| NETIF_F_TSO6;
	457	}
5539ae96	458	/* Individual feature bits: what can host handle? */
c45a6816	459	if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_HOST_TSO4))
5539ae96	460	dev->features \|= NETIF_F_TSO;
c45a6816	461	if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_HOST_TSO6))
5539ae96	462	dev->features \|= NETIF_F_TSO6;
c45a6816	463	if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_HOST_ECN))
5539ae96	464	dev->features \|= NETIF_F_TSO_ECN;
c45a6816	465	if (gso && virtio_has_feature(vdev, VIRTIO_NET_F_HOST_UFO))
5539ae96	466	dev->features \|= NETIF_F_UFO;
296f96fc RR	467	}
	468
	469	/* Configuration may specify what MAC to use. Otherwise random. */
c45a6816	470	if (virtio_has_feature(vdev, VIRTIO_NET_F_MAC)) {
a586d4f6 RR	471	vdev->config->get(vdev,
	472	offsetof(struct virtio_net_config, mac),
	473	dev->dev_addr, dev->addr_len);
296f96fc RR	474	} else
	475	random_ether_addr(dev->dev_addr);
	476
	477	/* Set up our device-specific information */
	478	vi = netdev_priv(dev);
6c0cd7c0	479	netif_napi_add(dev, &vi->napi, virtnet_poll, napi_weight);
296f96fc RR	480	vi->dev = dev;
296f96fc RR	481	vi->vdev = vdev;
d9d5dcc8	482	vdev->priv = vi;
296f96fc	483
363f1514 RR	484	/* If they give us a callback when all buffers are done, we don't need
	485	* the timer. */
	486	vi->free_in_tasklet = virtio_has_feature(vdev,VIRTIO_F_NOTIFY_ON_EMPTY);
	487
296f96fc	488	/* We expect two virtqueues, receive then send. */
a586d4f6	489	vi->rvq = vdev->config->find_vq(vdev, 0, skb_recv_done);
296f96fc RR	490	if (IS_ERR(vi->rvq)) {
	491	err = PTR_ERR(vi->rvq);
	492	goto free;
	493	}
	494
a586d4f6	495	vi->svq = vdev->config->find_vq(vdev, 1, skb_xmit_done);
296f96fc RR	496	if (IS_ERR(vi->svq)) {
	497	err = PTR_ERR(vi->svq);
	498	goto free_recv;
	499	}
	500
	501	/* Initialize our empty receive and send queues. */
	502	skb_queue_head_init(&vi->recv);
	503	skb_queue_head_init(&vi->send);
	504
11a3a154 RR	505	tasklet_init(&vi->tasklet, xmit_tasklet, (unsigned long)vi);
11a3a154 RR	506
363f1514 RR	507	if (!vi->free_in_tasklet)
363f1514 RR	508	setup_timer(&vi->xmit_free_timer, xmit_free, (unsigned long)vi);
14c998f0	509
296f96fc RR	510	err = register_netdev(dev);
	511	if (err) {
	512	pr_debug("virtio_net: registering device failed\n");
	513	goto free_send;
	514	}
b3369c1f RR	515
	516	/* Last of all, set up some receive buffers. */
	517	try_fill_recv(vi);
	518
	519	/* If we didn't even get one input buffer, we're useless. */
	520	if (vi->num == 0) {
	521	err = -ENOMEM;
	522	goto unregister;
	523	}
	524
296f96fc	525	pr_debug("virtnet: registered device %s\n", dev->name);
296f96fc RR	526	return 0;
296f96fc RR	527
b3369c1f RR	528	unregister:
b3369c1f RR	529	unregister_netdev(dev);
296f96fc RR	530	free_send:
	531	vdev->config->del_vq(vi->svq);
	532	free_recv:
	533	vdev->config->del_vq(vi->rvq);
	534	free:
	535	free_netdev(dev);
	536	return err;
	537	}
	538
	539	static void virtnet_remove(struct virtio_device *vdev)
	540	{
74b2553f	541	struct virtnet_info *vi = vdev->priv;
b3369c1f RR	542	struct sk_buff *skb;
b3369c1f RR	543
6e5aa7ef RR	544	/* Stop all the virtqueues. */
	545	vdev->config->reset(vdev);
	546
363f1514 RR	547	if (!vi->free_in_tasklet)
363f1514 RR	548	del_timer_sync(&vi->xmit_free_timer);
14c998f0	549
b3369c1f	550	/* Free our skbs in send and recv queues, if any. */
b3369c1f RR	551	while ((skb = __skb_dequeue(&vi->recv)) != NULL) {
	552	kfree_skb(skb);
	553	vi->num--;
	554	}
288369cc	555	__skb_queue_purge(&vi->send);
b3369c1f RR	556
b3369c1f RR	557	BUG_ON(vi->num != 0);
74b2553f RR	558
	559	vdev->config->del_vq(vi->svq);
	560	vdev->config->del_vq(vi->rvq);
	561	unregister_netdev(vi->dev);
	562	free_netdev(vi->dev);
296f96fc RR	563	}
	564
	565	static struct virtio_device_id id_table[] = {
	566	{ VIRTIO_ID_NET, VIRTIO_DEV_ANY_ID },
	567	{ 0 },
	568	};
	569
c45a6816	570	static unsigned int features[] = {
5e4fe5c4 MM	571	VIRTIO_NET_F_CSUM, VIRTIO_NET_F_GUEST_CSUM,
5e4fe5c4 MM	572	VIRTIO_NET_F_GSO, VIRTIO_NET_F_MAC,
c45a6816	573	VIRTIO_NET_F_HOST_TSO4, VIRTIO_NET_F_HOST_UFO, VIRTIO_NET_F_HOST_TSO6,
363f1514	574	VIRTIO_NET_F_HOST_ECN, VIRTIO_F_NOTIFY_ON_EMPTY,
c45a6816 RR	575	};
c45a6816 RR	576
296f96fc	577	static struct virtio_driver virtio_net = {
c45a6816 RR	578	.feature_table = features,
c45a6816 RR	579	.feature_table_size = ARRAY_SIZE(features),
296f96fc RR	580	.driver.name = KBUILD_MODNAME,
	581	.driver.owner = THIS_MODULE,
	582	.id_table = id_table,
	583	.probe = virtnet_probe,
	584	.remove = __devexit_p(virtnet_remove),
	585	};
	586
	587	static int __init init(void)
	588	{
	589	return register_virtio_driver(&virtio_net);
	590	}
	591
	592	static void __exit fini(void)
	593	{
	594	unregister_virtio_driver(&virtio_net);
	595	}
	596	module_init(init);
	597	module_exit(fini);
	598
	599	MODULE_DEVICE_TABLE(virtio, id_table);
	600	MODULE_DESCRIPTION("Virtio network driver");
	601	MODULE_LICENSE("GPL");