ifb.c 6.1 KB
Newer Older
1
/* drivers/net/ifb.c:
2 3 4 5 6 7 8 9 10

	The purpose of this driver is to provide a device that allows
	for sharing of resources:

	1) qdiscs/policies that are per device as opposed to system wide.
	ifb allows for a device which can be redirected to thus providing
	an impression of sharing.

	2) Allows for queueing incoming traffic for shaping instead of
11 12
	dropping.

13 14 15 16 17 18 19 20 21 22 23
	The original concept is based on what is known as the IMQ
	driver initially written by Martin Devera, later rewritten
	by Patrick McHardy and then maintained by Andre Correa.

	You need the tc action  mirror or redirect to feed this device
       	packets.

	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.
24

25
  	Authors:	Jamal Hadi Salim (2005)
26

27 28 29 30 31 32 33 34 35
*/


#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/init.h>
#include <linux/moduleparam.h>
36
#include <net/pkt_sched.h>
37
#include <net/net_namespace.h>
38 39 40 41 42 43 44 45 46

#define TX_Q_LIMIT    32
struct ifb_private {
	struct tasklet_struct   ifb_tasklet;
	int     tasklet_pending;
	struct sk_buff_head     rq;
	struct sk_buff_head     tq;
};

47
static int numifbs = 2;
48 49

static void ri_tasklet(unsigned long dev);
50
static netdev_tx_t ifb_xmit(struct sk_buff *skb, struct net_device *dev);
51 52 53
static int ifb_open(struct net_device *dev);
static int ifb_close(struct net_device *dev);

54
static void ri_tasklet(unsigned long dev)
55 56 57 58
{

	struct net_device *_dev = (struct net_device *)dev;
	struct ifb_private *dp = netdev_priv(_dev);
59
	struct net_device_stats *stats = &_dev->stats;
60
	struct netdev_queue *txq;
61 62
	struct sk_buff *skb;

63
	txq = netdev_get_tx_queue(_dev, 0);
64
	if ((skb = skb_peek(&dp->tq)) == NULL) {
65
		if (__netif_tx_trylock(txq)) {
66
			skb_queue_splice_tail_init(&dp->rq, &dp->tq);
67
			__netif_tx_unlock(txq);
68 69 70 71 72 73 74 75 76 77 78 79 80
		} else {
			/* reschedule */
			goto resched;
		}
	}

	while ((skb = skb_dequeue(&dp->tq)) != NULL) {
		u32 from = G_TC_FROM(skb->tc_verd);

		skb->tc_verd = 0;
		skb->tc_verd = SET_TC_NCLS(skb->tc_verd);
		stats->tx_packets++;
		stats->tx_bytes +=skb->len;
81

82
		rcu_read_lock();
83
		skb->dev = dev_get_by_index_rcu(&init_net, skb->skb_iif);
84
		if (!skb->dev) {
85
			rcu_read_unlock();
86 87
			dev_kfree_skb(skb);
			stats->tx_dropped++;
88 89
			if (skb_queue_len(&dp->tq) != 0)
				goto resched;
90 91
			break;
		}
92
		rcu_read_unlock();
93
		skb->skb_iif = _dev->ifindex;
94

95 96 97
		if (from & AT_EGRESS) {
			dev_queue_xmit(skb);
		} else if (from & AT_INGRESS) {
98
			skb_pull(skb, skb->dev->hard_header_len);
99
			netif_rx(skb);
100 101
		} else
			BUG();
102 103
	}

104
	if (__netif_tx_trylock(txq)) {
105 106 107 108 109
		if ((skb = skb_peek(&dp->rq)) == NULL) {
			dp->tasklet_pending = 0;
			if (netif_queue_stopped(_dev))
				netif_wake_queue(_dev);
		} else {
110
			__netif_tx_unlock(txq);
111 112
			goto resched;
		}
113
		__netif_tx_unlock(txq);
114 115 116 117 118 119 120 121
	} else {
resched:
		dp->tasklet_pending = 1;
		tasklet_schedule(&dp->ifb_tasklet);
	}

}

S
Stephen Hemminger 已提交
122 123 124
static const struct net_device_ops ifb_netdev_ops = {
	.ndo_open	= ifb_open,
	.ndo_stop	= ifb_close,
125 126
	.ndo_start_xmit	= ifb_xmit,
	.ndo_validate_addr = eth_validate_addr,
S
Stephen Hemminger 已提交
127 128
};

P
Patrick McHardy 已提交
129
static void ifb_setup(struct net_device *dev)
130 131
{
	/* Initialize the device structure. */
P
Patrick McHardy 已提交
132
	dev->destructor = free_netdev;
S
Stephen Hemminger 已提交
133
	dev->netdev_ops = &ifb_netdev_ops;
134 135 136 137

	/* Fill in device structure with ethernet-generic values. */
	ether_setup(dev);
	dev->tx_queue_len = TX_Q_LIMIT;
S
Stephen Hemminger 已提交
138

139 140
	dev->flags |= IFF_NOARP;
	dev->flags &= ~IFF_MULTICAST;
141
	dev->priv_flags &= ~IFF_XMIT_DST_RELEASE;
142 143 144
	random_ether_addr(dev->dev_addr);
}

145
static netdev_tx_t ifb_xmit(struct sk_buff *skb, struct net_device *dev)
146 147
{
	struct ifb_private *dp = netdev_priv(dev);
148
	struct net_device_stats *stats = &dev->stats;
149 150
	u32 from = G_TC_FROM(skb->tc_verd);

D
dean gaudet 已提交
151 152
	stats->rx_packets++;
	stats->rx_bytes+=skb->len;
153

154
	if (!(from & (AT_INGRESS|AT_EGRESS)) || !skb->skb_iif) {
155 156
		dev_kfree_skb(skb);
		stats->rx_dropped++;
157
		return NETDEV_TX_OK;
158 159 160 161 162 163
	}

	if (skb_queue_len(&dp->rq) >= dev->tx_queue_len) {
		netif_stop_queue(dev);
	}

164
	__skb_queue_tail(&dp->rq, skb);
165 166 167 168 169
	if (!dp->tasklet_pending) {
		dp->tasklet_pending = 1;
		tasklet_schedule(&dp->ifb_tasklet);
	}

170
	return NETDEV_TX_OK;
171 172 173 174 175 176 177 178
}

static int ifb_close(struct net_device *dev)
{
	struct ifb_private *dp = netdev_priv(dev);

	tasklet_kill(&dp->ifb_tasklet);
	netif_stop_queue(dev);
179 180
	__skb_queue_purge(&dp->rq);
	__skb_queue_purge(&dp->tq);
181 182 183 184 185 186 187 188
	return 0;
}

static int ifb_open(struct net_device *dev)
{
	struct ifb_private *dp = netdev_priv(dev);

	tasklet_init(&dp->ifb_tasklet, ri_tasklet, (unsigned long)dev);
189 190
	__skb_queue_head_init(&dp->rq);
	__skb_queue_head_init(&dp->tq);
191 192 193 194 195
	netif_start_queue(dev);

	return 0;
}

196 197 198 199 200 201 202 203 204 205 206
static int ifb_validate(struct nlattr *tb[], struct nlattr *data[])
{
	if (tb[IFLA_ADDRESS]) {
		if (nla_len(tb[IFLA_ADDRESS]) != ETH_ALEN)
			return -EINVAL;
		if (!is_valid_ether_addr(nla_data(tb[IFLA_ADDRESS])))
			return -EADDRNOTAVAIL;
	}
	return 0;
}

P
Patrick McHardy 已提交
207 208 209 210
static struct rtnl_link_ops ifb_link_ops __read_mostly = {
	.kind		= "ifb",
	.priv_size	= sizeof(struct ifb_private),
	.setup		= ifb_setup,
211
	.validate	= ifb_validate,
P
Patrick McHardy 已提交
212 213
};

214 215 216 217
/* Number of ifb devices to be set up by this module. */
module_param(numifbs, int, 0);
MODULE_PARM_DESC(numifbs, "Number of ifb devices");

218 219 220 221 222 223 224 225 226 227 228
static int __init ifb_init_one(int index)
{
	struct net_device *dev_ifb;
	int err;

	dev_ifb = alloc_netdev(sizeof(struct ifb_private),
				 "ifb%d", ifb_setup);

	if (!dev_ifb)
		return -ENOMEM;

P
Patrick McHardy 已提交
229 230 231
	err = dev_alloc_name(dev_ifb, dev_ifb->name);
	if (err < 0)
		goto err;
232

P
Patrick McHardy 已提交
233 234 235 236
	dev_ifb->rtnl_link_ops = &ifb_link_ops;
	err = register_netdevice(dev_ifb);
	if (err < 0)
		goto err;
237

P
Patrick McHardy 已提交
238
	return 0;
239

P
Patrick McHardy 已提交
240 241 242
err:
	free_netdev(dev_ifb);
	return err;
243
}
244 245

static int __init ifb_init_module(void)
246
{
P
Patrick McHardy 已提交
247 248 249 250
	int i, err;

	rtnl_lock();
	err = __rtnl_link_register(&ifb_link_ops);
251

252
	for (i = 0; i < numifbs && !err; i++)
253
		err = ifb_init_one(i);
254
	if (err)
P
Patrick McHardy 已提交
255 256
		__rtnl_link_unregister(&ifb_link_ops);
	rtnl_unlock();
257 258

	return err;
259
}
260 261 262

static void __exit ifb_cleanup_module(void)
{
263
	rtnl_link_unregister(&ifb_link_ops);
264 265 266 267 268 269
}

module_init(ifb_init_module);
module_exit(ifb_cleanup_module);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Jamal Hadi Salim");
P
Patrick McHardy 已提交
270
MODULE_ALIAS_RTNL_LINK("ifb");