ifb.c 7.1 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
*/


#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/init.h>
35
#include <linux/interrupt.h>
36
#include <linux/moduleparam.h>
37
#include <net/pkt_sched.h>
38
#include <net/net_namespace.h>
39 40 41 42 43

#define TX_Q_LIMIT    32
struct ifb_private {
	struct tasklet_struct   ifb_tasklet;
	int     tasklet_pending;
44 45

	struct u64_stats_sync	rsync;
46
	struct sk_buff_head     rq;
47 48 49 50
	u64 rx_packets;
	u64 rx_bytes;

	struct u64_stats_sync	tsync;
51
	struct sk_buff_head     tq;
52 53
	u64 tx_packets;
	u64 tx_bytes;
54 55
};

56
static int numifbs = 2;
57 58

static void ri_tasklet(unsigned long dev);
59
static netdev_tx_t ifb_xmit(struct sk_buff *skb, struct net_device *dev);
60 61 62
static int ifb_open(struct net_device *dev);
static int ifb_close(struct net_device *dev);

63
static void ri_tasklet(unsigned long dev)
64 65 66
{
	struct net_device *_dev = (struct net_device *)dev;
	struct ifb_private *dp = netdev_priv(_dev);
67
	struct netdev_queue *txq;
68 69
	struct sk_buff *skb;

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

E
Eric Dumazet 已提交
81
	while ((skb = __skb_dequeue(&dp->tq)) != NULL) {
82 83 84 85
		u32 from = G_TC_FROM(skb->tc_verd);

		skb->tc_verd = 0;
		skb->tc_verd = SET_TC_NCLS(skb->tc_verd);
86 87 88 89 90

		u64_stats_update_begin(&dp->tsync);
		dp->tx_packets++;
		dp->tx_bytes += skb->len;
		u64_stats_update_end(&dp->tsync);
91

92
		rcu_read_lock();
93
		skb->dev = dev_get_by_index_rcu(&init_net, skb->skb_iif);
94
		if (!skb->dev) {
95
			rcu_read_unlock();
96
			dev_kfree_skb(skb);
97
			_dev->stats.tx_dropped++;
98 99
			if (skb_queue_len(&dp->tq) != 0)
				goto resched;
100 101
			break;
		}
102
		rcu_read_unlock();
103
		skb->skb_iif = _dev->ifindex;
104

105 106 107
		if (from & AT_EGRESS) {
			dev_queue_xmit(skb);
		} else if (from & AT_INGRESS) {
108
			skb_pull(skb, skb->dev->hard_header_len);
109
			netif_receive_skb(skb);
110 111
		} else
			BUG();
112 113
	}

114
	if (__netif_tx_trylock(txq)) {
115 116 117 118 119
		if ((skb = skb_peek(&dp->rq)) == NULL) {
			dp->tasklet_pending = 0;
			if (netif_queue_stopped(_dev))
				netif_wake_queue(_dev);
		} else {
120
			__netif_tx_unlock(txq);
121 122
			goto resched;
		}
123
		__netif_tx_unlock(txq);
124 125 126 127 128 129 130 131
	} else {
resched:
		dp->tasklet_pending = 1;
		tasklet_schedule(&dp->ifb_tasklet);
	}

}

132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158
static struct rtnl_link_stats64 *ifb_stats64(struct net_device *dev,
					     struct rtnl_link_stats64 *stats)
{
	struct ifb_private *dp = netdev_priv(dev);
	unsigned int start;

	do {
		start = u64_stats_fetch_begin_bh(&dp->rsync);
		stats->rx_packets = dp->rx_packets;
		stats->rx_bytes = dp->rx_bytes;
	} while (u64_stats_fetch_retry_bh(&dp->rsync, start));

	do {
		start = u64_stats_fetch_begin_bh(&dp->tsync);

		stats->tx_packets = dp->tx_packets;
		stats->tx_bytes = dp->tx_bytes;

	} while (u64_stats_fetch_retry_bh(&dp->tsync, start));

	stats->rx_dropped = dev->stats.rx_dropped;
	stats->tx_dropped = dev->stats.tx_dropped;

	return stats;
}


159 160 161
static const struct net_device_ops ifb_netdev_ops = {
	.ndo_open	= ifb_open,
	.ndo_stop	= ifb_close,
162
	.ndo_get_stats64 = ifb_stats64,
163 164
	.ndo_start_xmit	= ifb_xmit,
	.ndo_validate_addr = eth_validate_addr,
165 166
};

167
#define IFB_FEATURES (NETIF_F_HW_CSUM | NETIF_F_SG  | NETIF_F_FRAGLIST	| \
E
Eric Dumazet 已提交
168 169 170
		      NETIF_F_TSO_ECN | NETIF_F_TSO | NETIF_F_TSO6	| \
		      NETIF_F_HIGHDMA | NETIF_F_HW_VLAN_TX)

171
static void ifb_setup(struct net_device *dev)
172 173
{
	/* Initialize the device structure. */
174
	dev->destructor = free_netdev;
175
	dev->netdev_ops = &ifb_netdev_ops;
176 177 178 179

	/* Fill in device structure with ethernet-generic values. */
	ether_setup(dev);
	dev->tx_queue_len = TX_Q_LIMIT;
180

E
Eric Dumazet 已提交
181 182 183
	dev->features |= IFB_FEATURES;
	dev->vlan_features |= IFB_FEATURES;

184 185
	dev->flags |= IFF_NOARP;
	dev->flags &= ~IFF_MULTICAST;
186
	dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_TX_SKB_SHARING);
187 188 189
	random_ether_addr(dev->dev_addr);
}

190
static netdev_tx_t ifb_xmit(struct sk_buff *skb, struct net_device *dev)
191 192 193 194
{
	struct ifb_private *dp = netdev_priv(dev);
	u32 from = G_TC_FROM(skb->tc_verd);

195 196 197 198
	u64_stats_update_begin(&dp->rsync);
	dp->rx_packets++;
	dp->rx_bytes += skb->len;
	u64_stats_update_end(&dp->rsync);
199

200
	if (!(from & (AT_INGRESS|AT_EGRESS)) || !skb->skb_iif) {
201
		dev_kfree_skb(skb);
202
		dev->stats.rx_dropped++;
203
		return NETDEV_TX_OK;
204 205 206 207 208 209
	}

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

210
	__skb_queue_tail(&dp->rq, skb);
211 212 213 214 215
	if (!dp->tasklet_pending) {
		dp->tasklet_pending = 1;
		tasklet_schedule(&dp->ifb_tasklet);
	}

216
	return NETDEV_TX_OK;
217 218 219 220 221 222 223 224
}

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

	tasklet_kill(&dp->ifb_tasklet);
	netif_stop_queue(dev);
225 226
	__skb_queue_purge(&dp->rq);
	__skb_queue_purge(&dp->tq);
227 228 229 230 231 232 233 234
	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);
235 236
	__skb_queue_head_init(&dp->rq);
	__skb_queue_head_init(&dp->tq);
237 238 239 240 241
	netif_start_queue(dev);

	return 0;
}

242 243 244 245 246 247 248 249 250 251 252
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;
}

253 254 255 256
static struct rtnl_link_ops ifb_link_ops __read_mostly = {
	.kind		= "ifb",
	.priv_size	= sizeof(struct ifb_private),
	.setup		= ifb_setup,
257
	.validate	= ifb_validate,
258 259
};

260 261 262 263
/* Number of ifb devices to be set up by this module. */
module_param(numifbs, int, 0);
MODULE_PARM_DESC(numifbs, "Number of ifb devices");

264 265 266 267 268 269 270 271 272 273 274
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;

275 276 277 278
	dev_ifb->rtnl_link_ops = &ifb_link_ops;
	err = register_netdevice(dev_ifb);
	if (err < 0)
		goto err;
279

280
	return 0;
281

282 283 284
err:
	free_netdev(dev_ifb);
	return err;
285
}
286 287

static int __init ifb_init_module(void)
288
{
289 290 291 292
	int i, err;

	rtnl_lock();
	err = __rtnl_link_register(&ifb_link_ops);
293

294
	for (i = 0; i < numifbs && !err; i++)
295
		err = ifb_init_one(i);
296
	if (err)
297 298
		__rtnl_link_unregister(&ifb_link_ops);
	rtnl_unlock();
299 300

	return err;
301
}
302 303 304

static void __exit ifb_cleanup_module(void)
{
305
	rtnl_link_unregister(&ifb_link_ops);
306 307 308 309 310 311
}

module_init(ifb_init_module);
module_exit(ifb_cleanup_module);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Jamal Hadi Salim");
312
MODULE_ALIAS_RTNL_LINK("ifb");