ipip.c

嵌入式系统设计与实验教材二源码linux内核移植与编译

/*
 *	Linux NET3:	IP/IP protocol decoder. 
 *
 *	Version: $Id: ipip.c,v 1.50 2001/10/02 02:22:36 davem Exp $
 *
 *	Authors:
 *		Sam Lantinga ([e-mail])  02/01/95
 *
 *	Fixes:
 *		Alan Cox	:	Merged and made usable non modular (its so tiny its silly as
 *					a module taking up 2 pages).
 *		Alan Cox	: 	Fixed bug with 1.3.18 and IPIP not working (now needs to set skb->h.iph)
 *					to keep ip_forward happy.
 *		Alan Cox	:	More fixes for 1.3.21, and firewall fix. Maybe this will work soon 8).
 *		Kai Schulte	:	Fixed #defines for IP_FIREWALL->FIREWALL
 *              David Woodhouse :       Perform some basic ICMP handling.
 *                                      IPIP Routing without decapsulation.
 *              Carlos Picoto   :       GRE over IP support
 *		Alexey Kuznetsov:	Reworked. Really, now it is truncated version of ipv4/ip_gre.c.
 *					I do not want to merge them together.
 *
 *	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.
 *
 */

/* tunnel.c: an IP tunnel driver

	The purpose of this driver is to provide an IP tunnel through
	which you can tunnel network traffic transparently across subnets.

	This was written by looking at Nick Holloway's dummy driver
	Thanks for the great code!

		-Sam Lantinga	([e-mail])  02/01/95
		
	Minor tweaks:
		Cleaned up the code a little and added some pre-1.3.0 tweaks.
		dev->hard_header/hard_header_len changed to use no headers.
		Comments/bracketing tweaked.
		Made the tunnels use dev->name not tunnel: when error reporting.
		Added tx_dropped stat
		
		-Alan Cox	([e-mail]) 21 March 95

	Reworked:
		Changed to tunnel to destination gateway in addition to the
			tunnel's pointopoint address
		Almost completely rewritten
		Note:  There is currently no firewall or ICMP handling done.

		-Sam Lantinga	([e-mail]) 02/13/96
		
*/

/* Things I wish I had known when writing the tunnel driver:

	When the tunnel_xmit() function is called, the skb contains the
	packet to be sent (plus a great deal of extra info), and dev
	contains the tunnel device that _we_ are.

	When we are passed a packet, we are expected to fill in the
	source address with our source IP address.

	What is the proper way to allocate, copy and free a buffer?
	After you allocate it, it is a "0 length" chunk of memory
	starting at zero.  If you want to add headers to the buffer
	later, you'll have to call "skb_reserve(skb, amount)" with
	the amount of memory you want reserved.  Then, you call
	"skb_put(skb, amount)" with the amount of space you want in
	the buffer.  skb_put() returns a pointer to the top (#0) of
	that buffer.  skb->len is set to the amount of space you have
	"allocated" with skb_put().  You can then write up to skb->len
	bytes to that buffer.  If you need more, you can call skb_put()
	again with the additional amount of space you need.  You can
	find out how much more space you can allocate by calling 
	"skb_tailroom(skb)".
	Now, to add header space, call "skb_push(skb, header_len)".
	This creates space at the beginning of the buffer and returns
	a pointer to this new space.  If later you need to strip a
	header from a buffer, call "skb_pull(skb, header_len)".
	skb_headroom() will return how much space is left at the top
	of the buffer (before the main data).  Remember, this headroom
	space must be reserved before the skb_put() function is called.
	*/

/*
   This version of net/ipv4/ipip.c is cloned of net/ipv4/ip_gre.c

   For comments look at net/ipv4/ip_gre.c --ANK
 */

 
#include <linux/config.h>
#include <linux/module.h>
#include <linux/types.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <asm/uaccess.h>
#include <linux/skbuff.h>
#include <linux/netdevice.h>
#include <linux/in.h>
#include <linux/tcp.h>
#include <linux/udp.h>
#include <linux/if_arp.h>
#include <linux/mroute.h>
#include <linux/init.h>
#include <linux/netfilter_ipv4.h>

#include <net/sock.h>
#include <net/ip.h>
#include <net/icmp.h>
#include <net/protocol.h>
#include <net/ipip.h>
#include <net/inet_ecn.h>

#define HASH_SIZE  16
#define HASH(addr) ((addr^(addr>>4))&0xF)

static int ipip_fb_tunnel_init(struct net_device *dev);
static int ipip_tunnel_init(struct net_device *dev);

static struct net_device ipip_fb_tunnel_dev = {
	name:	"tunl0",
	init:	ipip_fb_tunnel_init,
};

static struct ip_tunnel ipip_fb_tunnel = {
	dev:	&ipip_fb_tunnel_dev,
	parms:	{ name:	"tunl0", }
};

static struct ip_tunnel *tunnels_r_l[HASH_SIZE];
static struct ip_tunnel *tunnels_r[HASH_SIZE];
static struct ip_tunnel *tunnels_l[HASH_SIZE];
static struct ip_tunnel *tunnels_wc[1];
static struct ip_tunnel **tunnels[4] = { tunnels_wc, tunnels_l, tunnels_r, tunnels_r_l };

static rwlock_t ipip_lock = RW_LOCK_UNLOCKED;

static struct ip_tunnel * ipip_tunnel_lookup(u32 remote, u32 local)
{
	unsigned h0 = HASH(remote);
	unsigned h1 = HASH(local);
	struct ip_tunnel *t;

	for (t = tunnels_r_l[h0^h1]; t; t = t->next) {
		if (local == t->parms.iph.saddr &&
		    remote == t->parms.iph.daddr && (t->dev->flags&IFF_UP))
			return t;
	}
	for (t = tunnels_r[h0]; t; t = t->next) {
		if (remote == t->parms.iph.daddr && (t->dev->flags&IFF_UP))
			return t;
	}
	for (t = tunnels_l[h1]; t; t = t->next) {
		if (local == t->parms.iph.saddr && (t->dev->flags&IFF_UP))
			return t;
	}
	if ((t = tunnels_wc[0]) != NULL && (t->dev->flags&IFF_UP))
		return t;
	return NULL;
}

static struct ip_tunnel **ipip_bucket(struct ip_tunnel *t)
{
	u32 remote = t->parms.iph.daddr;
	u32 local = t->parms.iph.saddr;
	unsigned h = 0;
	int prio = 0;

	if (remote) {
		prio |= 2;
		h ^= HASH(remote);
	}
	if (local) {
		prio |= 1;
		h ^= HASH(local);
	}
	return &tunnels[prio][h];
}


static void ipip_tunnel_unlink(struct ip_tunnel *t)
{
	struct ip_tunnel **tp;

	for (tp = ipip_bucket(t); *tp; tp = &(*tp)->next) {
		if (t == *tp) {
			write_lock_bh(&ipip_lock);
			*tp = t->next;
			write_unlock_bh(&ipip_lock);
			break;
		}
	}
}

static void ipip_tunnel_link(struct ip_tunnel *t)
{
	struct ip_tunnel **tp = ipip_bucket(t);

	t->next = *tp;
	write_lock_bh(&ipip_lock);
	*tp = t;
	write_unlock_bh(&ipip_lock);
}

struct ip_tunnel * ipip_tunnel_locate(struct ip_tunnel_parm *parms, int create)
{
	u32 remote = parms->iph.daddr;
	u32 local = parms->iph.saddr;
	struct ip_tunnel *t, **tp, *nt;
	struct net_device *dev;
	unsigned h = 0;
	int prio = 0;

	if (remote) {
		prio |= 2;
		h ^= HASH(remote);
	}
	if (local) {
		prio |= 1;
		h ^= HASH(local);
	}
	for (tp = &tunnels[prio][h]; (t = *tp) != NULL; tp = &t->next) {
		if (local == t->parms.iph.saddr && remote == t->parms.iph.daddr)
			return t;
	}
	if (!create)
		return NULL;

	MOD_INC_USE_COUNT;
	dev = kmalloc(sizeof(*dev) + sizeof(*t), GFP_KERNEL);
	if (dev == NULL) {
		MOD_DEC_USE_COUNT;
		return NULL;
	}
	memset(dev, 0, sizeof(*dev) + sizeof(*t));
	dev->priv = (void*)(dev+1);
	nt = (struct ip_tunnel*)dev->priv;
	nt->dev = dev;
	dev->init = ipip_tunnel_init;
	dev->features |= NETIF_F_DYNALLOC;
	memcpy(&nt->parms, parms, sizeof(*parms));
	nt->parms.name[IFNAMSIZ-1] = '\0';
	strcpy(dev->name, nt->parms.name);
	if (dev->name[0] == 0) {
		int i;
		for (i=1; i<100; i++) {
			sprintf(dev->name, "tunl%d", i);
			if (__dev_get_by_name(dev->name) == NULL)
				break;
		}
		if (i==100)
			goto failed;
		memcpy(nt->parms.name, dev->name, IFNAMSIZ);
	}
	if (register_netdevice(dev) < 0)
		goto failed;

	dev_hold(dev);
	ipip_tunnel_link(nt);
	/* Do not decrement MOD_USE_COUNT here. */
	return nt;

failed:
	kfree(dev);
	MOD_DEC_USE_COUNT;
	return NULL;
}

static void ipip_tunnel_destructor(struct net_device *dev)
{
	if (dev != &ipip_fb_tunnel_dev) {
		MOD_DEC_USE_COUNT;
	}
}

static void ipip_tunnel_uninit(struct net_device *dev)
{
	if (dev == &ipip_fb_tunnel_dev) {
		write_lock_bh(&ipip_lock);
		tunnels_wc[0] = NULL;
		write_unlock_bh(&ipip_lock);
	} else
		ipip_tunnel_unlink((struct ip_tunnel*)dev->priv);
	dev_put(dev);
}

void ipip_err(struct sk_buff *skb, u32 info)
{
#ifndef I_WISH_WORLD_WERE_PERFECT

/* It is not :-( All the routers (except for Linux) return only
   8 bytes of packet payload. It means, that precise relaying of
   ICMP in the real Internet is absolutely infeasible.
 */
	struct iphdr *iph = (struct iphdr*)skb->data;
	int type = skb->h.icmph->type;
	int code = skb->h.icmph->code;
	struct ip_tunnel *t;

	switch (type) {
	default:
	case ICMP_PARAMETERPROB:
		return;

	case ICMP_DEST_UNREACH:
		switch (code) {
		case ICMP_SR_FAILED:
		case ICMP_PORT_UNREACH:
			/* Impossible event. */
			return;
		case ICMP_FRAG_NEEDED:
			/* Soft state for pmtu is maintained by IP core. */
			return;
		default:
			/* All others are translated to HOST_UNREACH.
			   rfc2003 contains "deep thoughts" about NET_UNREACH,
			   I believe they are just ether pollution. --ANK
			 */
			break;
		}
		break;
	case ICMP_TIME_EXCEEDED:
		if (code != ICMP_EXC_TTL)
			return;
		break;
	}

	read_lock(&ipip_lock);
	t = ipip_tunnel_lookup(iph->daddr, iph->saddr);
	if (t == NULL || t->parms.iph.daddr == 0)
		goto out;
	if (t->parms.iph.ttl == 0 && type == ICMP_TIME_EXCEEDED)
		goto out;

	if (jiffies - t->err_time < IPTUNNEL_ERR_TIMEO)
		t->err_count++;
	else
		t->err_count = 1;
	t->err_time = jiffies;
out:
	read_unlock(&ipip_lock);
	return;
#else
	struct iphdr *iph = (struct iphdr*)dp;
	int hlen = iph->ihl<<2;
	struct iphdr *eiph;
	int type = skb->h.icmph->type;
	int code = skb->h.icmph->code;
	int rel_type = 0;
	int rel_code = 0;
	int rel_info = 0;
	struct sk_buff *skb2;
	struct rtable *rt;

	if (len < hlen + sizeof(struct iphdr))
		return;
	eiph = (struct iphdr*)(dp + hlen);

	switch (type) {
	default:
		return;
	case ICMP_PARAMETERPROB:
		if (skb->h.icmph->un.gateway < hlen)
			return;

		/* So... This guy found something strange INSIDE encapsulated
		   packet. Well, he is fool, but what can we do ?
		 */
		rel_type = ICMP_PARAMETERPROB;
		rel_info = skb->h.icmph->un.gateway - hlen;
		break;

	case ICMP_DEST_UNREACH:
		switch (code) {
		case ICMP_SR_FAILED:
		case ICMP_PORT_UNREACH:
			/* Impossible event. */
			return;
		case ICMP_FRAG_NEEDED:
			/* And it is the only really necesary thing :-) */
			rel_info = ntohs(skb->h.icmph->un.frag.mtu);
			if (rel_info < hlen+68)
				return;
			rel_info -= hlen;
			/* BSD 4.2 MORE DOES NOT EXIST IN NATURE. */
			if (rel_info > ntohs(eiph->tot_len))
				return;
			break;
		default:
			/* All others are translated to HOST_UNREACH.
			   rfc2003 contains "deep thoughts" about NET_UNREACH,
			   I believe, it is just ether pollution. --ANK
			 */
			rel_type = ICMP_DEST_UNREACH;
			rel_code = ICMP_HOST_UNREACH;
			break;
		}
		break;
	case ICMP_TIME_EXCEEDED:
		if (code != ICMP_EXC_TTL)
			return;
		break;
	}

	/* Prepare fake skb to feed it to icmp_send */
	skb2 = skb_clone(skb, GFP_ATOMIC);
	if (skb2 == NULL)
		return;
	dst_release(skb2->dst);
	skb2->dst = NULL;
	skb_pull(skb2, skb->data - (u8*)eiph);
	skb2->nh.raw = skb2->data;

	/* Try to guess incoming interface */
	if (ip_route_output(&rt, eiph->saddr, 0, RT_TOS(eiph->tos), 0)) {
		kfree_skb(skb2);
		return;
	}
	skb2->dev = rt->u.dst.dev;

	/* route "incoming" packet */
	if (rt->rt_flags&RTCF_LOCAL) {
		ip_rt_put(rt);
		rt = NULL;
		if (ip_route_output(&rt, eiph->daddr, eiph->saddr, eiph->tos, 0) ||
		    rt->u.dst.dev->type != ARPHRD_IPGRE) {
			ip_rt_put(rt);
			kfree_skb(skb2);
			return;
		}
	} else {
		ip_rt_put(rt);
		if (ip_route_input(skb2, eiph->daddr, eiph->saddr, eiph->tos, skb2->dev) ||
		    skb2->dst->dev->type != ARPHRD_IPGRE) {
			kfree_skb(skb2);
			return;
		}
	}

	/* change mtu on this route */
	if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED) {
		if (rel_info > skb2->dst->pmtu) {
			kfree_skb(skb2);
			return;
		}
		skb2->dst->pmtu = rel_info;
		rel_info = htonl(rel_info);
	} else if (type == ICMP_TIME_EXCEEDED) {
		struct ip_tunnel *t = (struct ip_tunnel*)skb2->dev->priv;