xfrm4_input.c revision dbe5b4aaafc715b12dbbea309d3d17958d01fd65 
1/*
2 * xfrm4_input.c
3 *
4 * Changes:
5 *	YOSHIFUJI Hideaki @USAGI
6 *		Split up af-specific portion
7 *	Derek Atkins <derek@ihtfp.com>
8 *		Add Encapsulation support
9 *
10 */
11
12#include <linux/module.h>
13#include <linux/string.h>
14#include <linux/netfilter.h>
15#include <linux/netfilter_ipv4.h>
16#include <net/inet_ecn.h>
17#include <net/ip.h>
18#include <net/xfrm.h>
19
20int xfrm4_rcv(struct sk_buff *skb)
21{
22	return xfrm4_rcv_encap(skb, 0);
23}
24
25EXPORT_SYMBOL(xfrm4_rcv);
26
27static inline void ipip_ecn_decapsulate(struct sk_buff *skb)
28{
29	struct iphdr *outer_iph = skb->nh.iph;
30	struct iphdr *inner_iph = skb->h.ipiph;
31
32	if (INET_ECN_is_ce(outer_iph->tos))
33		IP_ECN_set_ce(inner_iph);
34}
35
36static int xfrm4_parse_spi(struct sk_buff *skb, u8 nexthdr, u32 *spi, u32 *seq)
37{
38	switch (nexthdr) {
39	case IPPROTO_IPIP:
40		if (!pskb_may_pull(skb, sizeof(struct iphdr)))
41			return -EINVAL;
42		*spi = skb->nh.iph->saddr;
43		*seq = 0;
44		return 0;
45	}
46
47	return xfrm_parse_spi(skb, nexthdr, spi, seq);
48}
49
50#ifdef CONFIG_NETFILTER
51static inline int xfrm4_rcv_encap_finish(struct sk_buff *skb)
52{
53	struct iphdr *iph = skb->nh.iph;
54
55	if (skb->dst == NULL) {
56		if (ip_route_input(skb, iph->daddr, iph->saddr, iph->tos,
57		                   skb->dev))
58			goto drop;
59	}
60	return dst_input(skb);
61drop:
62	kfree_skb(skb);
63	return NET_RX_DROP;
64}
65#endif
66
67int xfrm4_rcv_encap(struct sk_buff *skb, __u16 encap_type)
68{
69	int err;
70	u32 spi, seq;
71	struct xfrm_state *xfrm_vec[XFRM_MAX_DEPTH];
72	struct xfrm_state *x;
73	int xfrm_nr = 0;
74	int decaps = 0;
75
76	if ((err = xfrm4_parse_spi(skb, skb->nh.iph->protocol, &spi, &seq)) != 0)
77		goto drop;
78
79	do {
80		struct iphdr *iph = skb->nh.iph;
81
82		if (xfrm_nr == XFRM_MAX_DEPTH)
83			goto drop;
84
85		x = xfrm_state_lookup((xfrm_address_t *)&iph->daddr, spi, iph->protocol, AF_INET);
86		if (x == NULL)
87			goto drop;
88
89		spin_lock(&x->lock);
90		if (unlikely(x->km.state != XFRM_STATE_VALID))
91			goto drop_unlock;
92
93		if (x->encap->encap_type != encap_type)
94			goto drop_unlock;
95
96		if (x->props.replay_window && xfrm_replay_check(x, seq))
97			goto drop_unlock;
98
99		if (xfrm_state_check_expire(x))
100			goto drop_unlock;
101
102		if (x->type->input(x, skb))
103			goto drop_unlock;
104
105		/* only the first xfrm gets the encap type */
106		encap_type = 0;
107
108		if (x->props.replay_window)
109			xfrm_replay_advance(x, seq);
110
111		x->curlft.bytes += skb->len;
112		x->curlft.packets++;
113
114		spin_unlock(&x->lock);
115
116		xfrm_vec[xfrm_nr++] = x;
117
118		iph = skb->nh.iph;
119
120		if (x->props.mode) {
121			if (iph->protocol != IPPROTO_IPIP)
122				goto drop;
123			if (!pskb_may_pull(skb, sizeof(struct iphdr)))
124				goto drop;
125			if (skb_cloned(skb) &&
126			    pskb_expand_head(skb, 0, 0, GFP_ATOMIC))
127				goto drop;
128			if (x->props.flags & XFRM_STATE_DECAP_DSCP)
129				ipv4_copy_dscp(iph, skb->h.ipiph);
130			if (!(x->props.flags & XFRM_STATE_NOECN))
131				ipip_ecn_decapsulate(skb);
132			skb->mac.raw = memmove(skb->data - skb->mac_len,
133					       skb->mac.raw, skb->mac_len);
134			skb->nh.raw = skb->data;
135			memset(&(IPCB(skb)->opt), 0, sizeof(struct ip_options));
136			decaps = 1;
137			break;
138		}
139
140		if ((err = xfrm_parse_spi(skb, skb->nh.iph->protocol, &spi, &seq)) < 0)
141			goto drop;
142	} while (!err);
143
144	/* Allocate new secpath or COW existing one. */
145
146	if (!skb->sp || atomic_read(&skb->sp->refcnt) != 1) {
147		struct sec_path *sp;
148		sp = secpath_dup(skb->sp);
149		if (!sp)
150			goto drop;
151		if (skb->sp)
152			secpath_put(skb->sp);
153		skb->sp = sp;
154	}
155	if (xfrm_nr + skb->sp->len > XFRM_MAX_DEPTH)
156		goto drop;
157
158	memcpy(skb->sp->xvec + skb->sp->len, xfrm_vec,
159	       xfrm_nr * sizeof(xfrm_vec[0]));
160	skb->sp->len += xfrm_nr;
161
162	nf_reset(skb);
163
164	if (decaps) {
165		if (!(skb->dev->flags&IFF_LOOPBACK)) {
166			dst_release(skb->dst);
167			skb->dst = NULL;
168		}
169		netif_rx(skb);
170		return 0;
171	} else {
172#ifdef CONFIG_NETFILTER
173		__skb_push(skb, skb->data - skb->nh.raw);
174		skb->nh.iph->tot_len = htons(skb->len);
175		ip_send_check(skb->nh.iph);
176
177		NF_HOOK(PF_INET, NF_IP_PRE_ROUTING, skb, skb->dev, NULL,
178		        xfrm4_rcv_encap_finish);
179		return 0;
180#else
181		return -skb->nh.iph->protocol;
182#endif
183	}
184
185drop_unlock:
186	spin_unlock(&x->lock);
187	xfrm_state_put(x);
188drop:
189	while (--xfrm_nr >= 0)
190		xfrm_state_put(xfrm_vec[xfrm_nr]);
191
192	kfree_skb(skb);
193	return 0;
194}
195