xref: /openbmc/linux/net/mpls/mpls_iptunnel.c (revision aac5987a)
1 /*
2  * mpls tunnels	An implementation mpls tunnels using the light weight tunnel
3  *		infrastructure
4  *
5  * Authors:	Roopa Prabhu, <roopa@cumulusnetworks.com>
6  *
7  *		This program is free software; you can redistribute it and/or
8  *		modify it under the terms of the GNU General Public License
9  *		as published by the Free Software Foundation; either version
10  *		2 of the License, or (at your option) any later version.
11  *
12  */
13 #include <linux/types.h>
14 #include <linux/skbuff.h>
15 #include <linux/net.h>
16 #include <linux/module.h>
17 #include <linux/mpls.h>
18 #include <linux/vmalloc.h>
19 #include <net/ip.h>
20 #include <net/dst.h>
21 #include <net/lwtunnel.h>
22 #include <net/netevent.h>
23 #include <net/netns/generic.h>
24 #include <net/ip6_fib.h>
25 #include <net/route.h>
26 #include <net/mpls_iptunnel.h>
27 #include <linux/mpls_iptunnel.h>
28 #include "internal.h"
29 
30 static const struct nla_policy mpls_iptunnel_policy[MPLS_IPTUNNEL_MAX + 1] = {
31 	[MPLS_IPTUNNEL_DST]	= { .type = NLA_U32 },
32 };
33 
34 static unsigned int mpls_encap_size(struct mpls_iptunnel_encap *en)
35 {
36 	/* The size of the layer 2.5 labels to be added for this route */
37 	return en->labels * sizeof(struct mpls_shim_hdr);
38 }
39 
40 static int mpls_xmit(struct sk_buff *skb)
41 {
42 	struct mpls_iptunnel_encap *tun_encap_info;
43 	struct mpls_shim_hdr *hdr;
44 	struct net_device *out_dev;
45 	unsigned int hh_len;
46 	unsigned int new_header_size;
47 	unsigned int mtu;
48 	struct dst_entry *dst = skb_dst(skb);
49 	struct rtable *rt = NULL;
50 	struct rt6_info *rt6 = NULL;
51 	struct mpls_dev *out_mdev;
52 	int err = 0;
53 	bool bos;
54 	int i;
55 	unsigned int ttl;
56 
57 	/* Find the output device */
58 	out_dev = dst->dev;
59 
60 	/* Obtain the ttl */
61 	if (dst->ops->family == AF_INET) {
62 		ttl = ip_hdr(skb)->ttl;
63 		rt = (struct rtable *)dst;
64 	} else if (dst->ops->family == AF_INET6) {
65 		ttl = ipv6_hdr(skb)->hop_limit;
66 		rt6 = (struct rt6_info *)dst;
67 	} else {
68 		goto drop;
69 	}
70 
71 	skb_orphan(skb);
72 
73 	if (!mpls_output_possible(out_dev) ||
74 	    !dst->lwtstate || skb_warn_if_lro(skb))
75 		goto drop;
76 
77 	skb_forward_csum(skb);
78 
79 	tun_encap_info = mpls_lwtunnel_encap(dst->lwtstate);
80 
81 	/* Verify the destination can hold the packet */
82 	new_header_size = mpls_encap_size(tun_encap_info);
83 	mtu = mpls_dev_mtu(out_dev);
84 	if (mpls_pkt_too_big(skb, mtu - new_header_size))
85 		goto drop;
86 
87 	hh_len = LL_RESERVED_SPACE(out_dev);
88 	if (!out_dev->header_ops)
89 		hh_len = 0;
90 
91 	/* Ensure there is enough space for the headers in the skb */
92 	if (skb_cow(skb, hh_len + new_header_size))
93 		goto drop;
94 
95 	skb_set_inner_protocol(skb, skb->protocol);
96 	skb_reset_inner_network_header(skb);
97 
98 	skb_push(skb, new_header_size);
99 
100 	skb_reset_network_header(skb);
101 
102 	skb->dev = out_dev;
103 	skb->protocol = htons(ETH_P_MPLS_UC);
104 
105 	/* Push the new labels */
106 	hdr = mpls_hdr(skb);
107 	bos = true;
108 	for (i = tun_encap_info->labels - 1; i >= 0; i--) {
109 		hdr[i] = mpls_entry_encode(tun_encap_info->label[i],
110 					   ttl, 0, bos);
111 		bos = false;
112 	}
113 
114 	mpls_stats_inc_outucastpkts(out_dev, skb);
115 
116 	if (rt)
117 		err = neigh_xmit(NEIGH_ARP_TABLE, out_dev, &rt->rt_gateway,
118 				 skb);
119 	else if (rt6)
120 		err = neigh_xmit(NEIGH_ND_TABLE, out_dev, &rt6->rt6i_gateway,
121 				 skb);
122 	if (err)
123 		net_dbg_ratelimited("%s: packet transmission failed: %d\n",
124 				    __func__, err);
125 
126 	return LWTUNNEL_XMIT_DONE;
127 
128 drop:
129 	out_mdev = out_dev ? mpls_dev_get(out_dev) : NULL;
130 	if (out_mdev)
131 		MPLS_INC_STATS(out_mdev, tx_errors);
132 	kfree_skb(skb);
133 	return -EINVAL;
134 }
135 
136 static int mpls_build_state(struct nlattr *nla,
137 			    unsigned int family, const void *cfg,
138 			    struct lwtunnel_state **ts)
139 {
140 	struct mpls_iptunnel_encap *tun_encap_info;
141 	struct nlattr *tb[MPLS_IPTUNNEL_MAX + 1];
142 	struct lwtunnel_state *newts;
143 	int ret;
144 
145 	ret = nla_parse_nested(tb, MPLS_IPTUNNEL_MAX, nla,
146 			       mpls_iptunnel_policy);
147 	if (ret < 0)
148 		return ret;
149 
150 	if (!tb[MPLS_IPTUNNEL_DST])
151 		return -EINVAL;
152 
153 
154 	newts = lwtunnel_state_alloc(sizeof(*tun_encap_info));
155 	if (!newts)
156 		return -ENOMEM;
157 
158 	tun_encap_info = mpls_lwtunnel_encap(newts);
159 	ret = nla_get_labels(tb[MPLS_IPTUNNEL_DST], MAX_NEW_LABELS,
160 			     &tun_encap_info->labels, tun_encap_info->label);
161 	if (ret)
162 		goto errout;
163 	newts->type = LWTUNNEL_ENCAP_MPLS;
164 	newts->flags |= LWTUNNEL_STATE_XMIT_REDIRECT;
165 	newts->headroom = mpls_encap_size(tun_encap_info);
166 
167 	*ts = newts;
168 
169 	return 0;
170 
171 errout:
172 	kfree(newts);
173 	*ts = NULL;
174 
175 	return ret;
176 }
177 
178 static int mpls_fill_encap_info(struct sk_buff *skb,
179 				struct lwtunnel_state *lwtstate)
180 {
181 	struct mpls_iptunnel_encap *tun_encap_info;
182 
183 	tun_encap_info = mpls_lwtunnel_encap(lwtstate);
184 
185 	if (nla_put_labels(skb, MPLS_IPTUNNEL_DST, tun_encap_info->labels,
186 			   tun_encap_info->label))
187 		goto nla_put_failure;
188 
189 	return 0;
190 
191 nla_put_failure:
192 	return -EMSGSIZE;
193 }
194 
195 static int mpls_encap_nlsize(struct lwtunnel_state *lwtstate)
196 {
197 	struct mpls_iptunnel_encap *tun_encap_info;
198 
199 	tun_encap_info = mpls_lwtunnel_encap(lwtstate);
200 
201 	return nla_total_size(tun_encap_info->labels * 4);
202 }
203 
204 static int mpls_encap_cmp(struct lwtunnel_state *a, struct lwtunnel_state *b)
205 {
206 	struct mpls_iptunnel_encap *a_hdr = mpls_lwtunnel_encap(a);
207 	struct mpls_iptunnel_encap *b_hdr = mpls_lwtunnel_encap(b);
208 	int l;
209 
210 	if (a_hdr->labels != b_hdr->labels)
211 		return 1;
212 
213 	for (l = 0; l < MAX_NEW_LABELS; l++)
214 		if (a_hdr->label[l] != b_hdr->label[l])
215 			return 1;
216 	return 0;
217 }
218 
219 static const struct lwtunnel_encap_ops mpls_iptun_ops = {
220 	.build_state = mpls_build_state,
221 	.xmit = mpls_xmit,
222 	.fill_encap = mpls_fill_encap_info,
223 	.get_encap_size = mpls_encap_nlsize,
224 	.cmp_encap = mpls_encap_cmp,
225 	.owner = THIS_MODULE,
226 };
227 
228 static int __init mpls_iptunnel_init(void)
229 {
230 	return lwtunnel_encap_add_ops(&mpls_iptun_ops, LWTUNNEL_ENCAP_MPLS);
231 }
232 module_init(mpls_iptunnel_init);
233 
234 static void __exit mpls_iptunnel_exit(void)
235 {
236 	lwtunnel_encap_del_ops(&mpls_iptun_ops, LWTUNNEL_ENCAP_MPLS);
237 }
238 module_exit(mpls_iptunnel_exit);
239 
240 MODULE_ALIAS_RTNL_LWT(MPLS);
241 MODULE_DESCRIPTION("MultiProtocol Label Switching IP Tunnels");
242 MODULE_LICENSE("GPL v2");
243