History

route: Per route IP tunnel metadata via lightweight tunnel

This introduces a new IP tunnel lightweight tunnel type which allows to specify IP tunnel instructions per route. Only IPv4 is supported at this point.

Thomas Graf committed 11 years ago

Commit 3093fbe1 parent 1b7179d8 files · +157 −5Browse files

Unified · Split

8 files
  1. drivers/net/vxlan.c +8 −2
  2. include/net/dst_metadata.h +11 −1
  3. include/net/ip_tunnels.h +6 −1
  4. include/uapi/linux/lwtunnel.h +1 −0
  5. include/uapi/linux/rtnetlink.h +15 −0
  6. net/ipv4/ip_tunnel_core.c +114 −0
  7. net/ipv4/route.c +1 −1
  8. net/openvswitch/vport.h +1 −0
drivers/net/vxlan.c+8−2
@@ -1935,7 +1935,7 @@ static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan, · 1934 unmodified lines
1935static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,1935static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev,
1936 struct vxlan_rdst *rdst, bool did_rsc)1936 struct vxlan_rdst *rdst, bool did_rsc)
1937{1937{
1938 struct ip_tunnel_info *info = skb_tunnel_info(skb);1938 struct ip_tunnel_info *info;
1939 struct vxlan_dev *vxlan = netdev_priv(dev);1939 struct vxlan_dev *vxlan = netdev_priv(dev);
1940 struct sock *sk = vxlan->vn_sock->sock->sk;1940 struct sock *sk = vxlan->vn_sock->sock->sk;
1941 struct rtable *rt = NULL;1941 struct rtable *rt = NULL;
@@ -1952,6 +1952,9 @@ static void vxlan_xmit_one(struct sk_buff *skb, struct net_device *dev, · 10 unmodified lines
1952 int err;1952 int err;
1953 u32 flags = vxlan->flags;1953 u32 flags = vxlan->flags;
1954 1954
1955 /* FIXME: Support IPv6 */
1956 info = skb_tunnel_info(skb, AF_INET);
1957
1955 if (rdst) {1958 if (rdst) {
1956 dst_port = rdst->remote_port ? rdst->remote_port : vxlan->dst_port;1959 dst_port = rdst->remote_port ? rdst->remote_port : vxlan->dst_port;
1957 vni = rdst->remote_vni;1960 vni = rdst->remote_vni;
@@ -2141,12 +2144,15 @@ tx_free: · 183 unmodified lines
2141static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev)2144static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev)
2142{2145{
2143 struct vxlan_dev *vxlan = netdev_priv(dev);2146 struct vxlan_dev *vxlan = netdev_priv(dev);
2144 const struct ip_tunnel_info *info = skb_tunnel_info(skb);2147 const struct ip_tunnel_info *info;
2145 struct ethhdr *eth;2148 struct ethhdr *eth;
2146 bool did_rsc = false;2149 bool did_rsc = false;
2147 struct vxlan_rdst *rdst, *fdst = NULL;2150 struct vxlan_rdst *rdst, *fdst = NULL;
2148 struct vxlan_fdb *f;2151 struct vxlan_fdb *f;
2149 2152
2153 /* FIXME: Support IPv6 */
2154 info = skb_tunnel_info(skb, AF_INET);
2155
2150 skb_reset_mac_header(skb);2156 skb_reset_mac_header(skb);
2151 eth = eth_hdr(skb);2157 eth = eth_hdr(skb);
2152 2158
include/net/dst_metadata.h+11−1
@@ -23,13 +23,23 @@ static inline struct metadata_dst *skb_metadata_dst(struct sk_buff *skb) · 22 unmodified lines
23 return NULL;23 return NULL;
24}24}
25 25
26static inline struct ip_tunnel_info *skb_tunnel_info(struct sk_buff *skb)26static inline struct ip_tunnel_info *skb_tunnel_info(struct sk_buff *skb,
27 int family)
27{28{
28 struct metadata_dst *md_dst = skb_metadata_dst(skb);29 struct metadata_dst *md_dst = skb_metadata_dst(skb);
30 struct rtable *rt;
29 31
30 if (md_dst)32 if (md_dst)
31 return &md_dst->u.tun_info;33 return &md_dst->u.tun_info;
32 34
35 switch (family) {
36 case AF_INET:
37 rt = (struct rtable *)skb_dst(skb);
38 if (rt && rt->rt_lwtstate)
39 return lwt_tun_info(rt->rt_lwtstate);
40 break;
41 }
42
33 return NULL;43 return NULL;
34}44}
35 45
include/net/ip_tunnels.h+6−1
@@ -9,9 +9,9 @@ · 8 unmodified lines
9#include <net/dsfield.h>9#include <net/dsfield.h>
10#include <net/gro_cells.h>10#include <net/gro_cells.h>
11#include <net/inet_ecn.h>11#include <net/inet_ecn.h>
12#include <net/ip.h>
13#include <net/netns/generic.h>12#include <net/netns/generic.h>
14#include <net/rtnetlink.h>13#include <net/rtnetlink.h>
14#include <net/lwtunnel.h>
15 15
16#if IS_ENABLED(CONFIG_IPV6)16#if IS_ENABLED(CONFIG_IPV6)
17#include <net/ipv6.h>17#include <net/ipv6.h>
@@ -298,6 +298,11 @@ static inline void *ip_tunnel_info_opts(struct ip_tunnel_info *info, size_t n) · 280 unmodified lines
298 return info + 1;298 return info + 1;
299}299}
300 300
301static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
302{
303 return (struct ip_tunnel_info *)lwtstate->data;
304}
305
301#endif /* CONFIG_INET */306#endif /* CONFIG_INET */
302 307
303#endif /* __NET_IP_TUNNELS_H */308#endif /* __NET_IP_TUNNELS_H */
include/uapi/linux/lwtunnel.h+1−0
@@ -6,6 +6,7 @@ · 5 unmodified lines
6enum lwtunnel_encap_types {6enum lwtunnel_encap_types {
7 LWTUNNEL_ENCAP_NONE,7 LWTUNNEL_ENCAP_NONE,
8 LWTUNNEL_ENCAP_MPLS,8 LWTUNNEL_ENCAP_MPLS,
9 LWTUNNEL_ENCAP_IP,
9 __LWTUNNEL_ENCAP_MAX,10 __LWTUNNEL_ENCAP_MAX,
10};11};
11 12
include/uapi/linux/rtnetlink.h+15−0
@@ -286,6 +286,21 @@ enum rt_class_t { · 285 unmodified lines
286 286
287/* Routing message attributes */287/* Routing message attributes */
288 288
289enum ip_tunnel_t {
290 IP_TUN_UNSPEC,
291 IP_TUN_ID,
292 IP_TUN_DST,
293 IP_TUN_SRC,
294 IP_TUN_TTL,
295 IP_TUN_TOS,
296 IP_TUN_SPORT,
297 IP_TUN_DPORT,
298 IP_TUN_FLAGS,
299 __IP_TUN_MAX,
300};
301
302#define IP_TUN_MAX (__IP_TUN_MAX - 1)
303
289enum rtattr_type_t {304enum rtattr_type_t {
290 RTA_UNSPEC,305 RTA_UNSPEC,
291 RTA_DST,306 RTA_DST,
net/ipv4/ip_tunnel_core.c+114−0
@@ -190,3 +190,117 @@ struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev, · 189 unmodified lines
190 return tot;190 return tot;
191}191}
192EXPORT_SYMBOL_GPL(ip_tunnel_get_stats64);192EXPORT_SYMBOL_GPL(ip_tunnel_get_stats64);
193
194static const struct nla_policy ip_tun_policy[IP_TUN_MAX + 1] = {
195 [IP_TUN_ID] = { .type = NLA_U64 },
196 [IP_TUN_DST] = { .type = NLA_U32 },
197 [IP_TUN_SRC] = { .type = NLA_U32 },
198 [IP_TUN_TTL] = { .type = NLA_U8 },
199 [IP_TUN_TOS] = { .type = NLA_U8 },
200 [IP_TUN_SPORT] = { .type = NLA_U16 },
201 [IP_TUN_DPORT] = { .type = NLA_U16 },
202 [IP_TUN_FLAGS] = { .type = NLA_U16 },
203};
204
205static int ip_tun_build_state(struct net_device *dev, struct nlattr *attr,
206 struct lwtunnel_state **ts)
207{
208 struct ip_tunnel_info *tun_info;
209 struct lwtunnel_state *new_state;
210 struct nlattr *tb[IP_TUN_MAX + 1];
211 int err;
212
213 err = nla_parse_nested(tb, IP_TUN_MAX, attr, ip_tun_policy);
214 if (err < 0)
215 return err;
216
217 new_state = lwtunnel_state_alloc(sizeof(*tun_info));
218 if (!new_state)
219 return -ENOMEM;
220
221 new_state->type = LWTUNNEL_ENCAP_IP;
222
223 tun_info = lwt_tun_info(new_state);
224
225 if (tb[IP_TUN_ID])
226 tun_info->key.tun_id = nla_get_u64(tb[IP_TUN_ID]);
227
228 if (tb[IP_TUN_DST])
229 tun_info->key.ipv4_dst = nla_get_be32(tb[IP_TUN_DST]);
230
231 if (tb[IP_TUN_SRC])
232 tun_info->key.ipv4_src = nla_get_be32(tb[IP_TUN_SRC]);
233
234 if (tb[IP_TUN_TTL])
235 tun_info->key.ipv4_ttl = nla_get_u8(tb[IP_TUN_TTL]);
236
237 if (tb[IP_TUN_TOS])
238 tun_info->key.ipv4_tos = nla_get_u8(tb[IP_TUN_TOS]);
239
240 if (tb[IP_TUN_SPORT])
241 tun_info->key.tp_src = nla_get_be16(tb[IP_TUN_SPORT]);
242
243 if (tb[IP_TUN_DPORT])
244 tun_info->key.tp_dst = nla_get_be16(tb[IP_TUN_DPORT]);
245
246 if (tb[IP_TUN_FLAGS])
247 tun_info->key.tun_flags = nla_get_u16(tb[IP_TUN_FLAGS]);
248
249 tun_info->mode = IP_TUNNEL_INFO_TX;
250 tun_info->options = NULL;
251 tun_info->options_len = 0;
252
253 *ts = new_state;
254
255 return 0;
256}
257
258static int ip_tun_fill_encap_info(struct sk_buff *skb,
259 struct lwtunnel_state *lwtstate)
260{
261 struct ip_tunnel_info *tun_info = lwt_tun_info(lwtstate);
262
263 if (nla_put_u64(skb, IP_TUN_ID, tun_info->key.tun_id) ||
264 nla_put_be32(skb, IP_TUN_DST, tun_info->key.ipv4_dst) ||
265 nla_put_be32(skb, IP_TUN_SRC, tun_info->key.ipv4_src) ||
266 nla_put_u8(skb, IP_TUN_TOS, tun_info->key.ipv4_tos) ||
267 nla_put_u8(skb, IP_TUN_TTL, tun_info->key.ipv4_ttl) ||
268 nla_put_u16(skb, IP_TUN_SPORT, tun_info->key.tp_src) ||
269 nla_put_u16(skb, IP_TUN_DPORT, tun_info->key.tp_dst) ||
270 nla_put_u16(skb, IP_TUN_FLAGS, tun_info->key.tun_flags))
271 return -ENOMEM;
272
273 return 0;
274}
275
276static int ip_tun_encap_nlsize(struct lwtunnel_state *lwtstate)
277{
278 return nla_total_size(8) /* IP_TUN_ID */
279 + nla_total_size(4) /* IP_TUN_DST */
280 + nla_total_size(4) /* IP_TUN_SRC */
281 + nla_total_size(1) /* IP_TUN_TOS */
282 + nla_total_size(1) /* IP_TUN_TTL */
283 + nla_total_size(2) /* IP_TUN_SPORT */
284 + nla_total_size(2) /* IP_TUN_DPORT */
285 + nla_total_size(2); /* IP_TUN_FLAGS */
286}
287
288static const struct lwtunnel_encap_ops ip_tun_lwt_ops = {
289 .build_state = ip_tun_build_state,
290 .fill_encap = ip_tun_fill_encap_info,
291 .get_encap_size = ip_tun_encap_nlsize,
292};
293
294static int __init ip_tunnel_core_init(void)
295{
296 lwtunnel_encap_add_ops(&ip_tun_lwt_ops, LWTUNNEL_ENCAP_IP);
297
298 return 0;
299}
300module_init(ip_tunnel_core_init);
301
302static void __exit ip_tunnel_core_exit(void)
303{
304 lwtunnel_encap_del_ops(&ip_tun_lwt_ops, LWTUNNEL_ENCAP_IP);
305}
306module_exit(ip_tunnel_core_exit);
net/ipv4/route.c+1−1
@@ -1693,7 +1693,7 @@ static int ip_route_input_slow(struct sk_buff *skb, __be32 daddr, __be32 saddr, · 1692 unmodified lines
1693 by fib_lookup.1693 by fib_lookup.
1694 */1694 */
1695 1695
1696 tun_info = skb_tunnel_info(skb);1696 tun_info = skb_tunnel_info(skb, AF_INET);
1697 if (tun_info && tun_info->mode == IP_TUNNEL_INFO_RX)1697 if (tun_info && tun_info->mode == IP_TUNNEL_INFO_RX)
1698 fl4.flowi4_tun_key.tun_id = tun_info->key.tun_id;1698 fl4.flowi4_tun_key.tun_id = tun_info->key.tun_id;
1699 else1699 else
net/openvswitch/vport.h+1−0
@@ -27,6 +27,7 @@ · 26 unmodified lines
27#include <linux/skbuff.h>27#include <linux/skbuff.h>
28#include <linux/spinlock.h>28#include <linux/spinlock.h>
29#include <linux/u64_stats_sync.h>29#include <linux/u64_stats_sync.h>
30#include <net/route.h>
30 31
31#include "datapath.h"32#include "datapath.h"
32 33

Keyboard shortcuts

General

Show keyboard shortcuts
?
Focus the current search
/

Issues and pull requests

Open filters
f

Issues, pull requests and Inbox

Select next item
j↓
Select previous item
k↑
Open selected item
Enter
Dismiss filters or clear selection
Esc

Repository code

Go to file
t