5 * Kazunori MIYAZAWA @USAGI
6 * YOSHIFUJI Hideaki @USAGI
7 * Split up af-specific portion
11 #include <linux/err.h>
12 #include <linux/kernel.h>
13 #include <linux/inetdevice.h>
14 #include <linux/if_tunnel.h>
18 #include <net/l3mdev.h>
20 static struct xfrm_policy_afinfo xfrm4_policy_afinfo
;
22 static struct dst_entry
*__xfrm4_dst_lookup(struct net
*net
, struct flowi4
*fl4
,
24 const xfrm_address_t
*saddr
,
25 const xfrm_address_t
*daddr
)
29 memset(fl4
, 0, sizeof(*fl4
));
30 fl4
->daddr
= daddr
->a4
;
31 fl4
->flowi4_tos
= tos
;
32 fl4
->flowi4_oif
= oif
;
34 fl4
->saddr
= saddr
->a4
;
36 fl4
->flowi4_flags
= FLOWI_FLAG_SKIP_NH_OIF
;
38 rt
= __ip_route_output_key(net
, fl4
);
45 static struct dst_entry
*xfrm4_dst_lookup(struct net
*net
, int tos
, int oif
,
46 const xfrm_address_t
*saddr
,
47 const xfrm_address_t
*daddr
)
51 return __xfrm4_dst_lookup(net
, &fl4
, tos
, oif
, saddr
, daddr
);
54 static int xfrm4_get_saddr(struct net
*net
, int oif
,
55 xfrm_address_t
*saddr
, xfrm_address_t
*daddr
)
57 struct dst_entry
*dst
;
60 dst
= __xfrm4_dst_lookup(net
, &fl4
, 0, oif
, NULL
, daddr
);
64 saddr
->a4
= fl4
.saddr
;
69 static int xfrm4_get_tos(const struct flowi
*fl
)
71 return IPTOS_RT_MASK
& fl
->u
.ip4
.flowi4_tos
; /* Strip ECN bits */
74 static int xfrm4_init_path(struct xfrm_dst
*path
, struct dst_entry
*dst
,
80 static int xfrm4_fill_dst(struct xfrm_dst
*xdst
, struct net_device
*dev
,
81 const struct flowi
*fl
)
83 struct rtable
*rt
= (struct rtable
*)xdst
->route
;
84 const struct flowi4
*fl4
= &fl
->u
.ip4
;
86 xdst
->u
.rt
.rt_iif
= fl4
->flowi4_iif
;
88 xdst
->u
.dst
.dev
= dev
;
91 /* Sheit... I remember I did this right. Apparently,
92 * it was magically lost, so this code needs audit */
93 xdst
->u
.rt
.rt_is_input
= rt
->rt_is_input
;
94 xdst
->u
.rt
.rt_flags
= rt
->rt_flags
& (RTCF_BROADCAST
| RTCF_MULTICAST
|
96 xdst
->u
.rt
.rt_type
= rt
->rt_type
;
97 xdst
->u
.rt
.rt_gateway
= rt
->rt_gateway
;
98 xdst
->u
.rt
.rt_uses_gateway
= rt
->rt_uses_gateway
;
99 xdst
->u
.rt
.rt_pmtu
= rt
->rt_pmtu
;
100 xdst
->u
.rt
.rt_table_id
= rt
->rt_table_id
;
101 INIT_LIST_HEAD(&xdst
->u
.rt
.rt_uncached
);
107 _decode_session4(struct sk_buff
*skb
, struct flowi
*fl
, int reverse
)
109 const struct iphdr
*iph
= ip_hdr(skb
);
110 u8
*xprth
= skb_network_header(skb
) + iph
->ihl
* 4;
111 struct flowi4
*fl4
= &fl
->u
.ip4
;
115 oif
= l3mdev_fib_oif(skb_dst(skb
)->dev
);
117 memset(fl4
, 0, sizeof(struct flowi4
));
118 fl4
->flowi4_mark
= skb
->mark
;
119 fl4
->flowi4_oif
= reverse
? skb
->skb_iif
: oif
;
121 if (!ip_is_fragment(iph
)) {
122 switch (iph
->protocol
) {
124 case IPPROTO_UDPLITE
:
128 if (xprth
+ 4 < skb
->data
||
129 pskb_may_pull(skb
, xprth
+ 4 - skb
->data
)) {
132 xprth
= skb_network_header(skb
) + iph
->ihl
* 4;
133 ports
= (__be16
*)xprth
;
135 fl4
->fl4_sport
= ports
[!!reverse
];
136 fl4
->fl4_dport
= ports
[!reverse
];
141 if (xprth
+ 2 < skb
->data
||
142 pskb_may_pull(skb
, xprth
+ 2 - skb
->data
)) {
145 xprth
= skb_network_header(skb
) + iph
->ihl
* 4;
148 fl4
->fl4_icmp_type
= icmp
[0];
149 fl4
->fl4_icmp_code
= icmp
[1];
154 if (xprth
+ 4 < skb
->data
||
155 pskb_may_pull(skb
, xprth
+ 4 - skb
->data
)) {
158 xprth
= skb_network_header(skb
) + iph
->ihl
* 4;
159 ehdr
= (__be32
*)xprth
;
161 fl4
->fl4_ipsec_spi
= ehdr
[0];
166 if (xprth
+ 8 < skb
->data
||
167 pskb_may_pull(skb
, xprth
+ 8 - skb
->data
)) {
170 xprth
= skb_network_header(skb
) + iph
->ihl
* 4;
171 ah_hdr
= (__be32
*)xprth
;
173 fl4
->fl4_ipsec_spi
= ah_hdr
[1];
178 if (xprth
+ 4 < skb
->data
||
179 pskb_may_pull(skb
, xprth
+ 4 - skb
->data
)) {
182 xprth
= skb_network_header(skb
) + iph
->ihl
* 4;
183 ipcomp_hdr
= (__be16
*)xprth
;
185 fl4
->fl4_ipsec_spi
= htonl(ntohs(ipcomp_hdr
[1]));
190 if (xprth
+ 12 < skb
->data
||
191 pskb_may_pull(skb
, xprth
+ 12 - skb
->data
)) {
195 xprth
= skb_network_header(skb
) + iph
->ihl
* 4;
196 greflags
= (__be16
*)xprth
;
197 gre_hdr
= (__be32
*)xprth
;
199 if (greflags
[0] & GRE_KEY
) {
200 if (greflags
[0] & GRE_CSUM
)
202 fl4
->fl4_gre_key
= gre_hdr
[1];
208 fl4
->fl4_ipsec_spi
= 0;
212 fl4
->flowi4_proto
= iph
->protocol
;
213 fl4
->daddr
= reverse
? iph
->saddr
: iph
->daddr
;
214 fl4
->saddr
= reverse
? iph
->daddr
: iph
->saddr
;
215 fl4
->flowi4_tos
= iph
->tos
;
218 static inline int xfrm4_garbage_collect(struct dst_ops
*ops
)
220 struct net
*net
= container_of(ops
, struct net
, xfrm
.xfrm4_dst_ops
);
222 xfrm4_policy_afinfo
.garbage_collect(net
);
223 return (dst_entries_get_slow(ops
) > ops
->gc_thresh
* 2);
226 static void xfrm4_update_pmtu(struct dst_entry
*dst
, struct sock
*sk
,
227 struct sk_buff
*skb
, u32 mtu
)
229 struct xfrm_dst
*xdst
= (struct xfrm_dst
*)dst
;
230 struct dst_entry
*path
= xdst
->route
;
232 path
->ops
->update_pmtu(path
, sk
, skb
, mtu
);
235 static void xfrm4_redirect(struct dst_entry
*dst
, struct sock
*sk
,
238 struct xfrm_dst
*xdst
= (struct xfrm_dst
*)dst
;
239 struct dst_entry
*path
= xdst
->route
;
241 path
->ops
->redirect(path
, sk
, skb
);
244 static void xfrm4_dst_destroy(struct dst_entry
*dst
)
246 struct xfrm_dst
*xdst
= (struct xfrm_dst
*)dst
;
248 dst_destroy_metrics_generic(dst
);
250 xfrm_dst_destroy(xdst
);
253 static void xfrm4_dst_ifdown(struct dst_entry
*dst
, struct net_device
*dev
,
259 xfrm_dst_ifdown(dst
, dev
);
262 static struct dst_ops xfrm4_dst_ops_template
= {
264 .gc
= xfrm4_garbage_collect
,
265 .update_pmtu
= xfrm4_update_pmtu
,
266 .redirect
= xfrm4_redirect
,
267 .cow_metrics
= dst_cow_metrics_generic
,
268 .destroy
= xfrm4_dst_destroy
,
269 .ifdown
= xfrm4_dst_ifdown
,
270 .local_out
= __ip_local_out
,
271 .gc_thresh
= INT_MAX
,
274 static struct xfrm_policy_afinfo xfrm4_policy_afinfo
= {
276 .dst_ops
= &xfrm4_dst_ops_template
,
277 .dst_lookup
= xfrm4_dst_lookup
,
278 .get_saddr
= xfrm4_get_saddr
,
279 .decode_session
= _decode_session4
,
280 .get_tos
= xfrm4_get_tos
,
281 .init_path
= xfrm4_init_path
,
282 .fill_dst
= xfrm4_fill_dst
,
283 .blackhole_route
= ipv4_blackhole_route
,
287 static struct ctl_table xfrm4_policy_table
[] = {
289 .procname
= "xfrm4_gc_thresh",
290 .data
= &init_net
.xfrm
.xfrm4_dst_ops
.gc_thresh
,
291 .maxlen
= sizeof(int),
293 .proc_handler
= proc_dointvec
,
298 static int __net_init
xfrm4_net_sysctl_init(struct net
*net
)
300 struct ctl_table
*table
;
301 struct ctl_table_header
*hdr
;
303 table
= xfrm4_policy_table
;
304 if (!net_eq(net
, &init_net
)) {
305 table
= kmemdup(table
, sizeof(xfrm4_policy_table
), GFP_KERNEL
);
309 table
[0].data
= &net
->xfrm
.xfrm4_dst_ops
.gc_thresh
;
312 hdr
= register_net_sysctl(net
, "net/ipv4", table
);
316 net
->ipv4
.xfrm4_hdr
= hdr
;
320 if (!net_eq(net
, &init_net
))
326 static void __net_exit
xfrm4_net_sysctl_exit(struct net
*net
)
328 struct ctl_table
*table
;
330 if (!net
->ipv4
.xfrm4_hdr
)
333 table
= net
->ipv4
.xfrm4_hdr
->ctl_table_arg
;
334 unregister_net_sysctl_table(net
->ipv4
.xfrm4_hdr
);
335 if (!net_eq(net
, &init_net
))
338 #else /* CONFIG_SYSCTL */
339 static int inline xfrm4_net_sysctl_init(struct net
*net
)
344 static void inline xfrm4_net_sysctl_exit(struct net
*net
)
349 static int __net_init
xfrm4_net_init(struct net
*net
)
353 memcpy(&net
->xfrm
.xfrm4_dst_ops
, &xfrm4_dst_ops_template
,
354 sizeof(xfrm4_dst_ops_template
));
355 ret
= dst_entries_init(&net
->xfrm
.xfrm4_dst_ops
);
359 ret
= xfrm4_net_sysctl_init(net
);
361 dst_entries_destroy(&net
->xfrm
.xfrm4_dst_ops
);
366 static void __net_exit
xfrm4_net_exit(struct net
*net
)
368 xfrm4_net_sysctl_exit(net
);
369 dst_entries_destroy(&net
->xfrm
.xfrm4_dst_ops
);
372 static struct pernet_operations __net_initdata xfrm4_net_ops
= {
373 .init
= xfrm4_net_init
,
374 .exit
= xfrm4_net_exit
,
377 static void __init
xfrm4_policy_init(void)
379 xfrm_policy_register_afinfo(&xfrm4_policy_afinfo
);
382 void __init
xfrm4_init(void)
386 xfrm4_protocol_init();
387 register_pernet_subsys(&xfrm4_net_ops
);