include/net/route.h

   1 /*
   2  * INET         An implementation of the TCP/IP protocol suite for the LINUX
   3  *              operating system.  INET  is implemented using the  BSD Socket
   4  *              interface as the means of communication with the user level.
   5  *
   6  *              Definitions for the IP router.
   7  *
   8  * Version:     @(#)route.h     1.0.4   05/27/93
   9  *
  10  * Authors:     Ross Biro
  11  *              Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
  12  * Fixes:
  13  *              Alan Cox        :       Reformatted. Added ip_rt_local()
  14  *              Alan Cox        :       Support for TCP parameters.
  15  *              Alexey Kuznetsov:       Major changes for new routing code.
  16  *              Mike McLagan    :       Routing by source
  17  *              Robert Olsson   :       Added rt_cache statistics
  18  *
  19  *              This program is free software; you can redistribute it and/or
  20  *              modify it under the terms of the GNU General Public License
  21  *              as published by the Free Software Foundation; either version
  22  *              2 of the License, or (at your option) any later version.
  23  */
  24 #ifndef _ROUTE_H
  25 #define _ROUTE_H
  26
  27 #include <net/dst.h>
  28 #include <net/inetpeer.h>
  29 #include <net/flow.h>
  30 #include <net/inet_sock.h>
  31 #include <linux/in_route.h>
  32 #include <linux/rtnetlink.h>
  33 #include <linux/route.h>
  34 #include <linux/ip.h>
  35 #include <linux/cache.h>
  36 #include <linux/security.h>
  37
  38 #define RTO_ONLINK      0x01
  39
  40 #define RT_CONN_FLAGS(sk)   (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
  41
  42 struct fib_nh;
  43 struct inet_peer;
  44 struct fib_info;
  45 struct rtable {
  46         struct dst_entry        dst;
  47
  48         /* Lookup key. */
  49         __be32                  rt_key_dst;
  50         __be32                  rt_key_src;
  51
  52         int                     rt_genid;
  53         unsigned int            rt_flags;
  54         __u16                   rt_type;
  55         __u8                    rt_key_tos;
  56
  57         __be32                  rt_dst; /* Path destination     */
  58         __be32                  rt_src; /* Path source          */
  59         int                     rt_route_iif;
  60         int                     rt_iif;
  61         int                     rt_oif;
  62         __u32                   rt_mark;
  63
  64         /* Info on neighbour */
  65         __be32                  rt_gateway;
  66
  67         /* Miscellaneous cached information */
  68         __be32                  rt_spec_dst; /* RFC1122 specific destination */
  69         u32                     rt_peer_genid;
  70         unsigned long           _peer; /* long-living peer info */
  71         struct fib_info         *fi; /* for client ref to shared metrics */
  72 };
  73
  74 static inline struct inet_peer *rt_peer_ptr(struct rtable *rt)
  75 {
  76         return inetpeer_ptr(rt->_peer);
  77 }
  78
  79 static inline bool rt_has_peer(struct rtable *rt)
  80 {
  81         return inetpeer_ptr_is_peer(rt->_peer);
  82 }
  83
  84 static inline void __rt_set_peer(struct rtable *rt, struct inet_peer *peer)
  85 {
  86         __inetpeer_ptr_set_peer(&rt->_peer, peer);
  87 }
  88
  89 static inline bool rt_set_peer(struct rtable *rt, struct inet_peer *peer)
  90 {
  91         return inetpeer_ptr_set_peer(&rt->_peer, peer);
  92 }
  93
  94 static inline void rt_init_peer(struct rtable *rt, struct inet_peer_base *base)
  95 {
  96         inetpeer_init_ptr(&rt->_peer, base);
  97 }
  98
  99 static inline void rt_transfer_peer(struct rtable *rt, struct rtable *ort)
 100 {
 101         rt->_peer = ort->_peer;
 102         if (rt_has_peer(ort)) {
 103                 struct inet_peer *peer = rt_peer_ptr(ort);
 104                 atomic_inc(&peer->refcnt);
 105         }
 106 }
 107
 108 static inline bool rt_is_input_route(const struct rtable *rt)
 109 {
 110         return rt->rt_route_iif != 0;
 111 }
 112
 113 static inline bool rt_is_output_route(const struct rtable *rt)
 114 {
 115         return rt->rt_route_iif == 0;
 116 }
 117
 118 struct ip_rt_acct {
 119         __u32   o_bytes;
 120         __u32   o_packets;
 121         __u32   i_bytes;
 122         __u32   i_packets;
 123 };
 124
 125 struct rt_cache_stat {
 126         unsigned int in_hit;
 127         unsigned int in_slow_tot;
 128         unsigned int in_slow_mc;
 129         unsigned int in_no_route;
 130         unsigned int in_brd;
 131         unsigned int in_martian_dst;
 132         unsigned int in_martian_src;
 133         unsigned int out_hit;
 134         unsigned int out_slow_tot;
 135         unsigned int out_slow_mc;
 136         unsigned int gc_total;
 137         unsigned int gc_ignored;
 138         unsigned int gc_goal_miss;
 139         unsigned int gc_dst_overflow;
 140         unsigned int in_hlist_search;
 141         unsigned int out_hlist_search;
 142 };
 143
 144 extern struct ip_rt_acct __percpu *ip_rt_acct;
 145
 146 struct in_device;
 147 extern int              ip_rt_init(void);
 148 extern void             ip_rt_redirect(__be32 old_gw, __be32 dst, __be32 new_gw,
 149                                        __be32 src, struct net_device *dev);
 150 extern void             rt_cache_flush(struct net *net, int how);
 151 extern void             rt_cache_flush_batch(struct net *net);
 152 extern struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
 153 extern struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
 154                                            struct sock *sk);
 155 extern struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig);
 156
 157 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
 158 {
 159         return ip_route_output_flow(net, flp, NULL);
 160 }
 161
 162 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
 163                                              __be32 saddr, u8 tos, int oif)
 164 {
 165         struct flowi4 fl4 = {
 166                 .flowi4_oif = oif,
 167                 .flowi4_tos = tos,
 168                 .daddr = daddr,
 169                 .saddr = saddr,
 170         };
 171         return ip_route_output_key(net, &fl4);
 172 }
 173
 174 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
 175                                                    struct sock *sk,
 176                                                    __be32 daddr, __be32 saddr,
 177                                                    __be16 dport, __be16 sport,
 178                                                    __u8 proto, __u8 tos, int oif)
 179 {
 180         flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
 181                            RT_SCOPE_UNIVERSE, proto,
 182                            sk ? inet_sk_flowi_flags(sk) : 0,
 183                            daddr, saddr, dport, sport);
 184         if (sk)
 185                 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 186         return ip_route_output_flow(net, fl4, sk);
 187 }
 188
 189 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
 190                                                  __be32 daddr, __be32 saddr,
 191                                                  __be32 gre_key, __u8 tos, int oif)
 192 {
 193         memset(fl4, 0, sizeof(*fl4));
 194         fl4->flowi4_oif = oif;
 195         fl4->daddr = daddr;
 196         fl4->saddr = saddr;
 197         fl4->flowi4_tos = tos;
 198         fl4->flowi4_proto = IPPROTO_GRE;
 199         fl4->fl4_gre_key = gre_key;
 200         return ip_route_output_key(net, fl4);
 201 }
 202
 203 extern int ip_route_input_common(struct sk_buff *skb, __be32 dst, __be32 src,
 204                                  u8 tos, struct net_device *devin, bool noref);
 205
 206 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
 207                                  u8 tos, struct net_device *devin)
 208 {
 209         return ip_route_input_common(skb, dst, src, tos, devin, false);
 210 }
 211
 212 static inline int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
 213                                        u8 tos, struct net_device *devin)
 214 {
 215         return ip_route_input_common(skb, dst, src, tos, devin, true);
 216 }
 217
 218 extern void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu,
 219                              int oif, u32 mark, u8 protocol, int flow_flags);
 220 extern void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
 221 extern void ip_rt_send_redirect(struct sk_buff *skb);
 222
 223 extern unsigned int             inet_addr_type(struct net *net, __be32 addr);
 224 extern unsigned int             inet_dev_addr_type(struct net *net, const struct net_device *dev, __be32 addr);
 225 extern void             ip_rt_multicast_event(struct in_device *);
 226 extern int              ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
 227 extern void             ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
 228 extern int              ip_rt_dump(struct sk_buff *skb,  struct netlink_callback *cb);
 229
 230 struct in_ifaddr;
 231 extern void fib_add_ifaddr(struct in_ifaddr *);
 232 extern void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
 233
 234 static inline void ip_rt_put(struct rtable * rt)
 235 {
 236         if (rt)
 237                 dst_release(&rt->dst);
 238 }
 239
 240 #define IPTOS_RT_MASK   (IPTOS_TOS_MASK & ~3)
 241
 242 extern const __u8 ip_tos2prio[16];
 243
 244 static inline char rt_tos2priority(u8 tos)
 245 {
 246         return ip_tos2prio[IPTOS_TOS(tos)>>1];
 247 }
 248
 249 /* ip_route_connect() and ip_route_newports() work in tandem whilst
 250  * binding a socket for a new outgoing connection.
 251  *
 252  * In order to use IPSEC properly, we must, in the end, have a
 253  * route that was looked up using all available keys including source
 254  * and destination ports.
 255  *
 256  * However, if a source port needs to be allocated (the user specified
 257  * a wildcard source port) we need to obtain addressing information
 258  * in order to perform that allocation.
 259  *
 260  * So ip_route_connect() looks up a route using wildcarded source and
 261  * destination ports in the key, simply so that we can get a pair of
 262  * addresses to use for port allocation.
 263  *
 264  * Later, once the ports are allocated, ip_route_newports() will make
 265  * another route lookup if needed to make sure we catch any IPSEC
 266  * rules keyed on the port information.
 267  *
 268  * The callers allocate the flow key on their stack, and must pass in
 269  * the same flowi4 object to both the ip_route_connect() and the
 270  * ip_route_newports() calls.
 271  */
 272
 273 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
 274                                          u32 tos, int oif, u8 protocol,
 275                                          __be16 sport, __be16 dport,
 276                                          struct sock *sk, bool can_sleep)
 277 {
 278         __u8 flow_flags = 0;
 279
 280         if (inet_sk(sk)->transparent)
 281                 flow_flags |= FLOWI_FLAG_ANYSRC;
 282         if (protocol == IPPROTO_TCP)
 283                 flow_flags |= FLOWI_FLAG_PRECOW_METRICS;
 284         if (can_sleep)
 285                 flow_flags |= FLOWI_FLAG_CAN_SLEEP;
 286
 287         flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
 288                            protocol, flow_flags, dst, src, dport, sport);
 289 }
 290
 291 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
 292                                               __be32 dst, __be32 src, u32 tos,
 293                                               int oif, u8 protocol,
 294                                               __be16 sport, __be16 dport,
 295                                               struct sock *sk, bool can_sleep)
 296 {
 297         struct net *net = sock_net(sk);
 298         struct rtable *rt;
 299
 300         ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
 301                               sport, dport, sk, can_sleep);
 302
 303         if (!dst || !src) {
 304                 rt = __ip_route_output_key(net, fl4);
 305                 if (IS_ERR(rt))
 306                         return rt;
 307                 ip_rt_put(rt);
 308                 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
 309         }
 310         security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 311         return ip_route_output_flow(net, fl4, sk);
 312 }
 313
 314 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
 315                                                __be16 orig_sport, __be16 orig_dport,
 316                                                __be16 sport, __be16 dport,
 317                                                struct sock *sk)
 318 {
 319         if (sport != orig_sport || dport != orig_dport) {
 320                 fl4->fl4_dport = dport;
 321                 fl4->fl4_sport = sport;
 322                 ip_rt_put(rt);
 323                 flowi4_update_output(fl4, sk->sk_bound_dev_if,
 324                                      RT_CONN_FLAGS(sk), fl4->daddr,
 325                                      fl4->saddr);
 326                 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 327                 return ip_route_output_flow(sock_net(sk), fl4, sk);
 328         }
 329         return rt;
 330 }
 331
 332 extern void rt_bind_peer(struct rtable *rt, __be32 daddr, int create);
 333
 334 static inline struct inet_peer *__rt_get_peer(struct rtable *rt, __be32 daddr, int create)
 335 {
 336         if (rt_has_peer(rt))
 337                 return rt_peer_ptr(rt);
 338
 339         rt_bind_peer(rt, daddr, create);
 340         return (rt_has_peer(rt) ? rt_peer_ptr(rt) : NULL);
 341 }
 342
 343 static inline struct inet_peer *rt_get_peer(struct rtable *rt, __be32 daddr)
 344 {
 345         return __rt_get_peer(rt, daddr, 0);
 346 }
 347
 348 static inline struct inet_peer *rt_get_peer_create(struct rtable *rt, __be32 daddr)
 349 {
 350         return __rt_get_peer(rt, daddr, 1);
 351 }
 352
 353 static inline int inet_iif(const struct sk_buff *skb)
 354 {
 355         return skb_rtable(skb)->rt_iif;
 356 }
 357
 358 extern int sysctl_ip_default_ttl;
 359
 360 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
 361 {
 362         int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
 363
 364         if (hoplimit == 0)
 365                 hoplimit = sysctl_ip_default_ttl;
 366         return hoplimit;
 367 }
 368
 369 #endif  /* _ROUTE_H */