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/rcupdate.h>
  34 #include <linux/route.h>
  35 #include <linux/ip.h>
  36 #include <linux/cache.h>
  37 #include <linux/security.h>
  38
  39 /* IPv4 datagram length is stored into 16bit field (tot_len) */
  40 #define IP_MAX_MTU      0xFFFFU
  41
  42 #define RTO_ONLINK      0x01
  43
  44 #define RT_CONN_FLAGS(sk)   (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
  45 #define RT_CONN_FLAGS_TOS(sk,tos)   (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
  46
  47 struct fib_nh;
  48 struct fib_info;
  49 struct uncached_list;
  50 struct rtable {
  51         struct dst_entry        dst;
  52
  53         int                     rt_genid;
  54         unsigned int            rt_flags;
  55         __u16                   rt_type;
  56         __u8                    rt_is_input;
  57         __u8                    rt_uses_gateway;
  58
  59         int                     rt_iif;
  60
  61         /* Info on neighbour */
  62         __be32                  rt_gateway;
  63
  64         /* Miscellaneous cached information */
  65         u32                     rt_pmtu;
  66
  67         struct list_head        rt_uncached;
  68         struct uncached_list    *rt_uncached_list;
  69 };
  70
  71 static inline bool rt_is_input_route(const struct rtable *rt)
  72 {
  73         return rt->rt_is_input != 0;
  74 }
  75
  76 static inline bool rt_is_output_route(const struct rtable *rt)
  77 {
  78         return rt->rt_is_input == 0;
  79 }
  80
  81 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
  82 {
  83         if (rt->rt_gateway)
  84                 return rt->rt_gateway;
  85         return daddr;
  86 }
  87
  88 struct ip_rt_acct {
  89         __u32   o_bytes;
  90         __u32   o_packets;
  91         __u32   i_bytes;
  92         __u32   i_packets;
  93 };
  94
  95 struct rt_cache_stat {
  96         unsigned int in_slow_tot;
  97         unsigned int in_slow_mc;
  98         unsigned int in_no_route;
  99         unsigned int in_brd;
 100         unsigned int in_martian_dst;
 101         unsigned int in_martian_src;
 102         unsigned int out_slow_tot;
 103         unsigned int out_slow_mc;
 104 };
 105
 106 extern struct ip_rt_acct __percpu *ip_rt_acct;
 107
 108 struct in_device;
 109
 110 int ip_rt_init(void);
 111 void rt_cache_flush(struct net *net);
 112 void rt_flush_dev(struct net_device *dev);
 113 struct rtable *__ip_route_output_key(struct net *, struct flowi4 *flp);
 114 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
 115                                     struct sock *sk);
 116 struct dst_entry *ipv4_blackhole_route(struct net *net,
 117                                        struct dst_entry *dst_orig);
 118
 119 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
 120 {
 121         return ip_route_output_flow(net, flp, NULL);
 122 }
 123
 124 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
 125                                              __be32 saddr, u8 tos, int oif)
 126 {
 127         struct flowi4 fl4 = {
 128                 .flowi4_oif = oif,
 129                 .flowi4_tos = tos,
 130                 .daddr = daddr,
 131                 .saddr = saddr,
 132         };
 133         return ip_route_output_key(net, &fl4);
 134 }
 135
 136 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
 137                                                    struct sock *sk,
 138                                                    __be32 daddr, __be32 saddr,
 139                                                    __be16 dport, __be16 sport,
 140                                                    __u8 proto, __u8 tos, int oif)
 141 {
 142         flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
 143                            RT_SCOPE_UNIVERSE, proto,
 144                            sk ? inet_sk_flowi_flags(sk) : 0,
 145                            daddr, saddr, dport, sport);
 146         if (sk)
 147                 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 148         return ip_route_output_flow(net, fl4, sk);
 149 }
 150
 151 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
 152                                                  __be32 daddr, __be32 saddr,
 153                                                  __be32 gre_key, __u8 tos, int oif)
 154 {
 155         memset(fl4, 0, sizeof(*fl4));
 156         fl4->flowi4_oif = oif;
 157         fl4->daddr = daddr;
 158         fl4->saddr = saddr;
 159         fl4->flowi4_tos = tos;
 160         fl4->flowi4_proto = IPPROTO_GRE;
 161         fl4->fl4_gre_key = gre_key;
 162         return ip_route_output_key(net, fl4);
 163 }
 164
 165 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
 166                          u8 tos, struct net_device *devin);
 167
 168 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
 169                                  u8 tos, struct net_device *devin)
 170 {
 171         int err;
 172
 173         rcu_read_lock();
 174         err = ip_route_input_noref(skb, dst, src, tos, devin);
 175         if (!err)
 176                 skb_dst_force(skb);
 177         rcu_read_unlock();
 178
 179         return err;
 180 }
 181
 182 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
 183                       u32 mark, u8 protocol, int flow_flags);
 184 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
 185 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
 186                    u8 protocol, int flow_flags);
 187 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
 188 void ip_rt_send_redirect(struct sk_buff *skb);
 189
 190 unsigned int inet_addr_type(struct net *net, __be32 addr);
 191 unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
 192 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
 193                                 __be32 addr);
 194 unsigned int inet_addr_type_dev_table(struct net *net,
 195                                       const struct net_device *dev,
 196                                       __be32 addr);
 197 void ip_rt_multicast_event(struct in_device *);
 198 int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
 199 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
 200
 201 struct in_ifaddr;
 202 void fib_add_ifaddr(struct in_ifaddr *);
 203 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
 204
 205 static inline void ip_rt_put(struct rtable *rt)
 206 {
 207         /* dst_release() accepts a NULL parameter.
 208          * We rely on dst being first structure in struct rtable
 209          */
 210         BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
 211         dst_release(&rt->dst);
 212 }
 213
 214 #define IPTOS_RT_MASK   (IPTOS_TOS_MASK & ~3)
 215
 216 extern const __u8 ip_tos2prio[16];
 217
 218 static inline char rt_tos2priority(u8 tos)
 219 {
 220         return ip_tos2prio[IPTOS_TOS(tos)>>1];
 221 }
 222
 223 /* ip_route_connect() and ip_route_newports() work in tandem whilst
 224  * binding a socket for a new outgoing connection.
 225  *
 226  * In order to use IPSEC properly, we must, in the end, have a
 227  * route that was looked up using all available keys including source
 228  * and destination ports.
 229  *
 230  * However, if a source port needs to be allocated (the user specified
 231  * a wildcard source port) we need to obtain addressing information
 232  * in order to perform that allocation.
 233  *
 234  * So ip_route_connect() looks up a route using wildcarded source and
 235  * destination ports in the key, simply so that we can get a pair of
 236  * addresses to use for port allocation.
 237  *
 238  * Later, once the ports are allocated, ip_route_newports() will make
 239  * another route lookup if needed to make sure we catch any IPSEC
 240  * rules keyed on the port information.
 241  *
 242  * The callers allocate the flow key on their stack, and must pass in
 243  * the same flowi4 object to both the ip_route_connect() and the
 244  * ip_route_newports() calls.
 245  */
 246
 247 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
 248                                          u32 tos, int oif, u8 protocol,
 249                                          __be16 sport, __be16 dport,
 250                                          struct sock *sk)
 251 {
 252         __u8 flow_flags = 0;
 253
 254         if (inet_sk(sk)->transparent)
 255                 flow_flags |= FLOWI_FLAG_ANYSRC;
 256
 257         if (netif_index_is_vrf(sock_net(sk), oif))
 258                 flow_flags |= FLOWI_FLAG_VRFSRC;
 259
 260         flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
 261                            protocol, flow_flags, dst, src, dport, sport);
 262 }
 263
 264 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
 265                                               __be32 dst, __be32 src, u32 tos,
 266                                               int oif, u8 protocol,
 267                                               __be16 sport, __be16 dport,
 268                                               struct sock *sk)
 269 {
 270         struct net *net = sock_net(sk);
 271         struct rtable *rt;
 272
 273         ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
 274                               sport, dport, sk);
 275
 276         if (!dst || !src) {
 277                 rt = __ip_route_output_key(net, fl4);
 278                 if (IS_ERR(rt))
 279                         return rt;
 280                 ip_rt_put(rt);
 281                 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
 282         }
 283         security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 284         return ip_route_output_flow(net, fl4, sk);
 285 }
 286
 287 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
 288                                                __be16 orig_sport, __be16 orig_dport,
 289                                                __be16 sport, __be16 dport,
 290                                                struct sock *sk)
 291 {
 292         if (sport != orig_sport || dport != orig_dport) {
 293                 fl4->fl4_dport = dport;
 294                 fl4->fl4_sport = sport;
 295                 ip_rt_put(rt);
 296                 flowi4_update_output(fl4, sk->sk_bound_dev_if,
 297                                      RT_CONN_FLAGS(sk), fl4->daddr,
 298                                      fl4->saddr);
 299                 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 300                 return ip_route_output_flow(sock_net(sk), fl4, sk);
 301         }
 302         return rt;
 303 }
 304
 305 static inline int inet_iif(const struct sk_buff *skb)
 306 {
 307         int iif = skb_rtable(skb)->rt_iif;
 308
 309         if (iif)
 310                 return iif;
 311         return skb->skb_iif;
 312 }
 313
 314 extern int sysctl_ip_default_ttl;
 315
 316 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
 317 {
 318         int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
 319
 320         if (hoplimit == 0)
 321                 hoplimit = sysctl_ip_default_ttl;
 322         return hoplimit;
 323 }
 324
 325 #endif  /* _ROUTE_H */