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