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 <net/ip_fib.h>
  32 #include <net/l3mdev.h>
  33 #include <linux/in_route.h>
  34 #include <linux/rtnetlink.h>
  35 #include <linux/rcupdate.h>
  36 #include <linux/route.h>
  37 #include <linux/ip.h>
  38 #include <linux/cache.h>
  39 #include <linux/security.h>
  40
  41 /* IPv4 datagram length is stored into 16bit field (tot_len) */
  42 #define IP_MAX_MTU      0xFFFFU
  43
  44 #define RTO_ONLINK      0x01
  45
  46 #define RT_CONN_FLAGS(sk)   (RT_TOS(inet_sk(sk)->tos) | sock_flag(sk, SOCK_LOCALROUTE))
  47 #define RT_CONN_FLAGS_TOS(sk,tos)   (RT_TOS(tos) | sock_flag(sk, SOCK_LOCALROUTE))
  48
  49 struct fib_nh;
  50 struct fib_info;
  51 struct uncached_list;
  52 struct rtable {
  53         struct dst_entry        dst;
  54
  55         int                     rt_genid;
  56         unsigned int            rt_flags;
  57         __u16                   rt_type;
  58         __u8                    rt_is_input;
  59         __u8                    rt_uses_gateway;
  60
  61         int                     rt_iif;
  62
  63         /* Info on neighbour */
  64         __be32                  rt_gateway;
  65
  66         /* Miscellaneous cached information */
  67         u32                     rt_pmtu;
  68
  69         u32                     rt_table_id;
  70
  71         struct list_head        rt_uncached;
  72         struct uncached_list    *rt_uncached_list;
  73 };
  74
  75 static inline bool rt_is_input_route(const struct rtable *rt)
  76 {
  77         return rt->rt_is_input != 0;
  78 }
  79
  80 static inline bool rt_is_output_route(const struct rtable *rt)
  81 {
  82         return rt->rt_is_input == 0;
  83 }
  84
  85 static inline __be32 rt_nexthop(const struct rtable *rt, __be32 daddr)
  86 {
  87         if (rt->rt_gateway)
  88                 return rt->rt_gateway;
  89         return daddr;
  90 }
  91
  92 struct ip_rt_acct {
  93         __u32   o_bytes;
  94         __u32   o_packets;
  95         __u32   i_bytes;
  96         __u32   i_packets;
  97 };
  98
  99 struct rt_cache_stat {
 100         unsigned int in_slow_tot;
 101         unsigned int in_slow_mc;
 102         unsigned int in_no_route;
 103         unsigned int in_brd;
 104         unsigned int in_martian_dst;
 105         unsigned int in_martian_src;
 106         unsigned int out_slow_tot;
 107         unsigned int out_slow_mc;
 108 };
 109
 110 extern struct ip_rt_acct __percpu *ip_rt_acct;
 111
 112 struct in_device;
 113
 114 int ip_rt_init(void);
 115 void rt_cache_flush(struct net *net);
 116 void rt_flush_dev(struct net_device *dev);
 117 struct rtable *__ip_route_output_key_hash(struct net *, struct flowi4 *flp,
 118                                           int mp_hash);
 119
 120 static inline struct rtable *__ip_route_output_key(struct net *net,
 121                                                    struct flowi4 *flp)
 122 {
 123         return __ip_route_output_key_hash(net, flp, -1);
 124 }
 125
 126 struct rtable *ip_route_output_flow(struct net *, struct flowi4 *flp,
 127                                     const struct sock *sk);
 128 struct dst_entry *ipv4_blackhole_route(struct net *net,
 129                                        struct dst_entry *dst_orig);
 130
 131 static inline struct rtable *ip_route_output_key(struct net *net, struct flowi4 *flp)
 132 {
 133         return ip_route_output_flow(net, flp, NULL);
 134 }
 135
 136 static inline struct rtable *ip_route_output(struct net *net, __be32 daddr,
 137                                              __be32 saddr, u8 tos, int oif)
 138 {
 139         struct flowi4 fl4 = {
 140                 .flowi4_oif = oif,
 141                 .flowi4_tos = tos,
 142                 .daddr = daddr,
 143                 .saddr = saddr,
 144         };
 145         return ip_route_output_key(net, &fl4);
 146 }
 147
 148 static inline struct rtable *ip_route_output_ports(struct net *net, struct flowi4 *fl4,
 149                                                    struct sock *sk,
 150                                                    __be32 daddr, __be32 saddr,
 151                                                    __be16 dport, __be16 sport,
 152                                                    __u8 proto, __u8 tos, int oif)
 153 {
 154         flowi4_init_output(fl4, oif, sk ? sk->sk_mark : 0, tos,
 155                            RT_SCOPE_UNIVERSE, proto,
 156                            sk ? inet_sk_flowi_flags(sk) : 0,
 157                            daddr, saddr, dport, sport);
 158         if (sk)
 159                 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 160         return ip_route_output_flow(net, fl4, sk);
 161 }
 162
 163 static inline struct rtable *ip_route_output_gre(struct net *net, struct flowi4 *fl4,
 164                                                  __be32 daddr, __be32 saddr,
 165                                                  __be32 gre_key, __u8 tos, int oif)
 166 {
 167         memset(fl4, 0, sizeof(*fl4));
 168         fl4->flowi4_oif = oif;
 169         fl4->daddr = daddr;
 170         fl4->saddr = saddr;
 171         fl4->flowi4_tos = tos;
 172         fl4->flowi4_proto = IPPROTO_GRE;
 173         fl4->fl4_gre_key = gre_key;
 174         return ip_route_output_key(net, fl4);
 175 }
 176
 177 int ip_route_input_noref(struct sk_buff *skb, __be32 dst, __be32 src,
 178                          u8 tos, struct net_device *devin);
 179
 180 static inline int ip_route_input(struct sk_buff *skb, __be32 dst, __be32 src,
 181                                  u8 tos, struct net_device *devin)
 182 {
 183         int err;
 184
 185         rcu_read_lock();
 186         err = ip_route_input_noref(skb, dst, src, tos, devin);
 187         if (!err)
 188                 skb_dst_force(skb);
 189         rcu_read_unlock();
 190
 191         return err;
 192 }
 193
 194 void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, int oif,
 195                       u32 mark, u8 protocol, int flow_flags);
 196 void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu);
 197 void ipv4_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
 198                    u8 protocol, int flow_flags);
 199 void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk);
 200 void ip_rt_send_redirect(struct sk_buff *skb);
 201
 202 unsigned int inet_addr_type(struct net *net, __be32 addr);
 203 unsigned int inet_addr_type_table(struct net *net, __be32 addr, u32 tb_id);
 204 unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
 205                                 __be32 addr);
 206 unsigned int inet_addr_type_dev_table(struct net *net,
 207                                       const struct net_device *dev,
 208                                       __be32 addr);
 209 void ip_rt_multicast_event(struct in_device *);
 210 int ip_rt_ioctl(struct net *, unsigned int cmd, void __user *arg);
 211 void ip_rt_get_source(u8 *src, struct sk_buff *skb, struct rtable *rt);
 212 struct rtable *rt_dst_alloc(struct net_device *dev,
 213                              unsigned int flags, u16 type,
 214                              bool nopolicy, bool noxfrm, bool will_cache);
 215
 216 struct in_ifaddr;
 217 void fib_add_ifaddr(struct in_ifaddr *);
 218 void fib_del_ifaddr(struct in_ifaddr *, struct in_ifaddr *);
 219
 220 static inline void ip_rt_put(struct rtable *rt)
 221 {
 222         /* dst_release() accepts a NULL parameter.
 223          * We rely on dst being first structure in struct rtable
 224          */
 225         BUILD_BUG_ON(offsetof(struct rtable, dst) != 0);
 226         dst_release(&rt->dst);
 227 }
 228
 229 #define IPTOS_RT_MASK   (IPTOS_TOS_MASK & ~3)
 230
 231 extern const __u8 ip_tos2prio[16];
 232
 233 static inline char rt_tos2priority(u8 tos)
 234 {
 235         return ip_tos2prio[IPTOS_TOS(tos)>>1];
 236 }
 237
 238 /* ip_route_connect() and ip_route_newports() work in tandem whilst
 239  * binding a socket for a new outgoing connection.
 240  *
 241  * In order to use IPSEC properly, we must, in the end, have a
 242  * route that was looked up using all available keys including source
 243  * and destination ports.
 244  *
 245  * However, if a source port needs to be allocated (the user specified
 246  * a wildcard source port) we need to obtain addressing information
 247  * in order to perform that allocation.
 248  *
 249  * So ip_route_connect() looks up a route using wildcarded source and
 250  * destination ports in the key, simply so that we can get a pair of
 251  * addresses to use for port allocation.
 252  *
 253  * Later, once the ports are allocated, ip_route_newports() will make
 254  * another route lookup if needed to make sure we catch any IPSEC
 255  * rules keyed on the port information.
 256  *
 257  * The callers allocate the flow key on their stack, and must pass in
 258  * the same flowi4 object to both the ip_route_connect() and the
 259  * ip_route_newports() calls.
 260  */
 261
 262 static inline void ip_route_connect_init(struct flowi4 *fl4, __be32 dst, __be32 src,
 263                                          u32 tos, int oif, u8 protocol,
 264                                          __be16 sport, __be16 dport,
 265                                          struct sock *sk)
 266 {
 267         __u8 flow_flags = 0;
 268
 269         if (inet_sk(sk)->transparent)
 270                 flow_flags |= FLOWI_FLAG_ANYSRC;
 271
 272         flowi4_init_output(fl4, oif, sk->sk_mark, tos, RT_SCOPE_UNIVERSE,
 273                            protocol, flow_flags, dst, src, dport, sport);
 274 }
 275
 276 static inline struct rtable *ip_route_connect(struct flowi4 *fl4,
 277                                               __be32 dst, __be32 src, u32 tos,
 278                                               int oif, u8 protocol,
 279                                               __be16 sport, __be16 dport,
 280                                               struct sock *sk)
 281 {
 282         struct net *net = sock_net(sk);
 283         struct rtable *rt;
 284
 285         ip_route_connect_init(fl4, dst, src, tos, oif, protocol,
 286                               sport, dport, sk);
 287
 288         if (!src && oif) {
 289                 int rc;
 290
 291                 rc = l3mdev_get_saddr(net, oif, fl4);
 292                 if (rc < 0)
 293                         return ERR_PTR(rc);
 294
 295                 src = fl4->saddr;
 296         }
 297         if (!dst || !src) {
 298                 rt = __ip_route_output_key(net, fl4);
 299                 if (IS_ERR(rt))
 300                         return rt;
 301                 ip_rt_put(rt);
 302                 flowi4_update_output(fl4, oif, tos, fl4->daddr, fl4->saddr);
 303         }
 304         security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 305         return ip_route_output_flow(net, fl4, sk);
 306 }
 307
 308 static inline struct rtable *ip_route_newports(struct flowi4 *fl4, struct rtable *rt,
 309                                                __be16 orig_sport, __be16 orig_dport,
 310                                                __be16 sport, __be16 dport,
 311                                                struct sock *sk)
 312 {
 313         if (sport != orig_sport || dport != orig_dport) {
 314                 fl4->fl4_dport = dport;
 315                 fl4->fl4_sport = sport;
 316                 ip_rt_put(rt);
 317                 flowi4_update_output(fl4, sk->sk_bound_dev_if,
 318                                      RT_CONN_FLAGS(sk), fl4->daddr,
 319                                      fl4->saddr);
 320                 security_sk_classify_flow(sk, flowi4_to_flowi(fl4));
 321                 return ip_route_output_flow(sock_net(sk), fl4, sk);
 322         }
 323         return rt;
 324 }
 325
 326 static inline int inet_iif(const struct sk_buff *skb)
 327 {
 328         struct rtable *rt = skb_rtable(skb);
 329
 330         if (rt && rt->rt_iif)
 331                 return rt->rt_iif;
 332
 333         return skb->skb_iif;
 334 }
 335
 336 static inline int ip4_dst_hoplimit(const struct dst_entry *dst)
 337 {
 338         int hoplimit = dst_metric_raw(dst, RTAX_HOPLIMIT);
 339         struct net *net = dev_net(dst->dev);
 340
 341         if (hoplimit == 0)
 342                 hoplimit = net->ipv4.sysctl_ip_default_ttl;
 343         return hoplimit;
 344 }
 345
 346 #endif  /* _ROUTE_H */