include/net/ip6_route.h

   1 /* SPDX-License-Identifier: GPL-2.0 */
   2 #ifndef _NET_IP6_ROUTE_H
   3 #define _NET_IP6_ROUTE_H
   4
   5 #include <net/addrconf.h>
   6 #include <net/flow.h>
   7 #include <net/ip6_fib.h>
   8 #include <net/sock.h>
   9 #include <net/lwtunnel.h>
  10 #include <linux/ip.h>
  11 #include <linux/ipv6.h>
  12 #include <linux/route.h>
  13 #include <net/nexthop.h>
  14
  15 struct route_info {
  16         __u8                    type;
  17         __u8                    length;
  18         __u8                    prefix_len;
  19 #if defined(__BIG_ENDIAN_BITFIELD)
  20         __u8                    reserved_h:3,
  21                                 route_pref:2,
  22                                 reserved_l:3;
  23 #elif defined(__LITTLE_ENDIAN_BITFIELD)
  24         __u8                    reserved_l:3,
  25                                 route_pref:2,
  26                                 reserved_h:3;
  27 #endif
  28         __be32                  lifetime;
  29         __u8                    prefix[];       /* 0,8 or 16 */
  30 };
  31
  32 #define RT6_LOOKUP_F_IFACE              0x00000001
  33 #define RT6_LOOKUP_F_REACHABLE          0x00000002
  34 #define RT6_LOOKUP_F_HAS_SADDR          0x00000004
  35 #define RT6_LOOKUP_F_SRCPREF_TMP        0x00000008
  36 #define RT6_LOOKUP_F_SRCPREF_PUBLIC     0x00000010
  37 #define RT6_LOOKUP_F_SRCPREF_COA        0x00000020
  38 #define RT6_LOOKUP_F_IGNORE_LINKSTATE   0x00000040
  39 #define RT6_LOOKUP_F_DST_NOREF          0x00000080
  40
  41 /* We do not (yet ?) support IPv6 jumbograms (RFC 2675)
  42  * Unlike IPv4, hdr->seg_len doesn't include the IPv6 header
  43  */
  44 #define IP6_MAX_MTU (0xFFFF + sizeof(struct ipv6hdr))
  45
  46 /*
  47  * rt6_srcprefs2flags() and rt6_flags2srcprefs() translate
  48  * between IPV6_ADDR_PREFERENCES socket option values
  49  *      IPV6_PREFER_SRC_TMP    = 0x1
  50  *      IPV6_PREFER_SRC_PUBLIC = 0x2
  51  *      IPV6_PREFER_SRC_COA    = 0x4
  52  * and above RT6_LOOKUP_F_SRCPREF_xxx flags.
  53  */
  54 static inline int rt6_srcprefs2flags(unsigned int srcprefs)
  55 {
  56         return (srcprefs & IPV6_PREFER_SRC_MASK) << 3;
  57 }
  58
  59 static inline unsigned int rt6_flags2srcprefs(int flags)
  60 {
  61         return (flags >> 3) & IPV6_PREFER_SRC_MASK;
  62 }
  63
  64 static inline bool rt6_need_strict(const struct in6_addr *daddr)
  65 {
  66         return ipv6_addr_type(daddr) &
  67                 (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK);
  68 }
  69
  70 /* fib entries using a nexthop object can not be coalesced into
  71  * a multipath route
  72  */
  73 static inline bool rt6_qualify_for_ecmp(const struct fib6_info *f6i)
  74 {
  75         /* the RTF_ADDRCONF flag filters out RA's */
  76         return !(f6i->fib6_flags & RTF_ADDRCONF) && !f6i->nh &&
  77                 f6i->fib6_nh->fib_nh_gw_family;
  78 }
  79
  80 void ip6_route_input(struct sk_buff *skb);
  81 struct dst_entry *ip6_route_input_lookup(struct net *net,
  82                                          struct net_device *dev,
  83                                          struct flowi6 *fl6,
  84                                          const struct sk_buff *skb, int flags);
  85
  86 struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
  87                                          struct flowi6 *fl6, int flags);
  88
  89 static inline struct dst_entry *ip6_route_output(struct net *net,
  90                                                  const struct sock *sk,
  91                                                  struct flowi6 *fl6)
  92 {
  93         return ip6_route_output_flags(net, sk, fl6, 0);
  94 }
  95
  96 /* Only conditionally release dst if flags indicates
  97  * !RT6_LOOKUP_F_DST_NOREF or dst is in uncached_list.
  98  */
  99 static inline void ip6_rt_put_flags(struct rt6_info *rt, int flags)
 100 {
 101         if (!(flags & RT6_LOOKUP_F_DST_NOREF) ||
 102             !list_empty(&rt->dst.rt_uncached))
 103                 ip6_rt_put(rt);
 104 }
 105
 106 struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
 107                                    const struct sk_buff *skb, int flags);
 108 struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
 109                                int ifindex, struct flowi6 *fl6,
 110                                const struct sk_buff *skb, int flags);
 111
 112 void ip6_route_init_special_entries(void);
 113 int ip6_route_init(void);
 114 void ip6_route_cleanup(void);
 115
 116 int ipv6_route_ioctl(struct net *net, unsigned int cmd,
 117                 struct in6_rtmsg *rtmsg);
 118
 119 int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,
 120                   struct netlink_ext_ack *extack);
 121 int ip6_ins_rt(struct net *net, struct fib6_info *f6i);
 122 int ip6_del_rt(struct net *net, struct fib6_info *f6i, bool skip_notify);
 123
 124 void rt6_flush_exceptions(struct fib6_info *f6i);
 125 void rt6_age_exceptions(struct fib6_info *f6i, struct fib6_gc_args *gc_args,
 126                         unsigned long now);
 127
 128 static inline int ip6_route_get_saddr(struct net *net, struct fib6_info *f6i,
 129                                       const struct in6_addr *daddr,
 130                                       unsigned int prefs, int l3mdev_index,
 131                                       struct in6_addr *saddr)
 132 {
 133         struct net_device *l3mdev;
 134         struct net_device *dev;
 135         bool same_vrf;
 136         int err = 0;
 137
 138         rcu_read_lock();
 139
 140         l3mdev = dev_get_by_index_rcu(net, l3mdev_index);
 141         if (!f6i || !f6i->fib6_prefsrc.plen || l3mdev)
 142                 dev = f6i ? fib6_info_nh_dev(f6i) : NULL;
 143         same_vrf = !l3mdev || l3mdev_master_dev_rcu(dev) == l3mdev;
 144         if (f6i && f6i->fib6_prefsrc.plen && same_vrf)
 145                 *saddr = f6i->fib6_prefsrc.addr;
 146         else
 147                 err = ipv6_dev_get_saddr(net, same_vrf ? dev : l3mdev, daddr, prefs, saddr);
 148
 149         rcu_read_unlock();
 150
 151         return err;
 152 }
 153
 154 struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
 155                             const struct in6_addr *saddr, int oif,
 156                             const struct sk_buff *skb, int flags);
 157 u32 rt6_multipath_hash(const struct net *net, const struct flowi6 *fl6,
 158                        const struct sk_buff *skb, struct flow_keys *hkeys);
 159
 160 struct dst_entry *icmp6_dst_alloc(struct net_device *dev, struct flowi6 *fl6);
 161
 162 void fib6_force_start_gc(struct net *net);
 163
 164 struct fib6_info *addrconf_f6i_alloc(struct net *net, struct inet6_dev *idev,
 165                                      const struct in6_addr *addr, bool anycast,
 166                                      gfp_t gfp_flags, struct netlink_ext_ack *extack);
 167
 168 struct rt6_info *ip6_dst_alloc(struct net *net, struct net_device *dev,
 169                                int flags);
 170
 171 /*
 172  *      support functions for ND
 173  *
 174  */
 175 struct fib6_info *rt6_get_dflt_router(struct net *net,
 176                                      const struct in6_addr *addr,
 177                                      struct net_device *dev);
 178 struct fib6_info *rt6_add_dflt_router(struct net *net,
 179                                      const struct in6_addr *gwaddr,
 180                                      struct net_device *dev, unsigned int pref,
 181                                      u32 defrtr_usr_metric,
 182                                      int lifetime);
 183
 184 void rt6_purge_dflt_routers(struct net *net);
 185
 186 int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
 187                   const struct in6_addr *gwaddr);
 188
 189 void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, int oif,
 190                      u32 mark, kuid_t uid);
 191 void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu);
 192 void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
 193                   kuid_t uid);
 194 void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif);
 195 void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk);
 196
 197 struct netlink_callback;
 198
 199 struct rt6_rtnl_dump_arg {
 200         struct sk_buff *skb;
 201         struct netlink_callback *cb;
 202         struct net *net;
 203         struct fib_dump_filter filter;
 204 };
 205
 206 int rt6_dump_route(struct fib6_info *f6i, void *p_arg, unsigned int skip);
 207 void rt6_mtu_change(struct net_device *dev, unsigned int mtu);
 208 void rt6_remove_prefsrc(struct inet6_ifaddr *ifp);
 209 void rt6_clean_tohost(struct net *net, struct in6_addr *gateway);
 210 void rt6_sync_up(struct net_device *dev, unsigned char nh_flags);
 211 void rt6_disable_ip(struct net_device *dev, unsigned long event);
 212 void rt6_sync_down_dev(struct net_device *dev, unsigned long event);
 213 void rt6_multipath_rebalance(struct fib6_info *f6i);
 214
 215 void rt6_uncached_list_add(struct rt6_info *rt);
 216 void rt6_uncached_list_del(struct rt6_info *rt);
 217
 218 static inline const struct rt6_info *skb_rt6_info(const struct sk_buff *skb)
 219 {
 220         const struct dst_entry *dst = skb_dst(skb);
 221
 222         if (dst)
 223                 return dst_rt6_info(dst);
 224
 225         return NULL;
 226 }
 227
 228 /*
 229  *      Store a destination cache entry in a socket
 230  */
 231 static inline void ip6_dst_store(struct sock *sk, struct dst_entry *dst,
 232                                  const struct in6_addr *daddr,
 233                                  const struct in6_addr *saddr)
 234 {
 235         struct ipv6_pinfo *np = inet6_sk(sk);
 236
 237         np->dst_cookie = rt6_get_cookie(dst_rt6_info(dst));
 238         sk_setup_caps(sk, dst);
 239         np->daddr_cache = daddr;
 240 #ifdef CONFIG_IPV6_SUBTREES
 241         np->saddr_cache = saddr;
 242 #endif
 243 }
 244
 245 void ip6_sk_dst_store_flow(struct sock *sk, struct dst_entry *dst,
 246                            const struct flowi6 *fl6);
 247
 248 static inline bool ipv6_unicast_destination(const struct sk_buff *skb)
 249 {
 250         const struct rt6_info *rt = dst_rt6_info(skb_dst(skb));
 251
 252         return rt->rt6i_flags & RTF_LOCAL;
 253 }
 254
 255 static inline bool ipv6_anycast_destination(const struct dst_entry *dst,
 256                                             const struct in6_addr *daddr)
 257 {
 258         const struct rt6_info *rt = dst_rt6_info(dst);
 259
 260         return rt->rt6i_flags & RTF_ANYCAST ||
 261                 (rt->rt6i_dst.plen < 127 &&
 262                  !(rt->rt6i_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) &&
 263                  ipv6_addr_equal(&rt->rt6i_dst.addr, daddr));
 264 }
 265
 266 int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 267                  int (*output)(struct net *, struct sock *, struct sk_buff *));
 268
 269 static inline unsigned int ip6_skb_dst_mtu(const struct sk_buff *skb)
 270 {
 271         const struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
 272                                 inet6_sk(skb->sk) : NULL;
 273         const struct dst_entry *dst = skb_dst(skb);
 274         unsigned int mtu;
 275
 276         if (np && READ_ONCE(np->pmtudisc) >= IPV6_PMTUDISC_PROBE) {
 277                 mtu = READ_ONCE(dst->dev->mtu);
 278                 mtu -= lwtunnel_headroom(dst->lwtstate, mtu);
 279         } else {
 280                 mtu = dst_mtu(dst);
 281         }
 282         return mtu;
 283 }
 284
 285 static inline bool ip6_sk_accept_pmtu(const struct sock *sk)
 286 {
 287         u8 pmtudisc = READ_ONCE(inet6_sk(sk)->pmtudisc);
 288
 289         return pmtudisc != IPV6_PMTUDISC_INTERFACE &&
 290                pmtudisc != IPV6_PMTUDISC_OMIT;
 291 }
 292
 293 static inline bool ip6_sk_ignore_df(const struct sock *sk)
 294 {
 295         u8 pmtudisc = READ_ONCE(inet6_sk(sk)->pmtudisc);
 296
 297         return pmtudisc < IPV6_PMTUDISC_DO ||
 298                pmtudisc == IPV6_PMTUDISC_OMIT;
 299 }
 300
 301 static inline const struct in6_addr *rt6_nexthop(const struct rt6_info *rt,
 302                                                  const struct in6_addr *daddr)
 303 {
 304         if (rt->rt6i_flags & RTF_GATEWAY)
 305                 return &rt->rt6i_gateway;
 306         else if (unlikely(rt->rt6i_flags & RTF_CACHE))
 307                 return &rt->rt6i_dst.addr;
 308         else
 309                 return daddr;
 310 }
 311
 312 static inline bool rt6_duplicate_nexthop(struct fib6_info *a, struct fib6_info *b)
 313 {
 314         struct fib6_nh *nha, *nhb;
 315
 316         if (a->nh || b->nh)
 317                 return nexthop_cmp(a->nh, b->nh);
 318
 319         nha = a->fib6_nh;
 320         nhb = b->fib6_nh;
 321         return nha->fib_nh_dev == nhb->fib_nh_dev &&
 322                ipv6_addr_equal(&nha->fib_nh_gw6, &nhb->fib_nh_gw6) &&
 323                !lwtunnel_cmp_encap(nha->fib_nh_lws, nhb->fib_nh_lws);
 324 }
 325
 326 static inline unsigned int ip6_dst_mtu_maybe_forward(const struct dst_entry *dst,
 327                                                      bool forwarding)
 328 {
 329         struct inet6_dev *idev;
 330         unsigned int mtu;
 331
 332         if (!forwarding || dst_metric_locked(dst, RTAX_MTU)) {
 333                 mtu = dst_metric_raw(dst, RTAX_MTU);
 334                 if (mtu)
 335                         goto out;
 336         }
 337
 338         mtu = IPV6_MIN_MTU;
 339         rcu_read_lock();
 340         idev = __in6_dev_get(dst->dev);
 341         if (idev)
 342                 mtu = READ_ONCE(idev->cnf.mtu6);
 343         rcu_read_unlock();
 344
 345 out:
 346         return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
 347 }
 348
 349 u32 ip6_mtu_from_fib6(const struct fib6_result *res,
 350                       const struct in6_addr *daddr,
 351                       const struct in6_addr *saddr);
 352
 353 struct neighbour *ip6_neigh_lookup(const struct in6_addr *gw,
 354                                    struct net_device *dev, struct sk_buff *skb,
 355                                    const void *daddr);
 356 #endif