include/net/ip.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 module.
   7  *
   8  * Version:     @(#)ip.h        1.0.2   05/07/93
   9  *
  10  * Authors:     Ross Biro
  11  *              Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
  12  *              Alan Cox, <gw4pts@gw4pts.ampr.org>
  13  *
  14  * Changes:
  15  *              Mike McLagan    :       Routing by source
  16  *
  17  *              This program is free software; you can redistribute it and/or
  18  *              modify it under the terms of the GNU General Public License
  19  *              as published by the Free Software Foundation; either version
  20  *              2 of the License, or (at your option) any later version.
  21  */
  22 #ifndef _IP_H
  23 #define _IP_H
  24
  25 #include <linux/types.h>
  26 #include <linux/ip.h>
  27 #include <linux/in.h>
  28 #include <linux/skbuff.h>
  29
  30 #include <net/inet_sock.h>
  31 #include <net/route.h>
  32 #include <net/snmp.h>
  33 #include <net/flow.h>
  34 #include <net/flow_keys.h>
  35
  36 struct sock;
  37
  38 struct inet_skb_parm {
  39         struct ip_options       opt;            /* Compiled IP options          */
  40         unsigned char           flags;
  41
  42 #define IPSKB_FORWARDED         1
  43 #define IPSKB_XFRM_TUNNEL_SIZE  2
  44 #define IPSKB_XFRM_TRANSFORMED  4
  45 #define IPSKB_FRAG_COMPLETE     8
  46 #define IPSKB_REROUTED          16
  47
  48         u16                     frag_max_size;
  49 };
  50
  51 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
  52 {
  53         return ip_hdr(skb)->ihl * 4;
  54 }
  55
  56 struct ipcm_cookie {
  57         __be32                  addr;
  58         int                     oif;
  59         struct ip_options_rcu   *opt;
  60         __u8                    tx_flags;
  61         __u8                    ttl;
  62         __s16                   tos;
  63         char                    priority;
  64 };
  65
  66 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
  67 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
  68
  69 struct ip_ra_chain {
  70         struct ip_ra_chain __rcu *next;
  71         struct sock             *sk;
  72         union {
  73                 void                    (*destructor)(struct sock *);
  74                 struct sock             *saved_sk;
  75         };
  76         struct rcu_head         rcu;
  77 };
  78
  79 extern struct ip_ra_chain __rcu *ip_ra_chain;
  80
  81 /* IP flags. */
  82 #define IP_CE           0x8000          /* Flag: "Congestion"           */
  83 #define IP_DF           0x4000          /* Flag: "Don't Fragment"       */
  84 #define IP_MF           0x2000          /* Flag: "More Fragments"       */
  85 #define IP_OFFSET       0x1FFF          /* "Fragment Offset" part       */
  86
  87 #define IP_FRAG_TIME    (30 * HZ)               /* fragment lifetime    */
  88
  89 struct msghdr;
  90 struct net_device;
  91 struct packet_type;
  92 struct rtable;
  93 struct sockaddr;
  94
  95 int igmp_mc_init(void);
  96
  97 /*
  98  *      Functions provided by ip.c
  99  */
 100
 101 int ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
 102                           __be32 saddr, __be32 daddr,
 103                           struct ip_options_rcu *opt);
 104 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
 105            struct net_device *orig_dev);
 106 int ip_local_deliver(struct sk_buff *skb);
 107 int ip_mr_input(struct sk_buff *skb);
 108 int ip_output(struct sock *sk, struct sk_buff *skb);
 109 int ip_mc_output(struct sock *sk, struct sk_buff *skb);
 110 int ip_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
 111 int ip_do_nat(struct sk_buff *skb);
 112 void ip_send_check(struct iphdr *ip);
 113 int __ip_local_out(struct sk_buff *skb);
 114 int ip_local_out_sk(struct sock *sk, struct sk_buff *skb);
 115 static inline int ip_local_out(struct sk_buff *skb)
 116 {
 117         return ip_local_out_sk(skb->sk, skb);
 118 }
 119
 120 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
 121 void ip_init(void);
 122 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
 123                    int getfrag(void *from, char *to, int offset, int len,
 124                                int odd, struct sk_buff *skb),
 125                    void *from, int len, int protolen,
 126                    struct ipcm_cookie *ipc,
 127                    struct rtable **rt,
 128                    unsigned int flags);
 129 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
 130                        struct sk_buff *skb);
 131 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
 132                        int offset, size_t size, int flags);
 133 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
 134                               struct sk_buff_head *queue,
 135                               struct inet_cork *cork);
 136 int ip_send_skb(struct net *net, struct sk_buff *skb);
 137 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
 138 void ip_flush_pending_frames(struct sock *sk);
 139 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
 140                             int getfrag(void *from, char *to, int offset,
 141                                         int len, int odd, struct sk_buff *skb),
 142                             void *from, int length, int transhdrlen,
 143                             struct ipcm_cookie *ipc, struct rtable **rtp,
 144                             unsigned int flags);
 145
 146 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
 147 {
 148         return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
 149 }
 150
 151 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
 152 {
 153         return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
 154 }
 155
 156 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
 157 {
 158         return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
 159 }
 160
 161 /* datagram.c */
 162 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
 163
 164 void ip4_datagram_release_cb(struct sock *sk);
 165
 166 struct ip_reply_arg {
 167         struct kvec iov[1];
 168         int         flags;
 169         __wsum      csum;
 170         int         csumoffset; /* u16 offset of csum in iov[0].iov_base */
 171                                 /* -1 if not needed */
 172         int         bound_dev_if;
 173         u8          tos;
 174 };
 175
 176 #define IP_REPLY_ARG_NOSRCCHECK 1
 177
 178 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
 179 {
 180         return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
 181 }
 182
 183 void ip_send_unicast_reply(struct net *net, struct sk_buff *skb, __be32 daddr,
 184                            __be32 saddr, const struct ip_reply_arg *arg,
 185                            unsigned int len);
 186
 187 #define IP_INC_STATS(net, field)        SNMP_INC_STATS64((net)->mib.ip_statistics, field)
 188 #define IP_INC_STATS_BH(net, field)     SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
 189 #define IP_ADD_STATS(net, field, val)   SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
 190 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
 191 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
 192 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
 193 #define NET_INC_STATS(net, field)       SNMP_INC_STATS((net)->mib.net_statistics, field)
 194 #define NET_INC_STATS_BH(net, field)    SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
 195 #define NET_INC_STATS_USER(net, field)  SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
 196 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
 197 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
 198 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
 199
 200 unsigned long snmp_fold_field(void __percpu *mib, int offt);
 201 #if BITS_PER_LONG==32
 202 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
 203 #else
 204 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
 205 {
 206         return snmp_fold_field(mib, offt);
 207 }
 208 #endif
 209
 210 void inet_get_local_port_range(struct net *net, int *low, int *high);
 211
 212 #ifdef CONFIG_SYSCTL
 213 static inline int inet_is_local_reserved_port(struct net *net, int port)
 214 {
 215         if (!net->ipv4.sysctl_local_reserved_ports)
 216                 return 0;
 217         return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
 218 }
 219 #else
 220 static inline int inet_is_local_reserved_port(struct net *net, int port)
 221 {
 222         return 0;
 223 }
 224 #endif
 225
 226 extern int sysctl_ip_nonlocal_bind;
 227
 228 /* From inetpeer.c */
 229 extern int inet_peer_threshold;
 230 extern int inet_peer_minttl;
 231 extern int inet_peer_maxttl;
 232
 233 /* From ip_input.c */
 234 extern int sysctl_ip_early_demux;
 235
 236 /* From ip_output.c */
 237 extern int sysctl_ip_dynaddr;
 238
 239 void ipfrag_init(void);
 240
 241 void ip_static_sysctl_init(void);
 242
 243 #define IP4_REPLY_MARK(net, mark) \
 244         ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
 245
 246 static inline bool ip_is_fragment(const struct iphdr *iph)
 247 {
 248         return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
 249 }
 250
 251 #ifdef CONFIG_INET
 252 #include <net/dst.h>
 253
 254 /* The function in 2.2 was invalid, producing wrong result for
 255  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
 256 static inline
 257 int ip_decrease_ttl(struct iphdr *iph)
 258 {
 259         u32 check = (__force u32)iph->check;
 260         check += (__force u32)htons(0x0100);
 261         iph->check = (__force __sum16)(check + (check>=0xFFFF));
 262         return --iph->ttl;
 263 }
 264
 265 static inline
 266 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
 267 {
 268         return  inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
 269                 (inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
 270                  !(dst_metric_locked(dst, RTAX_MTU)));
 271 }
 272
 273 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
 274 {
 275         return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
 276                inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
 277 }
 278
 279 static inline bool ip_sk_use_pmtu(const struct sock *sk)
 280 {
 281         return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
 282 }
 283
 284 static inline bool ip_sk_ignore_df(const struct sock *sk)
 285 {
 286         return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
 287                inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
 288 }
 289
 290 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
 291                                                     bool forwarding)
 292 {
 293         struct net *net = dev_net(dst->dev);
 294
 295         if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
 296             dst_metric_locked(dst, RTAX_MTU) ||
 297             !forwarding)
 298                 return dst_mtu(dst);
 299
 300         return min(dst->dev->mtu, IP_MAX_MTU);
 301 }
 302
 303 static inline unsigned int ip_skb_dst_mtu(const struct sk_buff *skb)
 304 {
 305         if (!skb->sk || ip_sk_use_pmtu(skb->sk)) {
 306                 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
 307                 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
 308         } else {
 309                 return min(skb_dst(skb)->dev->mtu, IP_MAX_MTU);
 310         }
 311 }
 312
 313 #define IP_IDENTS_SZ 2048u
 314 extern atomic_t *ip_idents;
 315
 316 static inline u32 ip_idents_reserve(u32 hash, int segs)
 317 {
 318         atomic_t *id_ptr = ip_idents + hash % IP_IDENTS_SZ;
 319
 320         return atomic_add_return(segs, id_ptr) - segs;
 321 }
 322
 323 void __ip_select_ident(struct iphdr *iph, int segs);
 324
 325 static inline void ip_select_ident_segs(struct sk_buff *skb, struct sock *sk, int segs)
 326 {
 327         struct iphdr *iph = ip_hdr(skb);
 328
 329         if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
 330                 /* This is only to work around buggy Windows95/2000
 331                  * VJ compression implementations.  If the ID field
 332                  * does not change, they drop every other packet in
 333                  * a TCP stream using header compression.
 334                  */
 335                 if (sk && inet_sk(sk)->inet_daddr) {
 336                         iph->id = htons(inet_sk(sk)->inet_id);
 337                         inet_sk(sk)->inet_id += segs;
 338                 } else {
 339                         iph->id = 0;
 340                 }
 341         } else {
 342                 __ip_select_ident(iph, segs);
 343         }
 344 }
 345
 346 static inline void ip_select_ident(struct sk_buff *skb, struct sock *sk)
 347 {
 348         ip_select_ident_segs(skb, sk, 1);
 349 }
 350
 351 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
 352 {
 353         return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
 354                                   skb->len, proto, 0);
 355 }
 356
 357 static inline void inet_set_txhash(struct sock *sk)
 358 {
 359         struct inet_sock *inet = inet_sk(sk);
 360         struct flow_keys keys;
 361
 362         keys.src = inet->inet_saddr;
 363         keys.dst = inet->inet_daddr;
 364         keys.port16[0] = inet->inet_sport;
 365         keys.port16[1] = inet->inet_dport;
 366
 367         sk->sk_txhash = flow_hash_from_keys(&keys);
 368 }
 369
 370 /*
 371  *      Map a multicast IP onto multicast MAC for type ethernet.
 372  */
 373
 374 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
 375 {
 376         __u32 addr=ntohl(naddr);
 377         buf[0]=0x01;
 378         buf[1]=0x00;
 379         buf[2]=0x5e;
 380         buf[5]=addr&0xFF;
 381         addr>>=8;
 382         buf[4]=addr&0xFF;
 383         addr>>=8;
 384         buf[3]=addr&0x7F;
 385 }
 386
 387 /*
 388  *      Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
 389  *      Leave P_Key as 0 to be filled in by driver.
 390  */
 391
 392 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
 393 {
 394         __u32 addr;
 395         unsigned char scope = broadcast[5] & 0xF;
 396
 397         buf[0]  = 0;            /* Reserved */
 398         buf[1]  = 0xff;         /* Multicast QPN */
 399         buf[2]  = 0xff;
 400         buf[3]  = 0xff;
 401         addr    = ntohl(naddr);
 402         buf[4]  = 0xff;
 403         buf[5]  = 0x10 | scope; /* scope from broadcast address */
 404         buf[6]  = 0x40;         /* IPv4 signature */
 405         buf[7]  = 0x1b;
 406         buf[8]  = broadcast[8];         /* P_Key */
 407         buf[9]  = broadcast[9];
 408         buf[10] = 0;
 409         buf[11] = 0;
 410         buf[12] = 0;
 411         buf[13] = 0;
 412         buf[14] = 0;
 413         buf[15] = 0;
 414         buf[19] = addr & 0xff;
 415         addr  >>= 8;
 416         buf[18] = addr & 0xff;
 417         addr  >>= 8;
 418         buf[17] = addr & 0xff;
 419         addr  >>= 8;
 420         buf[16] = addr & 0x0f;
 421 }
 422
 423 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
 424 {
 425         if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
 426                 memcpy(buf, broadcast, 4);
 427         else
 428                 memcpy(buf, &naddr, sizeof(naddr));
 429 }
 430
 431 #if IS_ENABLED(CONFIG_IPV6)
 432 #include <linux/ipv6.h>
 433 #endif
 434
 435 static __inline__ void inet_reset_saddr(struct sock *sk)
 436 {
 437         inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
 438 #if IS_ENABLED(CONFIG_IPV6)
 439         if (sk->sk_family == PF_INET6) {
 440                 struct ipv6_pinfo *np = inet6_sk(sk);
 441
 442                 memset(&np->saddr, 0, sizeof(np->saddr));
 443                 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
 444         }
 445 #endif
 446 }
 447
 448 #endif
 449
 450 static inline int sk_mc_loop(struct sock *sk)
 451 {
 452         if (!sk)
 453                 return 1;
 454         switch (sk->sk_family) {
 455         case AF_INET:
 456                 return inet_sk(sk)->mc_loop;
 457 #if IS_ENABLED(CONFIG_IPV6)
 458         case AF_INET6:
 459                 return inet6_sk(sk)->mc_loop;
 460 #endif
 461         }
 462         WARN_ON(1);
 463         return 1;
 464 }
 465
 466 bool ip_call_ra_chain(struct sk_buff *skb);
 467
 468 /*
 469  *      Functions provided by ip_fragment.c
 470  */
 471
 472 enum ip_defrag_users {
 473         IP_DEFRAG_LOCAL_DELIVER,
 474         IP_DEFRAG_CALL_RA_CHAIN,
 475         IP_DEFRAG_CONNTRACK_IN,
 476         __IP_DEFRAG_CONNTRACK_IN_END    = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
 477         IP_DEFRAG_CONNTRACK_OUT,
 478         __IP_DEFRAG_CONNTRACK_OUT_END   = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
 479         IP_DEFRAG_CONNTRACK_BRIDGE_IN,
 480         __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
 481         IP_DEFRAG_VS_IN,
 482         IP_DEFRAG_VS_OUT,
 483         IP_DEFRAG_VS_FWD,
 484         IP_DEFRAG_AF_PACKET,
 485         IP_DEFRAG_MACVLAN,
 486 };
 487
 488 int ip_defrag(struct sk_buff *skb, u32 user);
 489 #ifdef CONFIG_INET
 490 struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user);
 491 #else
 492 static inline struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user)
 493 {
 494         return skb;
 495 }
 496 #endif
 497 int ip_frag_mem(struct net *net);
 498
 499 /*
 500  *      Functions provided by ip_forward.c
 501  */
 502
 503 int ip_forward(struct sk_buff *skb);
 504
 505 /*
 506  *      Functions provided by ip_options.c
 507  */
 508
 509 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
 510                       __be32 daddr, struct rtable *rt, int is_frag);
 511 int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb);
 512 void ip_options_fragment(struct sk_buff *skb);
 513 int ip_options_compile(struct net *net, struct ip_options *opt,
 514                        struct sk_buff *skb);
 515 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
 516                    unsigned char *data, int optlen);
 517 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
 518                              unsigned char __user *data, int optlen);
 519 void ip_options_undo(struct ip_options *opt);
 520 void ip_forward_options(struct sk_buff *skb);
 521 int ip_options_rcv_srr(struct sk_buff *skb);
 522
 523 /*
 524  *      Functions provided by ip_sockglue.c
 525  */
 526
 527 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
 528 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb);
 529 int ip_cmsg_send(struct net *net, struct msghdr *msg,
 530                  struct ipcm_cookie *ipc, bool allow_ipv6);
 531 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
 532                   unsigned int optlen);
 533 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
 534                   int __user *optlen);
 535 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
 536                          char __user *optval, unsigned int optlen);
 537 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
 538                          char __user *optval, int __user *optlen);
 539 int ip_ra_control(struct sock *sk, unsigned char on,
 540                   void (*destructor)(struct sock *));
 541
 542 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
 543 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
 544                    u32 info, u8 *payload);
 545 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
 546                     u32 info);
 547
 548 #ifdef CONFIG_PROC_FS
 549 int ip_misc_proc_init(void);
 550 #endif
 551
 552 #endif  /* _IP_H */