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/snmp.h>
  32 #include <net/flow.h>
  33
  34 struct sock;
  35
  36 struct inet_skb_parm {
  37         struct ip_options       opt;            /* Compiled IP options          */
  38         unsigned char           flags;
  39
  40 #define IPSKB_FORWARDED         1
  41 #define IPSKB_XFRM_TUNNEL_SIZE  2
  42 #define IPSKB_XFRM_TRANSFORMED  4
  43 #define IPSKB_FRAG_COMPLETE     8
  44 #define IPSKB_REROUTED          16
  45 };
  46
  47 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
  48 {
  49         return ip_hdr(skb)->ihl * 4;
  50 }
  51
  52 struct ipcm_cookie {
  53         __be32                  addr;
  54         int                     oif;
  55         struct ip_options       *opt;
  56         __u8                    tx_flags;
  57 };
  58
  59 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
  60
  61 struct ip_ra_chain {
  62         struct ip_ra_chain __rcu *next;
  63         struct sock             *sk;
  64         union {
  65                 void                    (*destructor)(struct sock *);
  66                 struct sock             *saved_sk;
  67         };
  68         struct rcu_head         rcu;
  69 };
  70
  71 extern struct ip_ra_chain __rcu *ip_ra_chain;
  72
  73 /* IP flags. */
  74 #define IP_CE           0x8000          /* Flag: "Congestion"           */
  75 #define IP_DF           0x4000          /* Flag: "Don't Fragment"       */
  76 #define IP_MF           0x2000          /* Flag: "More Fragments"       */
  77 #define IP_OFFSET       0x1FFF          /* "Fragment Offset" part       */
  78
  79 #define IP_FRAG_TIME    (30 * HZ)               /* fragment lifetime    */
  80
  81 struct msghdr;
  82 struct net_device;
  83 struct packet_type;
  84 struct rtable;
  85 struct sockaddr;
  86
  87 extern int              igmp_mc_proc_init(void);
  88
  89 /*
  90  *      Functions provided by ip.c
  91  */
  92
  93 extern int              ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
  94                                               __be32 saddr, __be32 daddr,
  95                                               struct ip_options *opt);
  96 extern int              ip_rcv(struct sk_buff *skb, struct net_device *dev,
  97                                struct packet_type *pt, struct net_device *orig_dev);
  98 extern int              ip_local_deliver(struct sk_buff *skb);
  99 extern int              ip_mr_input(struct sk_buff *skb);
 100 extern int              ip_output(struct sk_buff *skb);
 101 extern int              ip_mc_output(struct sk_buff *skb);
 102 extern int              ip_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
 103 extern int              ip_do_nat(struct sk_buff *skb);
 104 extern void             ip_send_check(struct iphdr *ip);
 105 extern int              __ip_local_out(struct sk_buff *skb);
 106 extern int              ip_local_out(struct sk_buff *skb);
 107 extern int              ip_queue_xmit(struct sk_buff *skb);
 108 extern void             ip_init(void);
 109 extern int              ip_append_data(struct sock *sk,
 110                                        int getfrag(void *from, char *to, int offset, int len,
 111                                                    int odd, struct sk_buff *skb),
 112                                 void *from, int len, int protolen,
 113                                 struct ipcm_cookie *ipc,
 114                                 struct rtable **rt,
 115                                 unsigned int flags);
 116 extern int              ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, struct sk_buff *skb);
 117 extern ssize_t          ip_append_page(struct sock *sk, struct page *page,
 118                                 int offset, size_t size, int flags);
 119 extern int              ip_push_pending_frames(struct sock *sk);
 120 extern void             ip_flush_pending_frames(struct sock *sk);
 121
 122 /* datagram.c */
 123 extern int              ip4_datagram_connect(struct sock *sk,
 124                                              struct sockaddr *uaddr, int addr_len);
 125
 126 /*
 127  *      Map a multicast IP onto multicast MAC for type Token Ring.
 128  *      This conforms to RFC1469 Option 2 Multicasting i.e.
 129  *      using a functional address to transmit / receive
 130  *      multicast packets.
 131  */
 132
 133 static inline void ip_tr_mc_map(__be32 addr, char *buf)
 134 {
 135         buf[0]=0xC0;
 136         buf[1]=0x00;
 137         buf[2]=0x00;
 138         buf[3]=0x04;
 139         buf[4]=0x00;
 140         buf[5]=0x00;
 141 }
 142
 143 struct ip_reply_arg {
 144         struct kvec iov[1];
 145         int         flags;
 146         __wsum      csum;
 147         int         csumoffset; /* u16 offset of csum in iov[0].iov_base */
 148                                 /* -1 if not needed */
 149         int         bound_dev_if;
 150 };
 151
 152 #define IP_REPLY_ARG_NOSRCCHECK 1
 153
 154 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
 155 {
 156         return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
 157 }
 158
 159 void ip_send_reply(struct sock *sk, struct sk_buff *skb, struct ip_reply_arg *arg,
 160                    unsigned int len);
 161
 162 struct ipv4_config {
 163         int     log_martians;
 164         int     no_pmtu_disc;
 165 };
 166
 167 extern struct ipv4_config ipv4_config;
 168 #define IP_INC_STATS(net, field)        SNMP_INC_STATS64((net)->mib.ip_statistics, field)
 169 #define IP_INC_STATS_BH(net, field)     SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
 170 #define IP_ADD_STATS(net, field, val)   SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
 171 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
 172 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
 173 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
 174 #define NET_INC_STATS(net, field)       SNMP_INC_STATS((net)->mib.net_statistics, field)
 175 #define NET_INC_STATS_BH(net, field)    SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
 176 #define NET_INC_STATS_USER(net, field)  SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
 177 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
 178 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
 179
 180 extern unsigned long snmp_fold_field(void __percpu *mib[], int offt);
 181 #if BITS_PER_LONG==32
 182 extern u64 snmp_fold_field64(void __percpu *mib[], int offt, size_t sync_off);
 183 #else
 184 static inline u64 snmp_fold_field64(void __percpu *mib[], int offt, size_t syncp_off)
 185 {
 186         return snmp_fold_field(mib, offt);
 187 }
 188 #endif
 189 extern int snmp_mib_init(void __percpu *ptr[2], size_t mibsize, size_t align);
 190 extern void snmp_mib_free(void __percpu *ptr[2]);
 191
 192 extern struct local_ports {
 193         seqlock_t       lock;
 194         int             range[2];
 195 } sysctl_local_ports;
 196 extern void inet_get_local_port_range(int *low, int *high);
 197
 198 extern unsigned long *sysctl_local_reserved_ports;
 199 static inline int inet_is_reserved_local_port(int port)
 200 {
 201         return test_bit(port, sysctl_local_reserved_ports);
 202 }
 203
 204 extern int sysctl_ip_default_ttl;
 205 extern int sysctl_ip_nonlocal_bind;
 206
 207 extern struct ctl_path net_core_path[];
 208 extern struct ctl_path net_ipv4_ctl_path[];
 209
 210 /* From inetpeer.c */
 211 extern int inet_peer_threshold;
 212 extern int inet_peer_minttl;
 213 extern int inet_peer_maxttl;
 214 extern int inet_peer_gc_mintime;
 215 extern int inet_peer_gc_maxtime;
 216
 217 /* From ip_output.c */
 218 extern int sysctl_ip_dynaddr;
 219
 220 extern void ipfrag_init(void);
 221
 222 extern void ip_static_sysctl_init(void);
 223
 224 #ifdef CONFIG_INET
 225 #include <net/dst.h>
 226
 227 /* The function in 2.2 was invalid, producing wrong result for
 228  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
 229 static inline
 230 int ip_decrease_ttl(struct iphdr *iph)
 231 {
 232         u32 check = (__force u32)iph->check;
 233         check += (__force u32)htons(0x0100);
 234         iph->check = (__force __sum16)(check + (check>=0xFFFF));
 235         return --iph->ttl;
 236 }
 237
 238 static inline
 239 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
 240 {
 241         return  inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
 242                 (inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
 243                  !(dst_metric_locked(dst, RTAX_MTU)));
 244 }
 245
 246 extern void __ip_select_ident(struct iphdr *iph, struct dst_entry *dst, int more);
 247
 248 static inline void ip_select_ident(struct iphdr *iph, struct dst_entry *dst, struct sock *sk)
 249 {
 250         if (iph->frag_off & htons(IP_DF)) {
 251                 /* This is only to work around buggy Windows95/2000
 252                  * VJ compression implementations.  If the ID field
 253                  * does not change, they drop every other packet in
 254                  * a TCP stream using header compression.
 255                  */
 256                 iph->id = (sk && inet_sk(sk)->inet_daddr) ?
 257                                         htons(inet_sk(sk)->inet_id++) : 0;
 258         } else
 259                 __ip_select_ident(iph, dst, 0);
 260 }
 261
 262 static inline void ip_select_ident_more(struct iphdr *iph, struct dst_entry *dst, struct sock *sk, int more)
 263 {
 264         if (iph->frag_off & htons(IP_DF)) {
 265                 if (sk && inet_sk(sk)->inet_daddr) {
 266                         iph->id = htons(inet_sk(sk)->inet_id);
 267                         inet_sk(sk)->inet_id += 1 + more;
 268                 } else
 269                         iph->id = 0;
 270         } else
 271                 __ip_select_ident(iph, dst, more);
 272 }
 273
 274 /*
 275  *      Map a multicast IP onto multicast MAC for type ethernet.
 276  */
 277
 278 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
 279 {
 280         __u32 addr=ntohl(naddr);
 281         buf[0]=0x01;
 282         buf[1]=0x00;
 283         buf[2]=0x5e;
 284         buf[5]=addr&0xFF;
 285         addr>>=8;
 286         buf[4]=addr&0xFF;
 287         addr>>=8;
 288         buf[3]=addr&0x7F;
 289 }
 290
 291 /*
 292  *      Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
 293  *      Leave P_Key as 0 to be filled in by driver.
 294  */
 295
 296 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
 297 {
 298         __u32 addr;
 299         unsigned char scope = broadcast[5] & 0xF;
 300
 301         buf[0]  = 0;            /* Reserved */
 302         buf[1]  = 0xff;         /* Multicast QPN */
 303         buf[2]  = 0xff;
 304         buf[3]  = 0xff;
 305         addr    = ntohl(naddr);
 306         buf[4]  = 0xff;
 307         buf[5]  = 0x10 | scope; /* scope from broadcast address */
 308         buf[6]  = 0x40;         /* IPv4 signature */
 309         buf[7]  = 0x1b;
 310         buf[8]  = broadcast[8];         /* P_Key */
 311         buf[9]  = broadcast[9];
 312         buf[10] = 0;
 313         buf[11] = 0;
 314         buf[12] = 0;
 315         buf[13] = 0;
 316         buf[14] = 0;
 317         buf[15] = 0;
 318         buf[19] = addr & 0xff;
 319         addr  >>= 8;
 320         buf[18] = addr & 0xff;
 321         addr  >>= 8;
 322         buf[17] = addr & 0xff;
 323         addr  >>= 8;
 324         buf[16] = addr & 0x0f;
 325 }
 326
 327 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 328 #include <linux/ipv6.h>
 329 #endif
 330
 331 static __inline__ void inet_reset_saddr(struct sock *sk)
 332 {
 333         inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
 334 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 335         if (sk->sk_family == PF_INET6) {
 336                 struct ipv6_pinfo *np = inet6_sk(sk);
 337
 338                 memset(&np->saddr, 0, sizeof(np->saddr));
 339                 memset(&np->rcv_saddr, 0, sizeof(np->rcv_saddr));
 340         }
 341 #endif
 342 }
 343
 344 #endif
 345
 346 static inline int sk_mc_loop(struct sock *sk)
 347 {
 348         if (!sk)
 349                 return 1;
 350         switch (sk->sk_family) {
 351         case AF_INET:
 352                 return inet_sk(sk)->mc_loop;
 353 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
 354         case AF_INET6:
 355                 return inet6_sk(sk)->mc_loop;
 356 #endif
 357         }
 358         WARN_ON(1);
 359         return 1;
 360 }
 361
 362 extern int      ip_call_ra_chain(struct sk_buff *skb);
 363
 364 /*
 365  *      Functions provided by ip_fragment.c
 366  */
 367
 368 enum ip_defrag_users {
 369         IP_DEFRAG_LOCAL_DELIVER,
 370         IP_DEFRAG_CALL_RA_CHAIN,
 371         IP_DEFRAG_CONNTRACK_IN,
 372         __IP_DEFRAG_CONNTRACK_IN_END    = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
 373         IP_DEFRAG_CONNTRACK_OUT,
 374         __IP_DEFRAG_CONNTRACK_OUT_END   = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
 375         IP_DEFRAG_CONNTRACK_BRIDGE_IN,
 376         __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
 377         IP_DEFRAG_VS_IN,
 378         IP_DEFRAG_VS_OUT,
 379         IP_DEFRAG_VS_FWD
 380 };
 381
 382 int ip_defrag(struct sk_buff *skb, u32 user);
 383 int ip_frag_mem(struct net *net);
 384 int ip_frag_nqueues(struct net *net);
 385
 386 /*
 387  *      Functions provided by ip_forward.c
 388  */
 389
 390 extern int ip_forward(struct sk_buff *skb);
 391
 392 /*
 393  *      Functions provided by ip_options.c
 394  */
 395
 396 extern void ip_options_build(struct sk_buff *skb, struct ip_options *opt, __be32 daddr, struct rtable *rt, int is_frag);
 397 extern int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb);
 398 extern void ip_options_fragment(struct sk_buff *skb);
 399 extern int ip_options_compile(struct net *net,
 400                               struct ip_options *opt, struct sk_buff *skb);
 401 extern int ip_options_get(struct net *net, struct ip_options **optp,
 402                           unsigned char *data, int optlen);
 403 extern int ip_options_get_from_user(struct net *net, struct ip_options **optp,
 404                                     unsigned char __user *data, int optlen);
 405 extern void ip_options_undo(struct ip_options * opt);
 406 extern void ip_forward_options(struct sk_buff *skb);
 407 extern int ip_options_rcv_srr(struct sk_buff *skb);
 408
 409 /*
 410  *      Functions provided by ip_sockglue.c
 411  */
 412
 413 extern int      ip_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
 414 extern void     ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb);
 415 extern int      ip_cmsg_send(struct net *net,
 416                              struct msghdr *msg, struct ipcm_cookie *ipc);
 417 extern int      ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval, unsigned int optlen);
 418 extern int      ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, int __user *optlen);
 419 extern int      compat_ip_setsockopt(struct sock *sk, int level,
 420                         int optname, char __user *optval, unsigned int optlen);
 421 extern int      compat_ip_getsockopt(struct sock *sk, int level,
 422                         int optname, char __user *optval, int __user *optlen);
 423 extern int      ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *));
 424
 425 extern int      ip_recv_error(struct sock *sk, struct msghdr *msg, int len);
 426 extern void     ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
 427                               __be16 port, u32 info, u8 *payload);
 428 extern void     ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
 429                                u32 info);
 430
 431 /* sysctl helpers - any sysctl which holds a value that ends up being
 432  * fed into the routing cache should use these handlers.
 433  */
 434 int ipv4_doint_and_flush(ctl_table *ctl, int write,
 435                          void __user *buffer,
 436                          size_t *lenp, loff_t *ppos);
 437 int ipv4_doint_and_flush_strategy(ctl_table *table,
 438                                   void __user *oldval, size_t __user *oldlenp,
 439                                   void __user *newval, size_t newlen);
 440 #ifdef CONFIG_PROC_FS
 441 extern int ip_misc_proc_init(void);
 442 #endif
 443
 444 #endif  /* _IP_H */