include/net/ip.h

   1 /* SPDX-License-Identifier: GPL-2.0-or-later */
   2 /*
   3  * INET         An implementation of the TCP/IP protocol suite for the LINUX
   4  *              operating system.  INET is implemented using the  BSD Socket
   5  *              interface as the means of communication with the user level.
   6  *
   7  *              Definitions for the IP module.
   8  *
   9  * Version:     @(#)ip.h        1.0.2   05/07/93
  10  *
  11  * Authors:     Ross Biro
  12  *              Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
  13  *              Alan Cox, <gw4pts@gw4pts.ampr.org>
  14  *
  15  * Changes:
  16  *              Mike McLagan    :       Routing by source
  17  */
  18 #ifndef _IP_H
  19 #define _IP_H
  20
  21 #include <linux/types.h>
  22 #include <linux/ip.h>
  23 #include <linux/in.h>
  24 #include <linux/skbuff.h>
  25 #include <linux/jhash.h>
  26 #include <linux/sockptr.h>
  27
  28 #include <net/inet_sock.h>
  29 #include <net/route.h>
  30 #include <net/snmp.h>
  31 #include <net/flow.h>
  32 #include <net/flow_dissector.h>
  33 #include <net/netns/hash.h>
  34
  35 #define IPV4_MAX_PMTU           65535U          /* RFC 2675, Section 5.1 */
  36 #define IPV4_MIN_MTU            68                      /* RFC 791 */
  37
  38 extern unsigned int sysctl_fib_sync_mem;
  39 extern unsigned int sysctl_fib_sync_mem_min;
  40 extern unsigned int sysctl_fib_sync_mem_max;
  41
  42 struct sock;
  43
  44 struct inet_skb_parm {
  45         int                     iif;
  46         struct ip_options       opt;            /* Compiled IP options          */
  47         u16                     flags;
  48
  49 #define IPSKB_FORWARDED         BIT(0)
  50 #define IPSKB_XFRM_TUNNEL_SIZE  BIT(1)
  51 #define IPSKB_XFRM_TRANSFORMED  BIT(2)
  52 #define IPSKB_FRAG_COMPLETE     BIT(3)
  53 #define IPSKB_REROUTED          BIT(4)
  54 #define IPSKB_DOREDIRECT        BIT(5)
  55 #define IPSKB_FRAG_PMTU         BIT(6)
  56 #define IPSKB_L3SLAVE           BIT(7)
  57
  58         u16                     frag_max_size;
  59 };
  60
  61 static inline bool ipv4_l3mdev_skb(u16 flags)
  62 {
  63         return !!(flags & IPSKB_L3SLAVE);
  64 }
  65
  66 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
  67 {
  68         return ip_hdr(skb)->ihl * 4;
  69 }
  70
  71 struct ipcm_cookie {
  72         struct sockcm_cookie    sockc;
  73         __be32                  addr;
  74         int                     oif;
  75         struct ip_options_rcu   *opt;
  76         __u8                    ttl;
  77         __s16                   tos;
  78         char                    priority;
  79         __u16                   gso_size;
  80 };
  81
  82 static inline void ipcm_init(struct ipcm_cookie *ipcm)
  83 {
  84         *ipcm = (struct ipcm_cookie) { .tos = -1 };
  85 }
  86
  87 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
  88                                 const struct inet_sock *inet)
  89 {
  90         ipcm_init(ipcm);
  91
  92         ipcm->sockc.mark = inet->sk.sk_mark;
  93         ipcm->sockc.tsflags = inet->sk.sk_tsflags;
  94         ipcm->oif = inet->sk.sk_bound_dev_if;
  95         ipcm->addr = inet->inet_saddr;
  96 }
  97
  98 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
  99 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
 100
 101 /* return enslaved device index if relevant */
 102 static inline int inet_sdif(struct sk_buff *skb)
 103 {
 104 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
 105         if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
 106                 return IPCB(skb)->iif;
 107 #endif
 108         return 0;
 109 }
 110
 111 /* Special input handler for packets caught by router alert option.
 112    They are selected only by protocol field, and then processed likely
 113    local ones; but only if someone wants them! Otherwise, router
 114    not running rsvpd will kill RSVP.
 115
 116    It is user level problem, what it will make with them.
 117    I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
 118    but receiver should be enough clever f.e. to forward mtrace requests,
 119    sent to multicast group to reach destination designated router.
 120  */
 121
 122 struct ip_ra_chain {
 123         struct ip_ra_chain __rcu *next;
 124         struct sock             *sk;
 125         union {
 126                 void                    (*destructor)(struct sock *);
 127                 struct sock             *saved_sk;
 128         };
 129         struct rcu_head         rcu;
 130 };
 131
 132 /* IP flags. */
 133 #define IP_CE           0x8000          /* Flag: "Congestion"           */
 134 #define IP_DF           0x4000          /* Flag: "Don't Fragment"       */
 135 #define IP_MF           0x2000          /* Flag: "More Fragments"       */
 136 #define IP_OFFSET       0x1FFF          /* "Fragment Offset" part       */
 137
 138 #define IP_FRAG_TIME    (30 * HZ)               /* fragment lifetime    */
 139
 140 struct msghdr;
 141 struct net_device;
 142 struct packet_type;
 143 struct rtable;
 144 struct sockaddr;
 145
 146 int igmp_mc_init(void);
 147
 148 /*
 149  *      Functions provided by ip.c
 150  */
 151
 152 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
 153                           __be32 saddr, __be32 daddr,
 154                           struct ip_options_rcu *opt);
 155 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
 156            struct net_device *orig_dev);
 157 void ip_list_rcv(struct list_head *head, struct packet_type *pt,
 158                  struct net_device *orig_dev);
 159 int ip_local_deliver(struct sk_buff *skb);
 160 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
 161 int ip_mr_input(struct sk_buff *skb);
 162 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
 163 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
 164 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
 165                    int (*output)(struct net *, struct sock *, struct sk_buff *));
 166
 167 struct ip_fraglist_iter {
 168         struct sk_buff  *frag;
 169         struct iphdr    *iph;
 170         int             offset;
 171         unsigned int    hlen;
 172 };
 173
 174 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
 175                       unsigned int hlen, struct ip_fraglist_iter *iter);
 176 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
 177
 178 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
 179 {
 180         struct sk_buff *skb = iter->frag;
 181
 182         iter->frag = skb->next;
 183         skb_mark_not_on_list(skb);
 184
 185         return skb;
 186 }
 187
 188 struct ip_frag_state {
 189         bool            DF;
 190         unsigned int    hlen;
 191         unsigned int    ll_rs;
 192         unsigned int    mtu;
 193         unsigned int    left;
 194         int             offset;
 195         int             ptr;
 196         __be16          not_last_frag;
 197 };
 198
 199 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
 200                   unsigned int mtu, bool DF, struct ip_frag_state *state);
 201 struct sk_buff *ip_frag_next(struct sk_buff *skb,
 202                              struct ip_frag_state *state);
 203
 204 void ip_send_check(struct iphdr *ip);
 205 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
 206 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
 207
 208 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
 209                     __u8 tos);
 210 void ip_init(void);
 211 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
 212                    int getfrag(void *from, char *to, int offset, int len,
 213                                int odd, struct sk_buff *skb),
 214                    void *from, int len, int protolen,
 215                    struct ipcm_cookie *ipc,
 216                    struct rtable **rt,
 217                    unsigned int flags);
 218 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
 219                        struct sk_buff *skb);
 220 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
 221                        int offset, size_t size, int flags);
 222 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
 223                               struct sk_buff_head *queue,
 224                               struct inet_cork *cork);
 225 int ip_send_skb(struct net *net, struct sk_buff *skb);
 226 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
 227 void ip_flush_pending_frames(struct sock *sk);
 228 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
 229                             int getfrag(void *from, char *to, int offset,
 230                                         int len, int odd, struct sk_buff *skb),
 231                             void *from, int length, int transhdrlen,
 232                             struct ipcm_cookie *ipc, struct rtable **rtp,
 233                             struct inet_cork *cork, unsigned int flags);
 234
 235 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
 236
 237 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
 238 {
 239         return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
 240 }
 241
 242 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
 243 {
 244         return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
 245 }
 246
 247 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
 248 {
 249         return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
 250 }
 251
 252 /* datagram.c */
 253 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
 254 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
 255
 256 void ip4_datagram_release_cb(struct sock *sk);
 257
 258 struct ip_reply_arg {
 259         struct kvec iov[1];
 260         int         flags;
 261         __wsum      csum;
 262         int         csumoffset; /* u16 offset of csum in iov[0].iov_base */
 263                                 /* -1 if not needed */
 264         int         bound_dev_if;
 265         u8          tos;
 266         kuid_t      uid;
 267 };
 268
 269 #define IP_REPLY_ARG_NOSRCCHECK 1
 270
 271 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
 272 {
 273         return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
 274 }
 275
 276 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
 277                            const struct ip_options *sopt,
 278                            __be32 daddr, __be32 saddr,
 279                            const struct ip_reply_arg *arg,
 280                            unsigned int len, u64 transmit_time);
 281
 282 #define IP_INC_STATS(net, field)        SNMP_INC_STATS64((net)->mib.ip_statistics, field)
 283 #define __IP_INC_STATS(net, field)      __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
 284 #define IP_ADD_STATS(net, field, val)   SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
 285 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
 286 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
 287 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
 288 #define NET_INC_STATS(net, field)       SNMP_INC_STATS((net)->mib.net_statistics, field)
 289 #define __NET_INC_STATS(net, field)     __SNMP_INC_STATS((net)->mib.net_statistics, field)
 290 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
 291 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
 292
 293 u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct);
 294 unsigned long snmp_fold_field(void __percpu *mib, int offt);
 295 #if BITS_PER_LONG==32
 296 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
 297                          size_t syncp_offset);
 298 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
 299 #else
 300 static inline u64  snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
 301                                         size_t syncp_offset)
 302 {
 303         return snmp_get_cpu_field(mib, cpu, offct);
 304
 305 }
 306
 307 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
 308 {
 309         return snmp_fold_field(mib, offt);
 310 }
 311 #endif
 312
 313 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
 314 { \
 315         int i, c; \
 316         for_each_possible_cpu(c) { \
 317                 for (i = 0; stats_list[i].name; i++) \
 318                         buff64[i] += snmp_get_cpu_field64( \
 319                                         mib_statistic, \
 320                                         c, stats_list[i].entry, \
 321                                         offset); \
 322         } \
 323 }
 324
 325 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
 326 { \
 327         int i, c; \
 328         for_each_possible_cpu(c) { \
 329                 for (i = 0; stats_list[i].name; i++) \
 330                         buff[i] += snmp_get_cpu_field( \
 331                                                 mib_statistic, \
 332                                                 c, stats_list[i].entry); \
 333         } \
 334 }
 335
 336 void inet_get_local_port_range(struct net *net, int *low, int *high);
 337
 338 #ifdef CONFIG_SYSCTL
 339 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
 340 {
 341         if (!net->ipv4.sysctl_local_reserved_ports)
 342                 return false;
 343         return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
 344 }
 345
 346 static inline bool sysctl_dev_name_is_allowed(const char *name)
 347 {
 348         return strcmp(name, "default") != 0  && strcmp(name, "all") != 0;
 349 }
 350
 351 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
 352 {
 353         return port < net->ipv4.sysctl_ip_prot_sock;
 354 }
 355
 356 #else
 357 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
 358 {
 359         return false;
 360 }
 361
 362 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
 363 {
 364         return port < PROT_SOCK;
 365 }
 366 #endif
 367
 368 __be32 inet_current_timestamp(void);
 369
 370 /* From inetpeer.c */
 371 extern int inet_peer_threshold;
 372 extern int inet_peer_minttl;
 373 extern int inet_peer_maxttl;
 374
 375 void ipfrag_init(void);
 376
 377 void ip_static_sysctl_init(void);
 378
 379 #define IP4_REPLY_MARK(net, mark) \
 380         ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
 381
 382 static inline bool ip_is_fragment(const struct iphdr *iph)
 383 {
 384         return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
 385 }
 386
 387 #ifdef CONFIG_INET
 388 #include <net/dst.h>
 389
 390 /* The function in 2.2 was invalid, producing wrong result for
 391  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
 392 static inline
 393 int ip_decrease_ttl(struct iphdr *iph)
 394 {
 395         u32 check = (__force u32)iph->check;
 396         check += (__force u32)htons(0x0100);
 397         iph->check = (__force __sum16)(check + (check>=0xFFFF));
 398         return --iph->ttl;
 399 }
 400
 401 static inline int ip_mtu_locked(const struct dst_entry *dst)
 402 {
 403         const struct rtable *rt = (const struct rtable *)dst;
 404
 405         return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
 406 }
 407
 408 static inline
 409 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
 410 {
 411         u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
 412
 413         return  pmtudisc == IP_PMTUDISC_DO ||
 414                 (pmtudisc == IP_PMTUDISC_WANT &&
 415                  !ip_mtu_locked(dst));
 416 }
 417
 418 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
 419 {
 420         return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
 421                inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
 422 }
 423
 424 static inline bool ip_sk_use_pmtu(const struct sock *sk)
 425 {
 426         return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
 427 }
 428
 429 static inline bool ip_sk_ignore_df(const struct sock *sk)
 430 {
 431         return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
 432                inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
 433 }
 434
 435 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
 436                                                     bool forwarding)
 437 {
 438         struct net *net = dev_net(dst->dev);
 439
 440         if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
 441             ip_mtu_locked(dst) ||
 442             !forwarding)
 443                 return dst_mtu(dst);
 444
 445         return min(READ_ONCE(dst->dev->mtu), IP_MAX_MTU);
 446 }
 447
 448 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
 449                                           const struct sk_buff *skb)
 450 {
 451         if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
 452                 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
 453
 454                 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
 455         }
 456
 457         return min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
 458 }
 459
 460 struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx,
 461                                         int fc_mx_len,
 462                                         struct netlink_ext_ack *extack);
 463 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
 464 {
 465         if (fib_metrics != &dst_default_metrics &&
 466             refcount_dec_and_test(&fib_metrics->refcnt))
 467                 kfree(fib_metrics);
 468 }
 469
 470 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
 471 static inline
 472 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
 473 {
 474         dst_init_metrics(dst, fib_metrics->metrics, true);
 475
 476         if (fib_metrics != &dst_default_metrics) {
 477                 dst->_metrics |= DST_METRICS_REFCOUNTED;
 478                 refcount_inc(&fib_metrics->refcnt);
 479         }
 480 }
 481
 482 static inline
 483 void ip_dst_metrics_put(struct dst_entry *dst)
 484 {
 485         struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
 486
 487         if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
 488                 kfree(p);
 489 }
 490
 491 u32 ip_idents_reserve(u32 hash, int segs);
 492 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
 493
 494 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
 495                                         struct sock *sk, int segs)
 496 {
 497         struct iphdr *iph = ip_hdr(skb);
 498
 499         if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
 500                 /* This is only to work around buggy Windows95/2000
 501                  * VJ compression implementations.  If the ID field
 502                  * does not change, they drop every other packet in
 503                  * a TCP stream using header compression.
 504                  */
 505                 if (sk && inet_sk(sk)->inet_daddr) {
 506                         iph->id = htons(inet_sk(sk)->inet_id);
 507                         inet_sk(sk)->inet_id += segs;
 508                 } else {
 509                         iph->id = 0;
 510                 }
 511         } else {
 512                 __ip_select_ident(net, iph, segs);
 513         }
 514 }
 515
 516 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
 517                                    struct sock *sk)
 518 {
 519         ip_select_ident_segs(net, skb, sk, 1);
 520 }
 521
 522 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
 523 {
 524         return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
 525                                   skb->len, proto, 0);
 526 }
 527
 528 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
 529  * Equivalent to :      flow->v4addrs.src = iph->saddr;
 530  *                      flow->v4addrs.dst = iph->daddr;
 531  */
 532 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
 533                                             const struct iphdr *iph)
 534 {
 535         BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
 536                      offsetof(typeof(flow->addrs), v4addrs.src) +
 537                               sizeof(flow->addrs.v4addrs.src));
 538         memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs));
 539         flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
 540 }
 541
 542 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
 543 {
 544         const struct iphdr *iph = skb_gro_network_header(skb);
 545
 546         return csum_tcpudp_nofold(iph->saddr, iph->daddr,
 547                                   skb_gro_len(skb), proto, 0);
 548 }
 549
 550 /*
 551  *      Map a multicast IP onto multicast MAC for type ethernet.
 552  */
 553
 554 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
 555 {
 556         __u32 addr=ntohl(naddr);
 557         buf[0]=0x01;
 558         buf[1]=0x00;
 559         buf[2]=0x5e;
 560         buf[5]=addr&0xFF;
 561         addr>>=8;
 562         buf[4]=addr&0xFF;
 563         addr>>=8;
 564         buf[3]=addr&0x7F;
 565 }
 566
 567 /*
 568  *      Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
 569  *      Leave P_Key as 0 to be filled in by driver.
 570  */
 571
 572 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
 573 {
 574         __u32 addr;
 575         unsigned char scope = broadcast[5] & 0xF;
 576
 577         buf[0]  = 0;            /* Reserved */
 578         buf[1]  = 0xff;         /* Multicast QPN */
 579         buf[2]  = 0xff;
 580         buf[3]  = 0xff;
 581         addr    = ntohl(naddr);
 582         buf[4]  = 0xff;
 583         buf[5]  = 0x10 | scope; /* scope from broadcast address */
 584         buf[6]  = 0x40;         /* IPv4 signature */
 585         buf[7]  = 0x1b;
 586         buf[8]  = broadcast[8];         /* P_Key */
 587         buf[9]  = broadcast[9];
 588         buf[10] = 0;
 589         buf[11] = 0;
 590         buf[12] = 0;
 591         buf[13] = 0;
 592         buf[14] = 0;
 593         buf[15] = 0;
 594         buf[19] = addr & 0xff;
 595         addr  >>= 8;
 596         buf[18] = addr & 0xff;
 597         addr  >>= 8;
 598         buf[17] = addr & 0xff;
 599         addr  >>= 8;
 600         buf[16] = addr & 0x0f;
 601 }
 602
 603 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
 604 {
 605         if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
 606                 memcpy(buf, broadcast, 4);
 607         else
 608                 memcpy(buf, &naddr, sizeof(naddr));
 609 }
 610
 611 #if IS_ENABLED(CONFIG_IPV6)
 612 #include <linux/ipv6.h>
 613 #endif
 614
 615 static __inline__ void inet_reset_saddr(struct sock *sk)
 616 {
 617         inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
 618 #if IS_ENABLED(CONFIG_IPV6)
 619         if (sk->sk_family == PF_INET6) {
 620                 struct ipv6_pinfo *np = inet6_sk(sk);
 621
 622                 memset(&np->saddr, 0, sizeof(np->saddr));
 623                 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
 624         }
 625 #endif
 626 }
 627
 628 #endif
 629
 630 static inline unsigned int ipv4_addr_hash(__be32 ip)
 631 {
 632         return (__force unsigned int) ip;
 633 }
 634
 635 static inline u32 ipv4_portaddr_hash(const struct net *net,
 636                                      __be32 saddr,
 637                                      unsigned int port)
 638 {
 639         return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
 640 }
 641
 642 bool ip_call_ra_chain(struct sk_buff *skb);
 643
 644 /*
 645  *      Functions provided by ip_fragment.c
 646  */
 647
 648 enum ip_defrag_users {
 649         IP_DEFRAG_LOCAL_DELIVER,
 650         IP_DEFRAG_CALL_RA_CHAIN,
 651         IP_DEFRAG_CONNTRACK_IN,
 652         __IP_DEFRAG_CONNTRACK_IN_END    = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
 653         IP_DEFRAG_CONNTRACK_OUT,
 654         __IP_DEFRAG_CONNTRACK_OUT_END   = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
 655         IP_DEFRAG_CONNTRACK_BRIDGE_IN,
 656         __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
 657         IP_DEFRAG_VS_IN,
 658         IP_DEFRAG_VS_OUT,
 659         IP_DEFRAG_VS_FWD,
 660         IP_DEFRAG_AF_PACKET,
 661         IP_DEFRAG_MACVLAN,
 662 };
 663
 664 /* Return true if the value of 'user' is between 'lower_bond'
 665  * and 'upper_bond' inclusively.
 666  */
 667 static inline bool ip_defrag_user_in_between(u32 user,
 668                                              enum ip_defrag_users lower_bond,
 669                                              enum ip_defrag_users upper_bond)
 670 {
 671         return user >= lower_bond && user <= upper_bond;
 672 }
 673
 674 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
 675 #ifdef CONFIG_INET
 676 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
 677 #else
 678 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
 679 {
 680         return skb;
 681 }
 682 #endif
 683
 684 /*
 685  *      Functions provided by ip_forward.c
 686  */
 687
 688 int ip_forward(struct sk_buff *skb);
 689
 690 /*
 691  *      Functions provided by ip_options.c
 692  */
 693
 694 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
 695                       __be32 daddr, struct rtable *rt, int is_frag);
 696
 697 int __ip_options_echo(struct net *net, struct ip_options *dopt,
 698                       struct sk_buff *skb, const struct ip_options *sopt);
 699 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
 700                                   struct sk_buff *skb)
 701 {
 702         return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
 703 }
 704
 705 void ip_options_fragment(struct sk_buff *skb);
 706 int __ip_options_compile(struct net *net, struct ip_options *opt,
 707                          struct sk_buff *skb, __be32 *info);
 708 int ip_options_compile(struct net *net, struct ip_options *opt,
 709                        struct sk_buff *skb);
 710 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
 711                    sockptr_t data, int optlen);
 712 void ip_options_undo(struct ip_options *opt);
 713 void ip_forward_options(struct sk_buff *skb);
 714 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
 715
 716 /*
 717  *      Functions provided by ip_sockglue.c
 718  */
 719
 720 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
 721 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
 722                          struct sk_buff *skb, int tlen, int offset);
 723 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
 724                  struct ipcm_cookie *ipc, bool allow_ipv6);
 725 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
 726                   unsigned int optlen);
 727 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
 728                   int __user *optlen);
 729 int ip_ra_control(struct sock *sk, unsigned char on,
 730                   void (*destructor)(struct sock *));
 731
 732 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
 733 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
 734                    u32 info, u8 *payload);
 735 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
 736                     u32 info);
 737
 738 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
 739 {
 740         ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
 741 }
 742
 743 bool icmp_global_allow(void);
 744 extern int sysctl_icmp_msgs_per_sec;
 745 extern int sysctl_icmp_msgs_burst;
 746
 747 #ifdef CONFIG_PROC_FS
 748 int ip_misc_proc_init(void);
 749 #endif
 750
 751 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
 752                                 struct netlink_ext_ack *extack);
 753
 754 static inline bool inetdev_valid_mtu(unsigned int mtu)
 755 {
 756         return likely(mtu >= IPV4_MIN_MTU);
 757 }
 758
 759 void ip_sock_set_freebind(struct sock *sk);
 760 int ip_sock_set_mtu_discover(struct sock *sk, int val);
 761 void ip_sock_set_pktinfo(struct sock *sk);
 762 void ip_sock_set_recverr(struct sock *sk);
 763 void ip_sock_set_tos(struct sock *sk, int val);
 764
 765 #endif  /* _IP_H */