net/netfilter/nf_nat_proto.c

   1 // SPDX-License-Identifier: GPL-2.0-only
   2 /* (C) 1999-2001 Paul `Rusty' Russell
   3  * (C) 2002-2006 Netfilter Core Team <[email protected]>
   4  */
   5
   6 #include <linux/types.h>
   7 #include <linux/export.h>
   8 #include <linux/init.h>
   9 #include <linux/udp.h>
  10 #include <linux/tcp.h>
  11 #include <linux/icmp.h>
  12 #include <linux/icmpv6.h>
  13
  14 #include <linux/dccp.h>
  15 #include <linux/sctp.h>
  16 #include <net/sctp/checksum.h>
  17
  18 #include <linux/netfilter.h>
  19 #include <net/netfilter/nf_nat.h>
  20
  21 #include <linux/ipv6.h>
  22 #include <linux/netfilter_ipv6.h>
  23 #include <net/checksum.h>
  24 #include <net/ip6_checksum.h>
  25 #include <net/ip6_route.h>
  26 #include <net/xfrm.h>
  27 #include <net/ipv6.h>
  28
  29 #include <net/netfilter/nf_conntrack_core.h>
  30 #include <net/netfilter/nf_conntrack.h>
  31 #include <linux/netfilter/nfnetlink_conntrack.h>
  32
  33 static void nf_csum_update(struct sk_buff *skb,
  34                            unsigned int iphdroff, __sum16 *check,
  35                            const struct nf_conntrack_tuple *t,
  36                            enum nf_nat_manip_type maniptype);
  37
  38 static void
  39 __udp_manip_pkt(struct sk_buff *skb,
  40                 unsigned int iphdroff, struct udphdr *hdr,
  41                 const struct nf_conntrack_tuple *tuple,
  42                 enum nf_nat_manip_type maniptype, bool do_csum)
  43 {
  44         __be16 *portptr, newport;
  45
  46         if (maniptype == NF_NAT_MANIP_SRC) {
  47                 /* Get rid of src port */
  48                 newport = tuple->src.u.udp.port;
  49                 portptr = &hdr->source;
  50         } else {
  51                 /* Get rid of dst port */
  52                 newport = tuple->dst.u.udp.port;
  53                 portptr = &hdr->dest;
  54         }
  55         if (do_csum) {
  56                 nf_csum_update(skb, iphdroff, &hdr->check, tuple, maniptype);
  57                 inet_proto_csum_replace2(&hdr->check, skb, *portptr, newport,
  58                                          false);
  59                 if (!hdr->check)
  60                         hdr->check = CSUM_MANGLED_0;
  61         }
  62         *portptr = newport;
  63 }
  64
  65 static bool udp_manip_pkt(struct sk_buff *skb,
  66                           unsigned int iphdroff, unsigned int hdroff,
  67                           const struct nf_conntrack_tuple *tuple,
  68                           enum nf_nat_manip_type maniptype)
  69 {
  70         struct udphdr *hdr;
  71         bool do_csum;
  72
  73         if (skb_ensure_writable(skb, hdroff + sizeof(*hdr)))
  74                 return false;
  75
  76         hdr = (struct udphdr *)(skb->data + hdroff);
  77         do_csum = hdr->check || skb->ip_summed == CHECKSUM_PARTIAL;
  78
  79         __udp_manip_pkt(skb, iphdroff, hdr, tuple, maniptype, do_csum);
  80         return true;
  81 }
  82
  83 static bool udplite_manip_pkt(struct sk_buff *skb,
  84                               unsigned int iphdroff, unsigned int hdroff,
  85                               const struct nf_conntrack_tuple *tuple,
  86                               enum nf_nat_manip_type maniptype)
  87 {
  88 #ifdef CONFIG_NF_CT_PROTO_UDPLITE
  89         struct udphdr *hdr;
  90
  91         if (skb_ensure_writable(skb, hdroff + sizeof(*hdr)))
  92                 return false;
  93
  94         hdr = (struct udphdr *)(skb->data + hdroff);
  95         __udp_manip_pkt(skb, iphdroff, hdr, tuple, maniptype, true);
  96 #endif
  97         return true;
  98 }
  99
 100 static bool
 101 sctp_manip_pkt(struct sk_buff *skb,
 102                unsigned int iphdroff, unsigned int hdroff,
 103                const struct nf_conntrack_tuple *tuple,
 104                enum nf_nat_manip_type maniptype)
 105 {
 106 #ifdef CONFIG_NF_CT_PROTO_SCTP
 107         struct sctphdr *hdr;
 108         int hdrsize = 8;
 109
 110         /* This could be an inner header returned in imcp packet; in such
 111          * cases we cannot update the checksum field since it is outside
 112          * of the 8 bytes of transport layer headers we are guaranteed.
 113          */
 114         if (skb->len >= hdroff + sizeof(*hdr))
 115                 hdrsize = sizeof(*hdr);
 116
 117         if (skb_ensure_writable(skb, hdroff + hdrsize))
 118                 return false;
 119
 120         hdr = (struct sctphdr *)(skb->data + hdroff);
 121
 122         if (maniptype == NF_NAT_MANIP_SRC) {
 123                 /* Get rid of src port */
 124                 hdr->source = tuple->src.u.sctp.port;
 125         } else {
 126                 /* Get rid of dst port */
 127                 hdr->dest = tuple->dst.u.sctp.port;
 128         }
 129
 130         if (hdrsize < sizeof(*hdr))
 131                 return true;
 132
 133         if (skb->ip_summed != CHECKSUM_PARTIAL) {
 134                 hdr->checksum = sctp_compute_cksum(skb, hdroff);
 135                 skb->ip_summed = CHECKSUM_NONE;
 136         }
 137
 138 #endif
 139         return true;
 140 }
 141
 142 static bool
 143 tcp_manip_pkt(struct sk_buff *skb,
 144               unsigned int iphdroff, unsigned int hdroff,
 145               const struct nf_conntrack_tuple *tuple,
 146               enum nf_nat_manip_type maniptype)
 147 {
 148         struct tcphdr *hdr;
 149         __be16 *portptr, newport, oldport;
 150         int hdrsize = 8; /* TCP connection tracking guarantees this much */
 151
 152         /* this could be a inner header returned in icmp packet; in such
 153            cases we cannot update the checksum field since it is outside of
 154            the 8 bytes of transport layer headers we are guaranteed */
 155         if (skb->len >= hdroff + sizeof(struct tcphdr))
 156                 hdrsize = sizeof(struct tcphdr);
 157
 158         if (skb_ensure_writable(skb, hdroff + hdrsize))
 159                 return false;
 160
 161         hdr = (struct tcphdr *)(skb->data + hdroff);
 162
 163         if (maniptype == NF_NAT_MANIP_SRC) {
 164                 /* Get rid of src port */
 165                 newport = tuple->src.u.tcp.port;
 166                 portptr = &hdr->source;
 167         } else {
 168                 /* Get rid of dst port */
 169                 newport = tuple->dst.u.tcp.port;
 170                 portptr = &hdr->dest;
 171         }
 172
 173         oldport = *portptr;
 174         *portptr = newport;
 175
 176         if (hdrsize < sizeof(*hdr))
 177                 return true;
 178
 179         nf_csum_update(skb, iphdroff, &hdr->check, tuple, maniptype);
 180         inet_proto_csum_replace2(&hdr->check, skb, oldport, newport, false);
 181         return true;
 182 }
 183
 184 static bool
 185 dccp_manip_pkt(struct sk_buff *skb,
 186                unsigned int iphdroff, unsigned int hdroff,
 187                const struct nf_conntrack_tuple *tuple,
 188                enum nf_nat_manip_type maniptype)
 189 {
 190 #ifdef CONFIG_NF_CT_PROTO_DCCP
 191         struct dccp_hdr *hdr;
 192         __be16 *portptr, oldport, newport;
 193         int hdrsize = 8; /* DCCP connection tracking guarantees this much */
 194
 195         if (skb->len >= hdroff + sizeof(struct dccp_hdr))
 196                 hdrsize = sizeof(struct dccp_hdr);
 197
 198         if (skb_ensure_writable(skb, hdroff + hdrsize))
 199                 return false;
 200
 201         hdr = (struct dccp_hdr *)(skb->data + hdroff);
 202
 203         if (maniptype == NF_NAT_MANIP_SRC) {
 204                 newport = tuple->src.u.dccp.port;
 205                 portptr = &hdr->dccph_sport;
 206         } else {
 207                 newport = tuple->dst.u.dccp.port;
 208                 portptr = &hdr->dccph_dport;
 209         }
 210
 211         oldport = *portptr;
 212         *portptr = newport;
 213
 214         if (hdrsize < sizeof(*hdr))
 215                 return true;
 216
 217         nf_csum_update(skb, iphdroff, &hdr->dccph_checksum, tuple, maniptype);
 218         inet_proto_csum_replace2(&hdr->dccph_checksum, skb, oldport, newport,
 219                                  false);
 220 #endif
 221         return true;
 222 }
 223
 224 static bool
 225 icmp_manip_pkt(struct sk_buff *skb,
 226                unsigned int iphdroff, unsigned int hdroff,
 227                const struct nf_conntrack_tuple *tuple,
 228                enum nf_nat_manip_type maniptype)
 229 {
 230         struct icmphdr *hdr;
 231
 232         if (skb_ensure_writable(skb, hdroff + sizeof(*hdr)))
 233                 return false;
 234
 235         hdr = (struct icmphdr *)(skb->data + hdroff);
 236         switch (hdr->type) {
 237         case ICMP_ECHO:
 238         case ICMP_ECHOREPLY:
 239         case ICMP_TIMESTAMP:
 240         case ICMP_TIMESTAMPREPLY:
 241         case ICMP_INFO_REQUEST:
 242         case ICMP_INFO_REPLY:
 243         case ICMP_ADDRESS:
 244         case ICMP_ADDRESSREPLY:
 245                 break;
 246         default:
 247                 return true;
 248         }
 249         inet_proto_csum_replace2(&hdr->checksum, skb,
 250                                  hdr->un.echo.id, tuple->src.u.icmp.id, false);
 251         hdr->un.echo.id = tuple->src.u.icmp.id;
 252         return true;
 253 }
 254
 255 static bool
 256 icmpv6_manip_pkt(struct sk_buff *skb,
 257                  unsigned int iphdroff, unsigned int hdroff,
 258                  const struct nf_conntrack_tuple *tuple,
 259                  enum nf_nat_manip_type maniptype)
 260 {
 261         struct icmp6hdr *hdr;
 262
 263         if (skb_ensure_writable(skb, hdroff + sizeof(*hdr)))
 264                 return false;
 265
 266         hdr = (struct icmp6hdr *)(skb->data + hdroff);
 267         nf_csum_update(skb, iphdroff, &hdr->icmp6_cksum, tuple, maniptype);
 268         if (hdr->icmp6_type == ICMPV6_ECHO_REQUEST ||
 269             hdr->icmp6_type == ICMPV6_ECHO_REPLY) {
 270                 inet_proto_csum_replace2(&hdr->icmp6_cksum, skb,
 271                                          hdr->icmp6_identifier,
 272                                          tuple->src.u.icmp.id, false);
 273                 hdr->icmp6_identifier = tuple->src.u.icmp.id;
 274         }
 275         return true;
 276 }
 277
 278 /* manipulate a GRE packet according to maniptype */
 279 static bool
 280 gre_manip_pkt(struct sk_buff *skb,
 281               unsigned int iphdroff, unsigned int hdroff,
 282               const struct nf_conntrack_tuple *tuple,
 283               enum nf_nat_manip_type maniptype)
 284 {
 285 #if IS_ENABLED(CONFIG_NF_CT_PROTO_GRE)
 286         const struct gre_base_hdr *greh;
 287         struct pptp_gre_header *pgreh;
 288
 289         /* pgreh includes two optional 32bit fields which are not required
 290          * to be there.  That's where the magic '8' comes from */
 291         if (skb_ensure_writable(skb, hdroff + sizeof(*pgreh) - 8))
 292                 return false;
 293
 294         greh = (void *)skb->data + hdroff;
 295         pgreh = (struct pptp_gre_header *)greh;
 296
 297         /* we only have destination manip of a packet, since 'source key'
 298          * is not present in the packet itself */
 299         if (maniptype != NF_NAT_MANIP_DST)
 300                 return true;
 301
 302         switch (greh->flags & GRE_VERSION) {
 303         case GRE_VERSION_0:
 304                 /* We do not currently NAT any GREv0 packets.
 305                  * Try to behave like "nf_nat_proto_unknown" */
 306                 break;
 307         case GRE_VERSION_1:
 308                 pr_debug("call_id -> 0x%04x\n", ntohs(tuple->dst.u.gre.key));
 309                 pgreh->call_id = tuple->dst.u.gre.key;
 310                 break;
 311         default:
 312                 pr_debug("can't nat unknown GRE version\n");
 313                 return false;
 314         }
 315 #endif
 316         return true;
 317 }
 318
 319 static bool l4proto_manip_pkt(struct sk_buff *skb,
 320                               unsigned int iphdroff, unsigned int hdroff,
 321                               const struct nf_conntrack_tuple *tuple,
 322                               enum nf_nat_manip_type maniptype)
 323 {
 324         switch (tuple->dst.protonum) {
 325         case IPPROTO_TCP:
 326                 return tcp_manip_pkt(skb, iphdroff, hdroff,
 327                                      tuple, maniptype);
 328         case IPPROTO_UDP:
 329                 return udp_manip_pkt(skb, iphdroff, hdroff,
 330                                      tuple, maniptype);
 331         case IPPROTO_UDPLITE:
 332                 return udplite_manip_pkt(skb, iphdroff, hdroff,
 333                                          tuple, maniptype);
 334         case IPPROTO_SCTP:
 335                 return sctp_manip_pkt(skb, iphdroff, hdroff,
 336                                       tuple, maniptype);
 337         case IPPROTO_ICMP:
 338                 return icmp_manip_pkt(skb, iphdroff, hdroff,
 339                                       tuple, maniptype);
 340         case IPPROTO_ICMPV6:
 341                 return icmpv6_manip_pkt(skb, iphdroff, hdroff,
 342                                         tuple, maniptype);
 343         case IPPROTO_DCCP:
 344                 return dccp_manip_pkt(skb, iphdroff, hdroff,
 345                                       tuple, maniptype);
 346         case IPPROTO_GRE:
 347                 return gre_manip_pkt(skb, iphdroff, hdroff,
 348                                      tuple, maniptype);
 349         }
 350
 351         /* If we don't know protocol -- no error, pass it unmodified. */
 352         return true;
 353 }
 354
 355 static bool nf_nat_ipv4_manip_pkt(struct sk_buff *skb,
 356                                   unsigned int iphdroff,
 357                                   const struct nf_conntrack_tuple *target,
 358                                   enum nf_nat_manip_type maniptype)
 359 {
 360         struct iphdr *iph;
 361         unsigned int hdroff;
 362
 363         if (skb_ensure_writable(skb, iphdroff + sizeof(*iph)))
 364                 return false;
 365
 366         iph = (void *)skb->data + iphdroff;
 367         hdroff = iphdroff + iph->ihl * 4;
 368
 369         if (!l4proto_manip_pkt(skb, iphdroff, hdroff, target, maniptype))
 370                 return false;
 371         iph = (void *)skb->data + iphdroff;
 372
 373         if (maniptype == NF_NAT_MANIP_SRC) {
 374                 csum_replace4(&iph->check, iph->saddr, target->src.u3.ip);
 375                 iph->saddr = target->src.u3.ip;
 376         } else {
 377                 csum_replace4(&iph->check, iph->daddr, target->dst.u3.ip);
 378                 iph->daddr = target->dst.u3.ip;
 379         }
 380         return true;
 381 }
 382
 383 static bool nf_nat_ipv6_manip_pkt(struct sk_buff *skb,
 384                                   unsigned int iphdroff,
 385                                   const struct nf_conntrack_tuple *target,
 386                                   enum nf_nat_manip_type maniptype)
 387 {
 388 #if IS_ENABLED(CONFIG_IPV6)
 389         struct ipv6hdr *ipv6h;
 390         __be16 frag_off;
 391         int hdroff;
 392         u8 nexthdr;
 393
 394         if (skb_ensure_writable(skb, iphdroff + sizeof(*ipv6h)))
 395                 return false;
 396
 397         ipv6h = (void *)skb->data + iphdroff;
 398         nexthdr = ipv6h->nexthdr;
 399         hdroff = ipv6_skip_exthdr(skb, iphdroff + sizeof(*ipv6h),
 400                                   &nexthdr, &frag_off);
 401         if (hdroff < 0)
 402                 goto manip_addr;
 403
 404         if ((frag_off & htons(~0x7)) == 0 &&
 405             !l4proto_manip_pkt(skb, iphdroff, hdroff, target, maniptype))
 406                 return false;
 407
 408         /* must reload, offset might have changed */
 409         ipv6h = (void *)skb->data + iphdroff;
 410
 411 manip_addr:
 412         if (maniptype == NF_NAT_MANIP_SRC)
 413                 ipv6h->saddr = target->src.u3.in6;
 414         else
 415                 ipv6h->daddr = target->dst.u3.in6;
 416
 417 #endif
 418         return true;
 419 }
 420
 421 unsigned int nf_nat_manip_pkt(struct sk_buff *skb, struct nf_conn *ct,
 422                               enum nf_nat_manip_type mtype,
 423                               enum ip_conntrack_dir dir)
 424 {
 425         struct nf_conntrack_tuple target;
 426
 427         /* We are aiming to look like inverse of other direction. */
 428         nf_ct_invert_tuple(&target, &ct->tuplehash[!dir].tuple);
 429
 430         switch (target.src.l3num) {
 431         case NFPROTO_IPV6:
 432                 if (nf_nat_ipv6_manip_pkt(skb, 0, &target, mtype))
 433                         return NF_ACCEPT;
 434                 break;
 435         case NFPROTO_IPV4:
 436                 if (nf_nat_ipv4_manip_pkt(skb, 0, &target, mtype))
 437                         return NF_ACCEPT;
 438                 break;
 439         default:
 440                 WARN_ON_ONCE(1);
 441                 break;
 442         }
 443
 444         return NF_DROP;
 445 }
 446
 447 static void nf_nat_ipv4_csum_update(struct sk_buff *skb,
 448                                     unsigned int iphdroff, __sum16 *check,
 449                                     const struct nf_conntrack_tuple *t,
 450                                     enum nf_nat_manip_type maniptype)
 451 {
 452         struct iphdr *iph = (struct iphdr *)(skb->data + iphdroff);
 453         __be32 oldip, newip;
 454
 455         if (maniptype == NF_NAT_MANIP_SRC) {
 456                 oldip = iph->saddr;
 457                 newip = t->src.u3.ip;
 458         } else {
 459                 oldip = iph->daddr;
 460                 newip = t->dst.u3.ip;
 461         }
 462         inet_proto_csum_replace4(check, skb, oldip, newip, true);
 463 }
 464
 465 static void nf_nat_ipv6_csum_update(struct sk_buff *skb,
 466                                     unsigned int iphdroff, __sum16 *check,
 467                                     const struct nf_conntrack_tuple *t,
 468                                     enum nf_nat_manip_type maniptype)
 469 {
 470 #if IS_ENABLED(CONFIG_IPV6)
 471         const struct ipv6hdr *ipv6h = (struct ipv6hdr *)(skb->data + iphdroff);
 472         const struct in6_addr *oldip, *newip;
 473
 474         if (maniptype == NF_NAT_MANIP_SRC) {
 475                 oldip = &ipv6h->saddr;
 476                 newip = &t->src.u3.in6;
 477         } else {
 478                 oldip = &ipv6h->daddr;
 479                 newip = &t->dst.u3.in6;
 480         }
 481         inet_proto_csum_replace16(check, skb, oldip->s6_addr32,
 482                                   newip->s6_addr32, true);
 483 #endif
 484 }
 485
 486 static void nf_csum_update(struct sk_buff *skb,
 487                            unsigned int iphdroff, __sum16 *check,
 488                            const struct nf_conntrack_tuple *t,
 489                            enum nf_nat_manip_type maniptype)
 490 {
 491         switch (t->src.l3num) {
 492         case NFPROTO_IPV4:
 493                 nf_nat_ipv4_csum_update(skb, iphdroff, check, t, maniptype);
 494                 return;
 495         case NFPROTO_IPV6:
 496                 nf_nat_ipv6_csum_update(skb, iphdroff, check, t, maniptype);
 497                 return;
 498         }
 499 }
 500
 501 static void nf_nat_ipv4_csum_recalc(struct sk_buff *skb,
 502                                     u8 proto, void *data, __sum16 *check,
 503                                     int datalen, int oldlen)
 504 {
 505         if (skb->ip_summed != CHECKSUM_PARTIAL) {
 506                 const struct iphdr *iph = ip_hdr(skb);
 507
 508                 skb->ip_summed = CHECKSUM_PARTIAL;
 509                 skb->csum_start = skb_headroom(skb) + skb_network_offset(skb) +
 510                         ip_hdrlen(skb);
 511                 skb->csum_offset = (void *)check - data;
 512                 *check = ~csum_tcpudp_magic(iph->saddr, iph->daddr, datalen,
 513                                             proto, 0);
 514         } else {
 515                 inet_proto_csum_replace2(check, skb,
 516                                          htons(oldlen), htons(datalen), true);
 517         }
 518 }
 519
 520 #if IS_ENABLED(CONFIG_IPV6)
 521 static void nf_nat_ipv6_csum_recalc(struct sk_buff *skb,
 522                                     u8 proto, void *data, __sum16 *check,
 523                                     int datalen, int oldlen)
 524 {
 525         if (skb->ip_summed != CHECKSUM_PARTIAL) {
 526                 const struct ipv6hdr *ipv6h = ipv6_hdr(skb);
 527
 528                 skb->ip_summed = CHECKSUM_PARTIAL;
 529                 skb->csum_start = skb_headroom(skb) + skb_network_offset(skb) +
 530                         (data - (void *)skb->data);
 531                 skb->csum_offset = (void *)check - data;
 532                 *check = ~csum_ipv6_magic(&ipv6h->saddr, &ipv6h->daddr,
 533                                           datalen, proto, 0);
 534         } else {
 535                 inet_proto_csum_replace2(check, skb,
 536                                          htons(oldlen), htons(datalen), true);
 537         }
 538 }
 539 #endif
 540
 541 void nf_nat_csum_recalc(struct sk_buff *skb,
 542                         u8 nfproto, u8 proto, void *data, __sum16 *check,
 543                         int datalen, int oldlen)
 544 {
 545         switch (nfproto) {
 546         case NFPROTO_IPV4:
 547                 nf_nat_ipv4_csum_recalc(skb, proto, data, check,
 548                                         datalen, oldlen);
 549                 return;
 550 #if IS_ENABLED(CONFIG_IPV6)
 551         case NFPROTO_IPV6:
 552                 nf_nat_ipv6_csum_recalc(skb, proto, data, check,
 553                                         datalen, oldlen);
 554                 return;
 555 #endif
 556         }
 557
 558         WARN_ON_ONCE(1);
 559 }
 560
 561 int nf_nat_icmp_reply_translation(struct sk_buff *skb,
 562                                   struct nf_conn *ct,
 563                                   enum ip_conntrack_info ctinfo,
 564                                   unsigned int hooknum)
 565 {
 566         struct {
 567                 struct icmphdr  icmp;
 568                 struct iphdr    ip;
 569         } *inside;
 570         enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
 571         enum nf_nat_manip_type manip = HOOK2MANIP(hooknum);
 572         unsigned int hdrlen = ip_hdrlen(skb);
 573         struct nf_conntrack_tuple target;
 574         unsigned long statusbit;
 575
 576         WARN_ON(ctinfo != IP_CT_RELATED && ctinfo != IP_CT_RELATED_REPLY);
 577
 578         if (skb_ensure_writable(skb, hdrlen + sizeof(*inside)))
 579                 return 0;
 580         if (nf_ip_checksum(skb, hooknum, hdrlen, IPPROTO_ICMP))
 581                 return 0;
 582
 583         inside = (void *)skb->data + hdrlen;
 584         if (inside->icmp.type == ICMP_REDIRECT) {
 585                 if ((ct->status & IPS_NAT_DONE_MASK) != IPS_NAT_DONE_MASK)
 586                         return 0;
 587                 if (ct->status & IPS_NAT_MASK)
 588                         return 0;
 589         }
 590
 591         if (manip == NF_NAT_MANIP_SRC)
 592                 statusbit = IPS_SRC_NAT;
 593         else
 594                 statusbit = IPS_DST_NAT;
 595
 596         /* Invert if this is reply direction */
 597         if (dir == IP_CT_DIR_REPLY)
 598                 statusbit ^= IPS_NAT_MASK;
 599
 600         if (!(ct->status & statusbit))
 601                 return 1;
 602
 603         if (!nf_nat_ipv4_manip_pkt(skb, hdrlen + sizeof(inside->icmp),
 604                                    &ct->tuplehash[!dir].tuple, !manip))
 605                 return 0;
 606
 607         if (skb->ip_summed != CHECKSUM_PARTIAL) {
 608                 /* Reloading "inside" here since manip_pkt may reallocate */
 609                 inside = (void *)skb->data + hdrlen;
 610                 inside->icmp.checksum = 0;
 611                 inside->icmp.checksum =
 612                         csum_fold(skb_checksum(skb, hdrlen,
 613                                                skb->len - hdrlen, 0));
 614         }
 615
 616         /* Change outer to look like the reply to an incoming packet */
 617         nf_ct_invert_tuple(&target, &ct->tuplehash[!dir].tuple);
 618         target.dst.protonum = IPPROTO_ICMP;
 619         if (!nf_nat_ipv4_manip_pkt(skb, 0, &target, manip))
 620                 return 0;
 621
 622         return 1;
 623 }
 624 EXPORT_SYMBOL_GPL(nf_nat_icmp_reply_translation);
 625
 626 static unsigned int
 627 nf_nat_ipv4_fn(void *priv, struct sk_buff *skb,
 628                const struct nf_hook_state *state)
 629 {
 630         struct nf_conn *ct;
 631         enum ip_conntrack_info ctinfo;
 632
 633         ct = nf_ct_get(skb, &ctinfo);
 634         if (!ct)
 635                 return NF_ACCEPT;
 636
 637         if (ctinfo == IP_CT_RELATED || ctinfo == IP_CT_RELATED_REPLY) {
 638                 if (ip_hdr(skb)->protocol == IPPROTO_ICMP) {
 639                         if (!nf_nat_icmp_reply_translation(skb, ct, ctinfo,
 640                                                            state->hook))
 641                                 return NF_DROP;
 642                         else
 643                                 return NF_ACCEPT;
 644                 }
 645         }
 646
 647         return nf_nat_inet_fn(priv, skb, state);
 648 }
 649
 650 static unsigned int
 651 nf_nat_ipv4_in(void *priv, struct sk_buff *skb,
 652                const struct nf_hook_state *state)
 653 {
 654         unsigned int ret;
 655         __be32 daddr = ip_hdr(skb)->daddr;
 656
 657         ret = nf_nat_ipv4_fn(priv, skb, state);
 658         if (ret == NF_ACCEPT && daddr != ip_hdr(skb)->daddr)
 659                 skb_dst_drop(skb);
 660
 661         return ret;
 662 }
 663
 664 static unsigned int
 665 nf_nat_ipv4_out(void *priv, struct sk_buff *skb,
 666                 const struct nf_hook_state *state)
 667 {
 668 #ifdef CONFIG_XFRM
 669         const struct nf_conn *ct;
 670         enum ip_conntrack_info ctinfo;
 671         int err;
 672 #endif
 673         unsigned int ret;
 674
 675         ret = nf_nat_ipv4_fn(priv, skb, state);
 676 #ifdef CONFIG_XFRM
 677         if (ret != NF_ACCEPT)
 678                 return ret;
 679
 680         if (IPCB(skb)->flags & IPSKB_XFRM_TRANSFORMED)
 681                 return ret;
 682
 683         ct = nf_ct_get(skb, &ctinfo);
 684         if (ct) {
 685                 enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
 686
 687                 if (ct->tuplehash[dir].tuple.src.u3.ip !=
 688                      ct->tuplehash[!dir].tuple.dst.u3.ip ||
 689                     (ct->tuplehash[dir].tuple.dst.protonum != IPPROTO_ICMP &&
 690                      ct->tuplehash[dir].tuple.src.u.all !=
 691                      ct->tuplehash[!dir].tuple.dst.u.all)) {
 692                         err = nf_xfrm_me_harder(state->net, skb, AF_INET);
 693                         if (err < 0)
 694                                 ret = NF_DROP_ERR(err);
 695                 }
 696         }
 697 #endif
 698         return ret;
 699 }
 700
 701 static unsigned int
 702 nf_nat_ipv4_local_fn(void *priv, struct sk_buff *skb,
 703                      const struct nf_hook_state *state)
 704 {
 705         const struct nf_conn *ct;
 706         enum ip_conntrack_info ctinfo;
 707         unsigned int ret;
 708         int err;
 709
 710         ret = nf_nat_ipv4_fn(priv, skb, state);
 711         if (ret != NF_ACCEPT)
 712                 return ret;
 713
 714         ct = nf_ct_get(skb, &ctinfo);
 715         if (ct) {
 716                 enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
 717
 718                 if (ct->tuplehash[dir].tuple.dst.u3.ip !=
 719                     ct->tuplehash[!dir].tuple.src.u3.ip) {
 720                         err = ip_route_me_harder(state->net, skb, RTN_UNSPEC);
 721                         if (err < 0)
 722                                 ret = NF_DROP_ERR(err);
 723                 }
 724 #ifdef CONFIG_XFRM
 725                 else if (!(IPCB(skb)->flags & IPSKB_XFRM_TRANSFORMED) &&
 726                          ct->tuplehash[dir].tuple.dst.protonum != IPPROTO_ICMP &&
 727                          ct->tuplehash[dir].tuple.dst.u.all !=
 728                          ct->tuplehash[!dir].tuple.src.u.all) {
 729                         err = nf_xfrm_me_harder(state->net, skb, AF_INET);
 730                         if (err < 0)
 731                                 ret = NF_DROP_ERR(err);
 732                 }
 733 #endif
 734         }
 735         return ret;
 736 }
 737
 738 static const struct nf_hook_ops nf_nat_ipv4_ops[] = {
 739         /* Before packet filtering, change destination */
 740         {
 741                 .hook           = nf_nat_ipv4_in,
 742                 .pf             = NFPROTO_IPV4,
 743                 .hooknum        = NF_INET_PRE_ROUTING,
 744                 .priority       = NF_IP_PRI_NAT_DST,
 745         },
 746         /* After packet filtering, change source */
 747         {
 748                 .hook           = nf_nat_ipv4_out,
 749                 .pf             = NFPROTO_IPV4,
 750                 .hooknum        = NF_INET_POST_ROUTING,
 751                 .priority       = NF_IP_PRI_NAT_SRC,
 752         },
 753         /* Before packet filtering, change destination */
 754         {
 755                 .hook           = nf_nat_ipv4_local_fn,
 756                 .pf             = NFPROTO_IPV4,
 757                 .hooknum        = NF_INET_LOCAL_OUT,
 758                 .priority       = NF_IP_PRI_NAT_DST,
 759         },
 760         /* After packet filtering, change source */
 761         {
 762                 .hook           = nf_nat_ipv4_fn,
 763                 .pf             = NFPROTO_IPV4,
 764                 .hooknum        = NF_INET_LOCAL_IN,
 765                 .priority       = NF_IP_PRI_NAT_SRC,
 766         },
 767 };
 768
 769 int nf_nat_ipv4_register_fn(struct net *net, const struct nf_hook_ops *ops)
 770 {
 771         return nf_nat_register_fn(net, ops->pf, ops, nf_nat_ipv4_ops,
 772                                   ARRAY_SIZE(nf_nat_ipv4_ops));
 773 }
 774 EXPORT_SYMBOL_GPL(nf_nat_ipv4_register_fn);
 775
 776 void nf_nat_ipv4_unregister_fn(struct net *net, const struct nf_hook_ops *ops)
 777 {
 778         nf_nat_unregister_fn(net, ops->pf, ops, ARRAY_SIZE(nf_nat_ipv4_ops));
 779 }
 780 EXPORT_SYMBOL_GPL(nf_nat_ipv4_unregister_fn);
 781
 782 #if IS_ENABLED(CONFIG_IPV6)
 783 int nf_nat_icmpv6_reply_translation(struct sk_buff *skb,
 784                                     struct nf_conn *ct,
 785                                     enum ip_conntrack_info ctinfo,
 786                                     unsigned int hooknum,
 787                                     unsigned int hdrlen)
 788 {
 789         struct {
 790                 struct icmp6hdr icmp6;
 791                 struct ipv6hdr  ip6;
 792         } *inside;
 793         enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
 794         enum nf_nat_manip_type manip = HOOK2MANIP(hooknum);
 795         struct nf_conntrack_tuple target;
 796         unsigned long statusbit;
 797
 798         WARN_ON(ctinfo != IP_CT_RELATED && ctinfo != IP_CT_RELATED_REPLY);
 799
 800         if (skb_ensure_writable(skb, hdrlen + sizeof(*inside)))
 801                 return 0;
 802         if (nf_ip6_checksum(skb, hooknum, hdrlen, IPPROTO_ICMPV6))
 803                 return 0;
 804
 805         inside = (void *)skb->data + hdrlen;
 806         if (inside->icmp6.icmp6_type == NDISC_REDIRECT) {
 807                 if ((ct->status & IPS_NAT_DONE_MASK) != IPS_NAT_DONE_MASK)
 808                         return 0;
 809                 if (ct->status & IPS_NAT_MASK)
 810                         return 0;
 811         }
 812
 813         if (manip == NF_NAT_MANIP_SRC)
 814                 statusbit = IPS_SRC_NAT;
 815         else
 816                 statusbit = IPS_DST_NAT;
 817
 818         /* Invert if this is reply direction */
 819         if (dir == IP_CT_DIR_REPLY)
 820                 statusbit ^= IPS_NAT_MASK;
 821
 822         if (!(ct->status & statusbit))
 823                 return 1;
 824
 825         if (!nf_nat_ipv6_manip_pkt(skb, hdrlen + sizeof(inside->icmp6),
 826                                    &ct->tuplehash[!dir].tuple, !manip))
 827                 return 0;
 828
 829         if (skb->ip_summed != CHECKSUM_PARTIAL) {
 830                 struct ipv6hdr *ipv6h = ipv6_hdr(skb);
 831
 832                 inside = (void *)skb->data + hdrlen;
 833                 inside->icmp6.icmp6_cksum = 0;
 834                 inside->icmp6.icmp6_cksum =
 835                         csum_ipv6_magic(&ipv6h->saddr, &ipv6h->daddr,
 836                                         skb->len - hdrlen, IPPROTO_ICMPV6,
 837                                         skb_checksum(skb, hdrlen,
 838                                                      skb->len - hdrlen, 0));
 839         }
 840
 841         nf_ct_invert_tuple(&target, &ct->tuplehash[!dir].tuple);
 842         target.dst.protonum = IPPROTO_ICMPV6;
 843         if (!nf_nat_ipv6_manip_pkt(skb, 0, &target, manip))
 844                 return 0;
 845
 846         return 1;
 847 }
 848 EXPORT_SYMBOL_GPL(nf_nat_icmpv6_reply_translation);
 849
 850 static unsigned int
 851 nf_nat_ipv6_fn(void *priv, struct sk_buff *skb,
 852                const struct nf_hook_state *state)
 853 {
 854         struct nf_conn *ct;
 855         enum ip_conntrack_info ctinfo;
 856         __be16 frag_off;
 857         int hdrlen;
 858         u8 nexthdr;
 859
 860         ct = nf_ct_get(skb, &ctinfo);
 861         /* Can't track?  It's not due to stress, or conntrack would
 862          * have dropped it.  Hence it's the user's responsibilty to
 863          * packet filter it out, or implement conntrack/NAT for that
 864          * protocol. 8) --RR
 865          */
 866         if (!ct)
 867                 return NF_ACCEPT;
 868
 869         if (ctinfo == IP_CT_RELATED || ctinfo == IP_CT_RELATED_REPLY) {
 870                 nexthdr = ipv6_hdr(skb)->nexthdr;
 871                 hdrlen = ipv6_skip_exthdr(skb, sizeof(struct ipv6hdr),
 872                                           &nexthdr, &frag_off);
 873
 874                 if (hdrlen >= 0 && nexthdr == IPPROTO_ICMPV6) {
 875                         if (!nf_nat_icmpv6_reply_translation(skb, ct, ctinfo,
 876                                                              state->hook,
 877                                                              hdrlen))
 878                                 return NF_DROP;
 879                         else
 880                                 return NF_ACCEPT;
 881                 }
 882         }
 883
 884         return nf_nat_inet_fn(priv, skb, state);
 885 }
 886
 887 static unsigned int
 888 nf_nat_ipv6_in(void *priv, struct sk_buff *skb,
 889                const struct nf_hook_state *state)
 890 {
 891         unsigned int ret;
 892         struct in6_addr daddr = ipv6_hdr(skb)->daddr;
 893
 894         ret = nf_nat_ipv6_fn(priv, skb, state);
 895         if (ret != NF_DROP && ret != NF_STOLEN &&
 896             ipv6_addr_cmp(&daddr, &ipv6_hdr(skb)->daddr))
 897                 skb_dst_drop(skb);
 898
 899         return ret;
 900 }
 901
 902 static unsigned int
 903 nf_nat_ipv6_out(void *priv, struct sk_buff *skb,
 904                 const struct nf_hook_state *state)
 905 {
 906 #ifdef CONFIG_XFRM
 907         const struct nf_conn *ct;
 908         enum ip_conntrack_info ctinfo;
 909         int err;
 910 #endif
 911         unsigned int ret;
 912
 913         ret = nf_nat_ipv6_fn(priv, skb, state);
 914 #ifdef CONFIG_XFRM
 915         if (ret != NF_ACCEPT)
 916                 return ret;
 917
 918         if (IP6CB(skb)->flags & IP6SKB_XFRM_TRANSFORMED)
 919                 return ret;
 920         ct = nf_ct_get(skb, &ctinfo);
 921         if (ct) {
 922                 enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
 923
 924                 if (!nf_inet_addr_cmp(&ct->tuplehash[dir].tuple.src.u3,
 925                                       &ct->tuplehash[!dir].tuple.dst.u3) ||
 926                     (ct->tuplehash[dir].tuple.dst.protonum != IPPROTO_ICMPV6 &&
 927                      ct->tuplehash[dir].tuple.src.u.all !=
 928                      ct->tuplehash[!dir].tuple.dst.u.all)) {
 929                         err = nf_xfrm_me_harder(state->net, skb, AF_INET6);
 930                         if (err < 0)
 931                                 ret = NF_DROP_ERR(err);
 932                 }
 933         }
 934 #endif
 935
 936         return ret;
 937 }
 938
 939 static unsigned int
 940 nf_nat_ipv6_local_fn(void *priv, struct sk_buff *skb,
 941                      const struct nf_hook_state *state)
 942 {
 943         const struct nf_conn *ct;
 944         enum ip_conntrack_info ctinfo;
 945         unsigned int ret;
 946         int err;
 947
 948         ret = nf_nat_ipv6_fn(priv, skb, state);
 949         if (ret != NF_ACCEPT)
 950                 return ret;
 951
 952         ct = nf_ct_get(skb, &ctinfo);
 953         if (ct) {
 954                 enum ip_conntrack_dir dir = CTINFO2DIR(ctinfo);
 955
 956                 if (!nf_inet_addr_cmp(&ct->tuplehash[dir].tuple.dst.u3,
 957                                       &ct->tuplehash[!dir].tuple.src.u3)) {
 958                         err = nf_ip6_route_me_harder(state->net, skb);
 959                         if (err < 0)
 960                                 ret = NF_DROP_ERR(err);
 961                 }
 962 #ifdef CONFIG_XFRM
 963                 else if (!(IP6CB(skb)->flags & IP6SKB_XFRM_TRANSFORMED) &&
 964                          ct->tuplehash[dir].tuple.dst.protonum != IPPROTO_ICMPV6 &&
 965                          ct->tuplehash[dir].tuple.dst.u.all !=
 966                          ct->tuplehash[!dir].tuple.src.u.all) {
 967                         err = nf_xfrm_me_harder(state->net, skb, AF_INET6);
 968                         if (err < 0)
 969                                 ret = NF_DROP_ERR(err);
 970                 }
 971 #endif
 972         }
 973
 974         return ret;
 975 }
 976
 977 static const struct nf_hook_ops nf_nat_ipv6_ops[] = {
 978         /* Before packet filtering, change destination */
 979         {
 980                 .hook           = nf_nat_ipv6_in,
 981                 .pf             = NFPROTO_IPV6,
 982                 .hooknum        = NF_INET_PRE_ROUTING,
 983                 .priority       = NF_IP6_PRI_NAT_DST,
 984         },
 985         /* After packet filtering, change source */
 986         {
 987                 .hook           = nf_nat_ipv6_out,
 988                 .pf             = NFPROTO_IPV6,
 989                 .hooknum        = NF_INET_POST_ROUTING,
 990                 .priority       = NF_IP6_PRI_NAT_SRC,
 991         },
 992         /* Before packet filtering, change destination */
 993         {
 994                 .hook           = nf_nat_ipv6_local_fn,
 995                 .pf             = NFPROTO_IPV6,
 996                 .hooknum        = NF_INET_LOCAL_OUT,
 997                 .priority       = NF_IP6_PRI_NAT_DST,
 998         },
 999         /* After packet filtering, change source */
1000         {
1001                 .hook           = nf_nat_ipv6_fn,
1002                 .pf             = NFPROTO_IPV6,
1003                 .hooknum        = NF_INET_LOCAL_IN,
1004                 .priority       = NF_IP6_PRI_NAT_SRC,
1005         },
1006 };
1007
1008 int nf_nat_ipv6_register_fn(struct net *net, const struct nf_hook_ops *ops)
1009 {
1010         return nf_nat_register_fn(net, ops->pf, ops, nf_nat_ipv6_ops,
1011                                   ARRAY_SIZE(nf_nat_ipv6_ops));
1012 }
1013 EXPORT_SYMBOL_GPL(nf_nat_ipv6_register_fn);
1014
1015 void nf_nat_ipv6_unregister_fn(struct net *net, const struct nf_hook_ops *ops)
1016 {
1017         nf_nat_unregister_fn(net, ops->pf, ops, ARRAY_SIZE(nf_nat_ipv6_ops));
1018 }
1019 EXPORT_SYMBOL_GPL(nf_nat_ipv6_unregister_fn);
1020 #endif /* CONFIG_IPV6 */
1021
1022 #if defined(CONFIG_NF_TABLES_INET) && IS_ENABLED(CONFIG_NFT_NAT)
1023 int nf_nat_inet_register_fn(struct net *net, const struct nf_hook_ops *ops)
1024 {
1025         int ret;
1026
1027         if (WARN_ON_ONCE(ops->pf != NFPROTO_INET))
1028                 return -EINVAL;
1029
1030         ret = nf_nat_register_fn(net, NFPROTO_IPV6, ops, nf_nat_ipv6_ops,
1031                                  ARRAY_SIZE(nf_nat_ipv6_ops));
1032         if (ret)
1033                 return ret;
1034
1035         ret = nf_nat_register_fn(net, NFPROTO_IPV4, ops, nf_nat_ipv4_ops,
1036                                  ARRAY_SIZE(nf_nat_ipv4_ops));
1037         if (ret)
1038                 nf_nat_ipv6_unregister_fn(net, ops);
1039
1040         return ret;
1041 }
1042 EXPORT_SYMBOL_GPL(nf_nat_inet_register_fn);
1043
1044 void nf_nat_inet_unregister_fn(struct net *net, const struct nf_hook_ops *ops)
1045 {
1046         nf_nat_unregister_fn(net, NFPROTO_IPV4, ops, ARRAY_SIZE(nf_nat_ipv4_ops));
1047         nf_nat_unregister_fn(net, NFPROTO_IPV6, ops, ARRAY_SIZE(nf_nat_ipv6_ops));
1048 }
1049 EXPORT_SYMBOL_GPL(nf_nat_inet_unregister_fn);
1050 #endif /* NFT INET NAT */