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 * The IP to API glue. 7 * 8 * Authors: see ip.c 9 * 10 * Fixes: 11 * Many : Split from ip.c , see ip.c for history. 12 * Martin Mares : TOS setting fixed. 13 * Alan Cox : Fixed a couple of oopses in Martin's 14 * TOS tweaks. 15 * Mike McLagan : Routing by source 16 */ 17 18 #include <linux/module.h> 19 #include <linux/types.h> 20 #include <linux/mm.h> 21 #include <linux/skbuff.h> 22 #include <linux/ip.h> 23 #include <linux/icmp.h> 24 #include <linux/inetdevice.h> 25 #include <linux/netdevice.h> 26 #include <linux/slab.h> 27 #include <net/sock.h> 28 #include <net/ip.h> 29 #include <net/icmp.h> 30 #include <net/tcp_states.h> 31 #include <linux/udp.h> 32 #include <linux/igmp.h> 33 #include <linux/netfilter.h> 34 #include <linux/route.h> 35 #include <linux/mroute.h> 36 #include <net/inet_ecn.h> 37 #include <net/route.h> 38 #include <net/xfrm.h> 39 #include <net/compat.h> 40 #if IS_ENABLED(CONFIG_IPV6) 41 #include <net/transp_v6.h> 42 #endif 43 #include <net/ip_fib.h> 44 45 #include <linux/errqueue.h> 46 #include <asm/uaccess.h> 47 48 #define IP_CMSG_PKTINFO 1 49 #define IP_CMSG_TTL 2 50 #define IP_CMSG_TOS 4 51 #define IP_CMSG_RECVOPTS 8 52 #define IP_CMSG_RETOPTS 16 53 #define IP_CMSG_PASSSEC 32 54 #define IP_CMSG_ORIGDSTADDR 64 55 56 /* 57 * SOL_IP control messages. 58 */ 59 60 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb) 61 { 62 struct in_pktinfo info = *PKTINFO_SKB_CB(skb); 63 64 info.ipi_addr.s_addr = ip_hdr(skb)->daddr; 65 66 put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 67 } 68 69 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb) 70 { 71 int ttl = ip_hdr(skb)->ttl; 72 put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl); 73 } 74 75 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb) 76 { 77 put_cmsg(msg, SOL_IP, IP_TOS, 1, &ip_hdr(skb)->tos); 78 } 79 80 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb) 81 { 82 if (IPCB(skb)->opt.optlen == 0) 83 return; 84 85 put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen, 86 ip_hdr(skb) + 1); 87 } 88 89 90 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb) 91 { 92 unsigned char optbuf[sizeof(struct ip_options) + 40]; 93 struct ip_options *opt = (struct ip_options *)optbuf; 94 95 if (IPCB(skb)->opt.optlen == 0) 96 return; 97 98 if (ip_options_echo(opt, skb)) { 99 msg->msg_flags |= MSG_CTRUNC; 100 return; 101 } 102 ip_options_undo(opt); 103 104 put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data); 105 } 106 107 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb) 108 { 109 char *secdata; 110 u32 seclen, secid; 111 int err; 112 113 err = security_socket_getpeersec_dgram(NULL, skb, &secid); 114 if (err) 115 return; 116 117 err = security_secid_to_secctx(secid, &secdata, &seclen); 118 if (err) 119 return; 120 121 put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata); 122 security_release_secctx(secdata, seclen); 123 } 124 125 static void ip_cmsg_recv_dstaddr(struct msghdr *msg, struct sk_buff *skb) 126 { 127 struct sockaddr_in sin; 128 const struct iphdr *iph = ip_hdr(skb); 129 __be16 *ports = (__be16 *)skb_transport_header(skb); 130 131 if (skb_transport_offset(skb) + 4 > skb->len) 132 return; 133 134 /* All current transport protocols have the port numbers in the 135 * first four bytes of the transport header and this function is 136 * written with this assumption in mind. 137 */ 138 139 sin.sin_family = AF_INET; 140 sin.sin_addr.s_addr = iph->daddr; 141 sin.sin_port = ports[1]; 142 memset(sin.sin_zero, 0, sizeof(sin.sin_zero)); 143 144 put_cmsg(msg, SOL_IP, IP_ORIGDSTADDR, sizeof(sin), &sin); 145 } 146 147 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 148 { 149 struct inet_sock *inet = inet_sk(skb->sk); 150 unsigned int flags = inet->cmsg_flags; 151 152 /* Ordered by supposed usage frequency */ 153 if (flags & 1) 154 ip_cmsg_recv_pktinfo(msg, skb); 155 if ((flags >>= 1) == 0) 156 return; 157 158 if (flags & 1) 159 ip_cmsg_recv_ttl(msg, skb); 160 if ((flags >>= 1) == 0) 161 return; 162 163 if (flags & 1) 164 ip_cmsg_recv_tos(msg, skb); 165 if ((flags >>= 1) == 0) 166 return; 167 168 if (flags & 1) 169 ip_cmsg_recv_opts(msg, skb); 170 if ((flags >>= 1) == 0) 171 return; 172 173 if (flags & 1) 174 ip_cmsg_recv_retopts(msg, skb); 175 if ((flags >>= 1) == 0) 176 return; 177 178 if (flags & 1) 179 ip_cmsg_recv_security(msg, skb); 180 181 if ((flags >>= 1) == 0) 182 return; 183 if (flags & 1) 184 ip_cmsg_recv_dstaddr(msg, skb); 185 186 } 187 EXPORT_SYMBOL(ip_cmsg_recv); 188 189 int ip_cmsg_send(struct net *net, struct msghdr *msg, struct ipcm_cookie *ipc, 190 bool allow_ipv6) 191 { 192 int err, val; 193 struct cmsghdr *cmsg; 194 195 for_each_cmsghdr(cmsg, msg) { 196 if (!CMSG_OK(msg, cmsg)) 197 return -EINVAL; 198 #if IS_ENABLED(CONFIG_IPV6) 199 if (allow_ipv6 && 200 cmsg->cmsg_level == SOL_IPV6 && 201 cmsg->cmsg_type == IPV6_PKTINFO) { 202 struct in6_pktinfo *src_info; 203 204 if (cmsg->cmsg_len < CMSG_LEN(sizeof(*src_info))) 205 return -EINVAL; 206 src_info = (struct in6_pktinfo *)CMSG_DATA(cmsg); 207 if (!ipv6_addr_v4mapped(&src_info->ipi6_addr)) 208 return -EINVAL; 209 ipc->oif = src_info->ipi6_ifindex; 210 ipc->addr = src_info->ipi6_addr.s6_addr32[3]; 211 continue; 212 } 213 #endif 214 if (cmsg->cmsg_level != SOL_IP) 215 continue; 216 switch (cmsg->cmsg_type) { 217 case IP_RETOPTS: 218 err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr)); 219 err = ip_options_get(net, &ipc->opt, CMSG_DATA(cmsg), 220 err < 40 ? err : 40); 221 if (err) 222 return err; 223 break; 224 case IP_PKTINFO: 225 { 226 struct in_pktinfo *info; 227 if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo))) 228 return -EINVAL; 229 info = (struct in_pktinfo *)CMSG_DATA(cmsg); 230 ipc->oif = info->ipi_ifindex; 231 ipc->addr = info->ipi_spec_dst.s_addr; 232 break; 233 } 234 case IP_TTL: 235 if (cmsg->cmsg_len != CMSG_LEN(sizeof(int))) 236 return -EINVAL; 237 val = *(int *)CMSG_DATA(cmsg); 238 if (val < 1 || val > 255) 239 return -EINVAL; 240 ipc->ttl = val; 241 break; 242 case IP_TOS: 243 if (cmsg->cmsg_len != CMSG_LEN(sizeof(int))) 244 return -EINVAL; 245 val = *(int *)CMSG_DATA(cmsg); 246 if (val < 0 || val > 255) 247 return -EINVAL; 248 ipc->tos = val; 249 ipc->priority = rt_tos2priority(ipc->tos); 250 break; 251 252 default: 253 return -EINVAL; 254 } 255 } 256 return 0; 257 } 258 259 260 /* Special input handler for packets caught by router alert option. 261 They are selected only by protocol field, and then processed likely 262 local ones; but only if someone wants them! Otherwise, router 263 not running rsvpd will kill RSVP. 264 265 It is user level problem, what it will make with them. 266 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), 267 but receiver should be enough clever f.e. to forward mtrace requests, 268 sent to multicast group to reach destination designated router. 269 */ 270 struct ip_ra_chain __rcu *ip_ra_chain; 271 static DEFINE_SPINLOCK(ip_ra_lock); 272 273 274 static void ip_ra_destroy_rcu(struct rcu_head *head) 275 { 276 struct ip_ra_chain *ra = container_of(head, struct ip_ra_chain, rcu); 277 278 sock_put(ra->saved_sk); 279 kfree(ra); 280 } 281 282 int ip_ra_control(struct sock *sk, unsigned char on, 283 void (*destructor)(struct sock *)) 284 { 285 struct ip_ra_chain *ra, *new_ra; 286 struct ip_ra_chain __rcu **rap; 287 288 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->inet_num == IPPROTO_RAW) 289 return -EINVAL; 290 291 new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL; 292 293 spin_lock_bh(&ip_ra_lock); 294 for (rap = &ip_ra_chain; 295 (ra = rcu_dereference_protected(*rap, 296 lockdep_is_held(&ip_ra_lock))) != NULL; 297 rap = &ra->next) { 298 if (ra->sk == sk) { 299 if (on) { 300 spin_unlock_bh(&ip_ra_lock); 301 kfree(new_ra); 302 return -EADDRINUSE; 303 } 304 /* dont let ip_call_ra_chain() use sk again */ 305 ra->sk = NULL; 306 RCU_INIT_POINTER(*rap, ra->next); 307 spin_unlock_bh(&ip_ra_lock); 308 309 if (ra->destructor) 310 ra->destructor(sk); 311 /* 312 * Delay sock_put(sk) and kfree(ra) after one rcu grace 313 * period. This guarantee ip_call_ra_chain() dont need 314 * to mess with socket refcounts. 315 */ 316 ra->saved_sk = sk; 317 call_rcu(&ra->rcu, ip_ra_destroy_rcu); 318 return 0; 319 } 320 } 321 if (new_ra == NULL) { 322 spin_unlock_bh(&ip_ra_lock); 323 return -ENOBUFS; 324 } 325 new_ra->sk = sk; 326 new_ra->destructor = destructor; 327 328 RCU_INIT_POINTER(new_ra->next, ra); 329 rcu_assign_pointer(*rap, new_ra); 330 sock_hold(sk); 331 spin_unlock_bh(&ip_ra_lock); 332 333 return 0; 334 } 335 336 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, 337 __be16 port, u32 info, u8 *payload) 338 { 339 struct sock_exterr_skb *serr; 340 341 skb = skb_clone(skb, GFP_ATOMIC); 342 if (!skb) 343 return; 344 345 serr = SKB_EXT_ERR(skb); 346 serr->ee.ee_errno = err; 347 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP; 348 serr->ee.ee_type = icmp_hdr(skb)->type; 349 serr->ee.ee_code = icmp_hdr(skb)->code; 350 serr->ee.ee_pad = 0; 351 serr->ee.ee_info = info; 352 serr->ee.ee_data = 0; 353 serr->addr_offset = (u8 *)&(((struct iphdr *)(icmp_hdr(skb) + 1))->daddr) - 354 skb_network_header(skb); 355 serr->port = port; 356 357 if (skb_pull(skb, payload - skb->data) != NULL) { 358 skb_reset_transport_header(skb); 359 if (sock_queue_err_skb(sk, skb) == 0) 360 return; 361 } 362 kfree_skb(skb); 363 } 364 365 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info) 366 { 367 struct inet_sock *inet = inet_sk(sk); 368 struct sock_exterr_skb *serr; 369 struct iphdr *iph; 370 struct sk_buff *skb; 371 372 if (!inet->recverr) 373 return; 374 375 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC); 376 if (!skb) 377 return; 378 379 skb_put(skb, sizeof(struct iphdr)); 380 skb_reset_network_header(skb); 381 iph = ip_hdr(skb); 382 iph->daddr = daddr; 383 384 serr = SKB_EXT_ERR(skb); 385 serr->ee.ee_errno = err; 386 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL; 387 serr->ee.ee_type = 0; 388 serr->ee.ee_code = 0; 389 serr->ee.ee_pad = 0; 390 serr->ee.ee_info = info; 391 serr->ee.ee_data = 0; 392 serr->addr_offset = (u8 *)&iph->daddr - skb_network_header(skb); 393 serr->port = port; 394 395 __skb_pull(skb, skb_tail_pointer(skb) - skb->data); 396 skb_reset_transport_header(skb); 397 398 if (sock_queue_err_skb(sk, skb)) 399 kfree_skb(skb); 400 } 401 402 static bool ipv4_pktinfo_prepare_errqueue(const struct sock *sk, 403 const struct sk_buff *skb, 404 int ee_origin) 405 { 406 struct in_pktinfo *info = PKTINFO_SKB_CB(skb); 407 408 if ((ee_origin != SO_EE_ORIGIN_TIMESTAMPING) || 409 (!(sk->sk_tsflags & SOF_TIMESTAMPING_OPT_CMSG)) || 410 (!skb->dev)) 411 return false; 412 413 info->ipi_spec_dst.s_addr = ip_hdr(skb)->saddr; 414 info->ipi_ifindex = skb->dev->ifindex; 415 return true; 416 } 417 418 /* 419 * Handle MSG_ERRQUEUE 420 */ 421 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len) 422 { 423 struct sock_exterr_skb *serr; 424 struct sk_buff *skb; 425 DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name); 426 struct { 427 struct sock_extended_err ee; 428 struct sockaddr_in offender; 429 } errhdr; 430 int err; 431 int copied; 432 433 WARN_ON_ONCE(sk->sk_family == AF_INET6); 434 435 err = -EAGAIN; 436 skb = sock_dequeue_err_skb(sk); 437 if (skb == NULL) 438 goto out; 439 440 copied = skb->len; 441 if (copied > len) { 442 msg->msg_flags |= MSG_TRUNC; 443 copied = len; 444 } 445 err = skb_copy_datagram_msg(skb, 0, msg, copied); 446 if (err) 447 goto out_free_skb; 448 449 sock_recv_timestamp(msg, sk, skb); 450 451 serr = SKB_EXT_ERR(skb); 452 453 if (sin) { 454 sin->sin_family = AF_INET; 455 sin->sin_addr.s_addr = *(__be32 *)(skb_network_header(skb) + 456 serr->addr_offset); 457 sin->sin_port = serr->port; 458 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 459 *addr_len = sizeof(*sin); 460 } 461 462 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err)); 463 sin = &errhdr.offender; 464 sin->sin_family = AF_UNSPEC; 465 466 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP || 467 ipv4_pktinfo_prepare_errqueue(sk, skb, serr->ee.ee_origin)) { 468 struct inet_sock *inet = inet_sk(sk); 469 470 sin->sin_family = AF_INET; 471 sin->sin_addr.s_addr = ip_hdr(skb)->saddr; 472 sin->sin_port = 0; 473 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 474 if (inet->cmsg_flags) 475 ip_cmsg_recv(msg, skb); 476 } 477 478 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr); 479 480 /* Now we could try to dump offended packet options */ 481 482 msg->msg_flags |= MSG_ERRQUEUE; 483 err = copied; 484 485 out_free_skb: 486 kfree_skb(skb); 487 out: 488 return err; 489 } 490 491 492 /* 493 * Socket option code for IP. This is the end of the line after any 494 * TCP,UDP etc options on an IP socket. 495 */ 496 497 static int do_ip_setsockopt(struct sock *sk, int level, 498 int optname, char __user *optval, unsigned int optlen) 499 { 500 struct inet_sock *inet = inet_sk(sk); 501 int val = 0, err; 502 503 switch (optname) { 504 case IP_PKTINFO: 505 case IP_RECVTTL: 506 case IP_RECVOPTS: 507 case IP_RECVTOS: 508 case IP_RETOPTS: 509 case IP_TOS: 510 case IP_TTL: 511 case IP_HDRINCL: 512 case IP_MTU_DISCOVER: 513 case IP_RECVERR: 514 case IP_ROUTER_ALERT: 515 case IP_FREEBIND: 516 case IP_PASSSEC: 517 case IP_TRANSPARENT: 518 case IP_MINTTL: 519 case IP_NODEFRAG: 520 case IP_UNICAST_IF: 521 case IP_MULTICAST_TTL: 522 case IP_MULTICAST_ALL: 523 case IP_MULTICAST_LOOP: 524 case IP_RECVORIGDSTADDR: 525 if (optlen >= sizeof(int)) { 526 if (get_user(val, (int __user *) optval)) 527 return -EFAULT; 528 } else if (optlen >= sizeof(char)) { 529 unsigned char ucval; 530 531 if (get_user(ucval, (unsigned char __user *) optval)) 532 return -EFAULT; 533 val = (int) ucval; 534 } 535 } 536 537 /* If optlen==0, it is equivalent to val == 0 */ 538 539 if (ip_mroute_opt(optname)) 540 return ip_mroute_setsockopt(sk, optname, optval, optlen); 541 542 err = 0; 543 lock_sock(sk); 544 545 switch (optname) { 546 case IP_OPTIONS: 547 { 548 struct ip_options_rcu *old, *opt = NULL; 549 550 if (optlen > 40) 551 goto e_inval; 552 err = ip_options_get_from_user(sock_net(sk), &opt, 553 optval, optlen); 554 if (err) 555 break; 556 old = rcu_dereference_protected(inet->inet_opt, 557 sock_owned_by_user(sk)); 558 if (inet->is_icsk) { 559 struct inet_connection_sock *icsk = inet_csk(sk); 560 #if IS_ENABLED(CONFIG_IPV6) 561 if (sk->sk_family == PF_INET || 562 (!((1 << sk->sk_state) & 563 (TCPF_LISTEN | TCPF_CLOSE)) && 564 inet->inet_daddr != LOOPBACK4_IPV6)) { 565 #endif 566 if (old) 567 icsk->icsk_ext_hdr_len -= old->opt.optlen; 568 if (opt) 569 icsk->icsk_ext_hdr_len += opt->opt.optlen; 570 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie); 571 #if IS_ENABLED(CONFIG_IPV6) 572 } 573 #endif 574 } 575 rcu_assign_pointer(inet->inet_opt, opt); 576 if (old) 577 kfree_rcu(old, rcu); 578 break; 579 } 580 case IP_PKTINFO: 581 if (val) 582 inet->cmsg_flags |= IP_CMSG_PKTINFO; 583 else 584 inet->cmsg_flags &= ~IP_CMSG_PKTINFO; 585 break; 586 case IP_RECVTTL: 587 if (val) 588 inet->cmsg_flags |= IP_CMSG_TTL; 589 else 590 inet->cmsg_flags &= ~IP_CMSG_TTL; 591 break; 592 case IP_RECVTOS: 593 if (val) 594 inet->cmsg_flags |= IP_CMSG_TOS; 595 else 596 inet->cmsg_flags &= ~IP_CMSG_TOS; 597 break; 598 case IP_RECVOPTS: 599 if (val) 600 inet->cmsg_flags |= IP_CMSG_RECVOPTS; 601 else 602 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS; 603 break; 604 case IP_RETOPTS: 605 if (val) 606 inet->cmsg_flags |= IP_CMSG_RETOPTS; 607 else 608 inet->cmsg_flags &= ~IP_CMSG_RETOPTS; 609 break; 610 case IP_PASSSEC: 611 if (val) 612 inet->cmsg_flags |= IP_CMSG_PASSSEC; 613 else 614 inet->cmsg_flags &= ~IP_CMSG_PASSSEC; 615 break; 616 case IP_RECVORIGDSTADDR: 617 if (val) 618 inet->cmsg_flags |= IP_CMSG_ORIGDSTADDR; 619 else 620 inet->cmsg_flags &= ~IP_CMSG_ORIGDSTADDR; 621 break; 622 case IP_TOS: /* This sets both TOS and Precedence */ 623 if (sk->sk_type == SOCK_STREAM) { 624 val &= ~INET_ECN_MASK; 625 val |= inet->tos & INET_ECN_MASK; 626 } 627 if (inet->tos != val) { 628 inet->tos = val; 629 sk->sk_priority = rt_tos2priority(val); 630 sk_dst_reset(sk); 631 } 632 break; 633 case IP_TTL: 634 if (optlen < 1) 635 goto e_inval; 636 if (val != -1 && (val < 1 || val > 255)) 637 goto e_inval; 638 inet->uc_ttl = val; 639 break; 640 case IP_HDRINCL: 641 if (sk->sk_type != SOCK_RAW) { 642 err = -ENOPROTOOPT; 643 break; 644 } 645 inet->hdrincl = val ? 1 : 0; 646 break; 647 case IP_NODEFRAG: 648 if (sk->sk_type != SOCK_RAW) { 649 err = -ENOPROTOOPT; 650 break; 651 } 652 inet->nodefrag = val ? 1 : 0; 653 break; 654 case IP_MTU_DISCOVER: 655 if (val < IP_PMTUDISC_DONT || val > IP_PMTUDISC_OMIT) 656 goto e_inval; 657 inet->pmtudisc = val; 658 break; 659 case IP_RECVERR: 660 inet->recverr = !!val; 661 if (!val) 662 skb_queue_purge(&sk->sk_error_queue); 663 break; 664 case IP_MULTICAST_TTL: 665 if (sk->sk_type == SOCK_STREAM) 666 goto e_inval; 667 if (optlen < 1) 668 goto e_inval; 669 if (val == -1) 670 val = 1; 671 if (val < 0 || val > 255) 672 goto e_inval; 673 inet->mc_ttl = val; 674 break; 675 case IP_MULTICAST_LOOP: 676 if (optlen < 1) 677 goto e_inval; 678 inet->mc_loop = !!val; 679 break; 680 case IP_UNICAST_IF: 681 { 682 struct net_device *dev = NULL; 683 int ifindex; 684 685 if (optlen != sizeof(int)) 686 goto e_inval; 687 688 ifindex = (__force int)ntohl((__force __be32)val); 689 if (ifindex == 0) { 690 inet->uc_index = 0; 691 err = 0; 692 break; 693 } 694 695 dev = dev_get_by_index(sock_net(sk), ifindex); 696 err = -EADDRNOTAVAIL; 697 if (!dev) 698 break; 699 dev_put(dev); 700 701 err = -EINVAL; 702 if (sk->sk_bound_dev_if) 703 break; 704 705 inet->uc_index = ifindex; 706 err = 0; 707 break; 708 } 709 case IP_MULTICAST_IF: 710 { 711 struct ip_mreqn mreq; 712 struct net_device *dev = NULL; 713 714 if (sk->sk_type == SOCK_STREAM) 715 goto e_inval; 716 /* 717 * Check the arguments are allowable 718 */ 719 720 if (optlen < sizeof(struct in_addr)) 721 goto e_inval; 722 723 err = -EFAULT; 724 if (optlen >= sizeof(struct ip_mreqn)) { 725 if (copy_from_user(&mreq, optval, sizeof(mreq))) 726 break; 727 } else { 728 memset(&mreq, 0, sizeof(mreq)); 729 if (optlen >= sizeof(struct ip_mreq)) { 730 if (copy_from_user(&mreq, optval, 731 sizeof(struct ip_mreq))) 732 break; 733 } else if (optlen >= sizeof(struct in_addr)) { 734 if (copy_from_user(&mreq.imr_address, optval, 735 sizeof(struct in_addr))) 736 break; 737 } 738 } 739 740 if (!mreq.imr_ifindex) { 741 if (mreq.imr_address.s_addr == htonl(INADDR_ANY)) { 742 inet->mc_index = 0; 743 inet->mc_addr = 0; 744 err = 0; 745 break; 746 } 747 dev = ip_dev_find(sock_net(sk), mreq.imr_address.s_addr); 748 if (dev) 749 mreq.imr_ifindex = dev->ifindex; 750 } else 751 dev = dev_get_by_index(sock_net(sk), mreq.imr_ifindex); 752 753 754 err = -EADDRNOTAVAIL; 755 if (!dev) 756 break; 757 dev_put(dev); 758 759 err = -EINVAL; 760 if (sk->sk_bound_dev_if && 761 mreq.imr_ifindex != sk->sk_bound_dev_if) 762 break; 763 764 inet->mc_index = mreq.imr_ifindex; 765 inet->mc_addr = mreq.imr_address.s_addr; 766 err = 0; 767 break; 768 } 769 770 case IP_ADD_MEMBERSHIP: 771 case IP_DROP_MEMBERSHIP: 772 { 773 struct ip_mreqn mreq; 774 775 err = -EPROTO; 776 if (inet_sk(sk)->is_icsk) 777 break; 778 779 if (optlen < sizeof(struct ip_mreq)) 780 goto e_inval; 781 err = -EFAULT; 782 if (optlen >= sizeof(struct ip_mreqn)) { 783 if (copy_from_user(&mreq, optval, sizeof(mreq))) 784 break; 785 } else { 786 memset(&mreq, 0, sizeof(mreq)); 787 if (copy_from_user(&mreq, optval, sizeof(struct ip_mreq))) 788 break; 789 } 790 791 if (optname == IP_ADD_MEMBERSHIP) 792 err = ip_mc_join_group(sk, &mreq); 793 else 794 err = ip_mc_leave_group(sk, &mreq); 795 break; 796 } 797 case IP_MSFILTER: 798 { 799 struct ip_msfilter *msf; 800 801 if (optlen < IP_MSFILTER_SIZE(0)) 802 goto e_inval; 803 if (optlen > sysctl_optmem_max) { 804 err = -ENOBUFS; 805 break; 806 } 807 msf = kmalloc(optlen, GFP_KERNEL); 808 if (!msf) { 809 err = -ENOBUFS; 810 break; 811 } 812 err = -EFAULT; 813 if (copy_from_user(msf, optval, optlen)) { 814 kfree(msf); 815 break; 816 } 817 /* numsrc >= (1G-4) overflow in 32 bits */ 818 if (msf->imsf_numsrc >= 0x3ffffffcU || 819 msf->imsf_numsrc > sysctl_igmp_max_msf) { 820 kfree(msf); 821 err = -ENOBUFS; 822 break; 823 } 824 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) { 825 kfree(msf); 826 err = -EINVAL; 827 break; 828 } 829 err = ip_mc_msfilter(sk, msf, 0); 830 kfree(msf); 831 break; 832 } 833 case IP_BLOCK_SOURCE: 834 case IP_UNBLOCK_SOURCE: 835 case IP_ADD_SOURCE_MEMBERSHIP: 836 case IP_DROP_SOURCE_MEMBERSHIP: 837 { 838 struct ip_mreq_source mreqs; 839 int omode, add; 840 841 if (optlen != sizeof(struct ip_mreq_source)) 842 goto e_inval; 843 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) { 844 err = -EFAULT; 845 break; 846 } 847 if (optname == IP_BLOCK_SOURCE) { 848 omode = MCAST_EXCLUDE; 849 add = 1; 850 } else if (optname == IP_UNBLOCK_SOURCE) { 851 omode = MCAST_EXCLUDE; 852 add = 0; 853 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) { 854 struct ip_mreqn mreq; 855 856 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr; 857 mreq.imr_address.s_addr = mreqs.imr_interface; 858 mreq.imr_ifindex = 0; 859 err = ip_mc_join_group(sk, &mreq); 860 if (err && err != -EADDRINUSE) 861 break; 862 omode = MCAST_INCLUDE; 863 add = 1; 864 } else /* IP_DROP_SOURCE_MEMBERSHIP */ { 865 omode = MCAST_INCLUDE; 866 add = 0; 867 } 868 err = ip_mc_source(add, omode, sk, &mreqs, 0); 869 break; 870 } 871 case MCAST_JOIN_GROUP: 872 case MCAST_LEAVE_GROUP: 873 { 874 struct group_req greq; 875 struct sockaddr_in *psin; 876 struct ip_mreqn mreq; 877 878 if (optlen < sizeof(struct group_req)) 879 goto e_inval; 880 err = -EFAULT; 881 if (copy_from_user(&greq, optval, sizeof(greq))) 882 break; 883 psin = (struct sockaddr_in *)&greq.gr_group; 884 if (psin->sin_family != AF_INET) 885 goto e_inval; 886 memset(&mreq, 0, sizeof(mreq)); 887 mreq.imr_multiaddr = psin->sin_addr; 888 mreq.imr_ifindex = greq.gr_interface; 889 890 if (optname == MCAST_JOIN_GROUP) 891 err = ip_mc_join_group(sk, &mreq); 892 else 893 err = ip_mc_leave_group(sk, &mreq); 894 break; 895 } 896 case MCAST_JOIN_SOURCE_GROUP: 897 case MCAST_LEAVE_SOURCE_GROUP: 898 case MCAST_BLOCK_SOURCE: 899 case MCAST_UNBLOCK_SOURCE: 900 { 901 struct group_source_req greqs; 902 struct ip_mreq_source mreqs; 903 struct sockaddr_in *psin; 904 int omode, add; 905 906 if (optlen != sizeof(struct group_source_req)) 907 goto e_inval; 908 if (copy_from_user(&greqs, optval, sizeof(greqs))) { 909 err = -EFAULT; 910 break; 911 } 912 if (greqs.gsr_group.ss_family != AF_INET || 913 greqs.gsr_source.ss_family != AF_INET) { 914 err = -EADDRNOTAVAIL; 915 break; 916 } 917 psin = (struct sockaddr_in *)&greqs.gsr_group; 918 mreqs.imr_multiaddr = psin->sin_addr.s_addr; 919 psin = (struct sockaddr_in *)&greqs.gsr_source; 920 mreqs.imr_sourceaddr = psin->sin_addr.s_addr; 921 mreqs.imr_interface = 0; /* use index for mc_source */ 922 923 if (optname == MCAST_BLOCK_SOURCE) { 924 omode = MCAST_EXCLUDE; 925 add = 1; 926 } else if (optname == MCAST_UNBLOCK_SOURCE) { 927 omode = MCAST_EXCLUDE; 928 add = 0; 929 } else if (optname == MCAST_JOIN_SOURCE_GROUP) { 930 struct ip_mreqn mreq; 931 932 psin = (struct sockaddr_in *)&greqs.gsr_group; 933 mreq.imr_multiaddr = psin->sin_addr; 934 mreq.imr_address.s_addr = 0; 935 mreq.imr_ifindex = greqs.gsr_interface; 936 err = ip_mc_join_group(sk, &mreq); 937 if (err && err != -EADDRINUSE) 938 break; 939 greqs.gsr_interface = mreq.imr_ifindex; 940 omode = MCAST_INCLUDE; 941 add = 1; 942 } else /* MCAST_LEAVE_SOURCE_GROUP */ { 943 omode = MCAST_INCLUDE; 944 add = 0; 945 } 946 err = ip_mc_source(add, omode, sk, &mreqs, 947 greqs.gsr_interface); 948 break; 949 } 950 case MCAST_MSFILTER: 951 { 952 struct sockaddr_in *psin; 953 struct ip_msfilter *msf = NULL; 954 struct group_filter *gsf = NULL; 955 int msize, i, ifindex; 956 957 if (optlen < GROUP_FILTER_SIZE(0)) 958 goto e_inval; 959 if (optlen > sysctl_optmem_max) { 960 err = -ENOBUFS; 961 break; 962 } 963 gsf = kmalloc(optlen, GFP_KERNEL); 964 if (!gsf) { 965 err = -ENOBUFS; 966 break; 967 } 968 err = -EFAULT; 969 if (copy_from_user(gsf, optval, optlen)) 970 goto mc_msf_out; 971 972 /* numsrc >= (4G-140)/128 overflow in 32 bits */ 973 if (gsf->gf_numsrc >= 0x1ffffff || 974 gsf->gf_numsrc > sysctl_igmp_max_msf) { 975 err = -ENOBUFS; 976 goto mc_msf_out; 977 } 978 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) { 979 err = -EINVAL; 980 goto mc_msf_out; 981 } 982 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc); 983 msf = kmalloc(msize, GFP_KERNEL); 984 if (!msf) { 985 err = -ENOBUFS; 986 goto mc_msf_out; 987 } 988 ifindex = gsf->gf_interface; 989 psin = (struct sockaddr_in *)&gsf->gf_group; 990 if (psin->sin_family != AF_INET) { 991 err = -EADDRNOTAVAIL; 992 goto mc_msf_out; 993 } 994 msf->imsf_multiaddr = psin->sin_addr.s_addr; 995 msf->imsf_interface = 0; 996 msf->imsf_fmode = gsf->gf_fmode; 997 msf->imsf_numsrc = gsf->gf_numsrc; 998 err = -EADDRNOTAVAIL; 999 for (i = 0; i < gsf->gf_numsrc; ++i) { 1000 psin = (struct sockaddr_in *)&gsf->gf_slist[i]; 1001 1002 if (psin->sin_family != AF_INET) 1003 goto mc_msf_out; 1004 msf->imsf_slist[i] = psin->sin_addr.s_addr; 1005 } 1006 kfree(gsf); 1007 gsf = NULL; 1008 1009 err = ip_mc_msfilter(sk, msf, ifindex); 1010 mc_msf_out: 1011 kfree(msf); 1012 kfree(gsf); 1013 break; 1014 } 1015 case IP_MULTICAST_ALL: 1016 if (optlen < 1) 1017 goto e_inval; 1018 if (val != 0 && val != 1) 1019 goto e_inval; 1020 inet->mc_all = val; 1021 break; 1022 case IP_ROUTER_ALERT: 1023 err = ip_ra_control(sk, val ? 1 : 0, NULL); 1024 break; 1025 1026 case IP_FREEBIND: 1027 if (optlen < 1) 1028 goto e_inval; 1029 inet->freebind = !!val; 1030 break; 1031 1032 case IP_IPSEC_POLICY: 1033 case IP_XFRM_POLICY: 1034 err = -EPERM; 1035 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)) 1036 break; 1037 err = xfrm_user_policy(sk, optname, optval, optlen); 1038 break; 1039 1040 case IP_TRANSPARENT: 1041 if (!!val && !ns_capable(sock_net(sk)->user_ns, CAP_NET_RAW) && 1042 !ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)) { 1043 err = -EPERM; 1044 break; 1045 } 1046 if (optlen < 1) 1047 goto e_inval; 1048 inet->transparent = !!val; 1049 break; 1050 1051 case IP_MINTTL: 1052 if (optlen < 1) 1053 goto e_inval; 1054 if (val < 0 || val > 255) 1055 goto e_inval; 1056 inet->min_ttl = val; 1057 break; 1058 1059 default: 1060 err = -ENOPROTOOPT; 1061 break; 1062 } 1063 release_sock(sk); 1064 return err; 1065 1066 e_inval: 1067 release_sock(sk); 1068 return -EINVAL; 1069 } 1070 1071 /** 1072 * ipv4_pktinfo_prepare - transfer some info from rtable to skb 1073 * @sk: socket 1074 * @skb: buffer 1075 * 1076 * To support IP_CMSG_PKTINFO option, we store rt_iif and specific 1077 * destination in skb->cb[] before dst drop. 1078 * This way, receiver doesn't make cache line misses to read rtable. 1079 */ 1080 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb) 1081 { 1082 struct in_pktinfo *pktinfo = PKTINFO_SKB_CB(skb); 1083 bool prepare = (inet_sk(sk)->cmsg_flags & IP_CMSG_PKTINFO) || 1084 ipv6_sk_rxinfo(sk); 1085 1086 if (prepare && skb_rtable(skb)) { 1087 pktinfo->ipi_ifindex = inet_iif(skb); 1088 pktinfo->ipi_spec_dst.s_addr = fib_compute_spec_dst(skb); 1089 } else { 1090 pktinfo->ipi_ifindex = 0; 1091 pktinfo->ipi_spec_dst.s_addr = 0; 1092 } 1093 skb_dst_drop(skb); 1094 } 1095 1096 int ip_setsockopt(struct sock *sk, int level, 1097 int optname, char __user *optval, unsigned int optlen) 1098 { 1099 int err; 1100 1101 if (level != SOL_IP) 1102 return -ENOPROTOOPT; 1103 1104 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 1105 #ifdef CONFIG_NETFILTER 1106 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1107 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 1108 optname != IP_IPSEC_POLICY && 1109 optname != IP_XFRM_POLICY && 1110 !ip_mroute_opt(optname)) { 1111 lock_sock(sk); 1112 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen); 1113 release_sock(sk); 1114 } 1115 #endif 1116 return err; 1117 } 1118 EXPORT_SYMBOL(ip_setsockopt); 1119 1120 #ifdef CONFIG_COMPAT 1121 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 1122 char __user *optval, unsigned int optlen) 1123 { 1124 int err; 1125 1126 if (level != SOL_IP) 1127 return -ENOPROTOOPT; 1128 1129 if (optname >= MCAST_JOIN_GROUP && optname <= MCAST_MSFILTER) 1130 return compat_mc_setsockopt(sk, level, optname, optval, optlen, 1131 ip_setsockopt); 1132 1133 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 1134 #ifdef CONFIG_NETFILTER 1135 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1136 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 1137 optname != IP_IPSEC_POLICY && 1138 optname != IP_XFRM_POLICY && 1139 !ip_mroute_opt(optname)) { 1140 lock_sock(sk); 1141 err = compat_nf_setsockopt(sk, PF_INET, optname, 1142 optval, optlen); 1143 release_sock(sk); 1144 } 1145 #endif 1146 return err; 1147 } 1148 EXPORT_SYMBOL(compat_ip_setsockopt); 1149 #endif 1150 1151 /* 1152 * Get the options. Note for future reference. The GET of IP options gets 1153 * the _received_ ones. The set sets the _sent_ ones. 1154 */ 1155 1156 static int do_ip_getsockopt(struct sock *sk, int level, int optname, 1157 char __user *optval, int __user *optlen, unsigned int flags) 1158 { 1159 struct inet_sock *inet = inet_sk(sk); 1160 int val; 1161 int len; 1162 1163 if (level != SOL_IP) 1164 return -EOPNOTSUPP; 1165 1166 if (ip_mroute_opt(optname)) 1167 return ip_mroute_getsockopt(sk, optname, optval, optlen); 1168 1169 if (get_user(len, optlen)) 1170 return -EFAULT; 1171 if (len < 0) 1172 return -EINVAL; 1173 1174 lock_sock(sk); 1175 1176 switch (optname) { 1177 case IP_OPTIONS: 1178 { 1179 unsigned char optbuf[sizeof(struct ip_options)+40]; 1180 struct ip_options *opt = (struct ip_options *)optbuf; 1181 struct ip_options_rcu *inet_opt; 1182 1183 inet_opt = rcu_dereference_protected(inet->inet_opt, 1184 sock_owned_by_user(sk)); 1185 opt->optlen = 0; 1186 if (inet_opt) 1187 memcpy(optbuf, &inet_opt->opt, 1188 sizeof(struct ip_options) + 1189 inet_opt->opt.optlen); 1190 release_sock(sk); 1191 1192 if (opt->optlen == 0) 1193 return put_user(0, optlen); 1194 1195 ip_options_undo(opt); 1196 1197 len = min_t(unsigned int, len, opt->optlen); 1198 if (put_user(len, optlen)) 1199 return -EFAULT; 1200 if (copy_to_user(optval, opt->__data, len)) 1201 return -EFAULT; 1202 return 0; 1203 } 1204 case IP_PKTINFO: 1205 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0; 1206 break; 1207 case IP_RECVTTL: 1208 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0; 1209 break; 1210 case IP_RECVTOS: 1211 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0; 1212 break; 1213 case IP_RECVOPTS: 1214 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0; 1215 break; 1216 case IP_RETOPTS: 1217 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0; 1218 break; 1219 case IP_PASSSEC: 1220 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0; 1221 break; 1222 case IP_RECVORIGDSTADDR: 1223 val = (inet->cmsg_flags & IP_CMSG_ORIGDSTADDR) != 0; 1224 break; 1225 case IP_TOS: 1226 val = inet->tos; 1227 break; 1228 case IP_TTL: 1229 val = (inet->uc_ttl == -1 ? 1230 sysctl_ip_default_ttl : 1231 inet->uc_ttl); 1232 break; 1233 case IP_HDRINCL: 1234 val = inet->hdrincl; 1235 break; 1236 case IP_NODEFRAG: 1237 val = inet->nodefrag; 1238 break; 1239 case IP_MTU_DISCOVER: 1240 val = inet->pmtudisc; 1241 break; 1242 case IP_MTU: 1243 { 1244 struct dst_entry *dst; 1245 val = 0; 1246 dst = sk_dst_get(sk); 1247 if (dst) { 1248 val = dst_mtu(dst); 1249 dst_release(dst); 1250 } 1251 if (!val) { 1252 release_sock(sk); 1253 return -ENOTCONN; 1254 } 1255 break; 1256 } 1257 case IP_RECVERR: 1258 val = inet->recverr; 1259 break; 1260 case IP_MULTICAST_TTL: 1261 val = inet->mc_ttl; 1262 break; 1263 case IP_MULTICAST_LOOP: 1264 val = inet->mc_loop; 1265 break; 1266 case IP_UNICAST_IF: 1267 val = (__force int)htonl((__u32) inet->uc_index); 1268 break; 1269 case IP_MULTICAST_IF: 1270 { 1271 struct in_addr addr; 1272 len = min_t(unsigned int, len, sizeof(struct in_addr)); 1273 addr.s_addr = inet->mc_addr; 1274 release_sock(sk); 1275 1276 if (put_user(len, optlen)) 1277 return -EFAULT; 1278 if (copy_to_user(optval, &addr, len)) 1279 return -EFAULT; 1280 return 0; 1281 } 1282 case IP_MSFILTER: 1283 { 1284 struct ip_msfilter msf; 1285 int err; 1286 1287 if (len < IP_MSFILTER_SIZE(0)) { 1288 release_sock(sk); 1289 return -EINVAL; 1290 } 1291 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) { 1292 release_sock(sk); 1293 return -EFAULT; 1294 } 1295 err = ip_mc_msfget(sk, &msf, 1296 (struct ip_msfilter __user *)optval, optlen); 1297 release_sock(sk); 1298 return err; 1299 } 1300 case MCAST_MSFILTER: 1301 { 1302 struct group_filter gsf; 1303 int err; 1304 1305 if (len < GROUP_FILTER_SIZE(0)) { 1306 release_sock(sk); 1307 return -EINVAL; 1308 } 1309 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) { 1310 release_sock(sk); 1311 return -EFAULT; 1312 } 1313 err = ip_mc_gsfget(sk, &gsf, 1314 (struct group_filter __user *)optval, 1315 optlen); 1316 release_sock(sk); 1317 return err; 1318 } 1319 case IP_MULTICAST_ALL: 1320 val = inet->mc_all; 1321 break; 1322 case IP_PKTOPTIONS: 1323 { 1324 struct msghdr msg; 1325 1326 release_sock(sk); 1327 1328 if (sk->sk_type != SOCK_STREAM) 1329 return -ENOPROTOOPT; 1330 1331 msg.msg_control = (__force void *) optval; 1332 msg.msg_controllen = len; 1333 msg.msg_flags = flags; 1334 1335 if (inet->cmsg_flags & IP_CMSG_PKTINFO) { 1336 struct in_pktinfo info; 1337 1338 info.ipi_addr.s_addr = inet->inet_rcv_saddr; 1339 info.ipi_spec_dst.s_addr = inet->inet_rcv_saddr; 1340 info.ipi_ifindex = inet->mc_index; 1341 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 1342 } 1343 if (inet->cmsg_flags & IP_CMSG_TTL) { 1344 int hlim = inet->mc_ttl; 1345 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim); 1346 } 1347 if (inet->cmsg_flags & IP_CMSG_TOS) { 1348 int tos = inet->rcv_tos; 1349 put_cmsg(&msg, SOL_IP, IP_TOS, sizeof(tos), &tos); 1350 } 1351 len -= msg.msg_controllen; 1352 return put_user(len, optlen); 1353 } 1354 case IP_FREEBIND: 1355 val = inet->freebind; 1356 break; 1357 case IP_TRANSPARENT: 1358 val = inet->transparent; 1359 break; 1360 case IP_MINTTL: 1361 val = inet->min_ttl; 1362 break; 1363 default: 1364 release_sock(sk); 1365 return -ENOPROTOOPT; 1366 } 1367 release_sock(sk); 1368 1369 if (len < sizeof(int) && len > 0 && val >= 0 && val <= 255) { 1370 unsigned char ucval = (unsigned char)val; 1371 len = 1; 1372 if (put_user(len, optlen)) 1373 return -EFAULT; 1374 if (copy_to_user(optval, &ucval, 1)) 1375 return -EFAULT; 1376 } else { 1377 len = min_t(unsigned int, sizeof(int), len); 1378 if (put_user(len, optlen)) 1379 return -EFAULT; 1380 if (copy_to_user(optval, &val, len)) 1381 return -EFAULT; 1382 } 1383 return 0; 1384 } 1385 1386 int ip_getsockopt(struct sock *sk, int level, 1387 int optname, char __user *optval, int __user *optlen) 1388 { 1389 int err; 1390 1391 err = do_ip_getsockopt(sk, level, optname, optval, optlen, 0); 1392 #ifdef CONFIG_NETFILTER 1393 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1394 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS && 1395 !ip_mroute_opt(optname)) { 1396 int len; 1397 1398 if (get_user(len, optlen)) 1399 return -EFAULT; 1400 1401 lock_sock(sk); 1402 err = nf_getsockopt(sk, PF_INET, optname, optval, 1403 &len); 1404 release_sock(sk); 1405 if (err >= 0) 1406 err = put_user(len, optlen); 1407 return err; 1408 } 1409 #endif 1410 return err; 1411 } 1412 EXPORT_SYMBOL(ip_getsockopt); 1413 1414 #ifdef CONFIG_COMPAT 1415 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 1416 char __user *optval, int __user *optlen) 1417 { 1418 int err; 1419 1420 if (optname == MCAST_MSFILTER) 1421 return compat_mc_getsockopt(sk, level, optname, optval, optlen, 1422 ip_getsockopt); 1423 1424 err = do_ip_getsockopt(sk, level, optname, optval, optlen, 1425 MSG_CMSG_COMPAT); 1426 1427 #ifdef CONFIG_NETFILTER 1428 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1429 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS && 1430 !ip_mroute_opt(optname)) { 1431 int len; 1432 1433 if (get_user(len, optlen)) 1434 return -EFAULT; 1435 1436 lock_sock(sk); 1437 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len); 1438 release_sock(sk); 1439 if (err >= 0) 1440 err = put_user(len, optlen); 1441 return err; 1442 } 1443 #endif 1444 return err; 1445 } 1446 EXPORT_SYMBOL(compat_ip_getsockopt); 1447 #endif 1448