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 * Version: $Id: ip_sockglue.c,v 1.62 2002/02/01 22:01:04 davem Exp $ 9 * 10 * Authors: see ip.c 11 * 12 * Fixes: 13 * Many : Split from ip.c , see ip.c for history. 14 * Martin Mares : TOS setting fixed. 15 * Alan Cox : Fixed a couple of oopses in Martin's 16 * TOS tweaks. 17 * Mike McLagan : Routing by source 18 */ 19 20 #include <linux/module.h> 21 #include <linux/types.h> 22 #include <linux/mm.h> 23 #include <linux/skbuff.h> 24 #include <linux/ip.h> 25 #include <linux/icmp.h> 26 #include <linux/inetdevice.h> 27 #include <linux/netdevice.h> 28 #include <net/sock.h> 29 #include <net/ip.h> 30 #include <net/icmp.h> 31 #include <net/tcp_states.h> 32 #include <linux/udp.h> 33 #include <linux/igmp.h> 34 #include <linux/netfilter.h> 35 #include <linux/route.h> 36 #include <linux/mroute.h> 37 #include <net/route.h> 38 #include <net/xfrm.h> 39 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 40 #include <net/transp_v6.h> 41 #endif 42 43 #include <linux/errqueue.h> 44 #include <asm/uaccess.h> 45 46 #define IP_CMSG_PKTINFO 1 47 #define IP_CMSG_TTL 2 48 #define IP_CMSG_TOS 4 49 #define IP_CMSG_RECVOPTS 8 50 #define IP_CMSG_RETOPTS 16 51 #define IP_CMSG_PASSSEC 32 52 53 /* 54 * SOL_IP control messages. 55 */ 56 57 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb) 58 { 59 struct in_pktinfo info; 60 struct rtable *rt = (struct rtable *)skb->dst; 61 62 info.ipi_addr.s_addr = ip_hdr(skb)->daddr; 63 if (rt) { 64 info.ipi_ifindex = rt->rt_iif; 65 info.ipi_spec_dst.s_addr = rt->rt_spec_dst; 66 } else { 67 info.ipi_ifindex = 0; 68 info.ipi_spec_dst.s_addr = 0; 69 } 70 71 put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 72 } 73 74 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb) 75 { 76 int ttl = ip_hdr(skb)->ttl; 77 put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl); 78 } 79 80 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb) 81 { 82 put_cmsg(msg, SOL_IP, IP_TOS, 1, &ip_hdr(skb)->tos); 83 } 84 85 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb) 86 { 87 if (IPCB(skb)->opt.optlen == 0) 88 return; 89 90 put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen, 91 ip_hdr(skb) + 1); 92 } 93 94 95 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb) 96 { 97 unsigned char optbuf[sizeof(struct ip_options) + 40]; 98 struct ip_options * opt = (struct ip_options*)optbuf; 99 100 if (IPCB(skb)->opt.optlen == 0) 101 return; 102 103 if (ip_options_echo(opt, skb)) { 104 msg->msg_flags |= MSG_CTRUNC; 105 return; 106 } 107 ip_options_undo(opt); 108 109 put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data); 110 } 111 112 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb) 113 { 114 char *secdata; 115 u32 seclen, secid; 116 int err; 117 118 err = security_socket_getpeersec_dgram(NULL, skb, &secid); 119 if (err) 120 return; 121 122 err = security_secid_to_secctx(secid, &secdata, &seclen); 123 if (err) 124 return; 125 126 put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata); 127 security_release_secctx(secdata, seclen); 128 } 129 130 131 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 132 { 133 struct inet_sock *inet = inet_sk(skb->sk); 134 unsigned flags = inet->cmsg_flags; 135 136 /* Ordered by supposed usage frequency */ 137 if (flags & 1) 138 ip_cmsg_recv_pktinfo(msg, skb); 139 if ((flags>>=1) == 0) 140 return; 141 142 if (flags & 1) 143 ip_cmsg_recv_ttl(msg, skb); 144 if ((flags>>=1) == 0) 145 return; 146 147 if (flags & 1) 148 ip_cmsg_recv_tos(msg, skb); 149 if ((flags>>=1) == 0) 150 return; 151 152 if (flags & 1) 153 ip_cmsg_recv_opts(msg, skb); 154 if ((flags>>=1) == 0) 155 return; 156 157 if (flags & 1) 158 ip_cmsg_recv_retopts(msg, skb); 159 if ((flags>>=1) == 0) 160 return; 161 162 if (flags & 1) 163 ip_cmsg_recv_security(msg, skb); 164 } 165 166 int ip_cmsg_send(struct msghdr *msg, struct ipcm_cookie *ipc) 167 { 168 int err; 169 struct cmsghdr *cmsg; 170 171 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) { 172 if (!CMSG_OK(msg, cmsg)) 173 return -EINVAL; 174 if (cmsg->cmsg_level != SOL_IP) 175 continue; 176 switch (cmsg->cmsg_type) { 177 case IP_RETOPTS: 178 err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr)); 179 err = ip_options_get(&ipc->opt, CMSG_DATA(cmsg), err < 40 ? err : 40); 180 if (err) 181 return err; 182 break; 183 case IP_PKTINFO: 184 { 185 struct in_pktinfo *info; 186 if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo))) 187 return -EINVAL; 188 info = (struct in_pktinfo *)CMSG_DATA(cmsg); 189 ipc->oif = info->ipi_ifindex; 190 ipc->addr = info->ipi_spec_dst.s_addr; 191 break; 192 } 193 default: 194 return -EINVAL; 195 } 196 } 197 return 0; 198 } 199 200 201 /* Special input handler for packets caught by router alert option. 202 They are selected only by protocol field, and then processed likely 203 local ones; but only if someone wants them! Otherwise, router 204 not running rsvpd will kill RSVP. 205 206 It is user level problem, what it will make with them. 207 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), 208 but receiver should be enough clever f.e. to forward mtrace requests, 209 sent to multicast group to reach destination designated router. 210 */ 211 struct ip_ra_chain *ip_ra_chain; 212 DEFINE_RWLOCK(ip_ra_lock); 213 214 int ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *)) 215 { 216 struct ip_ra_chain *ra, *new_ra, **rap; 217 218 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->num == IPPROTO_RAW) 219 return -EINVAL; 220 221 new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL; 222 223 write_lock_bh(&ip_ra_lock); 224 for (rap = &ip_ra_chain; (ra=*rap) != NULL; rap = &ra->next) { 225 if (ra->sk == sk) { 226 if (on) { 227 write_unlock_bh(&ip_ra_lock); 228 kfree(new_ra); 229 return -EADDRINUSE; 230 } 231 *rap = ra->next; 232 write_unlock_bh(&ip_ra_lock); 233 234 if (ra->destructor) 235 ra->destructor(sk); 236 sock_put(sk); 237 kfree(ra); 238 return 0; 239 } 240 } 241 if (new_ra == NULL) { 242 write_unlock_bh(&ip_ra_lock); 243 return -ENOBUFS; 244 } 245 new_ra->sk = sk; 246 new_ra->destructor = destructor; 247 248 new_ra->next = ra; 249 *rap = new_ra; 250 sock_hold(sk); 251 write_unlock_bh(&ip_ra_lock); 252 253 return 0; 254 } 255 256 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, 257 __be16 port, u32 info, u8 *payload) 258 { 259 struct inet_sock *inet = inet_sk(sk); 260 struct sock_exterr_skb *serr; 261 262 if (!inet->recverr) 263 return; 264 265 skb = skb_clone(skb, GFP_ATOMIC); 266 if (!skb) 267 return; 268 269 serr = SKB_EXT_ERR(skb); 270 serr->ee.ee_errno = err; 271 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP; 272 serr->ee.ee_type = icmp_hdr(skb)->type; 273 serr->ee.ee_code = icmp_hdr(skb)->code; 274 serr->ee.ee_pad = 0; 275 serr->ee.ee_info = info; 276 serr->ee.ee_data = 0; 277 serr->addr_offset = (u8 *)&(((struct iphdr *)(icmp_hdr(skb) + 1))->daddr) - 278 skb_network_header(skb); 279 serr->port = port; 280 281 if (skb_pull(skb, payload - skb->data) != NULL) { 282 skb_reset_transport_header(skb); 283 if (sock_queue_err_skb(sk, skb) == 0) 284 return; 285 } 286 kfree_skb(skb); 287 } 288 289 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info) 290 { 291 struct inet_sock *inet = inet_sk(sk); 292 struct sock_exterr_skb *serr; 293 struct iphdr *iph; 294 struct sk_buff *skb; 295 296 if (!inet->recverr) 297 return; 298 299 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC); 300 if (!skb) 301 return; 302 303 skb_put(skb, sizeof(struct iphdr)); 304 skb_reset_network_header(skb); 305 iph = ip_hdr(skb); 306 iph->daddr = daddr; 307 308 serr = SKB_EXT_ERR(skb); 309 serr->ee.ee_errno = err; 310 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL; 311 serr->ee.ee_type = 0; 312 serr->ee.ee_code = 0; 313 serr->ee.ee_pad = 0; 314 serr->ee.ee_info = info; 315 serr->ee.ee_data = 0; 316 serr->addr_offset = (u8 *)&iph->daddr - skb_network_header(skb); 317 serr->port = port; 318 319 __skb_pull(skb, skb_tail_pointer(skb) - skb->data); 320 skb_reset_transport_header(skb); 321 322 if (sock_queue_err_skb(sk, skb)) 323 kfree_skb(skb); 324 } 325 326 /* 327 * Handle MSG_ERRQUEUE 328 */ 329 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len) 330 { 331 struct sock_exterr_skb *serr; 332 struct sk_buff *skb, *skb2; 333 struct sockaddr_in *sin; 334 struct { 335 struct sock_extended_err ee; 336 struct sockaddr_in offender; 337 } errhdr; 338 int err; 339 int copied; 340 341 err = -EAGAIN; 342 skb = skb_dequeue(&sk->sk_error_queue); 343 if (skb == NULL) 344 goto out; 345 346 copied = skb->len; 347 if (copied > len) { 348 msg->msg_flags |= MSG_TRUNC; 349 copied = len; 350 } 351 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied); 352 if (err) 353 goto out_free_skb; 354 355 sock_recv_timestamp(msg, sk, skb); 356 357 serr = SKB_EXT_ERR(skb); 358 359 sin = (struct sockaddr_in *)msg->msg_name; 360 if (sin) { 361 sin->sin_family = AF_INET; 362 sin->sin_addr.s_addr = *(__be32 *)(skb_network_header(skb) + 363 serr->addr_offset); 364 sin->sin_port = serr->port; 365 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 366 } 367 368 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err)); 369 sin = &errhdr.offender; 370 sin->sin_family = AF_UNSPEC; 371 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) { 372 struct inet_sock *inet = inet_sk(sk); 373 374 sin->sin_family = AF_INET; 375 sin->sin_addr.s_addr = ip_hdr(skb)->saddr; 376 sin->sin_port = 0; 377 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 378 if (inet->cmsg_flags) 379 ip_cmsg_recv(msg, skb); 380 } 381 382 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr); 383 384 /* Now we could try to dump offended packet options */ 385 386 msg->msg_flags |= MSG_ERRQUEUE; 387 err = copied; 388 389 /* Reset and regenerate socket error */ 390 spin_lock_bh(&sk->sk_error_queue.lock); 391 sk->sk_err = 0; 392 if ((skb2 = skb_peek(&sk->sk_error_queue)) != NULL) { 393 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno; 394 spin_unlock_bh(&sk->sk_error_queue.lock); 395 sk->sk_error_report(sk); 396 } else 397 spin_unlock_bh(&sk->sk_error_queue.lock); 398 399 out_free_skb: 400 kfree_skb(skb); 401 out: 402 return err; 403 } 404 405 406 /* 407 * Socket option code for IP. This is the end of the line after any TCP,UDP etc options on 408 * an IP socket. 409 */ 410 411 static int do_ip_setsockopt(struct sock *sk, int level, 412 int optname, char __user *optval, int optlen) 413 { 414 struct inet_sock *inet = inet_sk(sk); 415 int val=0,err; 416 417 if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) | 418 (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) | 419 (1<<IP_RETOPTS) | (1<<IP_TOS) | 420 (1<<IP_TTL) | (1<<IP_HDRINCL) | 421 (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) | 422 (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND) | 423 (1<<IP_PASSSEC))) || 424 optname == IP_MULTICAST_TTL || 425 optname == IP_MULTICAST_LOOP) { 426 if (optlen >= sizeof(int)) { 427 if (get_user(val, (int __user *) optval)) 428 return -EFAULT; 429 } else if (optlen >= sizeof(char)) { 430 unsigned char ucval; 431 432 if (get_user(ucval, (unsigned char __user *) optval)) 433 return -EFAULT; 434 val = (int) ucval; 435 } 436 } 437 438 /* If optlen==0, it is equivalent to val == 0 */ 439 440 if (ip_mroute_opt(optname)) 441 return ip_mroute_setsockopt(sk,optname,optval,optlen); 442 443 err = 0; 444 lock_sock(sk); 445 446 switch (optname) { 447 case IP_OPTIONS: 448 { 449 struct ip_options * opt = NULL; 450 if (optlen > 40 || optlen < 0) 451 goto e_inval; 452 err = ip_options_get_from_user(&opt, optval, optlen); 453 if (err) 454 break; 455 if (inet->is_icsk) { 456 struct inet_connection_sock *icsk = inet_csk(sk); 457 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 458 if (sk->sk_family == PF_INET || 459 (!((1 << sk->sk_state) & 460 (TCPF_LISTEN | TCPF_CLOSE)) && 461 inet->daddr != LOOPBACK4_IPV6)) { 462 #endif 463 if (inet->opt) 464 icsk->icsk_ext_hdr_len -= inet->opt->optlen; 465 if (opt) 466 icsk->icsk_ext_hdr_len += opt->optlen; 467 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie); 468 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 469 } 470 #endif 471 } 472 opt = xchg(&inet->opt, opt); 473 kfree(opt); 474 break; 475 } 476 case IP_PKTINFO: 477 if (val) 478 inet->cmsg_flags |= IP_CMSG_PKTINFO; 479 else 480 inet->cmsg_flags &= ~IP_CMSG_PKTINFO; 481 break; 482 case IP_RECVTTL: 483 if (val) 484 inet->cmsg_flags |= IP_CMSG_TTL; 485 else 486 inet->cmsg_flags &= ~IP_CMSG_TTL; 487 break; 488 case IP_RECVTOS: 489 if (val) 490 inet->cmsg_flags |= IP_CMSG_TOS; 491 else 492 inet->cmsg_flags &= ~IP_CMSG_TOS; 493 break; 494 case IP_RECVOPTS: 495 if (val) 496 inet->cmsg_flags |= IP_CMSG_RECVOPTS; 497 else 498 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS; 499 break; 500 case IP_RETOPTS: 501 if (val) 502 inet->cmsg_flags |= IP_CMSG_RETOPTS; 503 else 504 inet->cmsg_flags &= ~IP_CMSG_RETOPTS; 505 break; 506 case IP_PASSSEC: 507 if (val) 508 inet->cmsg_flags |= IP_CMSG_PASSSEC; 509 else 510 inet->cmsg_flags &= ~IP_CMSG_PASSSEC; 511 break; 512 case IP_TOS: /* This sets both TOS and Precedence */ 513 if (sk->sk_type == SOCK_STREAM) { 514 val &= ~3; 515 val |= inet->tos & 3; 516 } 517 if (IPTOS_PREC(val) >= IPTOS_PREC_CRITIC_ECP && 518 !capable(CAP_NET_ADMIN)) { 519 err = -EPERM; 520 break; 521 } 522 if (inet->tos != val) { 523 inet->tos = val; 524 sk->sk_priority = rt_tos2priority(val); 525 sk_dst_reset(sk); 526 } 527 break; 528 case IP_TTL: 529 if (optlen<1) 530 goto e_inval; 531 if (val != -1 && (val < 1 || val>255)) 532 goto e_inval; 533 inet->uc_ttl = val; 534 break; 535 case IP_HDRINCL: 536 if (sk->sk_type != SOCK_RAW) { 537 err = -ENOPROTOOPT; 538 break; 539 } 540 inet->hdrincl = val ? 1 : 0; 541 break; 542 case IP_MTU_DISCOVER: 543 if (val<0 || val>3) 544 goto e_inval; 545 inet->pmtudisc = val; 546 break; 547 case IP_RECVERR: 548 inet->recverr = !!val; 549 if (!val) 550 skb_queue_purge(&sk->sk_error_queue); 551 break; 552 case IP_MULTICAST_TTL: 553 if (sk->sk_type == SOCK_STREAM) 554 goto e_inval; 555 if (optlen<1) 556 goto e_inval; 557 if (val==-1) 558 val = 1; 559 if (val < 0 || val > 255) 560 goto e_inval; 561 inet->mc_ttl = val; 562 break; 563 case IP_MULTICAST_LOOP: 564 if (optlen<1) 565 goto e_inval; 566 inet->mc_loop = !!val; 567 break; 568 case IP_MULTICAST_IF: 569 { 570 struct ip_mreqn mreq; 571 struct net_device *dev = NULL; 572 573 if (sk->sk_type == SOCK_STREAM) 574 goto e_inval; 575 /* 576 * Check the arguments are allowable 577 */ 578 579 err = -EFAULT; 580 if (optlen >= sizeof(struct ip_mreqn)) { 581 if (copy_from_user(&mreq,optval,sizeof(mreq))) 582 break; 583 } else { 584 memset(&mreq, 0, sizeof(mreq)); 585 if (optlen >= sizeof(struct in_addr) && 586 copy_from_user(&mreq.imr_address,optval,sizeof(struct in_addr))) 587 break; 588 } 589 590 if (!mreq.imr_ifindex) { 591 if (mreq.imr_address.s_addr == INADDR_ANY) { 592 inet->mc_index = 0; 593 inet->mc_addr = 0; 594 err = 0; 595 break; 596 } 597 dev = ip_dev_find(&init_net, mreq.imr_address.s_addr); 598 if (dev) { 599 mreq.imr_ifindex = dev->ifindex; 600 dev_put(dev); 601 } 602 } else 603 dev = __dev_get_by_index(&init_net, mreq.imr_ifindex); 604 605 606 err = -EADDRNOTAVAIL; 607 if (!dev) 608 break; 609 610 err = -EINVAL; 611 if (sk->sk_bound_dev_if && 612 mreq.imr_ifindex != sk->sk_bound_dev_if) 613 break; 614 615 inet->mc_index = mreq.imr_ifindex; 616 inet->mc_addr = mreq.imr_address.s_addr; 617 err = 0; 618 break; 619 } 620 621 case IP_ADD_MEMBERSHIP: 622 case IP_DROP_MEMBERSHIP: 623 { 624 struct ip_mreqn mreq; 625 626 err = -EPROTO; 627 if (inet_sk(sk)->is_icsk) 628 break; 629 630 if (optlen < sizeof(struct ip_mreq)) 631 goto e_inval; 632 err = -EFAULT; 633 if (optlen >= sizeof(struct ip_mreqn)) { 634 if (copy_from_user(&mreq,optval,sizeof(mreq))) 635 break; 636 } else { 637 memset(&mreq, 0, sizeof(mreq)); 638 if (copy_from_user(&mreq,optval,sizeof(struct ip_mreq))) 639 break; 640 } 641 642 if (optname == IP_ADD_MEMBERSHIP) 643 err = ip_mc_join_group(sk, &mreq); 644 else 645 err = ip_mc_leave_group(sk, &mreq); 646 break; 647 } 648 case IP_MSFILTER: 649 { 650 extern int sysctl_igmp_max_msf; 651 struct ip_msfilter *msf; 652 653 if (optlen < IP_MSFILTER_SIZE(0)) 654 goto e_inval; 655 if (optlen > sysctl_optmem_max) { 656 err = -ENOBUFS; 657 break; 658 } 659 msf = kmalloc(optlen, GFP_KERNEL); 660 if (!msf) { 661 err = -ENOBUFS; 662 break; 663 } 664 err = -EFAULT; 665 if (copy_from_user(msf, optval, optlen)) { 666 kfree(msf); 667 break; 668 } 669 /* numsrc >= (1G-4) overflow in 32 bits */ 670 if (msf->imsf_numsrc >= 0x3ffffffcU || 671 msf->imsf_numsrc > sysctl_igmp_max_msf) { 672 kfree(msf); 673 err = -ENOBUFS; 674 break; 675 } 676 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) { 677 kfree(msf); 678 err = -EINVAL; 679 break; 680 } 681 err = ip_mc_msfilter(sk, msf, 0); 682 kfree(msf); 683 break; 684 } 685 case IP_BLOCK_SOURCE: 686 case IP_UNBLOCK_SOURCE: 687 case IP_ADD_SOURCE_MEMBERSHIP: 688 case IP_DROP_SOURCE_MEMBERSHIP: 689 { 690 struct ip_mreq_source mreqs; 691 int omode, add; 692 693 if (optlen != sizeof(struct ip_mreq_source)) 694 goto e_inval; 695 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) { 696 err = -EFAULT; 697 break; 698 } 699 if (optname == IP_BLOCK_SOURCE) { 700 omode = MCAST_EXCLUDE; 701 add = 1; 702 } else if (optname == IP_UNBLOCK_SOURCE) { 703 omode = MCAST_EXCLUDE; 704 add = 0; 705 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) { 706 struct ip_mreqn mreq; 707 708 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr; 709 mreq.imr_address.s_addr = mreqs.imr_interface; 710 mreq.imr_ifindex = 0; 711 err = ip_mc_join_group(sk, &mreq); 712 if (err && err != -EADDRINUSE) 713 break; 714 omode = MCAST_INCLUDE; 715 add = 1; 716 } else /* IP_DROP_SOURCE_MEMBERSHIP */ { 717 omode = MCAST_INCLUDE; 718 add = 0; 719 } 720 err = ip_mc_source(add, omode, sk, &mreqs, 0); 721 break; 722 } 723 case MCAST_JOIN_GROUP: 724 case MCAST_LEAVE_GROUP: 725 { 726 struct group_req greq; 727 struct sockaddr_in *psin; 728 struct ip_mreqn mreq; 729 730 if (optlen < sizeof(struct group_req)) 731 goto e_inval; 732 err = -EFAULT; 733 if (copy_from_user(&greq, optval, sizeof(greq))) 734 break; 735 psin = (struct sockaddr_in *)&greq.gr_group; 736 if (psin->sin_family != AF_INET) 737 goto e_inval; 738 memset(&mreq, 0, sizeof(mreq)); 739 mreq.imr_multiaddr = psin->sin_addr; 740 mreq.imr_ifindex = greq.gr_interface; 741 742 if (optname == MCAST_JOIN_GROUP) 743 err = ip_mc_join_group(sk, &mreq); 744 else 745 err = ip_mc_leave_group(sk, &mreq); 746 break; 747 } 748 case MCAST_JOIN_SOURCE_GROUP: 749 case MCAST_LEAVE_SOURCE_GROUP: 750 case MCAST_BLOCK_SOURCE: 751 case MCAST_UNBLOCK_SOURCE: 752 { 753 struct group_source_req greqs; 754 struct ip_mreq_source mreqs; 755 struct sockaddr_in *psin; 756 int omode, add; 757 758 if (optlen != sizeof(struct group_source_req)) 759 goto e_inval; 760 if (copy_from_user(&greqs, optval, sizeof(greqs))) { 761 err = -EFAULT; 762 break; 763 } 764 if (greqs.gsr_group.ss_family != AF_INET || 765 greqs.gsr_source.ss_family != AF_INET) { 766 err = -EADDRNOTAVAIL; 767 break; 768 } 769 psin = (struct sockaddr_in *)&greqs.gsr_group; 770 mreqs.imr_multiaddr = psin->sin_addr.s_addr; 771 psin = (struct sockaddr_in *)&greqs.gsr_source; 772 mreqs.imr_sourceaddr = psin->sin_addr.s_addr; 773 mreqs.imr_interface = 0; /* use index for mc_source */ 774 775 if (optname == MCAST_BLOCK_SOURCE) { 776 omode = MCAST_EXCLUDE; 777 add = 1; 778 } else if (optname == MCAST_UNBLOCK_SOURCE) { 779 omode = MCAST_EXCLUDE; 780 add = 0; 781 } else if (optname == MCAST_JOIN_SOURCE_GROUP) { 782 struct ip_mreqn mreq; 783 784 psin = (struct sockaddr_in *)&greqs.gsr_group; 785 mreq.imr_multiaddr = psin->sin_addr; 786 mreq.imr_address.s_addr = 0; 787 mreq.imr_ifindex = greqs.gsr_interface; 788 err = ip_mc_join_group(sk, &mreq); 789 if (err && err != -EADDRINUSE) 790 break; 791 greqs.gsr_interface = mreq.imr_ifindex; 792 omode = MCAST_INCLUDE; 793 add = 1; 794 } else /* MCAST_LEAVE_SOURCE_GROUP */ { 795 omode = MCAST_INCLUDE; 796 add = 0; 797 } 798 err = ip_mc_source(add, omode, sk, &mreqs, 799 greqs.gsr_interface); 800 break; 801 } 802 case MCAST_MSFILTER: 803 { 804 extern int sysctl_igmp_max_msf; 805 struct sockaddr_in *psin; 806 struct ip_msfilter *msf = NULL; 807 struct group_filter *gsf = NULL; 808 int msize, i, ifindex; 809 810 if (optlen < GROUP_FILTER_SIZE(0)) 811 goto e_inval; 812 if (optlen > sysctl_optmem_max) { 813 err = -ENOBUFS; 814 break; 815 } 816 gsf = kmalloc(optlen,GFP_KERNEL); 817 if (!gsf) { 818 err = -ENOBUFS; 819 break; 820 } 821 err = -EFAULT; 822 if (copy_from_user(gsf, optval, optlen)) { 823 goto mc_msf_out; 824 } 825 /* numsrc >= (4G-140)/128 overflow in 32 bits */ 826 if (gsf->gf_numsrc >= 0x1ffffff || 827 gsf->gf_numsrc > sysctl_igmp_max_msf) { 828 err = -ENOBUFS; 829 goto mc_msf_out; 830 } 831 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) { 832 err = -EINVAL; 833 goto mc_msf_out; 834 } 835 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc); 836 msf = kmalloc(msize,GFP_KERNEL); 837 if (!msf) { 838 err = -ENOBUFS; 839 goto mc_msf_out; 840 } 841 ifindex = gsf->gf_interface; 842 psin = (struct sockaddr_in *)&gsf->gf_group; 843 if (psin->sin_family != AF_INET) { 844 err = -EADDRNOTAVAIL; 845 goto mc_msf_out; 846 } 847 msf->imsf_multiaddr = psin->sin_addr.s_addr; 848 msf->imsf_interface = 0; 849 msf->imsf_fmode = gsf->gf_fmode; 850 msf->imsf_numsrc = gsf->gf_numsrc; 851 err = -EADDRNOTAVAIL; 852 for (i=0; i<gsf->gf_numsrc; ++i) { 853 psin = (struct sockaddr_in *)&gsf->gf_slist[i]; 854 855 if (psin->sin_family != AF_INET) 856 goto mc_msf_out; 857 msf->imsf_slist[i] = psin->sin_addr.s_addr; 858 } 859 kfree(gsf); 860 gsf = NULL; 861 862 err = ip_mc_msfilter(sk, msf, ifindex); 863 mc_msf_out: 864 kfree(msf); 865 kfree(gsf); 866 break; 867 } 868 case IP_ROUTER_ALERT: 869 err = ip_ra_control(sk, val ? 1 : 0, NULL); 870 break; 871 872 case IP_FREEBIND: 873 if (optlen<1) 874 goto e_inval; 875 inet->freebind = !!val; 876 break; 877 878 case IP_IPSEC_POLICY: 879 case IP_XFRM_POLICY: 880 err = -EPERM; 881 if (!capable(CAP_NET_ADMIN)) 882 break; 883 err = xfrm_user_policy(sk, optname, optval, optlen); 884 break; 885 886 default: 887 err = -ENOPROTOOPT; 888 break; 889 } 890 release_sock(sk); 891 return err; 892 893 e_inval: 894 release_sock(sk); 895 return -EINVAL; 896 } 897 898 int ip_setsockopt(struct sock *sk, int level, 899 int optname, char __user *optval, int optlen) 900 { 901 int err; 902 903 if (level != SOL_IP) 904 return -ENOPROTOOPT; 905 906 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 907 #ifdef CONFIG_NETFILTER 908 /* we need to exclude all possible ENOPROTOOPTs except default case */ 909 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 910 optname != IP_IPSEC_POLICY && 911 optname != IP_XFRM_POLICY && 912 !ip_mroute_opt(optname)) { 913 lock_sock(sk); 914 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen); 915 release_sock(sk); 916 } 917 #endif 918 return err; 919 } 920 921 #ifdef CONFIG_COMPAT 922 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 923 char __user *optval, int optlen) 924 { 925 int err; 926 927 if (level != SOL_IP) 928 return -ENOPROTOOPT; 929 930 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 931 #ifdef CONFIG_NETFILTER 932 /* we need to exclude all possible ENOPROTOOPTs except default case */ 933 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 934 optname != IP_IPSEC_POLICY && 935 optname != IP_XFRM_POLICY && 936 !ip_mroute_opt(optname)) { 937 lock_sock(sk); 938 err = compat_nf_setsockopt(sk, PF_INET, optname, 939 optval, optlen); 940 release_sock(sk); 941 } 942 #endif 943 return err; 944 } 945 946 EXPORT_SYMBOL(compat_ip_setsockopt); 947 #endif 948 949 /* 950 * Get the options. Note for future reference. The GET of IP options gets the 951 * _received_ ones. The set sets the _sent_ ones. 952 */ 953 954 static int do_ip_getsockopt(struct sock *sk, int level, int optname, 955 char __user *optval, int __user *optlen) 956 { 957 struct inet_sock *inet = inet_sk(sk); 958 int val; 959 int len; 960 961 if (level != SOL_IP) 962 return -EOPNOTSUPP; 963 964 if (ip_mroute_opt(optname)) 965 return ip_mroute_getsockopt(sk,optname,optval,optlen); 966 967 if (get_user(len,optlen)) 968 return -EFAULT; 969 if (len < 0) 970 return -EINVAL; 971 972 lock_sock(sk); 973 974 switch (optname) { 975 case IP_OPTIONS: 976 { 977 unsigned char optbuf[sizeof(struct ip_options)+40]; 978 struct ip_options * opt = (struct ip_options*)optbuf; 979 opt->optlen = 0; 980 if (inet->opt) 981 memcpy(optbuf, inet->opt, 982 sizeof(struct ip_options)+ 983 inet->opt->optlen); 984 release_sock(sk); 985 986 if (opt->optlen == 0) 987 return put_user(0, optlen); 988 989 ip_options_undo(opt); 990 991 len = min_t(unsigned int, len, opt->optlen); 992 if (put_user(len, optlen)) 993 return -EFAULT; 994 if (copy_to_user(optval, opt->__data, len)) 995 return -EFAULT; 996 return 0; 997 } 998 case IP_PKTINFO: 999 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0; 1000 break; 1001 case IP_RECVTTL: 1002 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0; 1003 break; 1004 case IP_RECVTOS: 1005 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0; 1006 break; 1007 case IP_RECVOPTS: 1008 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0; 1009 break; 1010 case IP_RETOPTS: 1011 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0; 1012 break; 1013 case IP_PASSSEC: 1014 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0; 1015 break; 1016 case IP_TOS: 1017 val = inet->tos; 1018 break; 1019 case IP_TTL: 1020 val = (inet->uc_ttl == -1 ? 1021 sysctl_ip_default_ttl : 1022 inet->uc_ttl); 1023 break; 1024 case IP_HDRINCL: 1025 val = inet->hdrincl; 1026 break; 1027 case IP_MTU_DISCOVER: 1028 val = inet->pmtudisc; 1029 break; 1030 case IP_MTU: 1031 { 1032 struct dst_entry *dst; 1033 val = 0; 1034 dst = sk_dst_get(sk); 1035 if (dst) { 1036 val = dst_mtu(dst); 1037 dst_release(dst); 1038 } 1039 if (!val) { 1040 release_sock(sk); 1041 return -ENOTCONN; 1042 } 1043 break; 1044 } 1045 case IP_RECVERR: 1046 val = inet->recverr; 1047 break; 1048 case IP_MULTICAST_TTL: 1049 val = inet->mc_ttl; 1050 break; 1051 case IP_MULTICAST_LOOP: 1052 val = inet->mc_loop; 1053 break; 1054 case IP_MULTICAST_IF: 1055 { 1056 struct in_addr addr; 1057 len = min_t(unsigned int, len, sizeof(struct in_addr)); 1058 addr.s_addr = inet->mc_addr; 1059 release_sock(sk); 1060 1061 if (put_user(len, optlen)) 1062 return -EFAULT; 1063 if (copy_to_user(optval, &addr, len)) 1064 return -EFAULT; 1065 return 0; 1066 } 1067 case IP_MSFILTER: 1068 { 1069 struct ip_msfilter msf; 1070 int err; 1071 1072 if (len < IP_MSFILTER_SIZE(0)) { 1073 release_sock(sk); 1074 return -EINVAL; 1075 } 1076 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) { 1077 release_sock(sk); 1078 return -EFAULT; 1079 } 1080 err = ip_mc_msfget(sk, &msf, 1081 (struct ip_msfilter __user *)optval, optlen); 1082 release_sock(sk); 1083 return err; 1084 } 1085 case MCAST_MSFILTER: 1086 { 1087 struct group_filter gsf; 1088 int err; 1089 1090 if (len < GROUP_FILTER_SIZE(0)) { 1091 release_sock(sk); 1092 return -EINVAL; 1093 } 1094 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) { 1095 release_sock(sk); 1096 return -EFAULT; 1097 } 1098 err = ip_mc_gsfget(sk, &gsf, 1099 (struct group_filter __user *)optval, optlen); 1100 release_sock(sk); 1101 return err; 1102 } 1103 case IP_PKTOPTIONS: 1104 { 1105 struct msghdr msg; 1106 1107 release_sock(sk); 1108 1109 if (sk->sk_type != SOCK_STREAM) 1110 return -ENOPROTOOPT; 1111 1112 msg.msg_control = optval; 1113 msg.msg_controllen = len; 1114 msg.msg_flags = 0; 1115 1116 if (inet->cmsg_flags & IP_CMSG_PKTINFO) { 1117 struct in_pktinfo info; 1118 1119 info.ipi_addr.s_addr = inet->rcv_saddr; 1120 info.ipi_spec_dst.s_addr = inet->rcv_saddr; 1121 info.ipi_ifindex = inet->mc_index; 1122 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 1123 } 1124 if (inet->cmsg_flags & IP_CMSG_TTL) { 1125 int hlim = inet->mc_ttl; 1126 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim); 1127 } 1128 len -= msg.msg_controllen; 1129 return put_user(len, optlen); 1130 } 1131 case IP_FREEBIND: 1132 val = inet->freebind; 1133 break; 1134 default: 1135 release_sock(sk); 1136 return -ENOPROTOOPT; 1137 } 1138 release_sock(sk); 1139 1140 if (len < sizeof(int) && len > 0 && val>=0 && val<255) { 1141 unsigned char ucval = (unsigned char)val; 1142 len = 1; 1143 if (put_user(len, optlen)) 1144 return -EFAULT; 1145 if (copy_to_user(optval,&ucval,1)) 1146 return -EFAULT; 1147 } else { 1148 len = min_t(unsigned int, sizeof(int), len); 1149 if (put_user(len, optlen)) 1150 return -EFAULT; 1151 if (copy_to_user(optval,&val,len)) 1152 return -EFAULT; 1153 } 1154 return 0; 1155 } 1156 1157 int ip_getsockopt(struct sock *sk, int level, 1158 int optname, char __user *optval, int __user *optlen) 1159 { 1160 int err; 1161 1162 err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1163 #ifdef CONFIG_NETFILTER 1164 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1165 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS && 1166 !ip_mroute_opt(optname)) { 1167 int len; 1168 1169 if (get_user(len,optlen)) 1170 return -EFAULT; 1171 1172 lock_sock(sk); 1173 err = nf_getsockopt(sk, PF_INET, optname, optval, 1174 &len); 1175 release_sock(sk); 1176 if (err >= 0) 1177 err = put_user(len, optlen); 1178 return err; 1179 } 1180 #endif 1181 return err; 1182 } 1183 1184 #ifdef CONFIG_COMPAT 1185 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 1186 char __user *optval, int __user *optlen) 1187 { 1188 int err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1189 #ifdef CONFIG_NETFILTER 1190 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1191 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS && 1192 !ip_mroute_opt(optname)) { 1193 int len; 1194 1195 if (get_user(len, optlen)) 1196 return -EFAULT; 1197 1198 lock_sock(sk); 1199 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len); 1200 release_sock(sk); 1201 if (err >= 0) 1202 err = put_user(len, optlen); 1203 return err; 1204 } 1205 #endif 1206 return err; 1207 } 1208 1209 EXPORT_SYMBOL(compat_ip_getsockopt); 1210 #endif 1211 1212 EXPORT_SYMBOL(ip_cmsg_recv); 1213 1214 EXPORT_SYMBOL(ip_getsockopt); 1215 EXPORT_SYMBOL(ip_setsockopt); 1216