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 (inet->tos != val) { 518 inet->tos = val; 519 sk->sk_priority = rt_tos2priority(val); 520 sk_dst_reset(sk); 521 } 522 break; 523 case IP_TTL: 524 if (optlen<1) 525 goto e_inval; 526 if (val != -1 && (val < 1 || val>255)) 527 goto e_inval; 528 inet->uc_ttl = val; 529 break; 530 case IP_HDRINCL: 531 if (sk->sk_type != SOCK_RAW) { 532 err = -ENOPROTOOPT; 533 break; 534 } 535 inet->hdrincl = val ? 1 : 0; 536 break; 537 case IP_MTU_DISCOVER: 538 if (val<0 || val>3) 539 goto e_inval; 540 inet->pmtudisc = val; 541 break; 542 case IP_RECVERR: 543 inet->recverr = !!val; 544 if (!val) 545 skb_queue_purge(&sk->sk_error_queue); 546 break; 547 case IP_MULTICAST_TTL: 548 if (sk->sk_type == SOCK_STREAM) 549 goto e_inval; 550 if (optlen<1) 551 goto e_inval; 552 if (val==-1) 553 val = 1; 554 if (val < 0 || val > 255) 555 goto e_inval; 556 inet->mc_ttl = val; 557 break; 558 case IP_MULTICAST_LOOP: 559 if (optlen<1) 560 goto e_inval; 561 inet->mc_loop = !!val; 562 break; 563 case IP_MULTICAST_IF: 564 { 565 struct ip_mreqn mreq; 566 struct net_device *dev = NULL; 567 568 if (sk->sk_type == SOCK_STREAM) 569 goto e_inval; 570 /* 571 * Check the arguments are allowable 572 */ 573 574 err = -EFAULT; 575 if (optlen >= sizeof(struct ip_mreqn)) { 576 if (copy_from_user(&mreq,optval,sizeof(mreq))) 577 break; 578 } else { 579 memset(&mreq, 0, sizeof(mreq)); 580 if (optlen >= sizeof(struct in_addr) && 581 copy_from_user(&mreq.imr_address,optval,sizeof(struct in_addr))) 582 break; 583 } 584 585 if (!mreq.imr_ifindex) { 586 if (mreq.imr_address.s_addr == htonl(INADDR_ANY)) { 587 inet->mc_index = 0; 588 inet->mc_addr = 0; 589 err = 0; 590 break; 591 } 592 dev = ip_dev_find(&init_net, mreq.imr_address.s_addr); 593 if (dev) { 594 mreq.imr_ifindex = dev->ifindex; 595 dev_put(dev); 596 } 597 } else 598 dev = __dev_get_by_index(&init_net, mreq.imr_ifindex); 599 600 601 err = -EADDRNOTAVAIL; 602 if (!dev) 603 break; 604 605 err = -EINVAL; 606 if (sk->sk_bound_dev_if && 607 mreq.imr_ifindex != sk->sk_bound_dev_if) 608 break; 609 610 inet->mc_index = mreq.imr_ifindex; 611 inet->mc_addr = mreq.imr_address.s_addr; 612 err = 0; 613 break; 614 } 615 616 case IP_ADD_MEMBERSHIP: 617 case IP_DROP_MEMBERSHIP: 618 { 619 struct ip_mreqn mreq; 620 621 err = -EPROTO; 622 if (inet_sk(sk)->is_icsk) 623 break; 624 625 if (optlen < sizeof(struct ip_mreq)) 626 goto e_inval; 627 err = -EFAULT; 628 if (optlen >= sizeof(struct ip_mreqn)) { 629 if (copy_from_user(&mreq,optval,sizeof(mreq))) 630 break; 631 } else { 632 memset(&mreq, 0, sizeof(mreq)); 633 if (copy_from_user(&mreq,optval,sizeof(struct ip_mreq))) 634 break; 635 } 636 637 if (optname == IP_ADD_MEMBERSHIP) 638 err = ip_mc_join_group(sk, &mreq); 639 else 640 err = ip_mc_leave_group(sk, &mreq); 641 break; 642 } 643 case IP_MSFILTER: 644 { 645 extern int sysctl_igmp_max_msf; 646 struct ip_msfilter *msf; 647 648 if (optlen < IP_MSFILTER_SIZE(0)) 649 goto e_inval; 650 if (optlen > sysctl_optmem_max) { 651 err = -ENOBUFS; 652 break; 653 } 654 msf = kmalloc(optlen, GFP_KERNEL); 655 if (!msf) { 656 err = -ENOBUFS; 657 break; 658 } 659 err = -EFAULT; 660 if (copy_from_user(msf, optval, optlen)) { 661 kfree(msf); 662 break; 663 } 664 /* numsrc >= (1G-4) overflow in 32 bits */ 665 if (msf->imsf_numsrc >= 0x3ffffffcU || 666 msf->imsf_numsrc > sysctl_igmp_max_msf) { 667 kfree(msf); 668 err = -ENOBUFS; 669 break; 670 } 671 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) { 672 kfree(msf); 673 err = -EINVAL; 674 break; 675 } 676 err = ip_mc_msfilter(sk, msf, 0); 677 kfree(msf); 678 break; 679 } 680 case IP_BLOCK_SOURCE: 681 case IP_UNBLOCK_SOURCE: 682 case IP_ADD_SOURCE_MEMBERSHIP: 683 case IP_DROP_SOURCE_MEMBERSHIP: 684 { 685 struct ip_mreq_source mreqs; 686 int omode, add; 687 688 if (optlen != sizeof(struct ip_mreq_source)) 689 goto e_inval; 690 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) { 691 err = -EFAULT; 692 break; 693 } 694 if (optname == IP_BLOCK_SOURCE) { 695 omode = MCAST_EXCLUDE; 696 add = 1; 697 } else if (optname == IP_UNBLOCK_SOURCE) { 698 omode = MCAST_EXCLUDE; 699 add = 0; 700 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) { 701 struct ip_mreqn mreq; 702 703 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr; 704 mreq.imr_address.s_addr = mreqs.imr_interface; 705 mreq.imr_ifindex = 0; 706 err = ip_mc_join_group(sk, &mreq); 707 if (err && err != -EADDRINUSE) 708 break; 709 omode = MCAST_INCLUDE; 710 add = 1; 711 } else /* IP_DROP_SOURCE_MEMBERSHIP */ { 712 omode = MCAST_INCLUDE; 713 add = 0; 714 } 715 err = ip_mc_source(add, omode, sk, &mreqs, 0); 716 break; 717 } 718 case MCAST_JOIN_GROUP: 719 case MCAST_LEAVE_GROUP: 720 { 721 struct group_req greq; 722 struct sockaddr_in *psin; 723 struct ip_mreqn mreq; 724 725 if (optlen < sizeof(struct group_req)) 726 goto e_inval; 727 err = -EFAULT; 728 if (copy_from_user(&greq, optval, sizeof(greq))) 729 break; 730 psin = (struct sockaddr_in *)&greq.gr_group; 731 if (psin->sin_family != AF_INET) 732 goto e_inval; 733 memset(&mreq, 0, sizeof(mreq)); 734 mreq.imr_multiaddr = psin->sin_addr; 735 mreq.imr_ifindex = greq.gr_interface; 736 737 if (optname == MCAST_JOIN_GROUP) 738 err = ip_mc_join_group(sk, &mreq); 739 else 740 err = ip_mc_leave_group(sk, &mreq); 741 break; 742 } 743 case MCAST_JOIN_SOURCE_GROUP: 744 case MCAST_LEAVE_SOURCE_GROUP: 745 case MCAST_BLOCK_SOURCE: 746 case MCAST_UNBLOCK_SOURCE: 747 { 748 struct group_source_req greqs; 749 struct ip_mreq_source mreqs; 750 struct sockaddr_in *psin; 751 int omode, add; 752 753 if (optlen != sizeof(struct group_source_req)) 754 goto e_inval; 755 if (copy_from_user(&greqs, optval, sizeof(greqs))) { 756 err = -EFAULT; 757 break; 758 } 759 if (greqs.gsr_group.ss_family != AF_INET || 760 greqs.gsr_source.ss_family != AF_INET) { 761 err = -EADDRNOTAVAIL; 762 break; 763 } 764 psin = (struct sockaddr_in *)&greqs.gsr_group; 765 mreqs.imr_multiaddr = psin->sin_addr.s_addr; 766 psin = (struct sockaddr_in *)&greqs.gsr_source; 767 mreqs.imr_sourceaddr = psin->sin_addr.s_addr; 768 mreqs.imr_interface = 0; /* use index for mc_source */ 769 770 if (optname == MCAST_BLOCK_SOURCE) { 771 omode = MCAST_EXCLUDE; 772 add = 1; 773 } else if (optname == MCAST_UNBLOCK_SOURCE) { 774 omode = MCAST_EXCLUDE; 775 add = 0; 776 } else if (optname == MCAST_JOIN_SOURCE_GROUP) { 777 struct ip_mreqn mreq; 778 779 psin = (struct sockaddr_in *)&greqs.gsr_group; 780 mreq.imr_multiaddr = psin->sin_addr; 781 mreq.imr_address.s_addr = 0; 782 mreq.imr_ifindex = greqs.gsr_interface; 783 err = ip_mc_join_group(sk, &mreq); 784 if (err && err != -EADDRINUSE) 785 break; 786 greqs.gsr_interface = mreq.imr_ifindex; 787 omode = MCAST_INCLUDE; 788 add = 1; 789 } else /* MCAST_LEAVE_SOURCE_GROUP */ { 790 omode = MCAST_INCLUDE; 791 add = 0; 792 } 793 err = ip_mc_source(add, omode, sk, &mreqs, 794 greqs.gsr_interface); 795 break; 796 } 797 case MCAST_MSFILTER: 798 { 799 extern int sysctl_igmp_max_msf; 800 struct sockaddr_in *psin; 801 struct ip_msfilter *msf = NULL; 802 struct group_filter *gsf = NULL; 803 int msize, i, ifindex; 804 805 if (optlen < GROUP_FILTER_SIZE(0)) 806 goto e_inval; 807 if (optlen > sysctl_optmem_max) { 808 err = -ENOBUFS; 809 break; 810 } 811 gsf = kmalloc(optlen,GFP_KERNEL); 812 if (!gsf) { 813 err = -ENOBUFS; 814 break; 815 } 816 err = -EFAULT; 817 if (copy_from_user(gsf, optval, optlen)) { 818 goto mc_msf_out; 819 } 820 /* numsrc >= (4G-140)/128 overflow in 32 bits */ 821 if (gsf->gf_numsrc >= 0x1ffffff || 822 gsf->gf_numsrc > sysctl_igmp_max_msf) { 823 err = -ENOBUFS; 824 goto mc_msf_out; 825 } 826 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) { 827 err = -EINVAL; 828 goto mc_msf_out; 829 } 830 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc); 831 msf = kmalloc(msize,GFP_KERNEL); 832 if (!msf) { 833 err = -ENOBUFS; 834 goto mc_msf_out; 835 } 836 ifindex = gsf->gf_interface; 837 psin = (struct sockaddr_in *)&gsf->gf_group; 838 if (psin->sin_family != AF_INET) { 839 err = -EADDRNOTAVAIL; 840 goto mc_msf_out; 841 } 842 msf->imsf_multiaddr = psin->sin_addr.s_addr; 843 msf->imsf_interface = 0; 844 msf->imsf_fmode = gsf->gf_fmode; 845 msf->imsf_numsrc = gsf->gf_numsrc; 846 err = -EADDRNOTAVAIL; 847 for (i=0; i<gsf->gf_numsrc; ++i) { 848 psin = (struct sockaddr_in *)&gsf->gf_slist[i]; 849 850 if (psin->sin_family != AF_INET) 851 goto mc_msf_out; 852 msf->imsf_slist[i] = psin->sin_addr.s_addr; 853 } 854 kfree(gsf); 855 gsf = NULL; 856 857 err = ip_mc_msfilter(sk, msf, ifindex); 858 mc_msf_out: 859 kfree(msf); 860 kfree(gsf); 861 break; 862 } 863 case IP_ROUTER_ALERT: 864 err = ip_ra_control(sk, val ? 1 : 0, NULL); 865 break; 866 867 case IP_FREEBIND: 868 if (optlen<1) 869 goto e_inval; 870 inet->freebind = !!val; 871 break; 872 873 case IP_IPSEC_POLICY: 874 case IP_XFRM_POLICY: 875 err = -EPERM; 876 if (!capable(CAP_NET_ADMIN)) 877 break; 878 err = xfrm_user_policy(sk, optname, optval, optlen); 879 break; 880 881 default: 882 err = -ENOPROTOOPT; 883 break; 884 } 885 release_sock(sk); 886 return err; 887 888 e_inval: 889 release_sock(sk); 890 return -EINVAL; 891 } 892 893 int ip_setsockopt(struct sock *sk, int level, 894 int optname, char __user *optval, int optlen) 895 { 896 int err; 897 898 if (level != SOL_IP) 899 return -ENOPROTOOPT; 900 901 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 902 #ifdef CONFIG_NETFILTER 903 /* we need to exclude all possible ENOPROTOOPTs except default case */ 904 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 905 optname != IP_IPSEC_POLICY && 906 optname != IP_XFRM_POLICY && 907 !ip_mroute_opt(optname)) { 908 lock_sock(sk); 909 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen); 910 release_sock(sk); 911 } 912 #endif 913 return err; 914 } 915 916 #ifdef CONFIG_COMPAT 917 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 918 char __user *optval, int optlen) 919 { 920 int err; 921 922 if (level != SOL_IP) 923 return -ENOPROTOOPT; 924 925 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 926 #ifdef CONFIG_NETFILTER 927 /* we need to exclude all possible ENOPROTOOPTs except default case */ 928 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 929 optname != IP_IPSEC_POLICY && 930 optname != IP_XFRM_POLICY && 931 !ip_mroute_opt(optname)) { 932 lock_sock(sk); 933 err = compat_nf_setsockopt(sk, PF_INET, optname, 934 optval, optlen); 935 release_sock(sk); 936 } 937 #endif 938 return err; 939 } 940 941 EXPORT_SYMBOL(compat_ip_setsockopt); 942 #endif 943 944 /* 945 * Get the options. Note for future reference. The GET of IP options gets the 946 * _received_ ones. The set sets the _sent_ ones. 947 */ 948 949 static int do_ip_getsockopt(struct sock *sk, int level, int optname, 950 char __user *optval, int __user *optlen) 951 { 952 struct inet_sock *inet = inet_sk(sk); 953 int val; 954 int len; 955 956 if (level != SOL_IP) 957 return -EOPNOTSUPP; 958 959 if (ip_mroute_opt(optname)) 960 return ip_mroute_getsockopt(sk,optname,optval,optlen); 961 962 if (get_user(len,optlen)) 963 return -EFAULT; 964 if (len < 0) 965 return -EINVAL; 966 967 lock_sock(sk); 968 969 switch (optname) { 970 case IP_OPTIONS: 971 { 972 unsigned char optbuf[sizeof(struct ip_options)+40]; 973 struct ip_options * opt = (struct ip_options*)optbuf; 974 opt->optlen = 0; 975 if (inet->opt) 976 memcpy(optbuf, inet->opt, 977 sizeof(struct ip_options)+ 978 inet->opt->optlen); 979 release_sock(sk); 980 981 if (opt->optlen == 0) 982 return put_user(0, optlen); 983 984 ip_options_undo(opt); 985 986 len = min_t(unsigned int, len, opt->optlen); 987 if (put_user(len, optlen)) 988 return -EFAULT; 989 if (copy_to_user(optval, opt->__data, len)) 990 return -EFAULT; 991 return 0; 992 } 993 case IP_PKTINFO: 994 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0; 995 break; 996 case IP_RECVTTL: 997 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0; 998 break; 999 case IP_RECVTOS: 1000 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0; 1001 break; 1002 case IP_RECVOPTS: 1003 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0; 1004 break; 1005 case IP_RETOPTS: 1006 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0; 1007 break; 1008 case IP_PASSSEC: 1009 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0; 1010 break; 1011 case IP_TOS: 1012 val = inet->tos; 1013 break; 1014 case IP_TTL: 1015 val = (inet->uc_ttl == -1 ? 1016 sysctl_ip_default_ttl : 1017 inet->uc_ttl); 1018 break; 1019 case IP_HDRINCL: 1020 val = inet->hdrincl; 1021 break; 1022 case IP_MTU_DISCOVER: 1023 val = inet->pmtudisc; 1024 break; 1025 case IP_MTU: 1026 { 1027 struct dst_entry *dst; 1028 val = 0; 1029 dst = sk_dst_get(sk); 1030 if (dst) { 1031 val = dst_mtu(dst); 1032 dst_release(dst); 1033 } 1034 if (!val) { 1035 release_sock(sk); 1036 return -ENOTCONN; 1037 } 1038 break; 1039 } 1040 case IP_RECVERR: 1041 val = inet->recverr; 1042 break; 1043 case IP_MULTICAST_TTL: 1044 val = inet->mc_ttl; 1045 break; 1046 case IP_MULTICAST_LOOP: 1047 val = inet->mc_loop; 1048 break; 1049 case IP_MULTICAST_IF: 1050 { 1051 struct in_addr addr; 1052 len = min_t(unsigned int, len, sizeof(struct in_addr)); 1053 addr.s_addr = inet->mc_addr; 1054 release_sock(sk); 1055 1056 if (put_user(len, optlen)) 1057 return -EFAULT; 1058 if (copy_to_user(optval, &addr, len)) 1059 return -EFAULT; 1060 return 0; 1061 } 1062 case IP_MSFILTER: 1063 { 1064 struct ip_msfilter msf; 1065 int err; 1066 1067 if (len < IP_MSFILTER_SIZE(0)) { 1068 release_sock(sk); 1069 return -EINVAL; 1070 } 1071 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) { 1072 release_sock(sk); 1073 return -EFAULT; 1074 } 1075 err = ip_mc_msfget(sk, &msf, 1076 (struct ip_msfilter __user *)optval, optlen); 1077 release_sock(sk); 1078 return err; 1079 } 1080 case MCAST_MSFILTER: 1081 { 1082 struct group_filter gsf; 1083 int err; 1084 1085 if (len < GROUP_FILTER_SIZE(0)) { 1086 release_sock(sk); 1087 return -EINVAL; 1088 } 1089 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) { 1090 release_sock(sk); 1091 return -EFAULT; 1092 } 1093 err = ip_mc_gsfget(sk, &gsf, 1094 (struct group_filter __user *)optval, optlen); 1095 release_sock(sk); 1096 return err; 1097 } 1098 case IP_PKTOPTIONS: 1099 { 1100 struct msghdr msg; 1101 1102 release_sock(sk); 1103 1104 if (sk->sk_type != SOCK_STREAM) 1105 return -ENOPROTOOPT; 1106 1107 msg.msg_control = optval; 1108 msg.msg_controllen = len; 1109 msg.msg_flags = 0; 1110 1111 if (inet->cmsg_flags & IP_CMSG_PKTINFO) { 1112 struct in_pktinfo info; 1113 1114 info.ipi_addr.s_addr = inet->rcv_saddr; 1115 info.ipi_spec_dst.s_addr = inet->rcv_saddr; 1116 info.ipi_ifindex = inet->mc_index; 1117 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info); 1118 } 1119 if (inet->cmsg_flags & IP_CMSG_TTL) { 1120 int hlim = inet->mc_ttl; 1121 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim); 1122 } 1123 len -= msg.msg_controllen; 1124 return put_user(len, optlen); 1125 } 1126 case IP_FREEBIND: 1127 val = inet->freebind; 1128 break; 1129 default: 1130 release_sock(sk); 1131 return -ENOPROTOOPT; 1132 } 1133 release_sock(sk); 1134 1135 if (len < sizeof(int) && len > 0 && val>=0 && val<255) { 1136 unsigned char ucval = (unsigned char)val; 1137 len = 1; 1138 if (put_user(len, optlen)) 1139 return -EFAULT; 1140 if (copy_to_user(optval,&ucval,1)) 1141 return -EFAULT; 1142 } else { 1143 len = min_t(unsigned int, sizeof(int), len); 1144 if (put_user(len, optlen)) 1145 return -EFAULT; 1146 if (copy_to_user(optval,&val,len)) 1147 return -EFAULT; 1148 } 1149 return 0; 1150 } 1151 1152 int ip_getsockopt(struct sock *sk, int level, 1153 int optname, char __user *optval, int __user *optlen) 1154 { 1155 int err; 1156 1157 err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1158 #ifdef CONFIG_NETFILTER 1159 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1160 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS && 1161 !ip_mroute_opt(optname)) { 1162 int len; 1163 1164 if (get_user(len,optlen)) 1165 return -EFAULT; 1166 1167 lock_sock(sk); 1168 err = nf_getsockopt(sk, PF_INET, optname, optval, 1169 &len); 1170 release_sock(sk); 1171 if (err >= 0) 1172 err = put_user(len, optlen); 1173 return err; 1174 } 1175 #endif 1176 return err; 1177 } 1178 1179 #ifdef CONFIG_COMPAT 1180 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 1181 char __user *optval, int __user *optlen) 1182 { 1183 int err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1184 #ifdef CONFIG_NETFILTER 1185 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1186 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS && 1187 !ip_mroute_opt(optname)) { 1188 int len; 1189 1190 if (get_user(len, optlen)) 1191 return -EFAULT; 1192 1193 lock_sock(sk); 1194 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len); 1195 release_sock(sk); 1196 if (err >= 0) 1197 err = put_user(len, optlen); 1198 return err; 1199 } 1200 #endif 1201 return err; 1202 } 1203 1204 EXPORT_SYMBOL(compat_ip_getsockopt); 1205 #endif 1206 1207 EXPORT_SYMBOL(ip_cmsg_recv); 1208 1209 EXPORT_SYMBOL(ip_getsockopt); 1210 EXPORT_SYMBOL(ip_setsockopt); 1211