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 = skb->nh.iph->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 = skb->nh.iph->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, &skb->nh.iph->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, skb->nh.iph+1); 91 } 92 93 94 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb) 95 { 96 unsigned char optbuf[sizeof(struct ip_options) + 40]; 97 struct ip_options * opt = (struct ip_options*)optbuf; 98 99 if (IPCB(skb)->opt.optlen == 0) 100 return; 101 102 if (ip_options_echo(opt, skb)) { 103 msg->msg_flags |= MSG_CTRUNC; 104 return; 105 } 106 ip_options_undo(opt); 107 108 put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data); 109 } 110 111 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb) 112 { 113 char *secdata; 114 u32 seclen, secid; 115 int err; 116 117 err = security_socket_getpeersec_dgram(NULL, skb, &secid); 118 if (err) 119 return; 120 121 err = security_secid_to_secctx(secid, &secdata, &seclen); 122 if (err) 123 return; 124 125 put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata); 126 security_release_secctx(secdata, seclen); 127 } 128 129 130 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 131 { 132 struct inet_sock *inet = inet_sk(skb->sk); 133 unsigned flags = inet->cmsg_flags; 134 135 /* Ordered by supposed usage frequency */ 136 if (flags & 1) 137 ip_cmsg_recv_pktinfo(msg, skb); 138 if ((flags>>=1) == 0) 139 return; 140 141 if (flags & 1) 142 ip_cmsg_recv_ttl(msg, skb); 143 if ((flags>>=1) == 0) 144 return; 145 146 if (flags & 1) 147 ip_cmsg_recv_tos(msg, skb); 148 if ((flags>>=1) == 0) 149 return; 150 151 if (flags & 1) 152 ip_cmsg_recv_opts(msg, skb); 153 if ((flags>>=1) == 0) 154 return; 155 156 if (flags & 1) 157 ip_cmsg_recv_retopts(msg, skb); 158 if ((flags>>=1) == 0) 159 return; 160 161 if (flags & 1) 162 ip_cmsg_recv_security(msg, skb); 163 } 164 165 int ip_cmsg_send(struct msghdr *msg, struct ipcm_cookie *ipc) 166 { 167 int err; 168 struct cmsghdr *cmsg; 169 170 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) { 171 if (!CMSG_OK(msg, cmsg)) 172 return -EINVAL; 173 if (cmsg->cmsg_level != SOL_IP) 174 continue; 175 switch (cmsg->cmsg_type) { 176 case IP_RETOPTS: 177 err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr)); 178 err = ip_options_get(&ipc->opt, CMSG_DATA(cmsg), err < 40 ? err : 40); 179 if (err) 180 return err; 181 break; 182 case IP_PKTINFO: 183 { 184 struct in_pktinfo *info; 185 if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo))) 186 return -EINVAL; 187 info = (struct in_pktinfo *)CMSG_DATA(cmsg); 188 ipc->oif = info->ipi_ifindex; 189 ipc->addr = info->ipi_spec_dst.s_addr; 190 break; 191 } 192 default: 193 return -EINVAL; 194 } 195 } 196 return 0; 197 } 198 199 200 /* Special input handler for packets caught by router alert option. 201 They are selected only by protocol field, and then processed likely 202 local ones; but only if someone wants them! Otherwise, router 203 not running rsvpd will kill RSVP. 204 205 It is user level problem, what it will make with them. 206 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), 207 but receiver should be enough clever f.e. to forward mtrace requests, 208 sent to multicast group to reach destination designated router. 209 */ 210 struct ip_ra_chain *ip_ra_chain; 211 DEFINE_RWLOCK(ip_ra_lock); 212 213 int ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *)) 214 { 215 struct ip_ra_chain *ra, *new_ra, **rap; 216 217 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->num == IPPROTO_RAW) 218 return -EINVAL; 219 220 new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL; 221 222 write_lock_bh(&ip_ra_lock); 223 for (rap = &ip_ra_chain; (ra=*rap) != NULL; rap = &ra->next) { 224 if (ra->sk == sk) { 225 if (on) { 226 write_unlock_bh(&ip_ra_lock); 227 kfree(new_ra); 228 return -EADDRINUSE; 229 } 230 *rap = ra->next; 231 write_unlock_bh(&ip_ra_lock); 232 233 if (ra->destructor) 234 ra->destructor(sk); 235 sock_put(sk); 236 kfree(ra); 237 return 0; 238 } 239 } 240 if (new_ra == NULL) { 241 write_unlock_bh(&ip_ra_lock); 242 return -ENOBUFS; 243 } 244 new_ra->sk = sk; 245 new_ra->destructor = destructor; 246 247 new_ra->next = ra; 248 *rap = new_ra; 249 sock_hold(sk); 250 write_unlock_bh(&ip_ra_lock); 251 252 return 0; 253 } 254 255 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, 256 __be16 port, u32 info, u8 *payload) 257 { 258 struct inet_sock *inet = inet_sk(sk); 259 struct sock_exterr_skb *serr; 260 261 if (!inet->recverr) 262 return; 263 264 skb = skb_clone(skb, GFP_ATOMIC); 265 if (!skb) 266 return; 267 268 serr = SKB_EXT_ERR(skb); 269 serr->ee.ee_errno = err; 270 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP; 271 serr->ee.ee_type = skb->h.icmph->type; 272 serr->ee.ee_code = skb->h.icmph->code; 273 serr->ee.ee_pad = 0; 274 serr->ee.ee_info = info; 275 serr->ee.ee_data = 0; 276 serr->addr_offset = (u8*)&(((struct iphdr*)(skb->h.icmph+1))->daddr) - skb->nh.raw; 277 serr->port = port; 278 279 skb->h.raw = payload; 280 if (!skb_pull(skb, payload - skb->data) || 281 sock_queue_err_skb(sk, skb)) 282 kfree_skb(skb); 283 } 284 285 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info) 286 { 287 struct inet_sock *inet = inet_sk(sk); 288 struct sock_exterr_skb *serr; 289 struct iphdr *iph; 290 struct sk_buff *skb; 291 292 if (!inet->recverr) 293 return; 294 295 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC); 296 if (!skb) 297 return; 298 299 iph = (struct iphdr*)skb_put(skb, sizeof(struct iphdr)); 300 skb->nh.iph = iph; 301 iph->daddr = daddr; 302 303 serr = SKB_EXT_ERR(skb); 304 serr->ee.ee_errno = err; 305 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL; 306 serr->ee.ee_type = 0; 307 serr->ee.ee_code = 0; 308 serr->ee.ee_pad = 0; 309 serr->ee.ee_info = info; 310 serr->ee.ee_data = 0; 311 serr->addr_offset = (u8*)&iph->daddr - skb->nh.raw; 312 serr->port = port; 313 314 skb->h.raw = skb->tail; 315 __skb_pull(skb, skb->tail - skb->data); 316 317 if (sock_queue_err_skb(sk, skb)) 318 kfree_skb(skb); 319 } 320 321 /* 322 * Handle MSG_ERRQUEUE 323 */ 324 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len) 325 { 326 struct sock_exterr_skb *serr; 327 struct sk_buff *skb, *skb2; 328 struct sockaddr_in *sin; 329 struct { 330 struct sock_extended_err ee; 331 struct sockaddr_in offender; 332 } errhdr; 333 int err; 334 int copied; 335 336 err = -EAGAIN; 337 skb = skb_dequeue(&sk->sk_error_queue); 338 if (skb == NULL) 339 goto out; 340 341 copied = skb->len; 342 if (copied > len) { 343 msg->msg_flags |= MSG_TRUNC; 344 copied = len; 345 } 346 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied); 347 if (err) 348 goto out_free_skb; 349 350 sock_recv_timestamp(msg, sk, skb); 351 352 serr = SKB_EXT_ERR(skb); 353 354 sin = (struct sockaddr_in *)msg->msg_name; 355 if (sin) { 356 sin->sin_family = AF_INET; 357 sin->sin_addr.s_addr = *(__be32*)(skb->nh.raw + serr->addr_offset); 358 sin->sin_port = serr->port; 359 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 360 } 361 362 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err)); 363 sin = &errhdr.offender; 364 sin->sin_family = AF_UNSPEC; 365 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) { 366 struct inet_sock *inet = inet_sk(sk); 367 368 sin->sin_family = AF_INET; 369 sin->sin_addr.s_addr = skb->nh.iph->saddr; 370 sin->sin_port = 0; 371 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); 372 if (inet->cmsg_flags) 373 ip_cmsg_recv(msg, skb); 374 } 375 376 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr); 377 378 /* Now we could try to dump offended packet options */ 379 380 msg->msg_flags |= MSG_ERRQUEUE; 381 err = copied; 382 383 /* Reset and regenerate socket error */ 384 spin_lock_bh(&sk->sk_error_queue.lock); 385 sk->sk_err = 0; 386 if ((skb2 = skb_peek(&sk->sk_error_queue)) != NULL) { 387 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno; 388 spin_unlock_bh(&sk->sk_error_queue.lock); 389 sk->sk_error_report(sk); 390 } else 391 spin_unlock_bh(&sk->sk_error_queue.lock); 392 393 out_free_skb: 394 kfree_skb(skb); 395 out: 396 return err; 397 } 398 399 400 /* 401 * Socket option code for IP. This is the end of the line after any TCP,UDP etc options on 402 * an IP socket. 403 */ 404 405 static int do_ip_setsockopt(struct sock *sk, int level, 406 int optname, char __user *optval, int optlen) 407 { 408 struct inet_sock *inet = inet_sk(sk); 409 int val=0,err; 410 411 if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) | 412 (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) | 413 (1<<IP_RETOPTS) | (1<<IP_TOS) | 414 (1<<IP_TTL) | (1<<IP_HDRINCL) | 415 (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) | 416 (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND) | 417 (1<<IP_PASSSEC))) || 418 optname == IP_MULTICAST_TTL || 419 optname == IP_MULTICAST_LOOP) { 420 if (optlen >= sizeof(int)) { 421 if (get_user(val, (int __user *) optval)) 422 return -EFAULT; 423 } else if (optlen >= sizeof(char)) { 424 unsigned char ucval; 425 426 if (get_user(ucval, (unsigned char __user *) optval)) 427 return -EFAULT; 428 val = (int) ucval; 429 } 430 } 431 432 /* If optlen==0, it is equivalent to val == 0 */ 433 434 #ifdef CONFIG_IP_MROUTE 435 if (optname >= MRT_BASE && optname <= (MRT_BASE + 10)) 436 return ip_mroute_setsockopt(sk,optname,optval,optlen); 437 #endif 438 439 err = 0; 440 lock_sock(sk); 441 442 switch (optname) { 443 case IP_OPTIONS: 444 { 445 struct ip_options * opt = NULL; 446 if (optlen > 40 || optlen < 0) 447 goto e_inval; 448 err = ip_options_get_from_user(&opt, optval, optlen); 449 if (err) 450 break; 451 if (inet->is_icsk) { 452 struct inet_connection_sock *icsk = inet_csk(sk); 453 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 454 if (sk->sk_family == PF_INET || 455 (!((1 << sk->sk_state) & 456 (TCPF_LISTEN | TCPF_CLOSE)) && 457 inet->daddr != LOOPBACK4_IPV6)) { 458 #endif 459 if (inet->opt) 460 icsk->icsk_ext_hdr_len -= inet->opt->optlen; 461 if (opt) 462 icsk->icsk_ext_hdr_len += opt->optlen; 463 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie); 464 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) 465 } 466 #endif 467 } 468 opt = xchg(&inet->opt, opt); 469 kfree(opt); 470 break; 471 } 472 case IP_PKTINFO: 473 if (val) 474 inet->cmsg_flags |= IP_CMSG_PKTINFO; 475 else 476 inet->cmsg_flags &= ~IP_CMSG_PKTINFO; 477 break; 478 case IP_RECVTTL: 479 if (val) 480 inet->cmsg_flags |= IP_CMSG_TTL; 481 else 482 inet->cmsg_flags &= ~IP_CMSG_TTL; 483 break; 484 case IP_RECVTOS: 485 if (val) 486 inet->cmsg_flags |= IP_CMSG_TOS; 487 else 488 inet->cmsg_flags &= ~IP_CMSG_TOS; 489 break; 490 case IP_RECVOPTS: 491 if (val) 492 inet->cmsg_flags |= IP_CMSG_RECVOPTS; 493 else 494 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS; 495 break; 496 case IP_RETOPTS: 497 if (val) 498 inet->cmsg_flags |= IP_CMSG_RETOPTS; 499 else 500 inet->cmsg_flags &= ~IP_CMSG_RETOPTS; 501 break; 502 case IP_PASSSEC: 503 if (val) 504 inet->cmsg_flags |= IP_CMSG_PASSSEC; 505 else 506 inet->cmsg_flags &= ~IP_CMSG_PASSSEC; 507 break; 508 case IP_TOS: /* This sets both TOS and Precedence */ 509 if (sk->sk_type == SOCK_STREAM) { 510 val &= ~3; 511 val |= inet->tos & 3; 512 } 513 if (IPTOS_PREC(val) >= IPTOS_PREC_CRITIC_ECP && 514 !capable(CAP_NET_ADMIN)) { 515 err = -EPERM; 516 break; 517 } 518 if (inet->tos != val) { 519 inet->tos = val; 520 sk->sk_priority = rt_tos2priority(val); 521 sk_dst_reset(sk); 522 } 523 break; 524 case IP_TTL: 525 if (optlen<1) 526 goto e_inval; 527 if (val != -1 && (val < 1 || val>255)) 528 goto e_inval; 529 inet->uc_ttl = val; 530 break; 531 case IP_HDRINCL: 532 if (sk->sk_type != SOCK_RAW) { 533 err = -ENOPROTOOPT; 534 break; 535 } 536 inet->hdrincl = val ? 1 : 0; 537 break; 538 case IP_MTU_DISCOVER: 539 if (val<0 || val>2) 540 goto e_inval; 541 inet->pmtudisc = val; 542 break; 543 case IP_RECVERR: 544 inet->recverr = !!val; 545 if (!val) 546 skb_queue_purge(&sk->sk_error_queue); 547 break; 548 case IP_MULTICAST_TTL: 549 if (sk->sk_type == SOCK_STREAM) 550 goto e_inval; 551 if (optlen<1) 552 goto e_inval; 553 if (val==-1) 554 val = 1; 555 if (val < 0 || val > 255) 556 goto e_inval; 557 inet->mc_ttl = val; 558 break; 559 case IP_MULTICAST_LOOP: 560 if (optlen<1) 561 goto e_inval; 562 inet->mc_loop = !!val; 563 break; 564 case IP_MULTICAST_IF: 565 { 566 struct ip_mreqn mreq; 567 struct net_device *dev = NULL; 568 569 if (sk->sk_type == SOCK_STREAM) 570 goto e_inval; 571 /* 572 * Check the arguments are allowable 573 */ 574 575 err = -EFAULT; 576 if (optlen >= sizeof(struct ip_mreqn)) { 577 if (copy_from_user(&mreq,optval,sizeof(mreq))) 578 break; 579 } else { 580 memset(&mreq, 0, sizeof(mreq)); 581 if (optlen >= sizeof(struct in_addr) && 582 copy_from_user(&mreq.imr_address,optval,sizeof(struct in_addr))) 583 break; 584 } 585 586 if (!mreq.imr_ifindex) { 587 if (mreq.imr_address.s_addr == INADDR_ANY) { 588 inet->mc_index = 0; 589 inet->mc_addr = 0; 590 err = 0; 591 break; 592 } 593 dev = ip_dev_find(mreq.imr_address.s_addr); 594 if (dev) { 595 mreq.imr_ifindex = dev->ifindex; 596 dev_put(dev); 597 } 598 } else 599 dev = __dev_get_by_index(mreq.imr_ifindex); 600 601 602 err = -EADDRNOTAVAIL; 603 if (!dev) 604 break; 605 606 err = -EINVAL; 607 if (sk->sk_bound_dev_if && 608 mreq.imr_ifindex != sk->sk_bound_dev_if) 609 break; 610 611 inet->mc_index = mreq.imr_ifindex; 612 inet->mc_addr = mreq.imr_address.s_addr; 613 err = 0; 614 break; 615 } 616 617 case IP_ADD_MEMBERSHIP: 618 case IP_DROP_MEMBERSHIP: 619 { 620 struct ip_mreqn mreq; 621 622 if (optlen < sizeof(struct ip_mreq)) 623 goto e_inval; 624 err = -EFAULT; 625 if (optlen >= sizeof(struct ip_mreqn)) { 626 if(copy_from_user(&mreq,optval,sizeof(mreq))) 627 break; 628 } else { 629 memset(&mreq, 0, sizeof(mreq)); 630 if (copy_from_user(&mreq,optval,sizeof(struct ip_mreq))) 631 break; 632 } 633 634 if (optname == IP_ADD_MEMBERSHIP) 635 err = ip_mc_join_group(sk, &mreq); 636 else 637 err = ip_mc_leave_group(sk, &mreq); 638 break; 639 } 640 case IP_MSFILTER: 641 { 642 extern int sysctl_igmp_max_msf; 643 struct ip_msfilter *msf; 644 645 if (optlen < IP_MSFILTER_SIZE(0)) 646 goto e_inval; 647 if (optlen > sysctl_optmem_max) { 648 err = -ENOBUFS; 649 break; 650 } 651 msf = kmalloc(optlen, GFP_KERNEL); 652 if (msf == 0) { 653 err = -ENOBUFS; 654 break; 655 } 656 err = -EFAULT; 657 if (copy_from_user(msf, optval, optlen)) { 658 kfree(msf); 659 break; 660 } 661 /* numsrc >= (1G-4) overflow in 32 bits */ 662 if (msf->imsf_numsrc >= 0x3ffffffcU || 663 msf->imsf_numsrc > sysctl_igmp_max_msf) { 664 kfree(msf); 665 err = -ENOBUFS; 666 break; 667 } 668 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) { 669 kfree(msf); 670 err = -EINVAL; 671 break; 672 } 673 err = ip_mc_msfilter(sk, msf, 0); 674 kfree(msf); 675 break; 676 } 677 case IP_BLOCK_SOURCE: 678 case IP_UNBLOCK_SOURCE: 679 case IP_ADD_SOURCE_MEMBERSHIP: 680 case IP_DROP_SOURCE_MEMBERSHIP: 681 { 682 struct ip_mreq_source mreqs; 683 int omode, add; 684 685 if (optlen != sizeof(struct ip_mreq_source)) 686 goto e_inval; 687 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) { 688 err = -EFAULT; 689 break; 690 } 691 if (optname == IP_BLOCK_SOURCE) { 692 omode = MCAST_EXCLUDE; 693 add = 1; 694 } else if (optname == IP_UNBLOCK_SOURCE) { 695 omode = MCAST_EXCLUDE; 696 add = 0; 697 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) { 698 struct ip_mreqn mreq; 699 700 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr; 701 mreq.imr_address.s_addr = mreqs.imr_interface; 702 mreq.imr_ifindex = 0; 703 err = ip_mc_join_group(sk, &mreq); 704 if (err && err != -EADDRINUSE) 705 break; 706 omode = MCAST_INCLUDE; 707 add = 1; 708 } else /* IP_DROP_SOURCE_MEMBERSHIP */ { 709 omode = MCAST_INCLUDE; 710 add = 0; 711 } 712 err = ip_mc_source(add, omode, sk, &mreqs, 0); 713 break; 714 } 715 case MCAST_JOIN_GROUP: 716 case MCAST_LEAVE_GROUP: 717 { 718 struct group_req greq; 719 struct sockaddr_in *psin; 720 struct ip_mreqn mreq; 721 722 if (optlen < sizeof(struct group_req)) 723 goto e_inval; 724 err = -EFAULT; 725 if(copy_from_user(&greq, optval, sizeof(greq))) 726 break; 727 psin = (struct sockaddr_in *)&greq.gr_group; 728 if (psin->sin_family != AF_INET) 729 goto e_inval; 730 memset(&mreq, 0, sizeof(mreq)); 731 mreq.imr_multiaddr = psin->sin_addr; 732 mreq.imr_ifindex = greq.gr_interface; 733 734 if (optname == MCAST_JOIN_GROUP) 735 err = ip_mc_join_group(sk, &mreq); 736 else 737 err = ip_mc_leave_group(sk, &mreq); 738 break; 739 } 740 case MCAST_JOIN_SOURCE_GROUP: 741 case MCAST_LEAVE_SOURCE_GROUP: 742 case MCAST_BLOCK_SOURCE: 743 case MCAST_UNBLOCK_SOURCE: 744 { 745 struct group_source_req greqs; 746 struct ip_mreq_source mreqs; 747 struct sockaddr_in *psin; 748 int omode, add; 749 750 if (optlen != sizeof(struct group_source_req)) 751 goto e_inval; 752 if (copy_from_user(&greqs, optval, sizeof(greqs))) { 753 err = -EFAULT; 754 break; 755 } 756 if (greqs.gsr_group.ss_family != AF_INET || 757 greqs.gsr_source.ss_family != AF_INET) { 758 err = -EADDRNOTAVAIL; 759 break; 760 } 761 psin = (struct sockaddr_in *)&greqs.gsr_group; 762 mreqs.imr_multiaddr = psin->sin_addr.s_addr; 763 psin = (struct sockaddr_in *)&greqs.gsr_source; 764 mreqs.imr_sourceaddr = psin->sin_addr.s_addr; 765 mreqs.imr_interface = 0; /* use index for mc_source */ 766 767 if (optname == MCAST_BLOCK_SOURCE) { 768 omode = MCAST_EXCLUDE; 769 add = 1; 770 } else if (optname == MCAST_UNBLOCK_SOURCE) { 771 omode = MCAST_EXCLUDE; 772 add = 0; 773 } else if (optname == MCAST_JOIN_SOURCE_GROUP) { 774 struct ip_mreqn mreq; 775 776 psin = (struct sockaddr_in *)&greqs.gsr_group; 777 mreq.imr_multiaddr = psin->sin_addr; 778 mreq.imr_address.s_addr = 0; 779 mreq.imr_ifindex = greqs.gsr_interface; 780 err = ip_mc_join_group(sk, &mreq); 781 if (err && err != -EADDRINUSE) 782 break; 783 greqs.gsr_interface = mreq.imr_ifindex; 784 omode = MCAST_INCLUDE; 785 add = 1; 786 } else /* MCAST_LEAVE_SOURCE_GROUP */ { 787 omode = MCAST_INCLUDE; 788 add = 0; 789 } 790 err = ip_mc_source(add, omode, sk, &mreqs, 791 greqs.gsr_interface); 792 break; 793 } 794 case MCAST_MSFILTER: 795 { 796 extern int sysctl_igmp_max_msf; 797 struct sockaddr_in *psin; 798 struct ip_msfilter *msf = NULL; 799 struct group_filter *gsf = NULL; 800 int msize, i, ifindex; 801 802 if (optlen < GROUP_FILTER_SIZE(0)) 803 goto e_inval; 804 if (optlen > sysctl_optmem_max) { 805 err = -ENOBUFS; 806 break; 807 } 808 gsf = kmalloc(optlen,GFP_KERNEL); 809 if (gsf == 0) { 810 err = -ENOBUFS; 811 break; 812 } 813 err = -EFAULT; 814 if (copy_from_user(gsf, optval, optlen)) { 815 goto mc_msf_out; 816 } 817 /* numsrc >= (4G-140)/128 overflow in 32 bits */ 818 if (gsf->gf_numsrc >= 0x1ffffff || 819 gsf->gf_numsrc > sysctl_igmp_max_msf) { 820 err = -ENOBUFS; 821 goto mc_msf_out; 822 } 823 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) { 824 err = -EINVAL; 825 goto mc_msf_out; 826 } 827 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc); 828 msf = kmalloc(msize,GFP_KERNEL); 829 if (msf == 0) { 830 err = -ENOBUFS; 831 goto mc_msf_out; 832 } 833 ifindex = gsf->gf_interface; 834 psin = (struct sockaddr_in *)&gsf->gf_group; 835 if (psin->sin_family != AF_INET) { 836 err = -EADDRNOTAVAIL; 837 goto mc_msf_out; 838 } 839 msf->imsf_multiaddr = psin->sin_addr.s_addr; 840 msf->imsf_interface = 0; 841 msf->imsf_fmode = gsf->gf_fmode; 842 msf->imsf_numsrc = gsf->gf_numsrc; 843 err = -EADDRNOTAVAIL; 844 for (i=0; i<gsf->gf_numsrc; ++i) { 845 psin = (struct sockaddr_in *)&gsf->gf_slist[i]; 846 847 if (psin->sin_family != AF_INET) 848 goto mc_msf_out; 849 msf->imsf_slist[i] = psin->sin_addr.s_addr; 850 } 851 kfree(gsf); 852 gsf = NULL; 853 854 err = ip_mc_msfilter(sk, msf, ifindex); 855 mc_msf_out: 856 kfree(msf); 857 kfree(gsf); 858 break; 859 } 860 case IP_ROUTER_ALERT: 861 err = ip_ra_control(sk, val ? 1 : 0, NULL); 862 break; 863 864 case IP_FREEBIND: 865 if (optlen<1) 866 goto e_inval; 867 inet->freebind = !!val; 868 break; 869 870 case IP_IPSEC_POLICY: 871 case IP_XFRM_POLICY: 872 err = -EPERM; 873 if (!capable(CAP_NET_ADMIN)) 874 break; 875 err = xfrm_user_policy(sk, optname, optval, optlen); 876 break; 877 878 default: 879 err = -ENOPROTOOPT; 880 break; 881 } 882 release_sock(sk); 883 return err; 884 885 e_inval: 886 release_sock(sk); 887 return -EINVAL; 888 } 889 890 int ip_setsockopt(struct sock *sk, int level, 891 int optname, char __user *optval, int optlen) 892 { 893 int err; 894 895 if (level != SOL_IP) 896 return -ENOPROTOOPT; 897 898 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 899 #ifdef CONFIG_NETFILTER 900 /* we need to exclude all possible ENOPROTOOPTs except default case */ 901 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 902 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY 903 #ifdef CONFIG_IP_MROUTE 904 && (optname < MRT_BASE || optname > (MRT_BASE + 10)) 905 #endif 906 ) { 907 lock_sock(sk); 908 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen); 909 release_sock(sk); 910 } 911 #endif 912 return err; 913 } 914 915 #ifdef CONFIG_COMPAT 916 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 917 char __user *optval, int optlen) 918 { 919 int err; 920 921 if (level != SOL_IP) 922 return -ENOPROTOOPT; 923 924 err = do_ip_setsockopt(sk, level, optname, optval, optlen); 925 #ifdef CONFIG_NETFILTER 926 /* we need to exclude all possible ENOPROTOOPTs except default case */ 927 if (err == -ENOPROTOOPT && optname != IP_HDRINCL && 928 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY 929 #ifdef CONFIG_IP_MROUTE 930 && (optname < MRT_BASE || optname > (MRT_BASE + 10)) 931 #endif 932 ) { 933 lock_sock(sk); 934 err = compat_nf_setsockopt(sk, PF_INET, optname, 935 optval, optlen); 936 release_sock(sk); 937 } 938 #endif 939 return err; 940 } 941 942 EXPORT_SYMBOL(compat_ip_setsockopt); 943 #endif 944 945 /* 946 * Get the options. Note for future reference. The GET of IP options gets the 947 * _received_ ones. The set sets the _sent_ ones. 948 */ 949 950 static int do_ip_getsockopt(struct sock *sk, int level, int optname, 951 char __user *optval, int __user *optlen) 952 { 953 struct inet_sock *inet = inet_sk(sk); 954 int val; 955 int len; 956 957 if(level!=SOL_IP) 958 return -EOPNOTSUPP; 959 960 #ifdef CONFIG_IP_MROUTE 961 if(optname>=MRT_BASE && optname <=MRT_BASE+10) 962 { 963 return ip_mroute_getsockopt(sk,optname,optval,optlen); 964 } 965 #endif 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 #ifdef CONFIG_IP_MROUTE 1167 && (optname < MRT_BASE || optname > MRT_BASE+10) 1168 #endif 1169 ) { 1170 int len; 1171 1172 if(get_user(len,optlen)) 1173 return -EFAULT; 1174 1175 lock_sock(sk); 1176 err = nf_getsockopt(sk, PF_INET, optname, optval, 1177 &len); 1178 release_sock(sk); 1179 if (err >= 0) 1180 err = put_user(len, optlen); 1181 return err; 1182 } 1183 #endif 1184 return err; 1185 } 1186 1187 #ifdef CONFIG_COMPAT 1188 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 1189 char __user *optval, int __user *optlen) 1190 { 1191 int err = do_ip_getsockopt(sk, level, optname, optval, optlen); 1192 #ifdef CONFIG_NETFILTER 1193 /* we need to exclude all possible ENOPROTOOPTs except default case */ 1194 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS 1195 #ifdef CONFIG_IP_MROUTE 1196 && (optname < MRT_BASE || optname > MRT_BASE+10) 1197 #endif 1198 ) { 1199 int len; 1200 1201 if (get_user(len, optlen)) 1202 return -EFAULT; 1203 1204 lock_sock(sk); 1205 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len); 1206 release_sock(sk); 1207 if (err >= 0) 1208 err = put_user(len, optlen); 1209 return err; 1210 } 1211 #endif 1212 return err; 1213 } 1214 1215 EXPORT_SYMBOL(compat_ip_getsockopt); 1216 #endif 1217 1218 EXPORT_SYMBOL(ip_cmsg_recv); 1219 1220 EXPORT_SYMBOL(ip_getsockopt); 1221 EXPORT_SYMBOL(ip_setsockopt); 1222