net/ipv4/ip_sockglue.c

   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 <net/sock.h>
  27 #include <net/ip.h>
  28 #include <net/icmp.h>
  29 #include <net/tcp_states.h>
  30 #include <linux/udp.h>
  31 #include <linux/igmp.h>
  32 #include <linux/netfilter.h>
  33 #include <linux/route.h>
  34 #include <linux/mroute.h>
  35 #include <net/route.h>
  36 #include <net/xfrm.h>
  37 #include <net/compat.h>
  38 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
  39 #include <net/transp_v6.h>
  40 #endif
  41
  42 #include <linux/errqueue.h>
  43 #include <asm/uaccess.h>
  44
  45 #define IP_CMSG_PKTINFO         1
  46 #define IP_CMSG_TTL             2
  47 #define IP_CMSG_TOS             4
  48 #define IP_CMSG_RECVOPTS        8
  49 #define IP_CMSG_RETOPTS         16
  50 #define IP_CMSG_PASSSEC         32
  51
  52 /*
  53  *      SOL_IP control messages.
  54  */
  55
  56 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
  57 {
  58         struct in_pktinfo info;
  59         struct rtable *rt = skb->rtable;
  60
  61         info.ipi_addr.s_addr = ip_hdr(skb)->daddr;
  62         if (rt) {
  63                 info.ipi_ifindex = rt->rt_iif;
  64                 info.ipi_spec_dst.s_addr = rt->rt_spec_dst;
  65         } else {
  66                 info.ipi_ifindex = 0;
  67                 info.ipi_spec_dst.s_addr = 0;
  68         }
  69
  70         put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
  71 }
  72
  73 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb)
  74 {
  75         int ttl = ip_hdr(skb)->ttl;
  76         put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl);
  77 }
  78
  79 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb)
  80 {
  81         put_cmsg(msg, SOL_IP, IP_TOS, 1, &ip_hdr(skb)->tos);
  82 }
  83
  84 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb)
  85 {
  86         if (IPCB(skb)->opt.optlen == 0)
  87                 return;
  88
  89         put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen,
  90                  ip_hdr(skb) + 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 net *net, 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(net, &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 = icmp_hdr(skb)->type;
 272         serr->ee.ee_code = icmp_hdr(skb)->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 *)(icmp_hdr(skb) + 1))->daddr) -
 277                                    skb_network_header(skb);
 278         serr->port = port;
 279
 280         if (skb_pull(skb, payload - skb->data) != NULL) {
 281                 skb_reset_transport_header(skb);
 282                 if (sock_queue_err_skb(sk, skb) == 0)
 283                         return;
 284         }
 285         kfree_skb(skb);
 286 }
 287
 288 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info)
 289 {
 290         struct inet_sock *inet = inet_sk(sk);
 291         struct sock_exterr_skb *serr;
 292         struct iphdr *iph;
 293         struct sk_buff *skb;
 294
 295         if (!inet->recverr)
 296                 return;
 297
 298         skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC);
 299         if (!skb)
 300                 return;
 301
 302         skb_put(skb, sizeof(struct iphdr));
 303         skb_reset_network_header(skb);
 304         iph = ip_hdr(skb);
 305         iph->daddr = daddr;
 306
 307         serr = SKB_EXT_ERR(skb);
 308         serr->ee.ee_errno = err;
 309         serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
 310         serr->ee.ee_type = 0;
 311         serr->ee.ee_code = 0;
 312         serr->ee.ee_pad = 0;
 313         serr->ee.ee_info = info;
 314         serr->ee.ee_data = 0;
 315         serr->addr_offset = (u8 *)&iph->daddr - skb_network_header(skb);
 316         serr->port = port;
 317
 318         __skb_pull(skb, skb_tail_pointer(skb) - skb->data);
 319         skb_reset_transport_header(skb);
 320
 321         if (sock_queue_err_skb(sk, skb))
 322                 kfree_skb(skb);
 323 }
 324
 325 /*
 326  *      Handle MSG_ERRQUEUE
 327  */
 328 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len)
 329 {
 330         struct sock_exterr_skb *serr;
 331         struct sk_buff *skb, *skb2;
 332         struct sockaddr_in *sin;
 333         struct {
 334                 struct sock_extended_err ee;
 335                 struct sockaddr_in       offender;
 336         } errhdr;
 337         int err;
 338         int copied;
 339
 340         err = -EAGAIN;
 341         skb = skb_dequeue(&sk->sk_error_queue);
 342         if (skb == NULL)
 343                 goto out;
 344
 345         copied = skb->len;
 346         if (copied > len) {
 347                 msg->msg_flags |= MSG_TRUNC;
 348                 copied = len;
 349         }
 350         err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
 351         if (err)
 352                 goto out_free_skb;
 353
 354         sock_recv_timestamp(msg, sk, skb);
 355
 356         serr = SKB_EXT_ERR(skb);
 357
 358         sin = (struct sockaddr_in *)msg->msg_name;
 359         if (sin) {
 360                 sin->sin_family = AF_INET;
 361                 sin->sin_addr.s_addr = *(__be32 *)(skb_network_header(skb) +
 362                                                    serr->addr_offset);
 363                 sin->sin_port = serr->port;
 364                 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
 365         }
 366
 367         memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err));
 368         sin = &errhdr.offender;
 369         sin->sin_family = AF_UNSPEC;
 370         if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) {
 371                 struct inet_sock *inet = inet_sk(sk);
 372
 373                 sin->sin_family = AF_INET;
 374                 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
 375                 sin->sin_port = 0;
 376                 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
 377                 if (inet->cmsg_flags)
 378                         ip_cmsg_recv(msg, skb);
 379         }
 380
 381         put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr);
 382
 383         /* Now we could try to dump offended packet options */
 384
 385         msg->msg_flags |= MSG_ERRQUEUE;
 386         err = copied;
 387
 388         /* Reset and regenerate socket error */
 389         spin_lock_bh(&sk->sk_error_queue.lock);
 390         sk->sk_err = 0;
 391         if ((skb2 = skb_peek(&sk->sk_error_queue)) != NULL) {
 392                 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno;
 393                 spin_unlock_bh(&sk->sk_error_queue.lock);
 394                 sk->sk_error_report(sk);
 395         } else
 396                 spin_unlock_bh(&sk->sk_error_queue.lock);
 397
 398 out_free_skb:
 399         kfree_skb(skb);
 400 out:
 401         return err;
 402 }
 403
 404
 405 /*
 406  *      Socket option code for IP. This is the end of the line after any TCP,UDP etc options on
 407  *      an IP socket.
 408  */
 409
 410 static int do_ip_setsockopt(struct sock *sk, int level,
 411                             int optname, char __user *optval, int optlen)
 412 {
 413         struct inet_sock *inet = inet_sk(sk);
 414         int val=0,err;
 415
 416         if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) |
 417                              (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) |
 418                              (1<<IP_RETOPTS) | (1<<IP_TOS) |
 419                              (1<<IP_TTL) | (1<<IP_HDRINCL) |
 420                              (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) |
 421                              (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND) |
 422                              (1<<IP_PASSSEC) | (1<<IP_TRANSPARENT))) ||
 423             optname == IP_MULTICAST_TTL ||
 424             optname == IP_MULTICAST_LOOP) {
 425                 if (optlen >= sizeof(int)) {
 426                         if (get_user(val, (int __user *) optval))
 427                                 return -EFAULT;
 428                 } else if (optlen >= sizeof(char)) {
 429                         unsigned char ucval;
 430
 431                         if (get_user(ucval, (unsigned char __user *) optval))
 432                                 return -EFAULT;
 433                         val = (int) ucval;
 434                 }
 435         }
 436
 437         /* If optlen==0, it is equivalent to val == 0 */
 438
 439         if (ip_mroute_opt(optname))
 440                 return ip_mroute_setsockopt(sk,optname,optval,optlen);
 441
 442         err = 0;
 443         lock_sock(sk);
 444
 445         switch (optname) {
 446         case IP_OPTIONS:
 447         {
 448                 struct ip_options * opt = NULL;
 449                 if (optlen > 40 || optlen < 0)
 450                         goto e_inval;
 451                 err = ip_options_get_from_user(sock_net(sk), &opt,
 452                                                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(sock_net(sk), 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(sock_net(sk), 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         case IP_TRANSPARENT:
 882                 if (!capable(CAP_NET_ADMIN)) {
 883                         err = -EPERM;
 884                         break;
 885                 }
 886                 if (optlen < 1)
 887                         goto e_inval;
 888                 inet->transparent = !!val;
 889                 break;
 890
 891         default:
 892                 err = -ENOPROTOOPT;
 893                 break;
 894         }
 895         release_sock(sk);
 896         return err;
 897
 898 e_inval:
 899         release_sock(sk);
 900         return -EINVAL;
 901 }
 902
 903 int ip_setsockopt(struct sock *sk, int level,
 904                 int optname, char __user *optval, int optlen)
 905 {
 906         int err;
 907
 908         if (level != SOL_IP)
 909                 return -ENOPROTOOPT;
 910
 911         err = do_ip_setsockopt(sk, level, optname, optval, optlen);
 912 #ifdef CONFIG_NETFILTER
 913         /* we need to exclude all possible ENOPROTOOPTs except default case */
 914         if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
 915                         optname != IP_IPSEC_POLICY &&
 916                         optname != IP_XFRM_POLICY &&
 917                         !ip_mroute_opt(optname)) {
 918                 lock_sock(sk);
 919                 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen);
 920                 release_sock(sk);
 921         }
 922 #endif
 923         return err;
 924 }
 925
 926 #ifdef CONFIG_COMPAT
 927 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
 928                          char __user *optval, int optlen)
 929 {
 930         int err;
 931
 932         if (level != SOL_IP)
 933                 return -ENOPROTOOPT;
 934
 935         if (optname >= MCAST_JOIN_GROUP && optname <= MCAST_MSFILTER)
 936                 return compat_mc_setsockopt(sk, level, optname, optval, optlen,
 937                         ip_setsockopt);
 938
 939         err = do_ip_setsockopt(sk, level, optname, optval, optlen);
 940 #ifdef CONFIG_NETFILTER
 941         /* we need to exclude all possible ENOPROTOOPTs except default case */
 942         if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
 943                         optname != IP_IPSEC_POLICY &&
 944                         optname != IP_XFRM_POLICY &&
 945                         !ip_mroute_opt(optname)) {
 946                 lock_sock(sk);
 947                 err = compat_nf_setsockopt(sk, PF_INET, optname,
 948                                            optval, optlen);
 949                 release_sock(sk);
 950         }
 951 #endif
 952         return err;
 953 }
 954
 955 EXPORT_SYMBOL(compat_ip_setsockopt);
 956 #endif
 957
 958 /*
 959  *      Get the options. Note for future reference. The GET of IP options gets the
 960  *      _received_ ones. The set sets the _sent_ ones.
 961  */
 962
 963 static int do_ip_getsockopt(struct sock *sk, int level, int optname,
 964                             char __user *optval, int __user *optlen)
 965 {
 966         struct inet_sock *inet = inet_sk(sk);
 967         int val;
 968         int len;
 969
 970         if (level != SOL_IP)
 971                 return -EOPNOTSUPP;
 972
 973         if (ip_mroute_opt(optname))
 974                 return ip_mroute_getsockopt(sk,optname,optval,optlen);
 975
 976         if (get_user(len,optlen))
 977                 return -EFAULT;
 978         if (len < 0)
 979                 return -EINVAL;
 980
 981         lock_sock(sk);
 982
 983         switch (optname) {
 984         case IP_OPTIONS:
 985         {
 986                 unsigned char optbuf[sizeof(struct ip_options)+40];
 987                 struct ip_options * opt = (struct ip_options*)optbuf;
 988                 opt->optlen = 0;
 989                 if (inet->opt)
 990                         memcpy(optbuf, inet->opt,
 991                                sizeof(struct ip_options)+
 992                                inet->opt->optlen);
 993                 release_sock(sk);
 994
 995                 if (opt->optlen == 0)
 996                         return put_user(0, optlen);
 997
 998                 ip_options_undo(opt);
 999
1000                 len = min_t(unsigned int, len, opt->optlen);
1001                 if (put_user(len, optlen))
1002                         return -EFAULT;
1003                 if (copy_to_user(optval, opt->__data, len))
1004                         return -EFAULT;
1005                 return 0;
1006         }
1007         case IP_PKTINFO:
1008                 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0;
1009                 break;
1010         case IP_RECVTTL:
1011                 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0;
1012                 break;
1013         case IP_RECVTOS:
1014                 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0;
1015                 break;
1016         case IP_RECVOPTS:
1017                 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0;
1018                 break;
1019         case IP_RETOPTS:
1020                 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0;
1021                 break;
1022         case IP_PASSSEC:
1023                 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0;
1024                 break;
1025         case IP_TOS:
1026                 val = inet->tos;
1027                 break;
1028         case IP_TTL:
1029                 val = (inet->uc_ttl == -1 ?
1030                        sysctl_ip_default_ttl :
1031                        inet->uc_ttl);
1032                 break;
1033         case IP_HDRINCL:
1034                 val = inet->hdrincl;
1035                 break;
1036         case IP_MTU_DISCOVER:
1037                 val = inet->pmtudisc;
1038                 break;
1039         case IP_MTU:
1040         {
1041                 struct dst_entry *dst;
1042                 val = 0;
1043                 dst = sk_dst_get(sk);
1044                 if (dst) {
1045                         val = dst_mtu(dst);
1046                         dst_release(dst);
1047                 }
1048                 if (!val) {
1049                         release_sock(sk);
1050                         return -ENOTCONN;
1051                 }
1052                 break;
1053         }
1054         case IP_RECVERR:
1055                 val = inet->recverr;
1056                 break;
1057         case IP_MULTICAST_TTL:
1058                 val = inet->mc_ttl;
1059                 break;
1060         case IP_MULTICAST_LOOP:
1061                 val = inet->mc_loop;
1062                 break;
1063         case IP_MULTICAST_IF:
1064         {
1065                 struct in_addr addr;
1066                 len = min_t(unsigned int, len, sizeof(struct in_addr));
1067                 addr.s_addr = inet->mc_addr;
1068                 release_sock(sk);
1069
1070                 if (put_user(len, optlen))
1071                         return -EFAULT;
1072                 if (copy_to_user(optval, &addr, len))
1073                         return -EFAULT;
1074                 return 0;
1075         }
1076         case IP_MSFILTER:
1077         {
1078                 struct ip_msfilter msf;
1079                 int err;
1080
1081                 if (len < IP_MSFILTER_SIZE(0)) {
1082                         release_sock(sk);
1083                         return -EINVAL;
1084                 }
1085                 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) {
1086                         release_sock(sk);
1087                         return -EFAULT;
1088                 }
1089                 err = ip_mc_msfget(sk, &msf,
1090                                    (struct ip_msfilter __user *)optval, optlen);
1091                 release_sock(sk);
1092                 return err;
1093         }
1094         case MCAST_MSFILTER:
1095         {
1096                 struct group_filter gsf;
1097                 int err;
1098
1099                 if (len < GROUP_FILTER_SIZE(0)) {
1100                         release_sock(sk);
1101                         return -EINVAL;
1102                 }
1103                 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) {
1104                         release_sock(sk);
1105                         return -EFAULT;
1106                 }
1107                 err = ip_mc_gsfget(sk, &gsf,
1108                                    (struct group_filter __user *)optval, optlen);
1109                 release_sock(sk);
1110                 return err;
1111         }
1112         case IP_PKTOPTIONS:
1113         {
1114                 struct msghdr msg;
1115
1116                 release_sock(sk);
1117
1118                 if (sk->sk_type != SOCK_STREAM)
1119                         return -ENOPROTOOPT;
1120
1121                 msg.msg_control = optval;
1122                 msg.msg_controllen = len;
1123                 msg.msg_flags = 0;
1124
1125                 if (inet->cmsg_flags & IP_CMSG_PKTINFO) {
1126                         struct in_pktinfo info;
1127
1128                         info.ipi_addr.s_addr = inet->rcv_saddr;
1129                         info.ipi_spec_dst.s_addr = inet->rcv_saddr;
1130                         info.ipi_ifindex = inet->mc_index;
1131                         put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
1132                 }
1133                 if (inet->cmsg_flags & IP_CMSG_TTL) {
1134                         int hlim = inet->mc_ttl;
1135                         put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim);
1136                 }
1137                 len -= msg.msg_controllen;
1138                 return put_user(len, optlen);
1139         }
1140         case IP_FREEBIND:
1141                 val = inet->freebind;
1142                 break;
1143         case IP_TRANSPARENT:
1144                 val = inet->transparent;
1145                 break;
1146         default:
1147                 release_sock(sk);
1148                 return -ENOPROTOOPT;
1149         }
1150         release_sock(sk);
1151
1152         if (len < sizeof(int) && len > 0 && val>=0 && val<=255) {
1153                 unsigned char ucval = (unsigned char)val;
1154                 len = 1;
1155                 if (put_user(len, optlen))
1156                         return -EFAULT;
1157                 if (copy_to_user(optval,&ucval,1))
1158                         return -EFAULT;
1159         } else {
1160                 len = min_t(unsigned int, sizeof(int), len);
1161                 if (put_user(len, optlen))
1162                         return -EFAULT;
1163                 if (copy_to_user(optval,&val,len))
1164                         return -EFAULT;
1165         }
1166         return 0;
1167 }
1168
1169 int ip_getsockopt(struct sock *sk, int level,
1170                   int optname, char __user *optval, int __user *optlen)
1171 {
1172         int err;
1173
1174         err = do_ip_getsockopt(sk, level, optname, optval, optlen);
1175 #ifdef CONFIG_NETFILTER
1176         /* we need to exclude all possible ENOPROTOOPTs except default case */
1177         if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1178                         !ip_mroute_opt(optname)) {
1179                 int len;
1180
1181                 if (get_user(len,optlen))
1182                         return -EFAULT;
1183
1184                 lock_sock(sk);
1185                 err = nf_getsockopt(sk, PF_INET, optname, optval,
1186                                 &len);
1187                 release_sock(sk);
1188                 if (err >= 0)
1189                         err = put_user(len, optlen);
1190                 return err;
1191         }
1192 #endif
1193         return err;
1194 }
1195
1196 #ifdef CONFIG_COMPAT
1197 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
1198                          char __user *optval, int __user *optlen)
1199 {
1200         int err;
1201
1202         if (optname == MCAST_MSFILTER)
1203                 return compat_mc_getsockopt(sk, level, optname, optval, optlen,
1204                         ip_getsockopt);
1205
1206         err = do_ip_getsockopt(sk, level, optname, optval, optlen);
1207
1208 #ifdef CONFIG_NETFILTER
1209         /* we need to exclude all possible ENOPROTOOPTs except default case */
1210         if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1211                         !ip_mroute_opt(optname)) {
1212                 int len;
1213
1214                 if (get_user(len, optlen))
1215                         return -EFAULT;
1216
1217                 lock_sock(sk);
1218                 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len);
1219                 release_sock(sk);
1220                 if (err >= 0)
1221                         err = put_user(len, optlen);
1222                 return err;
1223         }
1224 #endif
1225         return err;
1226 }
1227
1228 EXPORT_SYMBOL(compat_ip_getsockopt);
1229 #endif
1230
1231 EXPORT_SYMBOL(ip_cmsg_recv);
1232
1233 EXPORT_SYMBOL(ip_getsockopt);
1234 EXPORT_SYMBOL(ip_setsockopt);