net/ipv4/ping.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  *              "Ping" sockets
   7  *
   8  *              This program is free software; you can redistribute it and/or
   9  *              modify it under the terms of the GNU General Public License
  10  *              as published by the Free Software Foundation; either version
  11  *              2 of the License, or (at your option) any later version.
  12  *
  13  * Based on ipv4/udp.c code.
  14  *
  15  * Authors:     Vasiliy Kulikov / Openwall (for Linux 2.6),
  16  *              Pavel Kankovsky (for Linux 2.4.32)
  17  *
  18  * Pavel gave all rights to bugs to Vasiliy,
  19  * none of the bugs are Pavel's now.
  20  *
  21  */
  22
  23 #include <linux/uaccess.h>
  24 #include <linux/types.h>
  25 #include <linux/fcntl.h>
  26 #include <linux/socket.h>
  27 #include <linux/sockios.h>
  28 #include <linux/in.h>
  29 #include <linux/errno.h>
  30 #include <linux/timer.h>
  31 #include <linux/mm.h>
  32 #include <linux/inet.h>
  33 #include <linux/netdevice.h>
  34 #include <net/snmp.h>
  35 #include <net/ip.h>
  36 #include <net/ipv6.h>
  37 #include <net/icmp.h>
  38 #include <net/protocol.h>
  39 #include <linux/skbuff.h>
  40 #include <linux/proc_fs.h>
  41 #include <linux/export.h>
  42 #include <net/sock.h>
  43 #include <net/ping.h>
  44 #include <net/udp.h>
  45 #include <net/route.h>
  46 #include <net/inet_common.h>
  47 #include <net/checksum.h>
  48
  49
  50 static struct ping_table ping_table;
  51
  52 static u16 ping_port_rover;
  53
  54 static inline int ping_hashfn(struct net *net, unsigned num, unsigned mask)
  55 {
  56         int res = (num + net_hash_mix(net)) & mask;
  57         pr_debug("hash(%d) = %d\n", num, res);
  58         return res;
  59 }
  60
  61 static inline struct hlist_nulls_head *ping_hashslot(struct ping_table *table,
  62                                              struct net *net, unsigned num)
  63 {
  64         return &table->hash[ping_hashfn(net, num, PING_HTABLE_MASK)];
  65 }
  66
  67 static int ping_v4_get_port(struct sock *sk, unsigned short ident)
  68 {
  69         struct hlist_nulls_node *node;
  70         struct hlist_nulls_head *hlist;
  71         struct inet_sock *isk, *isk2;
  72         struct sock *sk2 = NULL;
  73
  74         isk = inet_sk(sk);
  75         write_lock_bh(&ping_table.lock);
  76         if (ident == 0) {
  77                 u32 i;
  78                 u16 result = ping_port_rover + 1;
  79
  80                 for (i = 0; i < (1L << 16); i++, result++) {
  81                         if (!result)
  82                                 result++; /* avoid zero */
  83                         hlist = ping_hashslot(&ping_table, sock_net(sk),
  84                                             result);
  85                         ping_portaddr_for_each_entry(sk2, node, hlist) {
  86                                 isk2 = inet_sk(sk2);
  87
  88                                 if (isk2->inet_num == result)
  89                                         goto next_port;
  90                         }
  91
  92                         /* found */
  93                         ping_port_rover = ident = result;
  94                         break;
  95 next_port:
  96                         ;
  97                 }
  98                 if (i >= (1L << 16))
  99                         goto fail;
 100         } else {
 101                 hlist = ping_hashslot(&ping_table, sock_net(sk), ident);
 102                 ping_portaddr_for_each_entry(sk2, node, hlist) {
 103                         isk2 = inet_sk(sk2);
 104
 105                         if ((isk2->inet_num == ident) &&
 106                             (sk2 != sk) &&
 107                             (!sk2->sk_reuse || !sk->sk_reuse))
 108                                 goto fail;
 109                 }
 110         }
 111
 112         pr_debug("found port/ident = %d\n", ident);
 113         isk->inet_num = ident;
 114         if (sk_unhashed(sk)) {
 115                 pr_debug("was not hashed\n");
 116                 sock_hold(sk);
 117                 hlist_nulls_add_head(&sk->sk_nulls_node, hlist);
 118                 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
 119         }
 120         write_unlock_bh(&ping_table.lock);
 121         return 0;
 122
 123 fail:
 124         write_unlock_bh(&ping_table.lock);
 125         return 1;
 126 }
 127
 128 static void ping_v4_hash(struct sock *sk)
 129 {
 130         pr_debug("ping_v4_hash(sk->port=%u)\n", inet_sk(sk)->inet_num);
 131         BUG(); /* "Please do not press this button again." */
 132 }
 133
 134 static void ping_v4_unhash(struct sock *sk)
 135 {
 136         struct inet_sock *isk = inet_sk(sk);
 137         pr_debug("ping_v4_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
 138         if (sk_hashed(sk)) {
 139                 write_lock_bh(&ping_table.lock);
 140                 hlist_nulls_del(&sk->sk_nulls_node);
 141                 sock_put(sk);
 142                 isk->inet_num = 0;
 143                 isk->inet_sport = 0;
 144                 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
 145                 write_unlock_bh(&ping_table.lock);
 146         }
 147 }
 148
 149 static struct sock *ping_v4_lookup(struct net *net, __be32 saddr, __be32 daddr,
 150                                    u16 ident, int dif)
 151 {
 152         struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
 153         struct sock *sk = NULL;
 154         struct inet_sock *isk;
 155         struct hlist_nulls_node *hnode;
 156
 157         pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
 158                  (int)ident, &daddr, dif);
 159         read_lock_bh(&ping_table.lock);
 160
 161         ping_portaddr_for_each_entry(sk, hnode, hslot) {
 162                 isk = inet_sk(sk);
 163
 164                 pr_debug("found: %p: num = %d, daddr = %pI4, dif = %d\n", sk,
 165                          (int)isk->inet_num, &isk->inet_rcv_saddr,
 166                          sk->sk_bound_dev_if);
 167
 168                 pr_debug("iterate\n");
 169                 if (isk->inet_num != ident)
 170                         continue;
 171                 if (isk->inet_rcv_saddr && isk->inet_rcv_saddr != daddr)
 172                         continue;
 173                 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
 174                         continue;
 175
 176                 sock_hold(sk);
 177                 goto exit;
 178         }
 179
 180         sk = NULL;
 181 exit:
 182         read_unlock_bh(&ping_table.lock);
 183
 184         return sk;
 185 }
 186
 187 static void inet_get_ping_group_range_net(struct net *net, gid_t *low,
 188                                           gid_t *high)
 189 {
 190         gid_t *data = net->ipv4.sysctl_ping_group_range;
 191         unsigned seq;
 192         do {
 193                 seq = read_seqbegin(&sysctl_local_ports.lock);
 194
 195                 *low = data[0];
 196                 *high = data[1];
 197         } while (read_seqretry(&sysctl_local_ports.lock, seq));
 198 }
 199
 200
 201 static int ping_init_sock(struct sock *sk)
 202 {
 203         struct net *net = sock_net(sk);
 204         gid_t group = current_egid();
 205         gid_t range[2];
 206         struct group_info *group_info = get_current_groups();
 207         int i, j, count = group_info->ngroups;
 208         kgid_t low, high;
 209
 210         inet_get_ping_group_range_net(net, range, range+1);
 211         low = make_kgid(&init_user_ns, range[0]);
 212         high = make_kgid(&init_user_ns, range[1]);
 213         if (!gid_valid(low) || !gid_valid(high) || gid_lt(high, low))
 214                 return -EACCES;
 215
 216         if (range[0] <= group && group <= range[1])
 217                 return 0;
 218
 219         for (i = 0; i < group_info->nblocks; i++) {
 220                 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
 221                 for (j = 0; j < cp_count; j++) {
 222                         kgid_t gid = group_info->blocks[i][j];
 223                         if (gid_lte(low, gid) && gid_lte(gid, high))
 224                                 return 0;
 225                 }
 226
 227                 count -= cp_count;
 228         }
 229
 230         return -EACCES;
 231 }
 232
 233 static void ping_close(struct sock *sk, long timeout)
 234 {
 235         pr_debug("ping_close(sk=%p,sk->num=%u)\n",
 236                  inet_sk(sk), inet_sk(sk)->inet_num);
 237         pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
 238
 239         sk_common_release(sk);
 240 }
 241
 242 /*
 243  * We need our own bind because there are no privileged id's == local ports.
 244  * Moreover, we don't allow binding to multi- and broadcast addresses.
 245  */
 246
 247 static int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 248 {
 249         struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
 250         struct inet_sock *isk = inet_sk(sk);
 251         unsigned short snum;
 252         int chk_addr_ret;
 253         int err;
 254
 255         if (addr_len < sizeof(struct sockaddr_in))
 256                 return -EINVAL;
 257
 258         pr_debug("ping_v4_bind(sk=%p,sa_addr=%08x,sa_port=%d)\n",
 259                  sk, addr->sin_addr.s_addr, ntohs(addr->sin_port));
 260
 261         chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
 262         if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
 263                 chk_addr_ret = RTN_LOCAL;
 264
 265         if ((sysctl_ip_nonlocal_bind == 0 &&
 266             isk->freebind == 0 && isk->transparent == 0 &&
 267              chk_addr_ret != RTN_LOCAL) ||
 268             chk_addr_ret == RTN_MULTICAST ||
 269             chk_addr_ret == RTN_BROADCAST)
 270                 return -EADDRNOTAVAIL;
 271
 272         lock_sock(sk);
 273
 274         err = -EINVAL;
 275         if (isk->inet_num != 0)
 276                 goto out;
 277
 278         err = -EADDRINUSE;
 279         isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
 280         snum = ntohs(addr->sin_port);
 281         if (ping_v4_get_port(sk, snum) != 0) {
 282                 isk->inet_saddr = isk->inet_rcv_saddr = 0;
 283                 goto out;
 284         }
 285
 286         pr_debug("after bind(): num = %d, daddr = %pI4, dif = %d\n",
 287                  (int)isk->inet_num,
 288                  &isk->inet_rcv_saddr,
 289                  (int)sk->sk_bound_dev_if);
 290
 291         err = 0;
 292         if (isk->inet_rcv_saddr)
 293                 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
 294         if (snum)
 295                 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
 296         isk->inet_sport = htons(isk->inet_num);
 297         isk->inet_daddr = 0;
 298         isk->inet_dport = 0;
 299         sk_dst_reset(sk);
 300 out:
 301         release_sock(sk);
 302         pr_debug("ping_v4_bind -> %d\n", err);
 303         return err;
 304 }
 305
 306 /*
 307  * Is this a supported type of ICMP message?
 308  */
 309
 310 static inline int ping_supported(int type, int code)
 311 {
 312         if (type == ICMP_ECHO && code == 0)
 313                 return 1;
 314         return 0;
 315 }
 316
 317 /*
 318  * This routine is called by the ICMP module when it gets some
 319  * sort of error condition.
 320  */
 321
 322 static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
 323
 324 void ping_err(struct sk_buff *skb, u32 info)
 325 {
 326         struct iphdr *iph = (struct iphdr *)skb->data;
 327         struct icmphdr *icmph = (struct icmphdr *)(skb->data+(iph->ihl<<2));
 328         struct inet_sock *inet_sock;
 329         int type = icmph->type;
 330         int code = icmph->code;
 331         struct net *net = dev_net(skb->dev);
 332         struct sock *sk;
 333         int harderr;
 334         int err;
 335
 336         /* We assume the packet has already been checked by icmp_unreach */
 337
 338         if (!ping_supported(icmph->type, icmph->code))
 339                 return;
 340
 341         pr_debug("ping_err(type=%04x,code=%04x,id=%04x,seq=%04x)\n", type,
 342                  code, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
 343
 344         sk = ping_v4_lookup(net, iph->daddr, iph->saddr,
 345                             ntohs(icmph->un.echo.id), skb->dev->ifindex);
 346         if (sk == NULL) {
 347                 pr_debug("no socket, dropping\n");
 348                 return; /* No socket for error */
 349         }
 350         pr_debug("err on socket %p\n", sk);
 351
 352         err = 0;
 353         harderr = 0;
 354         inet_sock = inet_sk(sk);
 355
 356         switch (type) {
 357         default:
 358         case ICMP_TIME_EXCEEDED:
 359                 err = EHOSTUNREACH;
 360                 break;
 361         case ICMP_SOURCE_QUENCH:
 362                 /* This is not a real error but ping wants to see it.
 363                  * Report it with some fake errno. */
 364                 err = EREMOTEIO;
 365                 break;
 366         case ICMP_PARAMETERPROB:
 367                 err = EPROTO;
 368                 harderr = 1;
 369                 break;
 370         case ICMP_DEST_UNREACH:
 371                 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
 372                         if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
 373                                 err = EMSGSIZE;
 374                                 harderr = 1;
 375                                 break;
 376                         }
 377                         goto out;
 378                 }
 379                 err = EHOSTUNREACH;
 380                 if (code <= NR_ICMP_UNREACH) {
 381                         harderr = icmp_err_convert[code].fatal;
 382                         err = icmp_err_convert[code].errno;
 383                 }
 384                 break;
 385         case ICMP_REDIRECT:
 386                 /* See ICMP_SOURCE_QUENCH */
 387                 err = EREMOTEIO;
 388                 break;
 389         }
 390
 391         /*
 392          *      RFC1122: OK.  Passes ICMP errors back to application, as per
 393          *      4.1.3.3.
 394          */
 395         if (!inet_sock->recverr) {
 396                 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
 397                         goto out;
 398         } else {
 399                 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
 400                          info, (u8 *)icmph);
 401         }
 402         sk->sk_err = err;
 403         sk->sk_error_report(sk);
 404 out:
 405         sock_put(sk);
 406 }
 407
 408 /*
 409  *      Copy and checksum an ICMP Echo packet from user space into a buffer.
 410  */
 411
 412 struct pingfakehdr {
 413         struct icmphdr icmph;
 414         struct iovec *iov;
 415         __wsum wcheck;
 416 };
 417
 418 static int ping_getfrag(void *from, char * to,
 419                         int offset, int fraglen, int odd, struct sk_buff *skb)
 420 {
 421         struct pingfakehdr *pfh = (struct pingfakehdr *)from;
 422
 423         if (offset == 0) {
 424                 if (fraglen < sizeof(struct icmphdr))
 425                         BUG();
 426                 if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
 427                             pfh->iov, 0, fraglen - sizeof(struct icmphdr),
 428                             &pfh->wcheck))
 429                         return -EFAULT;
 430
 431                 return 0;
 432         }
 433         if (offset < sizeof(struct icmphdr))
 434                 BUG();
 435         if (csum_partial_copy_fromiovecend
 436                         (to, pfh->iov, offset - sizeof(struct icmphdr),
 437                          fraglen, &pfh->wcheck))
 438                 return -EFAULT;
 439         return 0;
 440 }
 441
 442 static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
 443                                     struct flowi4 *fl4)
 444 {
 445         struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
 446
 447         pfh->wcheck = csum_partial((char *)&pfh->icmph,
 448                 sizeof(struct icmphdr), pfh->wcheck);
 449         pfh->icmph.checksum = csum_fold(pfh->wcheck);
 450         memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
 451         skb->ip_summed = CHECKSUM_NONE;
 452         return ip_push_pending_frames(sk, fl4);
 453 }
 454
 455 static int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 456                         size_t len)
 457 {
 458         struct net *net = sock_net(sk);
 459         struct flowi4 fl4;
 460         struct inet_sock *inet = inet_sk(sk);
 461         struct ipcm_cookie ipc;
 462         struct icmphdr user_icmph;
 463         struct pingfakehdr pfh;
 464         struct rtable *rt = NULL;
 465         struct ip_options_data opt_copy;
 466         int free = 0;
 467         __be32 saddr, daddr, faddr;
 468         u8  tos;
 469         int err;
 470
 471         pr_debug("ping_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
 472
 473
 474         if (len > 0xFFFF)
 475                 return -EMSGSIZE;
 476
 477         /*
 478          *      Check the flags.
 479          */
 480
 481         /* Mirror BSD error message compatibility */
 482         if (msg->msg_flags & MSG_OOB)
 483                 return -EOPNOTSUPP;
 484
 485         /*
 486          *      Fetch the ICMP header provided by the userland.
 487          *      iovec is modified!
 488          */
 489
 490         if (memcpy_fromiovec((u8 *)&user_icmph, msg->msg_iov,
 491                              sizeof(struct icmphdr)))
 492                 return -EFAULT;
 493         if (!ping_supported(user_icmph.type, user_icmph.code))
 494                 return -EINVAL;
 495
 496         /*
 497          *      Get and verify the address.
 498          */
 499
 500         if (msg->msg_name) {
 501                 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
 502                 if (msg->msg_namelen < sizeof(*usin))
 503                         return -EINVAL;
 504                 if (usin->sin_family != AF_INET)
 505                         return -EINVAL;
 506                 daddr = usin->sin_addr.s_addr;
 507                 /* no remote port */
 508         } else {
 509                 if (sk->sk_state != TCP_ESTABLISHED)
 510                         return -EDESTADDRREQ;
 511                 daddr = inet->inet_daddr;
 512                 /* no remote port */
 513         }
 514
 515         ipc.addr = inet->inet_saddr;
 516         ipc.opt = NULL;
 517         ipc.oif = sk->sk_bound_dev_if;
 518         ipc.tx_flags = 0;
 519         err = sock_tx_timestamp(sk, &ipc.tx_flags);
 520         if (err)
 521                 return err;
 522
 523         if (msg->msg_controllen) {
 524                 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
 525                 if (err)
 526                         return err;
 527                 if (ipc.opt)
 528                         free = 1;
 529         }
 530         if (!ipc.opt) {
 531                 struct ip_options_rcu *inet_opt;
 532
 533                 rcu_read_lock();
 534                 inet_opt = rcu_dereference(inet->inet_opt);
 535                 if (inet_opt) {
 536                         memcpy(&opt_copy, inet_opt,
 537                                sizeof(*inet_opt) + inet_opt->opt.optlen);
 538                         ipc.opt = &opt_copy.opt;
 539                 }
 540                 rcu_read_unlock();
 541         }
 542
 543         saddr = ipc.addr;
 544         ipc.addr = faddr = daddr;
 545
 546         if (ipc.opt && ipc.opt->opt.srr) {
 547                 if (!daddr)
 548                         return -EINVAL;
 549                 faddr = ipc.opt->opt.faddr;
 550         }
 551         tos = RT_TOS(inet->tos);
 552         if (sock_flag(sk, SOCK_LOCALROUTE) ||
 553             (msg->msg_flags & MSG_DONTROUTE) ||
 554             (ipc.opt && ipc.opt->opt.is_strictroute)) {
 555                 tos |= RTO_ONLINK;
 556         }
 557
 558         if (ipv4_is_multicast(daddr)) {
 559                 if (!ipc.oif)
 560                         ipc.oif = inet->mc_index;
 561                 if (!saddr)
 562                         saddr = inet->mc_addr;
 563         } else if (!ipc.oif)
 564                 ipc.oif = inet->uc_index;
 565
 566         flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
 567                            RT_SCOPE_UNIVERSE, sk->sk_protocol,
 568                            inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
 569
 570         security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
 571         rt = ip_route_output_flow(net, &fl4, sk);
 572         if (IS_ERR(rt)) {
 573                 err = PTR_ERR(rt);
 574                 rt = NULL;
 575                 if (err == -ENETUNREACH)
 576                         IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
 577                 goto out;
 578         }
 579
 580         err = -EACCES;
 581         if ((rt->rt_flags & RTCF_BROADCAST) &&
 582             !sock_flag(sk, SOCK_BROADCAST))
 583                 goto out;
 584
 585         if (msg->msg_flags & MSG_CONFIRM)
 586                 goto do_confirm;
 587 back_from_confirm:
 588
 589         if (!ipc.addr)
 590                 ipc.addr = fl4.daddr;
 591
 592         lock_sock(sk);
 593
 594         pfh.icmph.type = user_icmph.type; /* already checked */
 595         pfh.icmph.code = user_icmph.code; /* ditto */
 596         pfh.icmph.checksum = 0;
 597         pfh.icmph.un.echo.id = inet->inet_sport;
 598         pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
 599         pfh.iov = msg->msg_iov;
 600         pfh.wcheck = 0;
 601
 602         err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
 603                         0, &ipc, &rt, msg->msg_flags);
 604         if (err)
 605                 ip_flush_pending_frames(sk);
 606         else
 607                 err = ping_push_pending_frames(sk, &pfh, &fl4);
 608         release_sock(sk);
 609
 610 out:
 611         ip_rt_put(rt);
 612         if (free)
 613                 kfree(ipc.opt);
 614         if (!err) {
 615                 icmp_out_count(sock_net(sk), user_icmph.type);
 616                 return len;
 617         }
 618         return err;
 619
 620 do_confirm:
 621         dst_confirm(&rt->dst);
 622         if (!(msg->msg_flags & MSG_PROBE) || len)
 623                 goto back_from_confirm;
 624         err = 0;
 625         goto out;
 626 }
 627
 628 static int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 629                         size_t len, int noblock, int flags, int *addr_len)
 630 {
 631         struct inet_sock *isk = inet_sk(sk);
 632         struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
 633         struct sk_buff *skb;
 634         int copied, err;
 635
 636         pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
 637
 638         err = -EOPNOTSUPP;
 639         if (flags & MSG_OOB)
 640                 goto out;
 641
 642         if (addr_len)
 643                 *addr_len = sizeof(*sin);
 644
 645         if (flags & MSG_ERRQUEUE)
 646                 return ip_recv_error(sk, msg, len);
 647
 648         skb = skb_recv_datagram(sk, flags, noblock, &err);
 649         if (!skb)
 650                 goto out;
 651
 652         copied = skb->len;
 653         if (copied > len) {
 654                 msg->msg_flags |= MSG_TRUNC;
 655                 copied = len;
 656         }
 657
 658         /* Don't bother checking the checksum */
 659         err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
 660         if (err)
 661                 goto done;
 662
 663         sock_recv_timestamp(msg, sk, skb);
 664
 665         /* Copy the address. */
 666         if (sin) {
 667                 sin->sin_family = AF_INET;
 668                 sin->sin_port = 0 /* skb->h.uh->source */;
 669                 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
 670                 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
 671         }
 672         if (isk->cmsg_flags)
 673                 ip_cmsg_recv(msg, skb);
 674         err = copied;
 675
 676 done:
 677         skb_free_datagram(sk, skb);
 678 out:
 679         pr_debug("ping_recvmsg -> %d\n", err);
 680         return err;
 681 }
 682
 683 static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
 684 {
 685         pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
 686                  inet_sk(sk), inet_sk(sk)->inet_num, skb);
 687         if (sock_queue_rcv_skb(sk, skb) < 0) {
 688                 kfree_skb(skb);
 689                 pr_debug("ping_queue_rcv_skb -> failed\n");
 690                 return -1;
 691         }
 692         return 0;
 693 }
 694
 695
 696 /*
 697  *      All we need to do is get the socket.
 698  */
 699
 700 void ping_rcv(struct sk_buff *skb)
 701 {
 702         struct sock *sk;
 703         struct net *net = dev_net(skb->dev);
 704         struct iphdr *iph = ip_hdr(skb);
 705         struct icmphdr *icmph = icmp_hdr(skb);
 706         __be32 saddr = iph->saddr;
 707         __be32 daddr = iph->daddr;
 708
 709         /* We assume the packet has already been checked by icmp_rcv */
 710
 711         pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
 712                  skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
 713
 714         /* Push ICMP header back */
 715         skb_push(skb, skb->data - (u8 *)icmph);
 716
 717         sk = ping_v4_lookup(net, saddr, daddr, ntohs(icmph->un.echo.id),
 718                             skb->dev->ifindex);
 719         if (sk != NULL) {
 720                 pr_debug("rcv on socket %p\n", sk);
 721                 ping_queue_rcv_skb(sk, skb_get(skb));
 722                 sock_put(sk);
 723                 return;
 724         }
 725         pr_debug("no socket, dropping\n");
 726
 727         /* We're called from icmp_rcv(). kfree_skb() is done there. */
 728 }
 729
 730 struct proto ping_prot = {
 731         .name =         "PING",
 732         .owner =        THIS_MODULE,
 733         .init =         ping_init_sock,
 734         .close =        ping_close,
 735         .connect =      ip4_datagram_connect,
 736         .disconnect =   udp_disconnect,
 737         .setsockopt =   ip_setsockopt,
 738         .getsockopt =   ip_getsockopt,
 739         .sendmsg =      ping_sendmsg,
 740         .recvmsg =      ping_recvmsg,
 741         .bind =         ping_bind,
 742         .backlog_rcv =  ping_queue_rcv_skb,
 743         .hash =         ping_v4_hash,
 744         .unhash =       ping_v4_unhash,
 745         .get_port =     ping_v4_get_port,
 746         .obj_size =     sizeof(struct inet_sock),
 747 };
 748 EXPORT_SYMBOL(ping_prot);
 749
 750 #ifdef CONFIG_PROC_FS
 751
 752 static struct sock *ping_get_first(struct seq_file *seq, int start)
 753 {
 754         struct sock *sk;
 755         struct ping_iter_state *state = seq->private;
 756         struct net *net = seq_file_net(seq);
 757
 758         for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
 759              ++state->bucket) {
 760                 struct hlist_nulls_node *node;
 761                 struct hlist_nulls_head *hslot;
 762
 763                 hslot = &ping_table.hash[state->bucket];
 764
 765                 if (hlist_nulls_empty(hslot))
 766                         continue;
 767
 768                 sk_nulls_for_each(sk, node, hslot) {
 769                         if (net_eq(sock_net(sk), net))
 770                                 goto found;
 771                 }
 772         }
 773         sk = NULL;
 774 found:
 775         return sk;
 776 }
 777
 778 static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
 779 {
 780         struct ping_iter_state *state = seq->private;
 781         struct net *net = seq_file_net(seq);
 782
 783         do {
 784                 sk = sk_nulls_next(sk);
 785         } while (sk && (!net_eq(sock_net(sk), net)));
 786
 787         if (!sk)
 788                 return ping_get_first(seq, state->bucket + 1);
 789         return sk;
 790 }
 791
 792 static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
 793 {
 794         struct sock *sk = ping_get_first(seq, 0);
 795
 796         if (sk)
 797                 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
 798                         --pos;
 799         return pos ? NULL : sk;
 800 }
 801
 802 static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
 803 {
 804         struct ping_iter_state *state = seq->private;
 805         state->bucket = 0;
 806
 807         read_lock_bh(&ping_table.lock);
 808
 809         return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
 810 }
 811
 812 static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
 813 {
 814         struct sock *sk;
 815
 816         if (v == SEQ_START_TOKEN)
 817                 sk = ping_get_idx(seq, 0);
 818         else
 819                 sk = ping_get_next(seq, v);
 820
 821         ++*pos;
 822         return sk;
 823 }
 824
 825 static void ping_seq_stop(struct seq_file *seq, void *v)
 826 {
 827         read_unlock_bh(&ping_table.lock);
 828 }
 829
 830 static void ping_format_sock(struct sock *sp, struct seq_file *f,
 831                 int bucket, int *len)
 832 {
 833         struct inet_sock *inet = inet_sk(sp);
 834         __be32 dest = inet->inet_daddr;
 835         __be32 src = inet->inet_rcv_saddr;
 836         __u16 destp = ntohs(inet->inet_dport);
 837         __u16 srcp = ntohs(inet->inet_sport);
 838
 839         seq_printf(f, "%5d: %08X:%04X %08X:%04X"
 840                 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
 841                 bucket, src, srcp, dest, destp, sp->sk_state,
 842                 sk_wmem_alloc_get(sp),
 843                 sk_rmem_alloc_get(sp),
 844                 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
 845                 atomic_read(&sp->sk_refcnt), sp,
 846                 atomic_read(&sp->sk_drops), len);
 847 }
 848
 849 static int ping_seq_show(struct seq_file *seq, void *v)
 850 {
 851         if (v == SEQ_START_TOKEN)
 852                 seq_printf(seq, "%-127s\n",
 853                            "  sl  local_address rem_address   st tx_queue "
 854                            "rx_queue tr tm->when retrnsmt   uid  timeout "
 855                            "inode ref pointer drops");
 856         else {
 857                 struct ping_iter_state *state = seq->private;
 858                 int len;
 859
 860                 ping_format_sock(v, seq, state->bucket, &len);
 861                 seq_printf(seq, "%*s\n", 127 - len, "");
 862         }
 863         return 0;
 864 }
 865
 866 static const struct seq_operations ping_seq_ops = {
 867         .show           = ping_seq_show,
 868         .start          = ping_seq_start,
 869         .next           = ping_seq_next,
 870         .stop           = ping_seq_stop,
 871 };
 872
 873 static int ping_seq_open(struct inode *inode, struct file *file)
 874 {
 875         return seq_open_net(inode, file, &ping_seq_ops,
 876                            sizeof(struct ping_iter_state));
 877 }
 878
 879 static const struct file_operations ping_seq_fops = {
 880         .open           = ping_seq_open,
 881         .read           = seq_read,
 882         .llseek         = seq_lseek,
 883         .release        = seq_release_net,
 884 };
 885
 886 static int ping_proc_register(struct net *net)
 887 {
 888         struct proc_dir_entry *p;
 889         int rc = 0;
 890
 891         p = proc_net_fops_create(net, "icmp", S_IRUGO, &ping_seq_fops);
 892         if (!p)
 893                 rc = -ENOMEM;
 894         return rc;
 895 }
 896
 897 static void ping_proc_unregister(struct net *net)
 898 {
 899         proc_net_remove(net, "icmp");
 900 }
 901
 902
 903 static int __net_init ping_proc_init_net(struct net *net)
 904 {
 905         return ping_proc_register(net);
 906 }
 907
 908 static void __net_exit ping_proc_exit_net(struct net *net)
 909 {
 910         ping_proc_unregister(net);
 911 }
 912
 913 static struct pernet_operations ping_net_ops = {
 914         .init = ping_proc_init_net,
 915         .exit = ping_proc_exit_net,
 916 };
 917
 918 int __init ping_proc_init(void)
 919 {
 920         return register_pernet_subsys(&ping_net_ops);
 921 }
 922
 923 void ping_proc_exit(void)
 924 {
 925         unregister_pernet_subsys(&ping_net_ops);
 926 }
 927
 928 #endif
 929
 930 void __init ping_init(void)
 931 {
 932         int i;
 933
 934         for (i = 0; i < PING_HTABLE_SIZE; i++)
 935                 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
 936         rwlock_init(&ping_table.lock);
 937 }