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 int num, unsigned int mask)
  55 {
  56         int res = (num + net_hash_mix(net)) & mask;
  57
  58         pr_debug("hash(%d) = %d\n", num, res);
  59         return res;
  60 }
  61
  62 static inline struct hlist_nulls_head *ping_hashslot(struct ping_table *table,
  63                                              struct net *net, unsigned int num)
  64 {
  65         return &table->hash[ping_hashfn(net, num, PING_HTABLE_MASK)];
  66 }
  67
  68 static int ping_v4_get_port(struct sock *sk, unsigned short ident)
  69 {
  70         struct hlist_nulls_node *node;
  71         struct hlist_nulls_head *hlist;
  72         struct inet_sock *isk, *isk2;
  73         struct sock *sk2 = NULL;
  74
  75         isk = inet_sk(sk);
  76         write_lock_bh(&ping_table.lock);
  77         if (ident == 0) {
  78                 u32 i;
  79                 u16 result = ping_port_rover + 1;
  80
  81                 for (i = 0; i < (1L << 16); i++, result++) {
  82                         if (!result)
  83                                 result++; /* avoid zero */
  84                         hlist = ping_hashslot(&ping_table, sock_net(sk),
  85                                             result);
  86                         ping_portaddr_for_each_entry(sk2, node, hlist) {
  87                                 isk2 = inet_sk(sk2);
  88
  89                                 if (isk2->inet_num == result)
  90                                         goto next_port;
  91                         }
  92
  93                         /* found */
  94                         ping_port_rover = ident = result;
  95                         break;
  96 next_port:
  97                         ;
  98                 }
  99                 if (i >= (1L << 16))
 100                         goto fail;
 101         } else {
 102                 hlist = ping_hashslot(&ping_table, sock_net(sk), ident);
 103                 ping_portaddr_for_each_entry(sk2, node, hlist) {
 104                         isk2 = inet_sk(sk2);
 105
 106                         if ((isk2->inet_num == ident) &&
 107                             (sk2 != sk) &&
 108                             (!sk2->sk_reuse || !sk->sk_reuse))
 109                                 goto fail;
 110                 }
 111         }
 112
 113         pr_debug("found port/ident = %d\n", ident);
 114         isk->inet_num = ident;
 115         if (sk_unhashed(sk)) {
 116                 pr_debug("was not hashed\n");
 117                 sock_hold(sk);
 118                 hlist_nulls_add_head(&sk->sk_nulls_node, hlist);
 119                 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
 120         }
 121         write_unlock_bh(&ping_table.lock);
 122         return 0;
 123
 124 fail:
 125         write_unlock_bh(&ping_table.lock);
 126         return 1;
 127 }
 128
 129 static void ping_v4_hash(struct sock *sk)
 130 {
 131         pr_debug("ping_v4_hash(sk->port=%u)\n", inet_sk(sk)->inet_num);
 132         BUG(); /* "Please do not press this button again." */
 133 }
 134
 135 static void ping_v4_unhash(struct sock *sk)
 136 {
 137         struct inet_sock *isk = inet_sk(sk);
 138         pr_debug("ping_v4_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
 139         if (sk_hashed(sk)) {
 140                 write_lock_bh(&ping_table.lock);
 141                 hlist_nulls_del(&sk->sk_nulls_node);
 142                 sock_put(sk);
 143                 isk->inet_num = 0;
 144                 isk->inet_sport = 0;
 145                 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
 146                 write_unlock_bh(&ping_table.lock);
 147         }
 148 }
 149
 150 static struct sock *ping_v4_lookup(struct net *net, __be32 saddr, __be32 daddr,
 151                                    u16 ident, int dif)
 152 {
 153         struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
 154         struct sock *sk = NULL;
 155         struct inet_sock *isk;
 156         struct hlist_nulls_node *hnode;
 157
 158         pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
 159                  (int)ident, &daddr, dif);
 160         read_lock_bh(&ping_table.lock);
 161
 162         ping_portaddr_for_each_entry(sk, hnode, hslot) {
 163                 isk = inet_sk(sk);
 164
 165                 pr_debug("found: %p: num = %d, daddr = %pI4, dif = %d\n", sk,
 166                          (int)isk->inet_num, &isk->inet_rcv_saddr,
 167                          sk->sk_bound_dev_if);
 168
 169                 pr_debug("iterate\n");
 170                 if (isk->inet_num != ident)
 171                         continue;
 172                 if (isk->inet_rcv_saddr && isk->inet_rcv_saddr != daddr)
 173                         continue;
 174                 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
 175                         continue;
 176
 177                 sock_hold(sk);
 178                 goto exit;
 179         }
 180
 181         sk = NULL;
 182 exit:
 183         read_unlock_bh(&ping_table.lock);
 184
 185         return sk;
 186 }
 187
 188 static void inet_get_ping_group_range_net(struct net *net, kgid_t *low,
 189                                           kgid_t *high)
 190 {
 191         kgid_t *data = net->ipv4.sysctl_ping_group_range;
 192         unsigned int seq;
 193
 194         do {
 195                 seq = read_seqbegin(&sysctl_local_ports.lock);
 196
 197                 *low = data[0];
 198                 *high = data[1];
 199         } while (read_seqretry(&sysctl_local_ports.lock, seq));
 200 }
 201
 202
 203 static int ping_init_sock(struct sock *sk)
 204 {
 205         struct net *net = sock_net(sk);
 206         kgid_t group = current_egid();
 207         struct group_info *group_info = get_current_groups();
 208         int i, j, count = group_info->ngroups;
 209         kgid_t low, high;
 210
 211         inet_get_ping_group_range_net(net, &low, &high);
 212         if (gid_lte(low, group) && gid_lte(group, high))
 213                 return 0;
 214
 215         for (i = 0; i < group_info->nblocks; i++) {
 216                 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
 217                 for (j = 0; j < cp_count; j++) {
 218                         kgid_t gid = group_info->blocks[i][j];
 219                         if (gid_lte(low, gid) && gid_lte(gid, high))
 220                                 return 0;
 221                 }
 222
 223                 count -= cp_count;
 224         }
 225
 226         return -EACCES;
 227 }
 228
 229 static void ping_close(struct sock *sk, long timeout)
 230 {
 231         pr_debug("ping_close(sk=%p,sk->num=%u)\n",
 232                  inet_sk(sk), inet_sk(sk)->inet_num);
 233         pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
 234
 235         sk_common_release(sk);
 236 }
 237
 238 /*
 239  * We need our own bind because there are no privileged id's == local ports.
 240  * Moreover, we don't allow binding to multi- and broadcast addresses.
 241  */
 242
 243 static int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 244 {
 245         struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
 246         struct inet_sock *isk = inet_sk(sk);
 247         unsigned short snum;
 248         int chk_addr_ret;
 249         int err;
 250
 251         if (addr_len < sizeof(struct sockaddr_in))
 252                 return -EINVAL;
 253
 254         pr_debug("ping_v4_bind(sk=%p,sa_addr=%08x,sa_port=%d)\n",
 255                  sk, addr->sin_addr.s_addr, ntohs(addr->sin_port));
 256
 257         chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
 258         if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
 259                 chk_addr_ret = RTN_LOCAL;
 260
 261         if ((sysctl_ip_nonlocal_bind == 0 &&
 262             isk->freebind == 0 && isk->transparent == 0 &&
 263              chk_addr_ret != RTN_LOCAL) ||
 264             chk_addr_ret == RTN_MULTICAST ||
 265             chk_addr_ret == RTN_BROADCAST)
 266                 return -EADDRNOTAVAIL;
 267
 268         lock_sock(sk);
 269
 270         err = -EINVAL;
 271         if (isk->inet_num != 0)
 272                 goto out;
 273
 274         err = -EADDRINUSE;
 275         isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
 276         snum = ntohs(addr->sin_port);
 277         if (ping_v4_get_port(sk, snum) != 0) {
 278                 isk->inet_saddr = isk->inet_rcv_saddr = 0;
 279                 goto out;
 280         }
 281
 282         pr_debug("after bind(): num = %d, daddr = %pI4, dif = %d\n",
 283                  (int)isk->inet_num,
 284                  &isk->inet_rcv_saddr,
 285                  (int)sk->sk_bound_dev_if);
 286
 287         err = 0;
 288         if (isk->inet_rcv_saddr)
 289                 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
 290         if (snum)
 291                 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
 292         isk->inet_sport = htons(isk->inet_num);
 293         isk->inet_daddr = 0;
 294         isk->inet_dport = 0;
 295         sk_dst_reset(sk);
 296 out:
 297         release_sock(sk);
 298         pr_debug("ping_v4_bind -> %d\n", err);
 299         return err;
 300 }
 301
 302 /*
 303  * Is this a supported type of ICMP message?
 304  */
 305
 306 static inline int ping_supported(int type, int code)
 307 {
 308         if (type == ICMP_ECHO && code == 0)
 309                 return 1;
 310         return 0;
 311 }
 312
 313 /*
 314  * This routine is called by the ICMP module when it gets some
 315  * sort of error condition.
 316  */
 317
 318 static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
 319
 320 void ping_err(struct sk_buff *skb, u32 info)
 321 {
 322         struct iphdr *iph = (struct iphdr *)skb->data;
 323         struct icmphdr *icmph = (struct icmphdr *)(skb->data+(iph->ihl<<2));
 324         struct inet_sock *inet_sock;
 325         int type = icmp_hdr(skb)->type;
 326         int code = icmp_hdr(skb)->code;
 327         struct net *net = dev_net(skb->dev);
 328         struct sock *sk;
 329         int harderr;
 330         int err;
 331
 332         /* We assume the packet has already been checked by icmp_unreach */
 333
 334         if (!ping_supported(icmph->type, icmph->code))
 335                 return;
 336
 337         pr_debug("ping_err(type=%04x,code=%04x,id=%04x,seq=%04x)\n", type,
 338                  code, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
 339
 340         sk = ping_v4_lookup(net, iph->daddr, iph->saddr,
 341                             ntohs(icmph->un.echo.id), skb->dev->ifindex);
 342         if (sk == NULL) {
 343                 pr_debug("no socket, dropping\n");
 344                 return; /* No socket for error */
 345         }
 346         pr_debug("err on socket %p\n", sk);
 347
 348         err = 0;
 349         harderr = 0;
 350         inet_sock = inet_sk(sk);
 351
 352         switch (type) {
 353         default:
 354         case ICMP_TIME_EXCEEDED:
 355                 err = EHOSTUNREACH;
 356                 break;
 357         case ICMP_SOURCE_QUENCH:
 358                 /* This is not a real error but ping wants to see it.
 359                  * Report it with some fake errno. */
 360                 err = EREMOTEIO;
 361                 break;
 362         case ICMP_PARAMETERPROB:
 363                 err = EPROTO;
 364                 harderr = 1;
 365                 break;
 366         case ICMP_DEST_UNREACH:
 367                 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
 368                         ipv4_sk_update_pmtu(skb, sk, info);
 369                         if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
 370                                 err = EMSGSIZE;
 371                                 harderr = 1;
 372                                 break;
 373                         }
 374                         goto out;
 375                 }
 376                 err = EHOSTUNREACH;
 377                 if (code <= NR_ICMP_UNREACH) {
 378                         harderr = icmp_err_convert[code].fatal;
 379                         err = icmp_err_convert[code].errno;
 380                 }
 381                 break;
 382         case ICMP_REDIRECT:
 383                 /* See ICMP_SOURCE_QUENCH */
 384                 ipv4_sk_redirect(skb, sk);
 385                 err = EREMOTEIO;
 386                 break;
 387         }
 388
 389         /*
 390          *      RFC1122: OK.  Passes ICMP errors back to application, as per
 391          *      4.1.3.3.
 392          */
 393         if (!inet_sock->recverr) {
 394                 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
 395                         goto out;
 396         } else {
 397                 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
 398                          info, (u8 *)icmph);
 399         }
 400         sk->sk_err = err;
 401         sk->sk_error_report(sk);
 402 out:
 403         sock_put(sk);
 404 }
 405
 406 /*
 407  *      Copy and checksum an ICMP Echo packet from user space into a buffer.
 408  */
 409
 410 struct pingfakehdr {
 411         struct icmphdr icmph;
 412         struct iovec *iov;
 413         __wsum wcheck;
 414 };
 415
 416 static int ping_getfrag(void *from, char *to,
 417                         int offset, int fraglen, int odd, struct sk_buff *skb)
 418 {
 419         struct pingfakehdr *pfh = (struct pingfakehdr *)from;
 420
 421         if (offset == 0) {
 422                 if (fraglen < sizeof(struct icmphdr))
 423                         BUG();
 424                 if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
 425                             pfh->iov, 0, fraglen - sizeof(struct icmphdr),
 426                             &pfh->wcheck))
 427                         return -EFAULT;
 428
 429                 return 0;
 430         }
 431         if (offset < sizeof(struct icmphdr))
 432                 BUG();
 433         if (csum_partial_copy_fromiovecend
 434                         (to, pfh->iov, offset - sizeof(struct icmphdr),
 435                          fraglen, &pfh->wcheck))
 436                 return -EFAULT;
 437         return 0;
 438 }
 439
 440 static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
 441                                     struct flowi4 *fl4)
 442 {
 443         struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
 444
 445         pfh->wcheck = csum_partial((char *)&pfh->icmph,
 446                 sizeof(struct icmphdr), pfh->wcheck);
 447         pfh->icmph.checksum = csum_fold(pfh->wcheck);
 448         memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
 449         skb->ip_summed = CHECKSUM_NONE;
 450         return ip_push_pending_frames(sk, fl4);
 451 }
 452
 453 static int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 454                         size_t len)
 455 {
 456         struct net *net = sock_net(sk);
 457         struct flowi4 fl4;
 458         struct inet_sock *inet = inet_sk(sk);
 459         struct ipcm_cookie ipc;
 460         struct icmphdr user_icmph;
 461         struct pingfakehdr pfh;
 462         struct rtable *rt = NULL;
 463         struct ip_options_data opt_copy;
 464         int free = 0;
 465         __be32 saddr, daddr, faddr;
 466         u8  tos;
 467         int err;
 468
 469         pr_debug("ping_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
 470
 471
 472         if (len > 0xFFFF)
 473                 return -EMSGSIZE;
 474
 475         /*
 476          *      Check the flags.
 477          */
 478
 479         /* Mirror BSD error message compatibility */
 480         if (msg->msg_flags & MSG_OOB)
 481                 return -EOPNOTSUPP;
 482
 483         /*
 484          *      Fetch the ICMP header provided by the userland.
 485          *      iovec is modified!
 486          */
 487
 488         if (memcpy_fromiovec((u8 *)&user_icmph, msg->msg_iov,
 489                              sizeof(struct icmphdr)))
 490                 return -EFAULT;
 491         if (!ping_supported(user_icmph.type, user_icmph.code))
 492                 return -EINVAL;
 493
 494         /*
 495          *      Get and verify the address.
 496          */
 497
 498         if (msg->msg_name) {
 499                 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
 500                 if (msg->msg_namelen < sizeof(*usin))
 501                         return -EINVAL;
 502                 if (usin->sin_family != AF_INET)
 503                         return -EINVAL;
 504                 daddr = usin->sin_addr.s_addr;
 505                 /* no remote port */
 506         } else {
 507                 if (sk->sk_state != TCP_ESTABLISHED)
 508                         return -EDESTADDRREQ;
 509                 daddr = inet->inet_daddr;
 510                 /* no remote port */
 511         }
 512
 513         ipc.addr = inet->inet_saddr;
 514         ipc.opt = NULL;
 515         ipc.oif = sk->sk_bound_dev_if;
 516         ipc.tx_flags = 0;
 517         err = sock_tx_timestamp(sk, &ipc.tx_flags);
 518         if (err)
 519                 return err;
 520
 521         if (msg->msg_controllen) {
 522                 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
 523                 if (err)
 524                         return err;
 525                 if (ipc.opt)
 526                         free = 1;
 527         }
 528         if (!ipc.opt) {
 529                 struct ip_options_rcu *inet_opt;
 530
 531                 rcu_read_lock();
 532                 inet_opt = rcu_dereference(inet->inet_opt);
 533                 if (inet_opt) {
 534                         memcpy(&opt_copy, inet_opt,
 535                                sizeof(*inet_opt) + inet_opt->opt.optlen);
 536                         ipc.opt = &opt_copy.opt;
 537                 }
 538                 rcu_read_unlock();
 539         }
 540
 541         saddr = ipc.addr;
 542         ipc.addr = faddr = daddr;
 543
 544         if (ipc.opt && ipc.opt->opt.srr) {
 545                 if (!daddr)
 546                         return -EINVAL;
 547                 faddr = ipc.opt->opt.faddr;
 548         }
 549         tos = RT_TOS(inet->tos);
 550         if (sock_flag(sk, SOCK_LOCALROUTE) ||
 551             (msg->msg_flags & MSG_DONTROUTE) ||
 552             (ipc.opt && ipc.opt->opt.is_strictroute)) {
 553                 tos |= RTO_ONLINK;
 554         }
 555
 556         if (ipv4_is_multicast(daddr)) {
 557                 if (!ipc.oif)
 558                         ipc.oif = inet->mc_index;
 559                 if (!saddr)
 560                         saddr = inet->mc_addr;
 561         } else if (!ipc.oif)
 562                 ipc.oif = inet->uc_index;
 563
 564         flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
 565                            RT_SCOPE_UNIVERSE, sk->sk_protocol,
 566                            inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
 567
 568         security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
 569         rt = ip_route_output_flow(net, &fl4, sk);
 570         if (IS_ERR(rt)) {
 571                 err = PTR_ERR(rt);
 572                 rt = NULL;
 573                 if (err == -ENETUNREACH)
 574                         IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
 575                 goto out;
 576         }
 577
 578         err = -EACCES;
 579         if ((rt->rt_flags & RTCF_BROADCAST) &&
 580             !sock_flag(sk, SOCK_BROADCAST))
 581                 goto out;
 582
 583         if (msg->msg_flags & MSG_CONFIRM)
 584                 goto do_confirm;
 585 back_from_confirm:
 586
 587         if (!ipc.addr)
 588                 ipc.addr = fl4.daddr;
 589
 590         lock_sock(sk);
 591
 592         pfh.icmph.type = user_icmph.type; /* already checked */
 593         pfh.icmph.code = user_icmph.code; /* ditto */
 594         pfh.icmph.checksum = 0;
 595         pfh.icmph.un.echo.id = inet->inet_sport;
 596         pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
 597         pfh.iov = msg->msg_iov;
 598         pfh.wcheck = 0;
 599
 600         err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
 601                         0, &ipc, &rt, msg->msg_flags);
 602         if (err)
 603                 ip_flush_pending_frames(sk);
 604         else
 605                 err = ping_push_pending_frames(sk, &pfh, &fl4);
 606         release_sock(sk);
 607
 608 out:
 609         ip_rt_put(rt);
 610         if (free)
 611                 kfree(ipc.opt);
 612         if (!err) {
 613                 icmp_out_count(sock_net(sk), user_icmph.type);
 614                 return len;
 615         }
 616         return err;
 617
 618 do_confirm:
 619         dst_confirm(&rt->dst);
 620         if (!(msg->msg_flags & MSG_PROBE) || len)
 621                 goto back_from_confirm;
 622         err = 0;
 623         goto out;
 624 }
 625
 626 static int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 627                         size_t len, int noblock, int flags, int *addr_len)
 628 {
 629         struct inet_sock *isk = inet_sk(sk);
 630         struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
 631         struct sk_buff *skb;
 632         int copied, err;
 633
 634         pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
 635
 636         err = -EOPNOTSUPP;
 637         if (flags & MSG_OOB)
 638                 goto out;
 639
 640         if (addr_len)
 641                 *addr_len = sizeof(*sin);
 642
 643         if (flags & MSG_ERRQUEUE)
 644                 return ip_recv_error(sk, msg, len);
 645
 646         skb = skb_recv_datagram(sk, flags, noblock, &err);
 647         if (!skb)
 648                 goto out;
 649
 650         copied = skb->len;
 651         if (copied > len) {
 652                 msg->msg_flags |= MSG_TRUNC;
 653                 copied = len;
 654         }
 655
 656         /* Don't bother checking the checksum */
 657         err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
 658         if (err)
 659                 goto done;
 660
 661         sock_recv_timestamp(msg, sk, skb);
 662
 663         /* Copy the address. */
 664         if (sin) {
 665                 sin->sin_family = AF_INET;
 666                 sin->sin_port = 0 /* skb->h.uh->source */;
 667                 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
 668                 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
 669         }
 670         if (isk->cmsg_flags)
 671                 ip_cmsg_recv(msg, skb);
 672         err = copied;
 673
 674 done:
 675         skb_free_datagram(sk, skb);
 676 out:
 677         pr_debug("ping_recvmsg -> %d\n", err);
 678         return err;
 679 }
 680
 681 static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
 682 {
 683         pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
 684                  inet_sk(sk), inet_sk(sk)->inet_num, skb);
 685         if (sock_queue_rcv_skb(sk, skb) < 0) {
 686                 kfree_skb(skb);
 687                 pr_debug("ping_queue_rcv_skb -> failed\n");
 688                 return -1;
 689         }
 690         return 0;
 691 }
 692
 693
 694 /*
 695  *      All we need to do is get the socket.
 696  */
 697
 698 void ping_rcv(struct sk_buff *skb)
 699 {
 700         struct sock *sk;
 701         struct net *net = dev_net(skb->dev);
 702         struct iphdr *iph = ip_hdr(skb);
 703         struct icmphdr *icmph = icmp_hdr(skb);
 704         __be32 saddr = iph->saddr;
 705         __be32 daddr = iph->daddr;
 706
 707         /* We assume the packet has already been checked by icmp_rcv */
 708
 709         pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
 710                  skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
 711
 712         /* Push ICMP header back */
 713         skb_push(skb, skb->data - (u8 *)icmph);
 714
 715         sk = ping_v4_lookup(net, saddr, daddr, ntohs(icmph->un.echo.id),
 716                             skb->dev->ifindex);
 717         if (sk != NULL) {
 718                 pr_debug("rcv on socket %p\n", sk);
 719                 ping_queue_rcv_skb(sk, skb_get(skb));
 720                 sock_put(sk);
 721                 return;
 722         }
 723         pr_debug("no socket, dropping\n");
 724
 725         /* We're called from icmp_rcv(). kfree_skb() is done there. */
 726 }
 727
 728 struct proto ping_prot = {
 729         .name =         "PING",
 730         .owner =        THIS_MODULE,
 731         .init =         ping_init_sock,
 732         .close =        ping_close,
 733         .connect =      ip4_datagram_connect,
 734         .disconnect =   udp_disconnect,
 735         .setsockopt =   ip_setsockopt,
 736         .getsockopt =   ip_getsockopt,
 737         .sendmsg =      ping_sendmsg,
 738         .recvmsg =      ping_recvmsg,
 739         .bind =         ping_bind,
 740         .backlog_rcv =  ping_queue_rcv_skb,
 741         .release_cb =   ip4_datagram_release_cb,
 742         .hash =         ping_v4_hash,
 743         .unhash =       ping_v4_unhash,
 744         .get_port =     ping_v4_get_port,
 745         .obj_size =     sizeof(struct inet_sock),
 746 };
 747 EXPORT_SYMBOL(ping_prot);
 748
 749 #ifdef CONFIG_PROC_FS
 750
 751 static struct sock *ping_get_first(struct seq_file *seq, int start)
 752 {
 753         struct sock *sk;
 754         struct ping_iter_state *state = seq->private;
 755         struct net *net = seq_file_net(seq);
 756
 757         for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
 758              ++state->bucket) {
 759                 struct hlist_nulls_node *node;
 760                 struct hlist_nulls_head *hslot;
 761
 762                 hslot = &ping_table.hash[state->bucket];
 763
 764                 if (hlist_nulls_empty(hslot))
 765                         continue;
 766
 767                 sk_nulls_for_each(sk, node, hslot) {
 768                         if (net_eq(sock_net(sk), net))
 769                                 goto found;
 770                 }
 771         }
 772         sk = NULL;
 773 found:
 774         return sk;
 775 }
 776
 777 static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
 778 {
 779         struct ping_iter_state *state = seq->private;
 780         struct net *net = seq_file_net(seq);
 781
 782         do {
 783                 sk = sk_nulls_next(sk);
 784         } while (sk && (!net_eq(sock_net(sk), net)));
 785
 786         if (!sk)
 787                 return ping_get_first(seq, state->bucket + 1);
 788         return sk;
 789 }
 790
 791 static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
 792 {
 793         struct sock *sk = ping_get_first(seq, 0);
 794
 795         if (sk)
 796                 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
 797                         --pos;
 798         return pos ? NULL : sk;
 799 }
 800
 801 static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
 802 {
 803         struct ping_iter_state *state = seq->private;
 804         state->bucket = 0;
 805
 806         read_lock_bh(&ping_table.lock);
 807
 808         return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
 809 }
 810
 811 static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
 812 {
 813         struct sock *sk;
 814
 815         if (v == SEQ_START_TOKEN)
 816                 sk = ping_get_idx(seq, 0);
 817         else
 818                 sk = ping_get_next(seq, v);
 819
 820         ++*pos;
 821         return sk;
 822 }
 823
 824 static void ping_seq_stop(struct seq_file *seq, void *v)
 825 {
 826         read_unlock_bh(&ping_table.lock);
 827 }
 828
 829 static void ping_format_sock(struct sock *sp, struct seq_file *f,
 830                 int bucket, int *len)
 831 {
 832         struct inet_sock *inet = inet_sk(sp);
 833         __be32 dest = inet->inet_daddr;
 834         __be32 src = inet->inet_rcv_saddr;
 835         __u16 destp = ntohs(inet->inet_dport);
 836         __u16 srcp = ntohs(inet->inet_sport);
 837
 838         seq_printf(f, "%5d: %08X:%04X %08X:%04X"
 839                 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
 840                 bucket, src, srcp, dest, destp, sp->sk_state,
 841                 sk_wmem_alloc_get(sp),
 842                 sk_rmem_alloc_get(sp),
 843                 0, 0L, 0,
 844                 from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
 845                 0, sock_i_ino(sp),
 846                 atomic_read(&sp->sk_refcnt), sp,
 847                 atomic_read(&sp->sk_drops), len);
 848 }
 849
 850 static int ping_seq_show(struct seq_file *seq, void *v)
 851 {
 852         if (v == SEQ_START_TOKEN)
 853                 seq_printf(seq, "%-127s\n",
 854                            "  sl  local_address rem_address   st tx_queue "
 855                            "rx_queue tr tm->when retrnsmt   uid  timeout "
 856                            "inode ref pointer drops");
 857         else {
 858                 struct ping_iter_state *state = seq->private;
 859                 int len;
 860
 861                 ping_format_sock(v, seq, state->bucket, &len);
 862                 seq_printf(seq, "%*s\n", 127 - len, "");
 863         }
 864         return 0;
 865 }
 866
 867 static const struct seq_operations ping_seq_ops = {
 868         .show           = ping_seq_show,
 869         .start          = ping_seq_start,
 870         .next           = ping_seq_next,
 871         .stop           = ping_seq_stop,
 872 };
 873
 874 static int ping_seq_open(struct inode *inode, struct file *file)
 875 {
 876         return seq_open_net(inode, file, &ping_seq_ops,
 877                            sizeof(struct ping_iter_state));
 878 }
 879
 880 static const struct file_operations ping_seq_fops = {
 881         .open           = ping_seq_open,
 882         .read           = seq_read,
 883         .llseek         = seq_lseek,
 884         .release        = seq_release_net,
 885 };
 886
 887 static int ping_proc_register(struct net *net)
 888 {
 889         struct proc_dir_entry *p;
 890         int rc = 0;
 891
 892         p = proc_create("icmp", S_IRUGO, net->proc_net, &ping_seq_fops);
 893         if (!p)
 894                 rc = -ENOMEM;
 895         return rc;
 896 }
 897
 898 static void ping_proc_unregister(struct net *net)
 899 {
 900         remove_proc_entry("icmp", net->proc_net);
 901 }
 902
 903
 904 static int __net_init ping_proc_init_net(struct net *net)
 905 {
 906         return ping_proc_register(net);
 907 }
 908
 909 static void __net_exit ping_proc_exit_net(struct net *net)
 910 {
 911         ping_proc_unregister(net);
 912 }
 913
 914 static struct pernet_operations ping_net_ops = {
 915         .init = ping_proc_init_net,
 916         .exit = ping_proc_exit_net,
 917 };
 918
 919 int __init ping_proc_init(void)
 920 {
 921         return register_pernet_subsys(&ping_net_ops);
 922 }
 923
 924 void ping_proc_exit(void)
 925 {
 926         unregister_pernet_subsys(&ping_net_ops);
 927 }
 928
 929 #endif
 930
 931 void __init ping_init(void)
 932 {
 933         int i;
 934
 935         for (i = 0; i < PING_HTABLE_SIZE; i++)
 936                 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
 937         rwlock_init(&ping_table.lock);
 938 }