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