linux/net/ipv4/ping.c
<<
>>
Prefs
   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
  50static struct ping_table ping_table;
  51
  52static u16 ping_port_rover;
  53
  54static 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
  62static 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
  68static 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;
  96next_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
 124fail:
 125        write_unlock_bh(&ping_table.lock);
 126        return 1;
 127}
 128
 129static 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
 135static void ping_v4_unhash(struct sock *sk)
 136{
 137        struct inet_sock *isk = inet_sk(sk);
 138
 139        pr_debug("ping_v4_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
 140        write_lock_bh(&ping_table.lock);
 141        if (sk_hashed(sk)) {
 142                hlist_nulls_del(&sk->sk_nulls_node);
 143                sk_nulls_node_init(&sk->sk_nulls_node);
 144                sock_put(sk);
 145                isk->inet_num = 0;
 146                isk->inet_sport = 0;
 147                sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
 148        }
 149        write_unlock_bh(&ping_table.lock);
 150}
 151
 152static struct sock *ping_v4_lookup(struct net *net, __be32 saddr, __be32 daddr,
 153                                   u16 ident, int dif)
 154{
 155        struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
 156        struct sock *sk = NULL;
 157        struct inet_sock *isk;
 158        struct hlist_nulls_node *hnode;
 159
 160        pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
 161                 (int)ident, &daddr, dif);
 162        read_lock_bh(&ping_table.lock);
 163
 164        ping_portaddr_for_each_entry(sk, hnode, hslot) {
 165                isk = inet_sk(sk);
 166
 167                pr_debug("found: %p: num = %d, daddr = %pI4, dif = %d\n", sk,
 168                         (int)isk->inet_num, &isk->inet_rcv_saddr,
 169                         sk->sk_bound_dev_if);
 170
 171                pr_debug("iterate\n");
 172                if (isk->inet_num != ident)
 173                        continue;
 174                if (isk->inet_rcv_saddr && isk->inet_rcv_saddr != daddr)
 175                        continue;
 176                if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
 177                        continue;
 178
 179                sock_hold(sk);
 180                goto exit;
 181        }
 182
 183        sk = NULL;
 184exit:
 185        read_unlock_bh(&ping_table.lock);
 186
 187        return sk;
 188}
 189
 190static void inet_get_ping_group_range_net(struct net *net, kgid_t *low,
 191                                          kgid_t *high)
 192{
 193        kgid_t *data = net->ipv4.sysctl_ping_group_range;
 194        unsigned int seq;
 195
 196        do {
 197                seq = read_seqbegin(&net->ipv4_sysctl_local_ports.lock);
 198
 199                *low = data[0];
 200                *high = data[1];
 201        } while (read_seqretry(&net->ipv4_sysctl_local_ports.lock, seq));
 202}
 203
 204
 205static int ping_init_sock(struct sock *sk)
 206{
 207        struct net *net = sock_net(sk);
 208        kgid_t group = current_egid();
 209        struct group_info *group_info;
 210        int i, j, count;
 211        kgid_t low, high;
 212        int ret = 0;
 213
 214        inet_get_ping_group_range_net(net, &low, &high);
 215        if (gid_lte(low, group) && gid_lte(group, high))
 216                return 0;
 217
 218        group_info = get_current_groups();
 219        count = group_info->ngroups;
 220        for (i = 0; i < group_info->nblocks; i++) {
 221                int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
 222                for (j = 0; j < cp_count; j++) {
 223                        kgid_t gid = group_info->blocks[i][j];
 224                        if (gid_lte(low, gid) && gid_lte(gid, high))
 225                                goto out_release_group;
 226                }
 227
 228                count -= cp_count;
 229        }
 230
 231        ret = -EACCES;
 232
 233out_release_group:
 234        put_group_info(group_info);
 235        return ret;
 236}
 237
 238static void ping_close(struct sock *sk, long timeout)
 239{
 240        pr_debug("ping_close(sk=%p,sk->num=%u)\n",
 241                 inet_sk(sk), inet_sk(sk)->inet_num);
 242        pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
 243
 244        sk_common_release(sk);
 245}
 246
 247/*
 248 * We need our own bind because there are no privileged id's == local ports.
 249 * Moreover, we don't allow binding to multi- and broadcast addresses.
 250 */
 251
 252static int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
 253{
 254        struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
 255        struct inet_sock *isk = inet_sk(sk);
 256        struct net *net = sock_net(sk);
 257        unsigned short snum;
 258        int chk_addr_ret;
 259        int err;
 260
 261        if (addr_len < sizeof(struct sockaddr_in))
 262                return -EINVAL;
 263
 264        pr_debug("ping_v4_bind(sk=%p,sa_addr=%08x,sa_port=%d)\n",
 265                 sk, addr->sin_addr.s_addr, ntohs(addr->sin_port));
 266
 267        chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
 268        if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
 269                chk_addr_ret = RTN_LOCAL;
 270
 271        if ((net->ipv4_sysctl_ip_nonlocal_bind == 0 &&
 272            isk->freebind == 0 && isk->transparent == 0 &&
 273             chk_addr_ret != RTN_LOCAL) ||
 274            chk_addr_ret == RTN_MULTICAST ||
 275            chk_addr_ret == RTN_BROADCAST)
 276                return -EADDRNOTAVAIL;
 277
 278        lock_sock(sk);
 279
 280        err = -EINVAL;
 281        if (isk->inet_num != 0)
 282                goto out;
 283
 284        err = -EADDRINUSE;
 285        isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
 286        snum = ntohs(addr->sin_port);
 287        if (ping_v4_get_port(sk, snum) != 0) {
 288                isk->inet_saddr = isk->inet_rcv_saddr = 0;
 289                goto out;
 290        }
 291
 292        pr_debug("after bind(): num = %d, daddr = %pI4, dif = %d\n",
 293                 (int)isk->inet_num,
 294                 &isk->inet_rcv_saddr,
 295                 (int)sk->sk_bound_dev_if);
 296
 297        err = 0;
 298        if (isk->inet_rcv_saddr)
 299                sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
 300        if (snum)
 301                sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
 302        isk->inet_sport = htons(isk->inet_num);
 303        isk->inet_daddr = 0;
 304        isk->inet_dport = 0;
 305        sk_dst_reset(sk);
 306out:
 307        release_sock(sk);
 308        pr_debug("ping_v4_bind -> %d\n", err);
 309        return err;
 310}
 311
 312/*
 313 * Is this a supported type of ICMP message?
 314 */
 315
 316static inline int ping_supported(int type, int code)
 317{
 318        if (type == ICMP_ECHO && code == 0)
 319                return 1;
 320        return 0;
 321}
 322
 323/*
 324 * This routine is called by the ICMP module when it gets some
 325 * sort of error condition.
 326 */
 327
 328static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
 329
 330void ping_err(struct sk_buff *skb, u32 info)
 331{
 332        struct iphdr *iph = (struct iphdr *)skb->data;
 333        struct icmphdr *icmph = (struct icmphdr *)(skb->data+(iph->ihl<<2));
 334        struct inet_sock *inet_sock;
 335        int type = icmp_hdr(skb)->type;
 336        int code = icmp_hdr(skb)->code;
 337        struct net *net = dev_net(skb->dev);
 338        struct sock *sk;
 339        int harderr;
 340        int err;
 341
 342        /* We assume the packet has already been checked by icmp_unreach */
 343
 344        if (!ping_supported(icmph->type, icmph->code))
 345                return;
 346
 347        pr_debug("ping_err(type=%04x,code=%04x,id=%04x,seq=%04x)\n", type,
 348                 code, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
 349
 350        sk = ping_v4_lookup(net, iph->daddr, iph->saddr,
 351                            ntohs(icmph->un.echo.id), skb->dev->ifindex);
 352        if (sk == NULL) {
 353                pr_debug("no socket, dropping\n");
 354                return; /* No socket for error */
 355        }
 356        pr_debug("err on socket %p\n", sk);
 357
 358        err = 0;
 359        harderr = 0;
 360        inet_sock = inet_sk(sk);
 361
 362        switch (type) {
 363        default:
 364        case ICMP_TIME_EXCEEDED:
 365                err = EHOSTUNREACH;
 366                break;
 367        case ICMP_SOURCE_QUENCH:
 368                /* This is not a real error but ping wants to see it.
 369                 * Report it with some fake errno. */
 370                err = EREMOTEIO;
 371                break;
 372        case ICMP_PARAMETERPROB:
 373                err = EPROTO;
 374                harderr = 1;
 375                break;
 376        case ICMP_DEST_UNREACH:
 377                if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
 378                        ipv4_sk_update_pmtu(skb, sk, info);
 379                        if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
 380                                err = EMSGSIZE;
 381                                harderr = 1;
 382                                break;
 383                        }
 384                        goto out;
 385                }
 386                err = EHOSTUNREACH;
 387                if (code <= NR_ICMP_UNREACH) {
 388                        harderr = icmp_err_convert[code].fatal;
 389                        err = icmp_err_convert[code].errno;
 390                }
 391                break;
 392        case ICMP_REDIRECT:
 393                /* See ICMP_SOURCE_QUENCH */
 394                ipv4_sk_redirect(skb, sk);
 395                err = EREMOTEIO;
 396                break;
 397        }
 398
 399        /*
 400         *      RFC1122: OK.  Passes ICMP errors back to application, as per
 401         *      4.1.3.3.
 402         */
 403        if (!inet_sock->recverr) {
 404                if (!harderr || sk->sk_state != TCP_ESTABLISHED)
 405                        goto out;
 406        } else {
 407                ip_icmp_error(sk, skb, err, 0 /* no remote port */,
 408                         info, (u8 *)icmph);
 409        }
 410        sk->sk_err = err;
 411        sk->sk_error_report(sk);
 412out:
 413        sock_put(sk);
 414}
 415
 416/*
 417 *      Copy and checksum an ICMP Echo packet from user space into a buffer.
 418 */
 419
 420struct pingfakehdr {
 421        struct icmphdr icmph;
 422        struct iovec *iov;
 423        __wsum wcheck;
 424};
 425
 426static int ping_getfrag(void *from, char *to,
 427                        int offset, int fraglen, int odd, struct sk_buff *skb)
 428{
 429        struct pingfakehdr *pfh = (struct pingfakehdr *)from;
 430
 431        if (offset == 0) {
 432                if (fraglen < sizeof(struct icmphdr))
 433                        BUG();
 434                if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
 435                            pfh->iov, 0, fraglen - sizeof(struct icmphdr),
 436                            &pfh->wcheck))
 437                        return -EFAULT;
 438
 439                return 0;
 440        }
 441        if (offset < sizeof(struct icmphdr))
 442                BUG();
 443        if (csum_partial_copy_fromiovecend
 444                        (to, pfh->iov, offset - sizeof(struct icmphdr),
 445                         fraglen, &pfh->wcheck))
 446                return -EFAULT;
 447        return 0;
 448}
 449
 450static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
 451                                    struct flowi4 *fl4)
 452{
 453        struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
 454
 455        if (!skb)
 456                return 0;
 457        pfh->wcheck = csum_partial((char *)&pfh->icmph,
 458                sizeof(struct icmphdr), pfh->wcheck);
 459        pfh->icmph.checksum = csum_fold(pfh->wcheck);
 460        memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
 461        skb->ip_summed = CHECKSUM_NONE;
 462        return ip_push_pending_frames(sk, fl4);
 463}
 464
 465static int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 466                        size_t len)
 467{
 468        struct net *net = sock_net(sk);
 469        struct flowi4 fl4;
 470        struct inet_sock *inet = inet_sk(sk);
 471        struct ipcm_cookie ipc;
 472        struct icmphdr user_icmph;
 473        struct pingfakehdr pfh;
 474        struct rtable *rt = NULL;
 475        struct ip_options_data opt_copy;
 476        int free = 0;
 477        __be32 saddr, daddr, faddr;
 478        u8  tos;
 479        int err;
 480
 481        pr_debug("ping_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
 482
 483
 484        if (len > 0xFFFF)
 485                return -EMSGSIZE;
 486
 487        /* Must have at least a full ICMP header. */
 488        if (len < sizeof(user_icmph))
 489                return -EINVAL;
 490
 491        /*
 492         *      Check the flags.
 493         */
 494
 495        /* Mirror BSD error message compatibility */
 496        if (msg->msg_flags & MSG_OOB)
 497                return -EOPNOTSUPP;
 498
 499        /*
 500         *      Fetch the ICMP header provided by the userland.
 501         *      iovec is modified!
 502         */
 503
 504        if (memcpy_fromiovec((u8 *)&user_icmph, msg->msg_iov,
 505                             sizeof(struct icmphdr)))
 506                return -EFAULT;
 507        if (!ping_supported(user_icmph.type, user_icmph.code))
 508                return -EINVAL;
 509
 510        /*
 511         *      Get and verify the address.
 512         */
 513
 514        if (msg->msg_name) {
 515                struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
 516                if (msg->msg_namelen < sizeof(*usin))
 517                        return -EINVAL;
 518                if (usin->sin_family != AF_INET)
 519                        return -EINVAL;
 520                daddr = usin->sin_addr.s_addr;
 521                /* no remote port */
 522        } else {
 523                if (sk->sk_state != TCP_ESTABLISHED)
 524                        return -EDESTADDRREQ;
 525                daddr = inet->inet_daddr;
 526                /* no remote port */
 527        }
 528
 529        ipc.addr = inet->inet_saddr;
 530        ipc.opt = NULL;
 531        ipc.oif = sk->sk_bound_dev_if;
 532        ipc.tx_flags = 0;
 533        ipc.ttl = 0;
 534        ipc.tos = -1;
 535
 536        sock_tx_timestamp(sk, &ipc.tx_flags);
 537
 538        if (msg->msg_controllen) {
 539                err = ip_cmsg_send(sock_net(sk), msg, &ipc, false);
 540                if (unlikely(err)) {
 541                        kfree(ipc.opt);
 542                        return err;
 543                }
 544                if (ipc.opt)
 545                        free = 1;
 546        }
 547        if (!ipc.opt) {
 548                struct ip_options_rcu *inet_opt;
 549
 550                rcu_read_lock();
 551                inet_opt = rcu_dereference(inet->inet_opt);
 552                if (inet_opt) {
 553                        memcpy(&opt_copy, inet_opt,
 554                               sizeof(*inet_opt) + inet_opt->opt.optlen);
 555                        ipc.opt = &opt_copy.opt;
 556                }
 557                rcu_read_unlock();
 558        }
 559
 560        saddr = ipc.addr;
 561        ipc.addr = faddr = daddr;
 562
 563        if (ipc.opt && ipc.opt->opt.srr) {
 564                if (!daddr) {
 565                        err = -EINVAL;
 566                        goto out_free;
 567                }
 568                faddr = ipc.opt->opt.faddr;
 569        }
 570        tos = get_rttos(&ipc, inet);
 571        if (sock_flag(sk, SOCK_LOCALROUTE) ||
 572            (msg->msg_flags & MSG_DONTROUTE) ||
 573            (ipc.opt && ipc.opt->opt.is_strictroute)) {
 574                tos |= RTO_ONLINK;
 575        }
 576
 577        if (ipv4_is_multicast(daddr)) {
 578                if (!ipc.oif)
 579                        ipc.oif = inet->mc_index;
 580                if (!saddr)
 581                        saddr = inet->mc_addr;
 582        } else if (!ipc.oif)
 583                ipc.oif = inet->uc_index;
 584
 585        flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
 586                           RT_SCOPE_UNIVERSE, sk->sk_protocol,
 587                           inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
 588
 589        security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
 590        rt = ip_route_output_flow(net, &fl4, sk);
 591        if (IS_ERR(rt)) {
 592                err = PTR_ERR(rt);
 593                rt = NULL;
 594                if (err == -ENETUNREACH)
 595                        IP_INC_STATS(net, IPSTATS_MIB_OUTNOROUTES);
 596                goto out;
 597        }
 598
 599        err = -EACCES;
 600        if ((rt->rt_flags & RTCF_BROADCAST) &&
 601            !sock_flag(sk, SOCK_BROADCAST))
 602                goto out;
 603
 604        if (msg->msg_flags & MSG_CONFIRM)
 605                goto do_confirm;
 606back_from_confirm:
 607
 608        if (!ipc.addr)
 609                ipc.addr = fl4.daddr;
 610
 611        lock_sock(sk);
 612
 613        pfh.icmph.type = user_icmph.type; /* already checked */
 614        pfh.icmph.code = user_icmph.code; /* ditto */
 615        pfh.icmph.checksum = 0;
 616        pfh.icmph.un.echo.id = inet->inet_sport;
 617        pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
 618        pfh.iov = msg->msg_iov;
 619        pfh.wcheck = 0;
 620
 621        err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
 622                        0, &ipc, &rt, msg->msg_flags);
 623        if (err)
 624                ip_flush_pending_frames(sk);
 625        else
 626                err = ping_push_pending_frames(sk, &pfh, &fl4);
 627        release_sock(sk);
 628
 629out:
 630        ip_rt_put(rt);
 631out_free:
 632        if (free)
 633                kfree(ipc.opt);
 634        if (!err) {
 635                icmp_out_count(sock_net(sk), user_icmph.type);
 636                return len;
 637        }
 638        return err;
 639
 640do_confirm:
 641        if (msg->msg_flags & MSG_PROBE)
 642                dst_confirm_neigh(&rt->dst, &fl4.daddr);
 643        if (!(msg->msg_flags & MSG_PROBE) || len)
 644                goto back_from_confirm;
 645        err = 0;
 646        goto out;
 647}
 648
 649static int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
 650                        size_t len, int noblock, int flags, int *addr_len)
 651{
 652        struct inet_sock *isk = inet_sk(sk);
 653        struct sk_buff *skb;
 654        int copied, err;
 655
 656        pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
 657
 658        err = -EOPNOTSUPP;
 659        if (flags & MSG_OOB)
 660                goto out;
 661
 662        if (flags & MSG_ERRQUEUE)
 663                return ip_recv_error(sk, msg, len, addr_len);
 664
 665        skb = skb_recv_datagram(sk, flags, noblock, &err);
 666        if (!skb)
 667                goto out;
 668
 669        copied = skb->len;
 670        if (copied > len) {
 671                msg->msg_flags |= MSG_TRUNC;
 672                copied = len;
 673        }
 674
 675        /* Don't bother checking the checksum */
 676        err = skb_copy_datagram_msg(skb, 0, msg, copied);
 677        if (err)
 678                goto done;
 679
 680        sock_recv_timestamp(msg, sk, skb);
 681
 682        /* Copy the address. */
 683        if (msg->msg_name) {
 684                struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
 685
 686                sin->sin_family = AF_INET;
 687                sin->sin_port = 0 /* skb->h.uh->source */;
 688                sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
 689                memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
 690                *addr_len = sizeof(*sin);
 691        }
 692        if (isk->cmsg_flags)
 693                ip_cmsg_recv(msg, skb);
 694        err = copied;
 695
 696done:
 697        skb_free_datagram(sk, skb);
 698out:
 699        pr_debug("ping_recvmsg -> %d\n", err);
 700        return err;
 701}
 702
 703static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
 704{
 705        pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
 706                 inet_sk(sk), inet_sk(sk)->inet_num, skb);
 707        if (sock_queue_rcv_skb(sk, skb) < 0) {
 708                kfree_skb(skb);
 709                pr_debug("ping_queue_rcv_skb -> failed\n");
 710                return -1;
 711        }
 712        return 0;
 713}
 714
 715
 716/*
 717 *      All we need to do is get the socket.
 718 */
 719
 720void ping_rcv(struct sk_buff *skb)
 721{
 722        struct sock *sk;
 723        struct net *net = dev_net(skb->dev);
 724        struct iphdr *iph = ip_hdr(skb);
 725        struct icmphdr *icmph = icmp_hdr(skb);
 726        __be32 saddr = iph->saddr;
 727        __be32 daddr = iph->daddr;
 728
 729        /* We assume the packet has already been checked by icmp_rcv */
 730
 731        pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
 732                 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
 733
 734        /* Push ICMP header back */
 735        skb_push(skb, skb->data - (u8 *)icmph);
 736
 737        sk = ping_v4_lookup(net, saddr, daddr, ntohs(icmph->un.echo.id),
 738                            skb->dev->ifindex);
 739        if (sk != NULL) {
 740                pr_debug("rcv on socket %p\n", sk);
 741                ping_queue_rcv_skb(sk, skb_get(skb));
 742                sock_put(sk);
 743                return;
 744        }
 745        pr_debug("no socket, dropping\n");
 746
 747        /* We're called from icmp_rcv(). kfree_skb() is done there. */
 748}
 749
 750struct proto ping_prot = {
 751        .name =         "PING",
 752        .owner =        THIS_MODULE,
 753        .init =         ping_init_sock,
 754        .close =        ping_close,
 755        .connect =      ip4_datagram_connect,
 756        .disconnect =   udp_disconnect,
 757        .setsockopt =   ip_setsockopt,
 758        .getsockopt =   ip_getsockopt,
 759        .sendmsg =      ping_sendmsg,
 760        .recvmsg =      ping_recvmsg,
 761        .bind =         ping_bind,
 762        .backlog_rcv =  ping_queue_rcv_skb,
 763        .release_cb =   ip4_datagram_release_cb,
 764        .hash =         ping_v4_hash,
 765        .unhash =       ping_v4_unhash,
 766        .get_port =     ping_v4_get_port,
 767        .obj_size =     sizeof(struct inet_sock),
 768};
 769EXPORT_SYMBOL(ping_prot);
 770
 771#ifdef CONFIG_PROC_FS
 772
 773static struct sock *ping_get_first(struct seq_file *seq, int start)
 774{
 775        struct sock *sk;
 776        struct ping_iter_state *state = seq->private;
 777        struct net *net = seq_file_net(seq);
 778
 779        for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
 780             ++state->bucket) {
 781                struct hlist_nulls_node *node;
 782                struct hlist_nulls_head *hslot;
 783
 784                hslot = &ping_table.hash[state->bucket];
 785
 786                if (hlist_nulls_empty(hslot))
 787                        continue;
 788
 789                sk_nulls_for_each(sk, node, hslot) {
 790                        if (net_eq(sock_net(sk), net))
 791                                goto found;
 792                }
 793        }
 794        sk = NULL;
 795found:
 796        return sk;
 797}
 798
 799static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
 800{
 801        struct ping_iter_state *state = seq->private;
 802        struct net *net = seq_file_net(seq);
 803
 804        do {
 805                sk = sk_nulls_next(sk);
 806        } while (sk && (!net_eq(sock_net(sk), net)));
 807
 808        if (!sk)
 809                return ping_get_first(seq, state->bucket + 1);
 810        return sk;
 811}
 812
 813static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
 814{
 815        struct sock *sk = ping_get_first(seq, 0);
 816
 817        if (sk)
 818                while (pos && (sk = ping_get_next(seq, sk)) != NULL)
 819                        --pos;
 820        return pos ? NULL : sk;
 821}
 822
 823static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
 824{
 825        struct ping_iter_state *state = seq->private;
 826        state->bucket = 0;
 827
 828        read_lock_bh(&ping_table.lock);
 829
 830        return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
 831}
 832
 833static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
 834{
 835        struct sock *sk;
 836
 837        if (v == SEQ_START_TOKEN)
 838                sk = ping_get_idx(seq, 0);
 839        else
 840                sk = ping_get_next(seq, v);
 841
 842        ++*pos;
 843        return sk;
 844}
 845
 846static void ping_seq_stop(struct seq_file *seq, void *v)
 847{
 848        read_unlock_bh(&ping_table.lock);
 849}
 850
 851static void ping_format_sock(struct sock *sp, struct seq_file *f,
 852                int bucket, int *len)
 853{
 854        struct inet_sock *inet = inet_sk(sp);
 855        __be32 dest = inet->inet_daddr;
 856        __be32 src = inet->inet_rcv_saddr;
 857        __u16 destp = ntohs(inet->inet_dport);
 858        __u16 srcp = ntohs(inet->inet_sport);
 859
 860        seq_printf(f, "%5d: %08X:%04X %08X:%04X"
 861                " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d%n",
 862                bucket, src, srcp, dest, destp, sp->sk_state,
 863                sk_wmem_alloc_get(sp),
 864                sk_rmem_alloc_get(sp),
 865                0, 0L, 0,
 866                from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
 867                0, sock_i_ino(sp),
 868                atomic_read(&sp->sk_refcnt), sp,
 869                atomic_read(&sp->sk_drops), len);
 870}
 871
 872static int ping_seq_show(struct seq_file *seq, void *v)
 873{
 874        if (v == SEQ_START_TOKEN)
 875                seq_printf(seq, "%-127s\n",
 876                           "  sl  local_address rem_address   st tx_queue "
 877                           "rx_queue tr tm->when retrnsmt   uid  timeout "
 878                           "inode ref pointer drops");
 879        else {
 880                struct ping_iter_state *state = seq->private;
 881                int len;
 882
 883                ping_format_sock(v, seq, state->bucket, &len);
 884                seq_printf(seq, "%*s\n", 127 - len, "");
 885        }
 886        return 0;
 887}
 888
 889static const struct seq_operations ping_seq_ops = {
 890        .show           = ping_seq_show,
 891        .start          = ping_seq_start,
 892        .next           = ping_seq_next,
 893        .stop           = ping_seq_stop,
 894};
 895
 896static int ping_seq_open(struct inode *inode, struct file *file)
 897{
 898        return seq_open_net(inode, file, &ping_seq_ops,
 899                           sizeof(struct ping_iter_state));
 900}
 901
 902static const struct file_operations ping_seq_fops = {
 903        .open           = ping_seq_open,
 904        .read           = seq_read,
 905        .llseek         = seq_lseek,
 906        .release        = seq_release_net,
 907};
 908
 909static int ping_proc_register(struct net *net)
 910{
 911        struct proc_dir_entry *p;
 912        int rc = 0;
 913
 914        p = proc_create("icmp", S_IRUGO, net->proc_net, &ping_seq_fops);
 915        if (!p)
 916                rc = -ENOMEM;
 917        return rc;
 918}
 919
 920static void ping_proc_unregister(struct net *net)
 921{
 922        remove_proc_entry("icmp", net->proc_net);
 923}
 924
 925
 926static int __net_init ping_proc_init_net(struct net *net)
 927{
 928        return ping_proc_register(net);
 929}
 930
 931static void __net_exit ping_proc_exit_net(struct net *net)
 932{
 933        ping_proc_unregister(net);
 934}
 935
 936static struct pernet_operations ping_net_ops = {
 937        .init = ping_proc_init_net,
 938        .exit = ping_proc_exit_net,
 939};
 940
 941int __init ping_proc_init(void)
 942{
 943        return register_pernet_subsys(&ping_net_ops);
 944}
 945
 946void ping_proc_exit(void)
 947{
 948        unregister_pernet_subsys(&ping_net_ops);
 949}
 950
 951#endif
 952
 953void __init ping_init(void)
 954{
 955        int i;
 956
 957        for (i = 0; i < PING_HTABLE_SIZE; i++)
 958                INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
 959        rwlock_init(&ping_table.lock);
 960}
 961