Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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> |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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> |
Paul Gortmaker | bc3b2d7 | 2011-07-15 11:47:34 -0400 | [diff] [blame] | 42 | #include <linux/export.h> |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 43 | #include <net/sock.h> |
| 44 | #include <net/ping.h> |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 45 | #include <net/udp.h> |
| 46 | #include <net/route.h> |
| 47 | #include <net/inet_common.h> |
| 48 | #include <net/checksum.h> |
| 49 | |
| 50 | |
Eric Dumazet | 1b1cb1f | 2011-05-13 22:59:19 +0000 | [diff] [blame] | 51 | static struct ping_table ping_table; |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 52 | |
Eric Dumazet | 1b1cb1f | 2011-05-13 22:59:19 +0000 | [diff] [blame] | 53 | static u16 ping_port_rover; |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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)) { |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 140 | write_lock_bh(&ping_table.lock); |
| 141 | hlist_nulls_del(&sk->sk_nulls_node); |
| 142 | sock_put(sk); |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 143 | isk->inet_num = 0; |
| 144 | isk->inet_sport = 0; |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 145 | sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1); |
| 146 | write_unlock_bh(&ping_table.lock); |
| 147 | } |
| 148 | } |
| 149 | |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 150 | static struct sock *ping_v4_lookup(struct net *net, __be32 saddr, __be32 daddr, |
Eric Dumazet | 1b1cb1f | 2011-05-13 22:59:19 +0000 | [diff] [blame] | 151 | u16 ident, int dif) |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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 | |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 158 | pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n", |
| 159 | (int)ident, &daddr, dif); |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 160 | read_lock_bh(&ping_table.lock); |
| 161 | |
| 162 | ping_portaddr_for_each_entry(sk, hnode, hslot) { |
| 163 | isk = inet_sk(sk); |
| 164 | |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 165 | pr_debug("found: %p: num = %d, daddr = %pI4, dif = %d\n", sk, |
| 166 | (int)isk->inet_num, &isk->inet_rcv_saddr, |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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 | |
Changli Gao | 75e308c | 2011-05-18 21:16:01 +0000 | [diff] [blame] | 188 | static void inet_get_ping_group_range_net(struct net *net, gid_t *low, |
| 189 | gid_t *high) |
Vasiliy Kulikov | f56e03e | 2011-05-17 00:16:56 +0000 | [diff] [blame] | 190 | { |
| 191 | gid_t *data = net->ipv4.sysctl_ping_group_range; |
| 192 | unsigned seq; |
| 193 | do { |
| 194 | seq = read_seqbegin(&sysctl_local_ports.lock); |
| 195 | |
| 196 | *low = data[0]; |
| 197 | *high = data[1]; |
| 198 | } while (read_seqretry(&sysctl_local_ports.lock, seq)); |
| 199 | } |
| 200 | |
| 201 | |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 202 | static int ping_init_sock(struct sock *sk) |
| 203 | { |
| 204 | struct net *net = sock_net(sk); |
| 205 | gid_t group = current_egid(); |
| 206 | gid_t range[2]; |
| 207 | struct group_info *group_info = get_current_groups(); |
| 208 | int i, j, count = group_info->ngroups; |
| 209 | |
| 210 | inet_get_ping_group_range_net(net, range, range+1); |
| 211 | if (range[0] <= group && group <= range[1]) |
| 212 | return 0; |
| 213 | |
| 214 | for (i = 0; i < group_info->nblocks; i++) { |
| 215 | int cp_count = min_t(int, NGROUPS_PER_BLOCK, count); |
| 216 | |
| 217 | for (j = 0; j < cp_count; j++) { |
| 218 | group = group_info->blocks[i][j]; |
| 219 | if (range[0] <= group && group <= range[1]) |
| 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); |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 258 | if (addr->sin_addr.s_addr == htonl(INADDR_ANY)) |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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 | |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 282 | pr_debug("after bind(): num = %d, daddr = %pI4, dif = %d\n", |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 283 | (int)isk->inet_num, |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 284 | &isk->inet_rcv_saddr, |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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 = icmph->type; |
| 326 | int code = icmph->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) { |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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; |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 411 | __wsum wcheck; |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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 | |
Changli Gao | 75e308c | 2011-05-18 21:16:01 +0000 | [diff] [blame] | 438 | static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh, |
| 439 | struct flowi4 *fl4) |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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 | |
Changli Gao | bb0cd2f | 2011-05-18 21:16:00 +0000 | [diff] [blame] | 451 | static int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg, |
| 452 | size_t len) |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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; |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 463 | __be32 saddr, daddr, faddr; |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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; |
Erich E. Hoover | 76e2105 | 2012-02-08 09:11:07 +0000 | [diff] [blame] | 559 | } else if (!ipc.oif) |
| 560 | ipc.oif = inet->uc_index; |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 561 | |
| 562 | flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos, |
| 563 | RT_SCOPE_UNIVERSE, sk->sk_protocol, |
| 564 | inet_sk_flowi_flags(sk), faddr, saddr, 0, 0); |
| 565 | |
| 566 | security_sk_classify_flow(sk, flowi4_to_flowi(&fl4)); |
| 567 | rt = ip_route_output_flow(net, &fl4, sk); |
| 568 | if (IS_ERR(rt)) { |
| 569 | err = PTR_ERR(rt); |
| 570 | rt = NULL; |
| 571 | if (err == -ENETUNREACH) |
| 572 | IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES); |
| 573 | goto out; |
| 574 | } |
| 575 | |
| 576 | err = -EACCES; |
| 577 | if ((rt->rt_flags & RTCF_BROADCAST) && |
| 578 | !sock_flag(sk, SOCK_BROADCAST)) |
| 579 | goto out; |
| 580 | |
| 581 | if (msg->msg_flags & MSG_CONFIRM) |
| 582 | goto do_confirm; |
| 583 | back_from_confirm: |
| 584 | |
| 585 | if (!ipc.addr) |
| 586 | ipc.addr = fl4.daddr; |
| 587 | |
| 588 | lock_sock(sk); |
| 589 | |
| 590 | pfh.icmph.type = user_icmph.type; /* already checked */ |
| 591 | pfh.icmph.code = user_icmph.code; /* ditto */ |
| 592 | pfh.icmph.checksum = 0; |
| 593 | pfh.icmph.un.echo.id = inet->inet_sport; |
| 594 | pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence; |
| 595 | pfh.iov = msg->msg_iov; |
| 596 | pfh.wcheck = 0; |
| 597 | |
| 598 | err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len, |
| 599 | 0, &ipc, &rt, msg->msg_flags); |
| 600 | if (err) |
| 601 | ip_flush_pending_frames(sk); |
| 602 | else |
| 603 | err = ping_push_pending_frames(sk, &pfh, &fl4); |
| 604 | release_sock(sk); |
| 605 | |
| 606 | out: |
| 607 | ip_rt_put(rt); |
| 608 | if (free) |
| 609 | kfree(ipc.opt); |
| 610 | if (!err) { |
| 611 | icmp_out_count(sock_net(sk), user_icmph.type); |
| 612 | return len; |
| 613 | } |
| 614 | return err; |
| 615 | |
| 616 | do_confirm: |
| 617 | dst_confirm(&rt->dst); |
| 618 | if (!(msg->msg_flags & MSG_PROBE) || len) |
| 619 | goto back_from_confirm; |
| 620 | err = 0; |
| 621 | goto out; |
| 622 | } |
| 623 | |
Changli Gao | bb0cd2f | 2011-05-18 21:16:00 +0000 | [diff] [blame] | 624 | static int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg, |
| 625 | size_t len, int noblock, int flags, int *addr_len) |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 626 | { |
| 627 | struct inet_sock *isk = inet_sk(sk); |
| 628 | struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name; |
| 629 | struct sk_buff *skb; |
| 630 | int copied, err; |
| 631 | |
| 632 | pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num); |
| 633 | |
| 634 | if (flags & MSG_OOB) |
| 635 | goto out; |
| 636 | |
| 637 | if (addr_len) |
| 638 | *addr_len = sizeof(*sin); |
| 639 | |
| 640 | if (flags & MSG_ERRQUEUE) |
| 641 | return ip_recv_error(sk, msg, len); |
| 642 | |
| 643 | skb = skb_recv_datagram(sk, flags, noblock, &err); |
| 644 | if (!skb) |
| 645 | goto out; |
| 646 | |
| 647 | copied = skb->len; |
| 648 | if (copied > len) { |
| 649 | msg->msg_flags |= MSG_TRUNC; |
| 650 | copied = len; |
| 651 | } |
| 652 | |
| 653 | /* Don't bother checking the checksum */ |
| 654 | err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied); |
| 655 | if (err) |
| 656 | goto done; |
| 657 | |
| 658 | sock_recv_timestamp(msg, sk, skb); |
| 659 | |
| 660 | /* Copy the address. */ |
| 661 | if (sin) { |
| 662 | sin->sin_family = AF_INET; |
| 663 | sin->sin_port = 0 /* skb->h.uh->source */; |
| 664 | sin->sin_addr.s_addr = ip_hdr(skb)->saddr; |
| 665 | memset(sin->sin_zero, 0, sizeof(sin->sin_zero)); |
| 666 | } |
| 667 | if (isk->cmsg_flags) |
| 668 | ip_cmsg_recv(msg, skb); |
| 669 | err = copied; |
| 670 | |
| 671 | done: |
| 672 | skb_free_datagram(sk, skb); |
| 673 | out: |
| 674 | pr_debug("ping_recvmsg -> %d\n", err); |
| 675 | return err; |
| 676 | } |
| 677 | |
| 678 | static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb) |
| 679 | { |
| 680 | pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n", |
| 681 | inet_sk(sk), inet_sk(sk)->inet_num, skb); |
| 682 | if (sock_queue_rcv_skb(sk, skb) < 0) { |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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); |
Eric Dumazet | 747465e | 2012-01-16 19:27:39 +0000 | [diff] [blame] | 701 | __be32 saddr = iph->saddr; |
| 702 | __be32 daddr = iph->daddr; |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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, |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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; |
Changli Gao | 75e308c | 2011-05-18 21:16:01 +0000 | [diff] [blame] | 756 | struct hlist_nulls_head *hslot; |
| 757 | |
| 758 | hslot = &ping_table.hash[state->bucket]; |
Vasiliy Kulikov | c319b4d | 2011-05-13 10:01:00 +0000 | [diff] [blame] | 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 | } |