blob: aea5a199c37a341be3fab08500f1c35a34cbf215 [file] [log] [blame]
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +00001/*
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 Kulikovc319b4d2011-05-13 10:01:00 +000025#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 Gortmakerbc3b2d72011-07-15 11:47:34 -040042#include <linux/export.h>
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +000043#include <net/sock.h>
44#include <net/ping.h>
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +000045#include <net/udp.h>
46#include <net/route.h>
47#include <net/inet_common.h>
48#include <net/checksum.h>
49
50
Eric Dumazet1b1cb1f2011-05-13 22:59:19 +000051static struct ping_table ping_table;
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +000052
Eric Dumazet1b1cb1f2011-05-13 22:59:19 +000053static u16 ping_port_rover;
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +000054
55static 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
62static 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
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 pr_debug("ping_v4_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
139 if (sk_hashed(sk)) {
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000140 write_lock_bh(&ping_table.lock);
141 hlist_nulls_del(&sk->sk_nulls_node);
142 sock_put(sk);
Eric Dumazet747465e2012-01-16 19:27:39 +0000143 isk->inet_num = 0;
144 isk->inet_sport = 0;
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000145 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
146 write_unlock_bh(&ping_table.lock);
147 }
148}
149
Eric Dumazet747465e2012-01-16 19:27:39 +0000150static struct sock *ping_v4_lookup(struct net *net, __be32 saddr, __be32 daddr,
Eric Dumazet1b1cb1f2011-05-13 22:59:19 +0000151 u16 ident, int dif)
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000152{
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 Dumazet747465e2012-01-16 19:27:39 +0000158 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
159 (int)ident, &daddr, dif);
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000160 read_lock_bh(&ping_table.lock);
161
162 ping_portaddr_for_each_entry(sk, hnode, hslot) {
163 isk = inet_sk(sk);
164
Eric Dumazet747465e2012-01-16 19:27:39 +0000165 pr_debug("found: %p: num = %d, daddr = %pI4, dif = %d\n", sk,
166 (int)isk->inet_num, &isk->inet_rcv_saddr,
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000167 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;
182exit:
183 read_unlock_bh(&ping_table.lock);
184
185 return sk;
186}
187
Changli Gao75e308c2011-05-18 21:16:01 +0000188static void inet_get_ping_group_range_net(struct net *net, gid_t *low,
189 gid_t *high)
Vasiliy Kulikovf56e03e2011-05-17 00:16:56 +0000190{
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 Kulikovc319b4d2011-05-13 10:01:00 +0000202static 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
229static 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
243static 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 Dumazet747465e2012-01-16 19:27:39 +0000258 if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000259 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 Dumazet747465e2012-01-16 19:27:39 +0000282 pr_debug("after bind(): num = %d, daddr = %pI4, dif = %d\n",
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000283 (int)isk->inet_num,
Eric Dumazet747465e2012-01-16 19:27:39 +0000284 &isk->inet_rcv_saddr,
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000285 (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);
296out:
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
306static 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
318static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
319
320void 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 Kulikovc319b4d2011-05-13 10:01:00 +0000343 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);
400out:
401 sock_put(sk);
402}
403
404/*
405 * Copy and checksum an ICMP Echo packet from user space into a buffer.
406 */
407
408struct pingfakehdr {
409 struct icmphdr icmph;
410 struct iovec *iov;
Eric Dumazet747465e2012-01-16 19:27:39 +0000411 __wsum wcheck;
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000412};
413
414static 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 Gao75e308c2011-05-18 21:16:01 +0000438static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
439 struct flowi4 *fl4)
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000440{
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 Gaobb0cd2f2011-05-18 21:16:00 +0000451static int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
452 size_t len)
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000453{
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 Dumazet747465e2012-01-16 19:27:39 +0000463 __be32 saddr, daddr, faddr;
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000464 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;
582back_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
605out:
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
615do_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
Changli Gaobb0cd2f2011-05-18 21:16:00 +0000623static int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
624 size_t len, int noblock, int flags, int *addr_len)
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000625{
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
670done:
671 skb_free_datagram(sk, skb);
672out:
673 pr_debug("ping_recvmsg -> %d\n", err);
674 return err;
675}
676
677static 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) {
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000682 kfree_skb(skb);
683 pr_debug("ping_queue_rcv_skb -> failed\n");
684 return -1;
685 }
686 return 0;
687}
688
689
690/*
691 * All we need to do is get the socket.
692 */
693
694void ping_rcv(struct sk_buff *skb)
695{
696 struct sock *sk;
697 struct net *net = dev_net(skb->dev);
698 struct iphdr *iph = ip_hdr(skb);
699 struct icmphdr *icmph = icmp_hdr(skb);
Eric Dumazet747465e2012-01-16 19:27:39 +0000700 __be32 saddr = iph->saddr;
701 __be32 daddr = iph->daddr;
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000702
703 /* We assume the packet has already been checked by icmp_rcv */
704
705 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
706 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
707
708 /* Push ICMP header back */
709 skb_push(skb, skb->data - (u8 *)icmph);
710
711 sk = ping_v4_lookup(net, saddr, daddr, ntohs(icmph->un.echo.id),
712 skb->dev->ifindex);
713 if (sk != NULL) {
714 pr_debug("rcv on socket %p\n", sk);
715 ping_queue_rcv_skb(sk, skb_get(skb));
716 sock_put(sk);
717 return;
718 }
719 pr_debug("no socket, dropping\n");
720
721 /* We're called from icmp_rcv(). kfree_skb() is done there. */
722}
723
724struct proto ping_prot = {
725 .name = "PING",
726 .owner = THIS_MODULE,
727 .init = ping_init_sock,
728 .close = ping_close,
729 .connect = ip4_datagram_connect,
730 .disconnect = udp_disconnect,
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000731 .setsockopt = ip_setsockopt,
732 .getsockopt = ip_getsockopt,
733 .sendmsg = ping_sendmsg,
734 .recvmsg = ping_recvmsg,
735 .bind = ping_bind,
736 .backlog_rcv = ping_queue_rcv_skb,
737 .hash = ping_v4_hash,
738 .unhash = ping_v4_unhash,
739 .get_port = ping_v4_get_port,
740 .obj_size = sizeof(struct inet_sock),
741};
742EXPORT_SYMBOL(ping_prot);
743
744#ifdef CONFIG_PROC_FS
745
746static struct sock *ping_get_first(struct seq_file *seq, int start)
747{
748 struct sock *sk;
749 struct ping_iter_state *state = seq->private;
750 struct net *net = seq_file_net(seq);
751
752 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
753 ++state->bucket) {
754 struct hlist_nulls_node *node;
Changli Gao75e308c2011-05-18 21:16:01 +0000755 struct hlist_nulls_head *hslot;
756
757 hslot = &ping_table.hash[state->bucket];
Vasiliy Kulikovc319b4d2011-05-13 10:01:00 +0000758
759 if (hlist_nulls_empty(hslot))
760 continue;
761
762 sk_nulls_for_each(sk, node, hslot) {
763 if (net_eq(sock_net(sk), net))
764 goto found;
765 }
766 }
767 sk = NULL;
768found:
769 return sk;
770}
771
772static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
773{
774 struct ping_iter_state *state = seq->private;
775 struct net *net = seq_file_net(seq);
776
777 do {
778 sk = sk_nulls_next(sk);
779 } while (sk && (!net_eq(sock_net(sk), net)));
780
781 if (!sk)
782 return ping_get_first(seq, state->bucket + 1);
783 return sk;
784}
785
786static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
787{
788 struct sock *sk = ping_get_first(seq, 0);
789
790 if (sk)
791 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
792 --pos;
793 return pos ? NULL : sk;
794}
795
796static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
797{
798 struct ping_iter_state *state = seq->private;
799 state->bucket = 0;
800
801 read_lock_bh(&ping_table.lock);
802
803 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
804}
805
806static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
807{
808 struct sock *sk;
809
810 if (v == SEQ_START_TOKEN)
811 sk = ping_get_idx(seq, 0);
812 else
813 sk = ping_get_next(seq, v);
814
815 ++*pos;
816 return sk;
817}
818
819static void ping_seq_stop(struct seq_file *seq, void *v)
820{
821 read_unlock_bh(&ping_table.lock);
822}
823
824static void ping_format_sock(struct sock *sp, struct seq_file *f,
825 int bucket, int *len)
826{
827 struct inet_sock *inet = inet_sk(sp);
828 __be32 dest = inet->inet_daddr;
829 __be32 src = inet->inet_rcv_saddr;
830 __u16 destp = ntohs(inet->inet_dport);
831 __u16 srcp = ntohs(inet->inet_sport);
832
833 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
834 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
835 bucket, src, srcp, dest, destp, sp->sk_state,
836 sk_wmem_alloc_get(sp),
837 sk_rmem_alloc_get(sp),
838 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
839 atomic_read(&sp->sk_refcnt), sp,
840 atomic_read(&sp->sk_drops), len);
841}
842
843static int ping_seq_show(struct seq_file *seq, void *v)
844{
845 if (v == SEQ_START_TOKEN)
846 seq_printf(seq, "%-127s\n",
847 " sl local_address rem_address st tx_queue "
848 "rx_queue tr tm->when retrnsmt uid timeout "
849 "inode ref pointer drops");
850 else {
851 struct ping_iter_state *state = seq->private;
852 int len;
853
854 ping_format_sock(v, seq, state->bucket, &len);
855 seq_printf(seq, "%*s\n", 127 - len, "");
856 }
857 return 0;
858}
859
860static const struct seq_operations ping_seq_ops = {
861 .show = ping_seq_show,
862 .start = ping_seq_start,
863 .next = ping_seq_next,
864 .stop = ping_seq_stop,
865};
866
867static int ping_seq_open(struct inode *inode, struct file *file)
868{
869 return seq_open_net(inode, file, &ping_seq_ops,
870 sizeof(struct ping_iter_state));
871}
872
873static const struct file_operations ping_seq_fops = {
874 .open = ping_seq_open,
875 .read = seq_read,
876 .llseek = seq_lseek,
877 .release = seq_release_net,
878};
879
880static int ping_proc_register(struct net *net)
881{
882 struct proc_dir_entry *p;
883 int rc = 0;
884
885 p = proc_net_fops_create(net, "icmp", S_IRUGO, &ping_seq_fops);
886 if (!p)
887 rc = -ENOMEM;
888 return rc;
889}
890
891static void ping_proc_unregister(struct net *net)
892{
893 proc_net_remove(net, "icmp");
894}
895
896
897static int __net_init ping_proc_init_net(struct net *net)
898{
899 return ping_proc_register(net);
900}
901
902static void __net_exit ping_proc_exit_net(struct net *net)
903{
904 ping_proc_unregister(net);
905}
906
907static struct pernet_operations ping_net_ops = {
908 .init = ping_proc_init_net,
909 .exit = ping_proc_exit_net,
910};
911
912int __init ping_proc_init(void)
913{
914 return register_pernet_subsys(&ping_net_ops);
915}
916
917void ping_proc_exit(void)
918{
919 unregister_pernet_subsys(&ping_net_ops);
920}
921
922#endif
923
924void __init ping_init(void)
925{
926 int i;
927
928 for (i = 0; i < PING_HTABLE_SIZE; i++)
929 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
930 rwlock_init(&ping_table.lock);
931}