net: kill duplicate included header
[linux-2.6-block.git] / net / ipv4 / ping.c
CommitLineData
c319b4d7
VK
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>
c319b4d7
VK
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>
bc3b2d7f 42#include <linux/export.h>
c319b4d7
VK
43#include <net/sock.h>
44#include <net/ping.h>
c319b4d7
VK
45#include <net/udp.h>
46#include <net/route.h>
47#include <net/inet_common.h>
48#include <net/checksum.h>
49
50
1b1cb1f7 51static struct ping_table ping_table;
c319b4d7 52
1b1cb1f7 53static u16 ping_port_rover;
c319b4d7
VK
54
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)) {
c319b4d7
VK
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
1b1cb1f7
ED
149static struct sock *ping_v4_lookup(struct net *net, u32 saddr, u32 daddr,
150 u16 ident, int dif)
c319b4d7
VK
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;
181exit:
182 read_unlock_bh(&ping_table.lock);
183
184 return sk;
185}
186
75e308c8
CG
187static void inet_get_ping_group_range_net(struct net *net, gid_t *low,
188 gid_t *high)
f56e03e8
VK
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
c319b4d7
VK
201static 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
228static 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
242static 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);
295out:
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
305static 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
317static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
318
319void 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) {
c319b4d7
VK
342 pr_debug("no socket, dropping\n");
343 return; /* No socket for error */
344 }
345 pr_debug("err on socket %p\n", sk);
346
347 err = 0;
348 harderr = 0;
349 inet_sock = inet_sk(sk);
350
351 switch (type) {
352 default:
353 case ICMP_TIME_EXCEEDED:
354 err = EHOSTUNREACH;
355 break;
356 case ICMP_SOURCE_QUENCH:
357 /* This is not a real error but ping wants to see it.
358 * Report it with some fake errno. */
359 err = EREMOTEIO;
360 break;
361 case ICMP_PARAMETERPROB:
362 err = EPROTO;
363 harderr = 1;
364 break;
365 case ICMP_DEST_UNREACH:
366 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
367 if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
368 err = EMSGSIZE;
369 harderr = 1;
370 break;
371 }
372 goto out;
373 }
374 err = EHOSTUNREACH;
375 if (code <= NR_ICMP_UNREACH) {
376 harderr = icmp_err_convert[code].fatal;
377 err = icmp_err_convert[code].errno;
378 }
379 break;
380 case ICMP_REDIRECT:
381 /* See ICMP_SOURCE_QUENCH */
382 err = EREMOTEIO;
383 break;
384 }
385
386 /*
387 * RFC1122: OK. Passes ICMP errors back to application, as per
388 * 4.1.3.3.
389 */
390 if (!inet_sock->recverr) {
391 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
392 goto out;
393 } else {
394 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
395 info, (u8 *)icmph);
396 }
397 sk->sk_err = err;
398 sk->sk_error_report(sk);
399out:
400 sock_put(sk);
401}
402
403/*
404 * Copy and checksum an ICMP Echo packet from user space into a buffer.
405 */
406
407struct pingfakehdr {
408 struct icmphdr icmph;
409 struct iovec *iov;
410 u32 wcheck;
411};
412
413static int ping_getfrag(void *from, char * to,
414 int offset, int fraglen, int odd, struct sk_buff *skb)
415{
416 struct pingfakehdr *pfh = (struct pingfakehdr *)from;
417
418 if (offset == 0) {
419 if (fraglen < sizeof(struct icmphdr))
420 BUG();
421 if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
422 pfh->iov, 0, fraglen - sizeof(struct icmphdr),
423 &pfh->wcheck))
424 return -EFAULT;
425
426 return 0;
427 }
428 if (offset < sizeof(struct icmphdr))
429 BUG();
430 if (csum_partial_copy_fromiovecend
431 (to, pfh->iov, offset - sizeof(struct icmphdr),
432 fraglen, &pfh->wcheck))
433 return -EFAULT;
434 return 0;
435}
436
75e308c8
CG
437static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
438 struct flowi4 *fl4)
c319b4d7
VK
439{
440 struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
441
442 pfh->wcheck = csum_partial((char *)&pfh->icmph,
443 sizeof(struct icmphdr), pfh->wcheck);
444 pfh->icmph.checksum = csum_fold(pfh->wcheck);
445 memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
446 skb->ip_summed = CHECKSUM_NONE;
447 return ip_push_pending_frames(sk, fl4);
448}
449
bb0cd2fb
CG
450static int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
451 size_t len)
c319b4d7
VK
452{
453 struct net *net = sock_net(sk);
454 struct flowi4 fl4;
455 struct inet_sock *inet = inet_sk(sk);
456 struct ipcm_cookie ipc;
457 struct icmphdr user_icmph;
458 struct pingfakehdr pfh;
459 struct rtable *rt = NULL;
460 struct ip_options_data opt_copy;
461 int free = 0;
462 u32 saddr, daddr, faddr;
463 u8 tos;
464 int err;
465
466 pr_debug("ping_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
467
468
469 if (len > 0xFFFF)
470 return -EMSGSIZE;
471
472 /*
473 * Check the flags.
474 */
475
476 /* Mirror BSD error message compatibility */
477 if (msg->msg_flags & MSG_OOB)
478 return -EOPNOTSUPP;
479
480 /*
481 * Fetch the ICMP header provided by the userland.
482 * iovec is modified!
483 */
484
485 if (memcpy_fromiovec((u8 *)&user_icmph, msg->msg_iov,
486 sizeof(struct icmphdr)))
487 return -EFAULT;
488 if (!ping_supported(user_icmph.type, user_icmph.code))
489 return -EINVAL;
490
491 /*
492 * Get and verify the address.
493 */
494
495 if (msg->msg_name) {
496 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
497 if (msg->msg_namelen < sizeof(*usin))
498 return -EINVAL;
499 if (usin->sin_family != AF_INET)
500 return -EINVAL;
501 daddr = usin->sin_addr.s_addr;
502 /* no remote port */
503 } else {
504 if (sk->sk_state != TCP_ESTABLISHED)
505 return -EDESTADDRREQ;
506 daddr = inet->inet_daddr;
507 /* no remote port */
508 }
509
510 ipc.addr = inet->inet_saddr;
511 ipc.opt = NULL;
512 ipc.oif = sk->sk_bound_dev_if;
513 ipc.tx_flags = 0;
514 err = sock_tx_timestamp(sk, &ipc.tx_flags);
515 if (err)
516 return err;
517
518 if (msg->msg_controllen) {
519 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
520 if (err)
521 return err;
522 if (ipc.opt)
523 free = 1;
524 }
525 if (!ipc.opt) {
526 struct ip_options_rcu *inet_opt;
527
528 rcu_read_lock();
529 inet_opt = rcu_dereference(inet->inet_opt);
530 if (inet_opt) {
531 memcpy(&opt_copy, inet_opt,
532 sizeof(*inet_opt) + inet_opt->opt.optlen);
533 ipc.opt = &opt_copy.opt;
534 }
535 rcu_read_unlock();
536 }
537
538 saddr = ipc.addr;
539 ipc.addr = faddr = daddr;
540
541 if (ipc.opt && ipc.opt->opt.srr) {
542 if (!daddr)
543 return -EINVAL;
544 faddr = ipc.opt->opt.faddr;
545 }
546 tos = RT_TOS(inet->tos);
547 if (sock_flag(sk, SOCK_LOCALROUTE) ||
548 (msg->msg_flags & MSG_DONTROUTE) ||
549 (ipc.opt && ipc.opt->opt.is_strictroute)) {
550 tos |= RTO_ONLINK;
551 }
552
553 if (ipv4_is_multicast(daddr)) {
554 if (!ipc.oif)
555 ipc.oif = inet->mc_index;
556 if (!saddr)
557 saddr = inet->mc_addr;
558 }
559
560 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
561 RT_SCOPE_UNIVERSE, sk->sk_protocol,
562 inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
563
564 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
565 rt = ip_route_output_flow(net, &fl4, sk);
566 if (IS_ERR(rt)) {
567 err = PTR_ERR(rt);
568 rt = NULL;
569 if (err == -ENETUNREACH)
570 IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
571 goto out;
572 }
573
574 err = -EACCES;
575 if ((rt->rt_flags & RTCF_BROADCAST) &&
576 !sock_flag(sk, SOCK_BROADCAST))
577 goto out;
578
579 if (msg->msg_flags & MSG_CONFIRM)
580 goto do_confirm;
581back_from_confirm:
582
583 if (!ipc.addr)
584 ipc.addr = fl4.daddr;
585
586 lock_sock(sk);
587
588 pfh.icmph.type = user_icmph.type; /* already checked */
589 pfh.icmph.code = user_icmph.code; /* ditto */
590 pfh.icmph.checksum = 0;
591 pfh.icmph.un.echo.id = inet->inet_sport;
592 pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
593 pfh.iov = msg->msg_iov;
594 pfh.wcheck = 0;
595
596 err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
597 0, &ipc, &rt, msg->msg_flags);
598 if (err)
599 ip_flush_pending_frames(sk);
600 else
601 err = ping_push_pending_frames(sk, &pfh, &fl4);
602 release_sock(sk);
603
604out:
605 ip_rt_put(rt);
606 if (free)
607 kfree(ipc.opt);
608 if (!err) {
609 icmp_out_count(sock_net(sk), user_icmph.type);
610 return len;
611 }
612 return err;
613
614do_confirm:
615 dst_confirm(&rt->dst);
616 if (!(msg->msg_flags & MSG_PROBE) || len)
617 goto back_from_confirm;
618 err = 0;
619 goto out;
620}
621
bb0cd2fb
CG
622static int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
623 size_t len, int noblock, int flags, int *addr_len)
c319b4d7
VK
624{
625 struct inet_sock *isk = inet_sk(sk);
626 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
627 struct sk_buff *skb;
628 int copied, err;
629
630 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
631
632 if (flags & MSG_OOB)
633 goto out;
634
635 if (addr_len)
636 *addr_len = sizeof(*sin);
637
638 if (flags & MSG_ERRQUEUE)
639 return ip_recv_error(sk, msg, len);
640
641 skb = skb_recv_datagram(sk, flags, noblock, &err);
642 if (!skb)
643 goto out;
644
645 copied = skb->len;
646 if (copied > len) {
647 msg->msg_flags |= MSG_TRUNC;
648 copied = len;
649 }
650
651 /* Don't bother checking the checksum */
652 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
653 if (err)
654 goto done;
655
656 sock_recv_timestamp(msg, sk, skb);
657
658 /* Copy the address. */
659 if (sin) {
660 sin->sin_family = AF_INET;
661 sin->sin_port = 0 /* skb->h.uh->source */;
662 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
663 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
664 }
665 if (isk->cmsg_flags)
666 ip_cmsg_recv(msg, skb);
667 err = copied;
668
669done:
670 skb_free_datagram(sk, skb);
671out:
672 pr_debug("ping_recvmsg -> %d\n", err);
673 return err;
674}
675
676static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
677{
678 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
679 inet_sk(sk), inet_sk(sk)->inet_num, skb);
680 if (sock_queue_rcv_skb(sk, skb) < 0) {
c319b4d7
VK
681 kfree_skb(skb);
682 pr_debug("ping_queue_rcv_skb -> failed\n");
683 return -1;
684 }
685 return 0;
686}
687
688
689/*
690 * All we need to do is get the socket.
691 */
692
693void ping_rcv(struct sk_buff *skb)
694{
695 struct sock *sk;
696 struct net *net = dev_net(skb->dev);
697 struct iphdr *iph = ip_hdr(skb);
698 struct icmphdr *icmph = icmp_hdr(skb);
699 u32 saddr = iph->saddr;
700 u32 daddr = iph->daddr;
701
702 /* We assume the packet has already been checked by icmp_rcv */
703
704 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
705 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
706
707 /* Push ICMP header back */
708 skb_push(skb, skb->data - (u8 *)icmph);
709
710 sk = ping_v4_lookup(net, saddr, daddr, ntohs(icmph->un.echo.id),
711 skb->dev->ifindex);
712 if (sk != NULL) {
713 pr_debug("rcv on socket %p\n", sk);
714 ping_queue_rcv_skb(sk, skb_get(skb));
715 sock_put(sk);
716 return;
717 }
718 pr_debug("no socket, dropping\n");
719
720 /* We're called from icmp_rcv(). kfree_skb() is done there. */
721}
722
723struct proto ping_prot = {
724 .name = "PING",
725 .owner = THIS_MODULE,
726 .init = ping_init_sock,
727 .close = ping_close,
728 .connect = ip4_datagram_connect,
729 .disconnect = udp_disconnect,
c319b4d7
VK
730 .setsockopt = ip_setsockopt,
731 .getsockopt = ip_getsockopt,
732 .sendmsg = ping_sendmsg,
733 .recvmsg = ping_recvmsg,
734 .bind = ping_bind,
735 .backlog_rcv = ping_queue_rcv_skb,
736 .hash = ping_v4_hash,
737 .unhash = ping_v4_unhash,
738 .get_port = ping_v4_get_port,
739 .obj_size = sizeof(struct inet_sock),
740};
741EXPORT_SYMBOL(ping_prot);
742
743#ifdef CONFIG_PROC_FS
744
745static struct sock *ping_get_first(struct seq_file *seq, int start)
746{
747 struct sock *sk;
748 struct ping_iter_state *state = seq->private;
749 struct net *net = seq_file_net(seq);
750
751 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
752 ++state->bucket) {
753 struct hlist_nulls_node *node;
75e308c8
CG
754 struct hlist_nulls_head *hslot;
755
756 hslot = &ping_table.hash[state->bucket];
c319b4d7
VK
757
758 if (hlist_nulls_empty(hslot))
759 continue;
760
761 sk_nulls_for_each(sk, node, hslot) {
762 if (net_eq(sock_net(sk), net))
763 goto found;
764 }
765 }
766 sk = NULL;
767found:
768 return sk;
769}
770
771static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
772{
773 struct ping_iter_state *state = seq->private;
774 struct net *net = seq_file_net(seq);
775
776 do {
777 sk = sk_nulls_next(sk);
778 } while (sk && (!net_eq(sock_net(sk), net)));
779
780 if (!sk)
781 return ping_get_first(seq, state->bucket + 1);
782 return sk;
783}
784
785static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
786{
787 struct sock *sk = ping_get_first(seq, 0);
788
789 if (sk)
790 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
791 --pos;
792 return pos ? NULL : sk;
793}
794
795static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
796{
797 struct ping_iter_state *state = seq->private;
798 state->bucket = 0;
799
800 read_lock_bh(&ping_table.lock);
801
802 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
803}
804
805static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
806{
807 struct sock *sk;
808
809 if (v == SEQ_START_TOKEN)
810 sk = ping_get_idx(seq, 0);
811 else
812 sk = ping_get_next(seq, v);
813
814 ++*pos;
815 return sk;
816}
817
818static void ping_seq_stop(struct seq_file *seq, void *v)
819{
820 read_unlock_bh(&ping_table.lock);
821}
822
823static void ping_format_sock(struct sock *sp, struct seq_file *f,
824 int bucket, int *len)
825{
826 struct inet_sock *inet = inet_sk(sp);
827 __be32 dest = inet->inet_daddr;
828 __be32 src = inet->inet_rcv_saddr;
829 __u16 destp = ntohs(inet->inet_dport);
830 __u16 srcp = ntohs(inet->inet_sport);
831
832 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
833 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
834 bucket, src, srcp, dest, destp, sp->sk_state,
835 sk_wmem_alloc_get(sp),
836 sk_rmem_alloc_get(sp),
837 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
838 atomic_read(&sp->sk_refcnt), sp,
839 atomic_read(&sp->sk_drops), len);
840}
841
842static int ping_seq_show(struct seq_file *seq, void *v)
843{
844 if (v == SEQ_START_TOKEN)
845 seq_printf(seq, "%-127s\n",
846 " sl local_address rem_address st tx_queue "
847 "rx_queue tr tm->when retrnsmt uid timeout "
848 "inode ref pointer drops");
849 else {
850 struct ping_iter_state *state = seq->private;
851 int len;
852
853 ping_format_sock(v, seq, state->bucket, &len);
854 seq_printf(seq, "%*s\n", 127 - len, "");
855 }
856 return 0;
857}
858
859static const struct seq_operations ping_seq_ops = {
860 .show = ping_seq_show,
861 .start = ping_seq_start,
862 .next = ping_seq_next,
863 .stop = ping_seq_stop,
864};
865
866static int ping_seq_open(struct inode *inode, struct file *file)
867{
868 return seq_open_net(inode, file, &ping_seq_ops,
869 sizeof(struct ping_iter_state));
870}
871
872static const struct file_operations ping_seq_fops = {
873 .open = ping_seq_open,
874 .read = seq_read,
875 .llseek = seq_lseek,
876 .release = seq_release_net,
877};
878
879static int ping_proc_register(struct net *net)
880{
881 struct proc_dir_entry *p;
882 int rc = 0;
883
884 p = proc_net_fops_create(net, "icmp", S_IRUGO, &ping_seq_fops);
885 if (!p)
886 rc = -ENOMEM;
887 return rc;
888}
889
890static void ping_proc_unregister(struct net *net)
891{
892 proc_net_remove(net, "icmp");
893}
894
895
896static int __net_init ping_proc_init_net(struct net *net)
897{
898 return ping_proc_register(net);
899}
900
901static void __net_exit ping_proc_exit_net(struct net *net)
902{
903 ping_proc_unregister(net);
904}
905
906static struct pernet_operations ping_net_ops = {
907 .init = ping_proc_init_net,
908 .exit = ping_proc_exit_net,
909};
910
911int __init ping_proc_init(void)
912{
913 return register_pernet_subsys(&ping_net_ops);
914}
915
916void ping_proc_exit(void)
917{
918 unregister_pernet_subsys(&ping_net_ops);
919}
920
921#endif
922
923void __init ping_init(void)
924{
925 int i;
926
927 for (i = 0; i < PING_HTABLE_SIZE; i++)
928 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
929 rwlock_init(&ping_table.lock);
930}