projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
userns: Print out socket uids in a user namespace aware fashion.
[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
c319b4d7 23#include <linux/uaccess.h>
c319b4d7
VK		 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>
bc3b2d7f 41#include <linux/export.h>
c319b4d7
VK		 42#include <net/sock.h>
43#include <net/ping.h>
c319b4d7
VK		 44#include <net/udp.h>
45#include <net/route.h>
46#include <net/inet_common.h>
47#include <net/checksum.h>
48
49
1b1cb1f7 50static struct ping_table ping_table;
c319b4d7 51
1b1cb1f7 52static u16 ping_port_rover;
c319b4d7 53
95c96174 54static inline int ping_hashfn(struct net *net, unsigned int num, unsigned int mask)
c319b4d7
VK		 55{
56 int res = (num + net_hash_mix(net)) & mask;
95c96174 57
c319b4d7
VK		 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,
95c96174 63 struct net *net, unsigned int num)
c319b4d7
VK		 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);
747465ef
ED		 143 isk->inet_num = 0;
144 isk->inet_sport = 0;
c319b4d7
VK		 145 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
146 write_unlock_bh(&ping_table.lock);
147 }
148}
149
747465ef 150static struct sock *ping_v4_lookup(struct net *net, __be32 saddr, __be32 daddr,
1b1cb1f7 151 u16 ident, int dif)
c319b4d7
VK		 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
747465ef 158 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
058bd4d2 159 (int)ident, &daddr, dif);
c319b4d7
VK		 160 read_lock_bh(&ping_table.lock);
161
162 ping_portaddr_for_each_entry(sk, hnode, hslot) {
163 isk = inet_sk(sk);
164
747465ef
ED		 165 pr_debug("found: %p: num = %d, daddr = %pI4, dif = %d\n", sk,
166 (int)isk->inet_num, &isk->inet_rcv_saddr,
c319b4d7
VK		 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;
182exit:
183 read_unlock_bh(&ping_table.lock);
184
185 return sk;
186}
187
75e308c8
CG		 188static void inet_get_ping_group_range_net(struct net *net, gid_t *low,
189 gid_t *high)
f56e03e8
VK		 190{
191 gid_t *data = net->ipv4.sysctl_ping_group_range;
95c96174
ED		 192 unsigned int seq;
193
f56e03e8
VK		 194 do {
195 seq = read_seqbegin(&sysctl_local_ports.lock);
196
197 *low = data[0];
198 *high = data[1];
199 } while (read_seqretry(&sysctl_local_ports.lock, seq));
200}
201
202
c319b4d7
VK		 203static int ping_init_sock(struct sock *sk)
204{
205 struct net *net = sock_net(sk);
206 gid_t group = current_egid();
207 gid_t range[2];
208 struct group_info *group_info = get_current_groups();
209 int i, j, count = group_info->ngroups;
ae2975bc 210 kgid_t low, high;
c319b4d7
VK		 211
212 inet_get_ping_group_range_net(net, range, range+1);
ae2975bc
EB		 213 low = make_kgid(&init_user_ns, range[0]);
214 high = make_kgid(&init_user_ns, range[1]);
215 if (!gid_valid(low) || !gid_valid(high) || gid_lt(high, low))
216 return -EACCES;
217
c319b4d7
VK		 218 if (range[0] <= group && group <= range[1])
219 return 0;
220
221 for (i = 0; i < group_info->nblocks; i++) {
222 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
c319b4d7 223 for (j = 0; j < cp_count; j++) {
ae2975bc
EB		 224 kgid_t gid = group_info->blocks[i][j];
225 if (gid_lte(low, gid) && gid_lte(gid, high))
c319b4d7
VK		 226 return 0;
227 }
228
229 count -= cp_count;
230 }
231
232 return -EACCES;
233}
234
235static void ping_close(struct sock *sk, long timeout)
236{
237 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
058bd4d2 238 inet_sk(sk), inet_sk(sk)->inet_num);
c319b4d7
VK		 239 pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
240
241 sk_common_release(sk);
242}
243
244/*
245 * We need our own bind because there are no privileged id's == local ports.
246 * Moreover, we don't allow binding to multi- and broadcast addresses.
247 */
248
249static int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
250{
251 struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
252 struct inet_sock *isk = inet_sk(sk);
253 unsigned short snum;
254 int chk_addr_ret;
255 int err;
256
257 if (addr_len < sizeof(struct sockaddr_in))
258 return -EINVAL;
259
260 pr_debug("ping_v4_bind(sk=%p,sa_addr=%08x,sa_port=%d)\n",
058bd4d2 261 sk, addr->sin_addr.s_addr, ntohs(addr->sin_port));
c319b4d7
VK		 262
263 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
747465ef 264 if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
c319b4d7
VK		 265 chk_addr_ret = RTN_LOCAL;
266
267 if ((sysctl_ip_nonlocal_bind == 0 &&
268 isk->freebind == 0 && isk->transparent == 0 &&
269 chk_addr_ret != RTN_LOCAL) ||
270 chk_addr_ret == RTN_MULTICAST ||
271 chk_addr_ret == RTN_BROADCAST)
272 return -EADDRNOTAVAIL;
273
274 lock_sock(sk);
275
276 err = -EINVAL;
277 if (isk->inet_num != 0)
278 goto out;
279
280 err = -EADDRINUSE;
281 isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
282 snum = ntohs(addr->sin_port);
283 if (ping_v4_get_port(sk, snum) != 0) {
284 isk->inet_saddr = isk->inet_rcv_saddr = 0;
285 goto out;
286 }
287
747465ef 288 pr_debug("after bind(): num = %d, daddr = %pI4, dif = %d\n",
058bd4d2
JP		 289 (int)isk->inet_num,
290 &isk->inet_rcv_saddr,
291 (int)sk->sk_bound_dev_if);
c319b4d7
VK		 292
293 err = 0;
294 if (isk->inet_rcv_saddr)
295 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
296 if (snum)
297 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
298 isk->inet_sport = htons(isk->inet_num);
299 isk->inet_daddr = 0;
300 isk->inet_dport = 0;
301 sk_dst_reset(sk);
302out:
303 release_sock(sk);
304 pr_debug("ping_v4_bind -> %d\n", err);
305 return err;
306}
307
308/*
309 * Is this a supported type of ICMP message?
310 */
311
312static inline int ping_supported(int type, int code)
313{
314 if (type == ICMP_ECHO && code == 0)
315 return 1;
316 return 0;
317}
318
319/*
320 * This routine is called by the ICMP module when it gets some
321 * sort of error condition.
322 */
323
324static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
325
326void ping_err(struct sk_buff *skb, u32 info)
327{
328 struct iphdr *iph = (struct iphdr *)skb->data;
329 struct icmphdr *icmph = (struct icmphdr *)(skb->data+(iph->ihl<<2));
330 struct inet_sock *inet_sock;
331 int type = icmph->type;
332 int code = icmph->code;
333 struct net *net = dev_net(skb->dev);
334 struct sock *sk;
335 int harderr;
336 int err;
337
338 /* We assume the packet has already been checked by icmp_unreach */
339
340 if (!ping_supported(icmph->type, icmph->code))
341 return;
342
343 pr_debug("ping_err(type=%04x,code=%04x,id=%04x,seq=%04x)\n", type,
058bd4d2 344 code, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
c319b4d7
VK		 345
346 sk = ping_v4_lookup(net, iph->daddr, iph->saddr,
347 ntohs(icmph->un.echo.id), skb->dev->ifindex);
348 if (sk == NULL) {
c319b4d7
VK		 349 pr_debug("no socket, dropping\n");
350 return; /* No socket for error */
351 }
352 pr_debug("err on socket %p\n", sk);
353
354 err = 0;
355 harderr = 0;
356 inet_sock = inet_sk(sk);
357
358 switch (type) {
359 default:
360 case ICMP_TIME_EXCEEDED:
361 err = EHOSTUNREACH;
362 break;
363 case ICMP_SOURCE_QUENCH:
364 /* This is not a real error but ping wants to see it.
365 * Report it with some fake errno. */
366 err = EREMOTEIO;
367 break;
368 case ICMP_PARAMETERPROB:
369 err = EPROTO;
370 harderr = 1;
371 break;
372 case ICMP_DEST_UNREACH:
373 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
36393395 374 ipv4_sk_update_pmtu(skb, sk, info);
c319b4d7
VK		 375 if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
376 err = EMSGSIZE;
377 harderr = 1;
378 break;
379 }
380 goto out;
381 }
382 err = EHOSTUNREACH;
383 if (code <= NR_ICMP_UNREACH) {
384 harderr = icmp_err_convert[code].fatal;
385 err = icmp_err_convert[code].errno;
386 }
387 break;
388 case ICMP_REDIRECT:
389 /* See ICMP_SOURCE_QUENCH */
55be7a9c 390 ipv4_sk_redirect(skb, sk);
c319b4d7
VK		 391 err = EREMOTEIO;
392 break;
393 }
394
395 /*
396 * RFC1122: OK. Passes ICMP errors back to application, as per
397 * 4.1.3.3.
398 */
399 if (!inet_sock->recverr) {
400 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
401 goto out;
402 } else {
403 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
404 info, (u8 *)icmph);
405 }
406 sk->sk_err = err;
407 sk->sk_error_report(sk);
408out:
409 sock_put(sk);
410}
411
412/*
413 * Copy and checksum an ICMP Echo packet from user space into a buffer.
414 */
415
416struct pingfakehdr {
417 struct icmphdr icmph;
418 struct iovec *iov;
747465ef 419 __wsum wcheck;
c319b4d7
VK		 420};
421
5e73ea1a 422static int ping_getfrag(void *from, char *to,
c319b4d7
VK		 423 int offset, int fraglen, int odd, struct sk_buff *skb)
424{
425 struct pingfakehdr *pfh = (struct pingfakehdr *)from;
426
427 if (offset == 0) {
428 if (fraglen < sizeof(struct icmphdr))
429 BUG();
430 if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
431 pfh->iov, 0, fraglen - sizeof(struct icmphdr),
432 &pfh->wcheck))
433 return -EFAULT;
434
435 return 0;
436 }
437 if (offset < sizeof(struct icmphdr))
438 BUG();
439 if (csum_partial_copy_fromiovecend
440 (to, pfh->iov, offset - sizeof(struct icmphdr),
441 fraglen, &pfh->wcheck))
442 return -EFAULT;
443 return 0;
444}
445
75e308c8
CG		 446static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
447 struct flowi4 *fl4)
c319b4d7
VK		 448{
449 struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
450
451 pfh->wcheck = csum_partial((char *)&pfh->icmph,
452 sizeof(struct icmphdr), pfh->wcheck);
453 pfh->icmph.checksum = csum_fold(pfh->wcheck);
454 memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
455 skb->ip_summed = CHECKSUM_NONE;
456 return ip_push_pending_frames(sk, fl4);
457}
458
bb0cd2fb
CG		 459static int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
460 size_t len)
c319b4d7
VK		 461{
462 struct net *net = sock_net(sk);
463 struct flowi4 fl4;
464 struct inet_sock *inet = inet_sk(sk);
465 struct ipcm_cookie ipc;
466 struct icmphdr user_icmph;
467 struct pingfakehdr pfh;
468 struct rtable *rt = NULL;
469 struct ip_options_data opt_copy;
470 int free = 0;
747465ef 471 __be32 saddr, daddr, faddr;
c319b4d7
VK		 472 u8 tos;
473 int err;
474
475 pr_debug("ping_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
476
477
478 if (len > 0xFFFF)
479 return -EMSGSIZE;
480
481 /*
482 * Check the flags.
483 */
484
485 /* Mirror BSD error message compatibility */
486 if (msg->msg_flags & MSG_OOB)
487 return -EOPNOTSUPP;
488
489 /*
490 * Fetch the ICMP header provided by the userland.
491 * iovec is modified!
492 */
493
494 if (memcpy_fromiovec((u8 *)&user_icmph, msg->msg_iov,
495 sizeof(struct icmphdr)))
496 return -EFAULT;
497 if (!ping_supported(user_icmph.type, user_icmph.code))
498 return -EINVAL;
499
500 /*
501 * Get and verify the address.
502 */
503
504 if (msg->msg_name) {
505 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
506 if (msg->msg_namelen < sizeof(*usin))
507 return -EINVAL;
508 if (usin->sin_family != AF_INET)
509 return -EINVAL;
510 daddr = usin->sin_addr.s_addr;
511 /* no remote port */
512 } else {
513 if (sk->sk_state != TCP_ESTABLISHED)
514 return -EDESTADDRREQ;
515 daddr = inet->inet_daddr;
516 /* no remote port */
517 }
518
519 ipc.addr = inet->inet_saddr;
520 ipc.opt = NULL;
521 ipc.oif = sk->sk_bound_dev_if;
522 ipc.tx_flags = 0;
523 err = sock_tx_timestamp(sk, &ipc.tx_flags);
524 if (err)
525 return err;
526
527 if (msg->msg_controllen) {
528 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
529 if (err)
530 return err;
531 if (ipc.opt)
532 free = 1;
533 }
534 if (!ipc.opt) {
535 struct ip_options_rcu *inet_opt;
536
537 rcu_read_lock();
538 inet_opt = rcu_dereference(inet->inet_opt);
539 if (inet_opt) {
540 memcpy(&opt_copy, inet_opt,
541 sizeof(*inet_opt) + inet_opt->opt.optlen);
542 ipc.opt = &opt_copy.opt;
543 }
544 rcu_read_unlock();
545 }
546
547 saddr = ipc.addr;
548 ipc.addr = faddr = daddr;
549
550 if (ipc.opt && ipc.opt->opt.srr) {
551 if (!daddr)
552 return -EINVAL;
553 faddr = ipc.opt->opt.faddr;
554 }
555 tos = RT_TOS(inet->tos);
556 if (sock_flag(sk, SOCK_LOCALROUTE) ||
557 (msg->msg_flags & MSG_DONTROUTE) ||
558 (ipc.opt && ipc.opt->opt.is_strictroute)) {
559 tos |= RTO_ONLINK;
560 }
561
562 if (ipv4_is_multicast(daddr)) {
563 if (!ipc.oif)
564 ipc.oif = inet->mc_index;
565 if (!saddr)
566 saddr = inet->mc_addr;
76e21053
EH		 567 } else if (!ipc.oif)
568 ipc.oif = inet->uc_index;
c319b4d7
VK		 569
570 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
571 RT_SCOPE_UNIVERSE, sk->sk_protocol,
572 inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
573
574 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
575 rt = ip_route_output_flow(net, &fl4, sk);
576 if (IS_ERR(rt)) {
577 err = PTR_ERR(rt);
578 rt = NULL;
579 if (err == -ENETUNREACH)
580 IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
581 goto out;
582 }
583
584 err = -EACCES;
585 if ((rt->rt_flags & RTCF_BROADCAST) &&
586 !sock_flag(sk, SOCK_BROADCAST))
587 goto out;
588
589 if (msg->msg_flags & MSG_CONFIRM)
590 goto do_confirm;
591back_from_confirm:
592
593 if (!ipc.addr)
594 ipc.addr = fl4.daddr;
595
596 lock_sock(sk);
597
598 pfh.icmph.type = user_icmph.type; /* already checked */
599 pfh.icmph.code = user_icmph.code; /* ditto */
600 pfh.icmph.checksum = 0;
601 pfh.icmph.un.echo.id = inet->inet_sport;
602 pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
603 pfh.iov = msg->msg_iov;
604 pfh.wcheck = 0;
605
606 err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
607 0, &ipc, &rt, msg->msg_flags);
608 if (err)
609 ip_flush_pending_frames(sk);
610 else
611 err = ping_push_pending_frames(sk, &pfh, &fl4);
612 release_sock(sk);
613
614out:
615 ip_rt_put(rt);
616 if (free)
617 kfree(ipc.opt);
618 if (!err) {
619 icmp_out_count(sock_net(sk), user_icmph.type);
620 return len;
621 }
622 return err;
623
624do_confirm:
625 dst_confirm(&rt->dst);
626 if (!(msg->msg_flags & MSG_PROBE) || len)
627 goto back_from_confirm;
628 err = 0;
629 goto out;
630}
631
bb0cd2fb
CG		 632static int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
633 size_t len, int noblock, int flags, int *addr_len)
c319b4d7
VK		 634{
635 struct inet_sock *isk = inet_sk(sk);
636 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
637 struct sk_buff *skb;
638 int copied, err;
639
640 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
641
a5e7424d 642 err = -EOPNOTSUPP;
c319b4d7
VK		 643 if (flags & MSG_OOB)
644 goto out;
645
646 if (addr_len)
647 *addr_len = sizeof(*sin);
648
649 if (flags & MSG_ERRQUEUE)
650 return ip_recv_error(sk, msg, len);
651
652 skb = skb_recv_datagram(sk, flags, noblock, &err);
653 if (!skb)
654 goto out;
655
656 copied = skb->len;
657 if (copied > len) {
658 msg->msg_flags |= MSG_TRUNC;
659 copied = len;
660 }
661
662 /* Don't bother checking the checksum */
663 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
664 if (err)
665 goto done;
666
667 sock_recv_timestamp(msg, sk, skb);
668
669 /* Copy the address. */
670 if (sin) {
671 sin->sin_family = AF_INET;
672 sin->sin_port = 0 /* skb->h.uh->source */;
673 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
674 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
675 }
676 if (isk->cmsg_flags)
677 ip_cmsg_recv(msg, skb);
678 err = copied;
679
680done:
681 skb_free_datagram(sk, skb);
682out:
683 pr_debug("ping_recvmsg -> %d\n", err);
684 return err;
685}
686
687static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
688{
689 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
058bd4d2 690 inet_sk(sk), inet_sk(sk)->inet_num, skb);
c319b4d7 691 if (sock_queue_rcv_skb(sk, skb) < 0) {
c319b4d7
VK		 692 kfree_skb(skb);
693 pr_debug("ping_queue_rcv_skb -> failed\n");
694 return -1;
695 }
696 return 0;
697}
698
699
700/*
701 * All we need to do is get the socket.
702 */
703
704void ping_rcv(struct sk_buff *skb)
705{
706 struct sock *sk;
707 struct net *net = dev_net(skb->dev);
708 struct iphdr *iph = ip_hdr(skb);
709 struct icmphdr *icmph = icmp_hdr(skb);
747465ef
ED		 710 __be32 saddr = iph->saddr;
711 __be32 daddr = iph->daddr;
c319b4d7
VK		 712
713 /* We assume the packet has already been checked by icmp_rcv */
714
715 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
058bd4d2 716 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
c319b4d7
VK		 717
718 /* Push ICMP header back */
719 skb_push(skb, skb->data - (u8 *)icmph);
720
721 sk = ping_v4_lookup(net, saddr, daddr, ntohs(icmph->un.echo.id),
722 skb->dev->ifindex);
723 if (sk != NULL) {
724 pr_debug("rcv on socket %p\n", sk);
725 ping_queue_rcv_skb(sk, skb_get(skb));
726 sock_put(sk);
727 return;
728 }
729 pr_debug("no socket, dropping\n");
730
731 /* We're called from icmp_rcv(). kfree_skb() is done there. */
732}
733
734struct proto ping_prot = {
735 .name = "PING",
736 .owner = THIS_MODULE,
737 .init = ping_init_sock,
738 .close = ping_close,
739 .connect = ip4_datagram_connect,
740 .disconnect = udp_disconnect,
c319b4d7
VK		 741 .setsockopt = ip_setsockopt,
742 .getsockopt = ip_getsockopt,
743 .sendmsg = ping_sendmsg,
744 .recvmsg = ping_recvmsg,
745 .bind = ping_bind,
746 .backlog_rcv = ping_queue_rcv_skb,
747 .hash = ping_v4_hash,
748 .unhash = ping_v4_unhash,
749 .get_port = ping_v4_get_port,
750 .obj_size = sizeof(struct inet_sock),
751};
752EXPORT_SYMBOL(ping_prot);
753
754#ifdef CONFIG_PROC_FS
755
756static struct sock *ping_get_first(struct seq_file *seq, int start)
757{
758 struct sock *sk;
759 struct ping_iter_state *state = seq->private;
760 struct net *net = seq_file_net(seq);
761
762 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
763 ++state->bucket) {
764 struct hlist_nulls_node *node;
75e308c8
CG		 765 struct hlist_nulls_head *hslot;
766
767 hslot = &ping_table.hash[state->bucket];
c319b4d7
VK		 768
769 if (hlist_nulls_empty(hslot))
770 continue;
771
772 sk_nulls_for_each(sk, node, hslot) {
773 if (net_eq(sock_net(sk), net))
774 goto found;
775 }
776 }
777 sk = NULL;
778found:
779 return sk;
780}
781
782static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
783{
784 struct ping_iter_state *state = seq->private;
785 struct net *net = seq_file_net(seq);
786
787 do {
788 sk = sk_nulls_next(sk);
789 } while (sk && (!net_eq(sock_net(sk), net)));
790
791 if (!sk)
792 return ping_get_first(seq, state->bucket + 1);
793 return sk;
794}
795
796static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
797{
798 struct sock *sk = ping_get_first(seq, 0);
799
800 if (sk)
801 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
802 --pos;
803 return pos ? NULL : sk;
804}
805
806static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
807{
808 struct ping_iter_state *state = seq->private;
809 state->bucket = 0;
810
811 read_lock_bh(&ping_table.lock);
812
813 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
814}
815
816static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
817{
818 struct sock *sk;
819
820 if (v == SEQ_START_TOKEN)
821 sk = ping_get_idx(seq, 0);
822 else
823 sk = ping_get_next(seq, v);
824
825 ++*pos;
826 return sk;
827}
828
829static void ping_seq_stop(struct seq_file *seq, void *v)
830{
831 read_unlock_bh(&ping_table.lock);
832}
833
834static void ping_format_sock(struct sock *sp, struct seq_file *f,
835 int bucket, int *len)
836{
837 struct inet_sock *inet = inet_sk(sp);
838 __be32 dest = inet->inet_daddr;
839 __be32 src = inet->inet_rcv_saddr;
840 __u16 destp = ntohs(inet->inet_dport);
841 __u16 srcp = ntohs(inet->inet_sport);
842
843 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
844 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
845 bucket, src, srcp, dest, destp, sp->sk_state,
846 sk_wmem_alloc_get(sp),
847 sk_rmem_alloc_get(sp),
a7cb5a49
EB		 848 0, 0L, 0,
849 from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
850 0, sock_i_ino(sp),
c319b4d7
VK		 851 atomic_read(&sp->sk_refcnt), sp,
852 atomic_read(&sp->sk_drops), len);
853}
854
855static int ping_seq_show(struct seq_file *seq, void *v)
856{
857 if (v == SEQ_START_TOKEN)
858 seq_printf(seq, "%-127s\n",
859 " sl local_address rem_address st tx_queue "
860 "rx_queue tr tm->when retrnsmt uid timeout "
861 "inode ref pointer drops");
862 else {
863 struct ping_iter_state *state = seq->private;
864 int len;
865
866 ping_format_sock(v, seq, state->bucket, &len);
867 seq_printf(seq, "%*s\n", 127 - len, "");
868 }
869 return 0;
870}
871
872static const struct seq_operations ping_seq_ops = {
873 .show = ping_seq_show,
874 .start = ping_seq_start,
875 .next = ping_seq_next,
876 .stop = ping_seq_stop,
877};
878
879static int ping_seq_open(struct inode *inode, struct file *file)
880{
881 return seq_open_net(inode, file, &ping_seq_ops,
882 sizeof(struct ping_iter_state));
883}
884
885static const struct file_operations ping_seq_fops = {
886 .open = ping_seq_open,
887 .read = seq_read,
888 .llseek = seq_lseek,
889 .release = seq_release_net,
890};
891
892static int ping_proc_register(struct net *net)
893{
894 struct proc_dir_entry *p;
895 int rc = 0;
896
897 p = proc_net_fops_create(net, "icmp", S_IRUGO, &ping_seq_fops);
898 if (!p)
899 rc = -ENOMEM;
900 return rc;
901}
902
903static void ping_proc_unregister(struct net *net)
904{
905 proc_net_remove(net, "icmp");
906}
907
908
909static int __net_init ping_proc_init_net(struct net *net)
910{
911 return ping_proc_register(net);
912}
913
914static void __net_exit ping_proc_exit_net(struct net *net)
915{
916 ping_proc_unregister(net);
917}
918
919static struct pernet_operations ping_net_ops = {
920 .init = ping_proc_init_net,
921 .exit = ping_proc_exit_net,
922};
923
924int __init ping_proc_init(void)
925{
926 return register_pernet_subsys(&ping_net_ops);
927}
928
929void ping_proc_exit(void)
930{
931 unregister_pernet_subsys(&ping_net_ops);
932}
933
934#endif
935
936void __init ping_init(void)
937{
938 int i;
939
940 for (i = 0; i < PING_HTABLE_SIZE; i++)
941 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
942 rwlock_init(&ping_table.lock);
943}
Linux 6.x block layer and io_uring tree(s)