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.
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.
13 * Based on ipv4/udp.c code.
15 * Authors: Vasiliy Kulikov / Openwall (for Linux 2.6),
16 * Pavel Kankovsky (for Linux 2.4.32)
18 * Pavel gave all rights to bugs to Vasiliy,
19 * none of the bugs are Pavel's now.
23 #include <linux/uaccess.h>
24 #include <linux/types.h>
25 #include <linux/fcntl.h>
26 #include <linux/socket.h>
27 #include <linux/sockios.h>
29 #include <linux/errno.h>
30 #include <linux/timer.h>
32 #include <linux/inet.h>
33 #include <linux/netdevice.h>
37 #include <net/protocol.h>
38 #include <linux/skbuff.h>
39 #include <linux/proc_fs.h>
40 #include <linux/export.h>
44 #include <net/route.h>
45 #include <net/inet_common.h>
46 #include <net/checksum.h>
48 #if IS_ENABLED(CONFIG_IPV6)
49 #include <linux/in6.h>
50 #include <linux/icmpv6.h>
51 #include <net/addrconf.h>
53 #include <net/transp_v6.h>
57 struct hlist_nulls_head hash[PING_HTABLE_SIZE];
61 static struct ping_table ping_table;
62 struct pingv6_ops pingv6_ops;
63 EXPORT_SYMBOL_GPL(pingv6_ops);
65 static u16 ping_port_rover;
67 static inline u32 ping_hashfn(const struct net *net, u32 num, u32 mask)
69 u32 res = (num + net_hash_mix(net)) & mask;
71 pr_debug("hash(%u) = %u\n", num, res);
74 EXPORT_SYMBOL_GPL(ping_hash);
76 static inline struct hlist_nulls_head *ping_hashslot(struct ping_table *table,
77 struct net *net, unsigned int num)
79 return &table->hash[ping_hashfn(net, num, PING_HTABLE_MASK)];
82 int ping_get_port(struct sock *sk, unsigned short ident)
84 struct hlist_nulls_node *node;
85 struct hlist_nulls_head *hlist;
86 struct inet_sock *isk, *isk2;
87 struct sock *sk2 = NULL;
90 write_lock_bh(&ping_table.lock);
93 u16 result = ping_port_rover + 1;
95 for (i = 0; i < (1L << 16); i++, result++) {
97 result++; /* avoid zero */
98 hlist = ping_hashslot(&ping_table, sock_net(sk),
100 ping_portaddr_for_each_entry(sk2, node, hlist) {
103 if (isk2->inet_num == result)
108 ping_port_rover = ident = result;
116 hlist = ping_hashslot(&ping_table, sock_net(sk), ident);
117 ping_portaddr_for_each_entry(sk2, node, hlist) {
120 /* BUG? Why is this reuse and not reuseaddr? ping.c
121 * doesn't turn off SO_REUSEADDR, and it doesn't expect
122 * that other ping processes can steal its packets.
124 if ((isk2->inet_num == ident) &&
126 (!sk2->sk_reuse || !sk->sk_reuse))
131 pr_debug("found port/ident = %d\n", ident);
132 isk->inet_num = ident;
133 if (sk_unhashed(sk)) {
134 pr_debug("was not hashed\n");
136 hlist_nulls_add_head(&sk->sk_nulls_node, hlist);
137 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
139 write_unlock_bh(&ping_table.lock);
143 write_unlock_bh(&ping_table.lock);
146 EXPORT_SYMBOL_GPL(ping_get_port);
148 void ping_hash(struct sock *sk)
150 pr_debug("ping_hash(sk->port=%u)\n", inet_sk(sk)->inet_num);
151 BUG(); /* "Please do not press this button again." */
154 void ping_unhash(struct sock *sk)
156 struct inet_sock *isk = inet_sk(sk);
157 pr_debug("ping_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
159 write_lock_bh(&ping_table.lock);
160 hlist_nulls_del(&sk->sk_nulls_node);
161 sk_nulls_node_init(&sk->sk_nulls_node);
165 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
166 write_unlock_bh(&ping_table.lock);
169 EXPORT_SYMBOL_GPL(ping_unhash);
171 static struct sock *ping_lookup(struct net *net, struct sk_buff *skb, u16 ident)
173 struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
174 struct sock *sk = NULL;
175 struct inet_sock *isk;
176 struct hlist_nulls_node *hnode;
177 int dif = skb->dev->ifindex;
179 if (skb->protocol == htons(ETH_P_IP)) {
180 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
181 (int)ident, &ip_hdr(skb)->daddr, dif);
182 #if IS_ENABLED(CONFIG_IPV6)
183 } else if (skb->protocol == htons(ETH_P_IPV6)) {
184 pr_debug("try to find: num = %d, daddr = %pI6c, dif = %d\n",
185 (int)ident, &ipv6_hdr(skb)->daddr, dif);
189 read_lock_bh(&ping_table.lock);
191 ping_portaddr_for_each_entry(sk, hnode, hslot) {
194 pr_debug("iterate\n");
195 if (isk->inet_num != ident)
198 if (skb->protocol == htons(ETH_P_IP) &&
199 sk->sk_family == AF_INET) {
200 pr_debug("found: %p: num=%d, daddr=%pI4, dif=%d\n", sk,
201 (int) isk->inet_num, &isk->inet_rcv_saddr,
202 sk->sk_bound_dev_if);
204 if (isk->inet_rcv_saddr &&
205 isk->inet_rcv_saddr != ip_hdr(skb)->daddr)
207 #if IS_ENABLED(CONFIG_IPV6)
208 } else if (skb->protocol == htons(ETH_P_IPV6) &&
209 sk->sk_family == AF_INET6) {
211 pr_debug("found: %p: num=%d, daddr=%pI6c, dif=%d\n", sk,
213 &sk->sk_v6_rcv_saddr,
214 sk->sk_bound_dev_if);
216 if (!ipv6_addr_any(&sk->sk_v6_rcv_saddr) &&
217 !ipv6_addr_equal(&sk->sk_v6_rcv_saddr,
218 &ipv6_hdr(skb)->daddr))
225 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
234 read_unlock_bh(&ping_table.lock);
239 static void inet_get_ping_group_range_net(struct net *net, kgid_t *low,
242 kgid_t *data = net->ipv4.ping_group_range.range;
246 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
250 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
254 int ping_init_sock(struct sock *sk)
256 struct net *net = sock_net(sk);
257 kgid_t group = current_egid();
258 struct group_info *group_info;
263 if (sk->sk_family == AF_INET6)
266 inet_get_ping_group_range_net(net, &low, &high);
267 if (gid_lte(low, group) && gid_lte(group, high))
270 group_info = get_current_groups();
271 count = group_info->ngroups;
272 for (i = 0; i < group_info->nblocks; i++) {
273 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
274 for (j = 0; j < cp_count; j++) {
275 kgid_t gid = group_info->blocks[i][j];
276 if (gid_lte(low, gid) && gid_lte(gid, high))
277 goto out_release_group;
286 put_group_info(group_info);
289 EXPORT_SYMBOL_GPL(ping_init_sock);
291 void ping_close(struct sock *sk, long timeout)
293 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
294 inet_sk(sk), inet_sk(sk)->inet_num);
295 pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
297 sk_common_release(sk);
299 EXPORT_SYMBOL_GPL(ping_close);
301 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
302 static int ping_check_bind_addr(struct sock *sk, struct inet_sock *isk,
303 struct sockaddr *uaddr, int addr_len) {
304 struct net *net = sock_net(sk);
305 if (sk->sk_family == AF_INET) {
306 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
309 if (addr_len < sizeof(*addr))
312 if (addr->sin_family != AF_INET &&
313 !(addr->sin_family == AF_UNSPEC &&
314 addr->sin_addr.s_addr == htonl(INADDR_ANY)))
315 return -EAFNOSUPPORT;
317 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI4,port=%d)\n",
318 sk, &addr->sin_addr.s_addr, ntohs(addr->sin_port));
320 chk_addr_ret = inet_addr_type(net, addr->sin_addr.s_addr);
322 if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
323 chk_addr_ret = RTN_LOCAL;
325 if ((net->ipv4.sysctl_ip_nonlocal_bind == 0 &&
326 isk->freebind == 0 && isk->transparent == 0 &&
327 chk_addr_ret != RTN_LOCAL) ||
328 chk_addr_ret == RTN_MULTICAST ||
329 chk_addr_ret == RTN_BROADCAST)
330 return -EADDRNOTAVAIL;
332 #if IS_ENABLED(CONFIG_IPV6)
333 } else if (sk->sk_family == AF_INET6) {
334 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) uaddr;
335 int addr_type, scoped, has_addr;
336 struct net_device *dev = NULL;
338 if (addr_len < sizeof(*addr))
341 if (addr->sin6_family != AF_INET6)
342 return -EAFNOSUPPORT;
344 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI6c,port=%d)\n",
345 sk, addr->sin6_addr.s6_addr, ntohs(addr->sin6_port));
347 addr_type = ipv6_addr_type(&addr->sin6_addr);
348 scoped = __ipv6_addr_needs_scope_id(addr_type);
349 if ((addr_type != IPV6_ADDR_ANY &&
350 !(addr_type & IPV6_ADDR_UNICAST)) ||
351 (scoped && !addr->sin6_scope_id))
355 if (addr->sin6_scope_id) {
356 dev = dev_get_by_index_rcu(net, addr->sin6_scope_id);
362 has_addr = pingv6_ops.ipv6_chk_addr(net, &addr->sin6_addr, dev,
366 if (!(isk->freebind || isk->transparent || has_addr ||
367 addr_type == IPV6_ADDR_ANY))
368 return -EADDRNOTAVAIL;
371 sk->sk_bound_dev_if = addr->sin6_scope_id;
374 return -EAFNOSUPPORT;
379 static void ping_set_saddr(struct sock *sk, struct sockaddr *saddr)
381 if (saddr->sa_family == AF_INET) {
382 struct inet_sock *isk = inet_sk(sk);
383 struct sockaddr_in *addr = (struct sockaddr_in *) saddr;
384 isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
385 #if IS_ENABLED(CONFIG_IPV6)
386 } else if (saddr->sa_family == AF_INET6) {
387 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) saddr;
388 struct ipv6_pinfo *np = inet6_sk(sk);
389 sk->sk_v6_rcv_saddr = np->saddr = addr->sin6_addr;
394 static void ping_clear_saddr(struct sock *sk, int dif)
396 sk->sk_bound_dev_if = dif;
397 if (sk->sk_family == AF_INET) {
398 struct inet_sock *isk = inet_sk(sk);
399 isk->inet_rcv_saddr = isk->inet_saddr = 0;
400 #if IS_ENABLED(CONFIG_IPV6)
401 } else if (sk->sk_family == AF_INET6) {
402 struct ipv6_pinfo *np = inet6_sk(sk);
403 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
404 memset(&np->saddr, 0, sizeof(np->saddr));
409 * We need our own bind because there are no privileged id's == local ports.
410 * Moreover, we don't allow binding to multi- and broadcast addresses.
413 int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
415 struct inet_sock *isk = inet_sk(sk);
418 int dif = sk->sk_bound_dev_if;
420 err = ping_check_bind_addr(sk, isk, uaddr, addr_len);
427 if (isk->inet_num != 0)
431 ping_set_saddr(sk, uaddr);
432 snum = ntohs(((struct sockaddr_in *)uaddr)->sin_port);
433 if (ping_get_port(sk, snum) != 0) {
434 ping_clear_saddr(sk, dif);
438 pr_debug("after bind(): num = %d, dif = %d\n",
440 (int)sk->sk_bound_dev_if);
443 if (sk->sk_family == AF_INET && isk->inet_rcv_saddr)
444 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
445 #if IS_ENABLED(CONFIG_IPV6)
446 if (sk->sk_family == AF_INET6 && !ipv6_addr_any(&sk->sk_v6_rcv_saddr))
447 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
451 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
452 isk->inet_sport = htons(isk->inet_num);
456 #if IS_ENABLED(CONFIG_IPV6)
457 if (sk->sk_family == AF_INET6)
458 memset(&sk->sk_v6_daddr, 0, sizeof(sk->sk_v6_daddr));
464 pr_debug("ping_v4_bind -> %d\n", err);
467 EXPORT_SYMBOL_GPL(ping_bind);
470 * Is this a supported type of ICMP message?
473 static inline int ping_supported(int family, int type, int code)
475 return (family == AF_INET && type == ICMP_ECHO && code == 0) ||
476 (family == AF_INET6 && type == ICMPV6_ECHO_REQUEST && code == 0);
480 * This routine is called by the ICMP module when it gets some
481 * sort of error condition.
484 void ping_err(struct sk_buff *skb, int offset, u32 info)
487 struct icmphdr *icmph;
488 struct inet_sock *inet_sock;
491 struct net *net = dev_net(skb->dev);
496 if (skb->protocol == htons(ETH_P_IP)) {
498 type = icmp_hdr(skb)->type;
499 code = icmp_hdr(skb)->code;
500 icmph = (struct icmphdr *)(skb->data + offset);
501 } else if (skb->protocol == htons(ETH_P_IPV6)) {
503 type = icmp6_hdr(skb)->icmp6_type;
504 code = icmp6_hdr(skb)->icmp6_code;
505 icmph = (struct icmphdr *) (skb->data + offset);
510 /* We assume the packet has already been checked by icmp_unreach */
512 if (!ping_supported(family, icmph->type, icmph->code))
515 pr_debug("ping_err(proto=0x%x,type=%d,code=%d,id=%04x,seq=%04x)\n",
516 skb->protocol, type, code, ntohs(icmph->un.echo.id),
517 ntohs(icmph->un.echo.sequence));
519 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
521 pr_debug("no socket, dropping\n");
522 return; /* No socket for error */
524 pr_debug("err on socket %p\n", sk);
528 inet_sock = inet_sk(sk);
530 if (skb->protocol == htons(ETH_P_IP)) {
533 case ICMP_TIME_EXCEEDED:
536 case ICMP_SOURCE_QUENCH:
537 /* This is not a real error but ping wants to see it.
538 * Report it with some fake errno.
542 case ICMP_PARAMETERPROB:
546 case ICMP_DEST_UNREACH:
547 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
548 ipv4_sk_update_pmtu(skb, sk, info);
549 if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
557 if (code <= NR_ICMP_UNREACH) {
558 harderr = icmp_err_convert[code].fatal;
559 err = icmp_err_convert[code].errno;
563 /* See ICMP_SOURCE_QUENCH */
564 ipv4_sk_redirect(skb, sk);
568 #if IS_ENABLED(CONFIG_IPV6)
569 } else if (skb->protocol == htons(ETH_P_IPV6)) {
570 harderr = pingv6_ops.icmpv6_err_convert(type, code, &err);
575 * RFC1122: OK. Passes ICMP errors back to application, as per
578 if ((family == AF_INET && !inet_sock->recverr) ||
579 (family == AF_INET6 && !inet6_sk(sk)->recverr)) {
580 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
583 if (family == AF_INET) {
584 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
586 #if IS_ENABLED(CONFIG_IPV6)
587 } else if (family == AF_INET6) {
588 pingv6_ops.ipv6_icmp_error(sk, skb, err, 0,
594 sk->sk_error_report(sk);
598 EXPORT_SYMBOL_GPL(ping_err);
601 * Copy and checksum an ICMP Echo packet from user space into a buffer
602 * starting from the payload.
605 int ping_getfrag(void *from, char *to,
606 int offset, int fraglen, int odd, struct sk_buff *skb)
608 struct pingfakehdr *pfh = (struct pingfakehdr *)from;
611 fraglen -= sizeof(struct icmphdr);
614 if (csum_and_copy_from_iter(to + sizeof(struct icmphdr),
615 fraglen, &pfh->wcheck,
616 &pfh->msg->msg_iter) != fraglen)
618 } else if (offset < sizeof(struct icmphdr)) {
621 if (csum_and_copy_from_iter(to, fraglen, &pfh->wcheck,
622 &pfh->msg->msg_iter) != fraglen)
626 #if IS_ENABLED(CONFIG_IPV6)
627 /* For IPv6, checksum each skb as we go along, as expected by
628 * icmpv6_push_pending_frames. For IPv4, accumulate the checksum in
629 * wcheck, it will be finalized in ping_v4_push_pending_frames.
631 if (pfh->family == AF_INET6) {
632 skb->csum = pfh->wcheck;
633 skb->ip_summed = CHECKSUM_NONE;
640 EXPORT_SYMBOL_GPL(ping_getfrag);
642 static int ping_v4_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
645 struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
647 pfh->wcheck = csum_partial((char *)&pfh->icmph,
648 sizeof(struct icmphdr), pfh->wcheck);
649 pfh->icmph.checksum = csum_fold(pfh->wcheck);
650 memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
651 skb->ip_summed = CHECKSUM_NONE;
652 return ip_push_pending_frames(sk, fl4);
655 int ping_common_sendmsg(int family, struct msghdr *msg, size_t len,
656 void *user_icmph, size_t icmph_len) {
666 /* Mirror BSD error message compatibility */
667 if (msg->msg_flags & MSG_OOB)
671 * Fetch the ICMP header provided by the userland.
672 * iovec is modified! The ICMP header is consumed.
674 if (memcpy_from_msg(user_icmph, msg, icmph_len))
677 if (family == AF_INET) {
678 type = ((struct icmphdr *) user_icmph)->type;
679 code = ((struct icmphdr *) user_icmph)->code;
680 #if IS_ENABLED(CONFIG_IPV6)
681 } else if (family == AF_INET6) {
682 type = ((struct icmp6hdr *) user_icmph)->icmp6_type;
683 code = ((struct icmp6hdr *) user_icmph)->icmp6_code;
689 if (!ping_supported(family, type, code))
694 EXPORT_SYMBOL_GPL(ping_common_sendmsg);
696 static int ping_v4_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
698 struct net *net = sock_net(sk);
700 struct inet_sock *inet = inet_sk(sk);
701 struct ipcm_cookie ipc;
702 struct icmphdr user_icmph;
703 struct pingfakehdr pfh;
704 struct rtable *rt = NULL;
705 struct ip_options_data opt_copy;
707 __be32 saddr, daddr, faddr;
711 pr_debug("ping_v4_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
713 err = ping_common_sendmsg(AF_INET, msg, len, &user_icmph,
719 * Get and verify the address.
723 DECLARE_SOCKADDR(struct sockaddr_in *, usin, msg->msg_name);
724 if (msg->msg_namelen < sizeof(*usin))
726 if (usin->sin_family != AF_INET)
727 return -EAFNOSUPPORT;
728 daddr = usin->sin_addr.s_addr;
731 if (sk->sk_state != TCP_ESTABLISHED)
732 return -EDESTADDRREQ;
733 daddr = inet->inet_daddr;
737 ipc.addr = inet->inet_saddr;
739 ipc.oif = sk->sk_bound_dev_if;
744 sock_tx_timestamp(sk, &ipc.tx_flags);
746 if (msg->msg_controllen) {
747 err = ip_cmsg_send(sock_net(sk), msg, &ipc, false);
754 struct ip_options_rcu *inet_opt;
757 inet_opt = rcu_dereference(inet->inet_opt);
759 memcpy(&opt_copy, inet_opt,
760 sizeof(*inet_opt) + inet_opt->opt.optlen);
761 ipc.opt = &opt_copy.opt;
767 ipc.addr = faddr = daddr;
769 if (ipc.opt && ipc.opt->opt.srr) {
772 faddr = ipc.opt->opt.faddr;
774 tos = get_rttos(&ipc, inet);
775 if (sock_flag(sk, SOCK_LOCALROUTE) ||
776 (msg->msg_flags & MSG_DONTROUTE) ||
777 (ipc.opt && ipc.opt->opt.is_strictroute)) {
781 if (ipv4_is_multicast(daddr)) {
783 ipc.oif = inet->mc_index;
785 saddr = inet->mc_addr;
787 ipc.oif = inet->uc_index;
789 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
790 RT_SCOPE_UNIVERSE, sk->sk_protocol,
791 inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
793 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
794 rt = ip_route_output_flow(net, &fl4, sk);
798 if (err == -ENETUNREACH)
799 IP_INC_STATS(net, IPSTATS_MIB_OUTNOROUTES);
804 if ((rt->rt_flags & RTCF_BROADCAST) &&
805 !sock_flag(sk, SOCK_BROADCAST))
808 if (msg->msg_flags & MSG_CONFIRM)
813 ipc.addr = fl4.daddr;
817 pfh.icmph.type = user_icmph.type; /* already checked */
818 pfh.icmph.code = user_icmph.code; /* ditto */
819 pfh.icmph.checksum = 0;
820 pfh.icmph.un.echo.id = inet->inet_sport;
821 pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
824 pfh.family = AF_INET;
826 err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
827 0, &ipc, &rt, msg->msg_flags);
829 ip_flush_pending_frames(sk);
831 err = ping_v4_push_pending_frames(sk, &pfh, &fl4);
839 icmp_out_count(sock_net(sk), user_icmph.type);
845 dst_confirm(&rt->dst);
846 if (!(msg->msg_flags & MSG_PROBE) || len)
847 goto back_from_confirm;
852 int ping_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int noblock,
853 int flags, int *addr_len)
855 struct inet_sock *isk = inet_sk(sk);
856 int family = sk->sk_family;
860 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
866 if (flags & MSG_ERRQUEUE)
867 return inet_recv_error(sk, msg, len, addr_len);
869 skb = skb_recv_datagram(sk, flags, noblock, &err);
875 msg->msg_flags |= MSG_TRUNC;
879 /* Don't bother checking the checksum */
880 err = skb_copy_datagram_msg(skb, 0, msg, copied);
884 sock_recv_timestamp(msg, sk, skb);
886 /* Copy the address and add cmsg data. */
887 if (family == AF_INET) {
888 DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name);
891 sin->sin_family = AF_INET;
892 sin->sin_port = 0 /* skb->h.uh->source */;
893 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
894 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
895 *addr_len = sizeof(*sin);
899 ip_cmsg_recv(msg, skb);
901 #if IS_ENABLED(CONFIG_IPV6)
902 } else if (family == AF_INET6) {
903 struct ipv6_pinfo *np = inet6_sk(sk);
904 struct ipv6hdr *ip6 = ipv6_hdr(skb);
905 DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
908 sin6->sin6_family = AF_INET6;
910 sin6->sin6_addr = ip6->saddr;
911 sin6->sin6_flowinfo = 0;
913 sin6->sin6_flowinfo = ip6_flowinfo(ip6);
914 sin6->sin6_scope_id =
915 ipv6_iface_scope_id(&sin6->sin6_addr,
917 *addr_len = sizeof(*sin6);
920 if (inet6_sk(sk)->rxopt.all)
921 pingv6_ops.ip6_datagram_recv_common_ctl(sk, msg, skb);
922 if (skb->protocol == htons(ETH_P_IPV6) &&
923 inet6_sk(sk)->rxopt.all)
924 pingv6_ops.ip6_datagram_recv_specific_ctl(sk, msg, skb);
925 else if (skb->protocol == htons(ETH_P_IP) && isk->cmsg_flags)
926 ip_cmsg_recv(msg, skb);
935 skb_free_datagram(sk, skb);
937 pr_debug("ping_recvmsg -> %d\n", err);
940 EXPORT_SYMBOL_GPL(ping_recvmsg);
942 int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
944 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
945 inet_sk(sk), inet_sk(sk)->inet_num, skb);
946 if (sock_queue_rcv_skb(sk, skb) < 0) {
948 pr_debug("ping_queue_rcv_skb -> failed\n");
953 EXPORT_SYMBOL_GPL(ping_queue_rcv_skb);
957 * All we need to do is get the socket.
960 bool ping_rcv(struct sk_buff *skb)
963 struct net *net = dev_net(skb->dev);
964 struct icmphdr *icmph = icmp_hdr(skb);
966 /* We assume the packet has already been checked by icmp_rcv */
968 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
969 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
971 /* Push ICMP header back */
972 skb_push(skb, skb->data - (u8 *)icmph);
974 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
976 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
978 pr_debug("rcv on socket %p\n", sk);
980 ping_queue_rcv_skb(sk, skb2);
984 pr_debug("no socket, dropping\n");
988 EXPORT_SYMBOL_GPL(ping_rcv);
990 struct proto ping_prot = {
992 .owner = THIS_MODULE,
993 .init = ping_init_sock,
995 .connect = ip4_datagram_connect,
996 .disconnect = udp_disconnect,
997 .setsockopt = ip_setsockopt,
998 .getsockopt = ip_getsockopt,
999 .sendmsg = ping_v4_sendmsg,
1000 .recvmsg = ping_recvmsg,
1002 .backlog_rcv = ping_queue_rcv_skb,
1003 .release_cb = ip4_datagram_release_cb,
1005 .unhash = ping_unhash,
1006 .get_port = ping_get_port,
1007 .obj_size = sizeof(struct inet_sock),
1009 EXPORT_SYMBOL(ping_prot);
1011 #ifdef CONFIG_PROC_FS
1013 static struct sock *ping_get_first(struct seq_file *seq, int start)
1016 struct ping_iter_state *state = seq->private;
1017 struct net *net = seq_file_net(seq);
1019 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
1021 struct hlist_nulls_node *node;
1022 struct hlist_nulls_head *hslot;
1024 hslot = &ping_table.hash[state->bucket];
1026 if (hlist_nulls_empty(hslot))
1029 sk_nulls_for_each(sk, node, hslot) {
1030 if (net_eq(sock_net(sk), net) &&
1031 sk->sk_family == state->family)
1040 static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
1042 struct ping_iter_state *state = seq->private;
1043 struct net *net = seq_file_net(seq);
1046 sk = sk_nulls_next(sk);
1047 } while (sk && (!net_eq(sock_net(sk), net)));
1050 return ping_get_first(seq, state->bucket + 1);
1054 static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
1056 struct sock *sk = ping_get_first(seq, 0);
1059 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
1061 return pos ? NULL : sk;
1064 void *ping_seq_start(struct seq_file *seq, loff_t *pos, sa_family_t family)
1066 struct ping_iter_state *state = seq->private;
1068 state->family = family;
1070 read_lock_bh(&ping_table.lock);
1072 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
1074 EXPORT_SYMBOL_GPL(ping_seq_start);
1076 static void *ping_v4_seq_start(struct seq_file *seq, loff_t *pos)
1078 return ping_seq_start(seq, pos, AF_INET);
1081 void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1085 if (v == SEQ_START_TOKEN)
1086 sk = ping_get_idx(seq, 0);
1088 sk = ping_get_next(seq, v);
1093 EXPORT_SYMBOL_GPL(ping_seq_next);
1095 void ping_seq_stop(struct seq_file *seq, void *v)
1097 read_unlock_bh(&ping_table.lock);
1099 EXPORT_SYMBOL_GPL(ping_seq_stop);
1101 static void ping_v4_format_sock(struct sock *sp, struct seq_file *f,
1104 struct inet_sock *inet = inet_sk(sp);
1105 __be32 dest = inet->inet_daddr;
1106 __be32 src = inet->inet_rcv_saddr;
1107 __u16 destp = ntohs(inet->inet_dport);
1108 __u16 srcp = ntohs(inet->inet_sport);
1110 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
1111 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d",
1112 bucket, src, srcp, dest, destp, sp->sk_state,
1113 sk_wmem_alloc_get(sp),
1114 sk_rmem_alloc_get(sp),
1116 from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
1118 atomic_read(&sp->sk_refcnt), sp,
1119 atomic_read(&sp->sk_drops));
1122 static int ping_v4_seq_show(struct seq_file *seq, void *v)
1124 seq_setwidth(seq, 127);
1125 if (v == SEQ_START_TOKEN)
1126 seq_puts(seq, " sl local_address rem_address st tx_queue "
1127 "rx_queue tr tm->when retrnsmt uid timeout "
1128 "inode ref pointer drops");
1130 struct ping_iter_state *state = seq->private;
1132 ping_v4_format_sock(v, seq, state->bucket);
1138 static const struct seq_operations ping_v4_seq_ops = {
1139 .show = ping_v4_seq_show,
1140 .start = ping_v4_seq_start,
1141 .next = ping_seq_next,
1142 .stop = ping_seq_stop,
1145 static int ping_seq_open(struct inode *inode, struct file *file)
1147 struct ping_seq_afinfo *afinfo = PDE_DATA(inode);
1148 return seq_open_net(inode, file, &afinfo->seq_ops,
1149 sizeof(struct ping_iter_state));
1152 const struct file_operations ping_seq_fops = {
1153 .open = ping_seq_open,
1155 .llseek = seq_lseek,
1156 .release = seq_release_net,
1158 EXPORT_SYMBOL_GPL(ping_seq_fops);
1160 static struct ping_seq_afinfo ping_v4_seq_afinfo = {
1163 .seq_fops = &ping_seq_fops,
1165 .start = ping_v4_seq_start,
1166 .show = ping_v4_seq_show,
1167 .next = ping_seq_next,
1168 .stop = ping_seq_stop,
1172 int ping_proc_register(struct net *net, struct ping_seq_afinfo *afinfo)
1174 struct proc_dir_entry *p;
1175 p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
1176 afinfo->seq_fops, afinfo);
1181 EXPORT_SYMBOL_GPL(ping_proc_register);
1183 void ping_proc_unregister(struct net *net, struct ping_seq_afinfo *afinfo)
1185 remove_proc_entry(afinfo->name, net->proc_net);
1187 EXPORT_SYMBOL_GPL(ping_proc_unregister);
1189 static int __net_init ping_v4_proc_init_net(struct net *net)
1191 return ping_proc_register(net, &ping_v4_seq_afinfo);
1194 static void __net_exit ping_v4_proc_exit_net(struct net *net)
1196 ping_proc_unregister(net, &ping_v4_seq_afinfo);
1199 static struct pernet_operations ping_v4_net_ops = {
1200 .init = ping_v4_proc_init_net,
1201 .exit = ping_v4_proc_exit_net,
1204 int __init ping_proc_init(void)
1206 return register_pernet_subsys(&ping_v4_net_ops);
1209 void ping_proc_exit(void)
1211 unregister_pernet_subsys(&ping_v4_net_ops);
1216 void __init ping_init(void)
1220 for (i = 0; i < PING_HTABLE_SIZE; i++)
1221 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
1222 rwlock_init(&ping_table.lock);