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 ping_table ping_table;
58 struct pingv6_ops pingv6_ops;
59 EXPORT_SYMBOL_GPL(pingv6_ops);
61 static u16 ping_port_rover;
63 static inline int ping_hashfn(struct net *net, unsigned int num, unsigned int mask)
65 int res = (num + net_hash_mix(net)) & mask;
67 pr_debug("hash(%d) = %d\n", num, res);
70 EXPORT_SYMBOL_GPL(ping_hash);
72 static inline struct hlist_nulls_head *ping_hashslot(struct ping_table *table,
73 struct net *net, unsigned int num)
75 return &table->hash[ping_hashfn(net, num, PING_HTABLE_MASK)];
78 int ping_get_port(struct sock *sk, unsigned short ident)
80 struct hlist_nulls_node *node;
81 struct hlist_nulls_head *hlist;
82 struct inet_sock *isk, *isk2;
83 struct sock *sk2 = NULL;
86 write_lock_bh(&ping_table.lock);
89 u16 result = ping_port_rover + 1;
91 for (i = 0; i < (1L << 16); i++, result++) {
93 result++; /* avoid zero */
94 hlist = ping_hashslot(&ping_table, sock_net(sk),
96 ping_portaddr_for_each_entry(sk2, node, hlist) {
99 if (isk2->inet_num == result)
104 ping_port_rover = ident = result;
112 hlist = ping_hashslot(&ping_table, sock_net(sk), ident);
113 ping_portaddr_for_each_entry(sk2, node, hlist) {
116 /* BUG? Why is this reuse and not reuseaddr? ping.c
117 * doesn't turn off SO_REUSEADDR, and it doesn't expect
118 * that other ping processes can steal its packets.
120 if ((isk2->inet_num == ident) &&
122 (!sk2->sk_reuse || !sk->sk_reuse))
127 pr_debug("found port/ident = %d\n", ident);
128 isk->inet_num = ident;
129 if (sk_unhashed(sk)) {
130 pr_debug("was not hashed\n");
132 hlist_nulls_add_head(&sk->sk_nulls_node, hlist);
133 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
135 write_unlock_bh(&ping_table.lock);
139 write_unlock_bh(&ping_table.lock);
142 EXPORT_SYMBOL_GPL(ping_get_port);
144 void ping_hash(struct sock *sk)
146 pr_debug("ping_hash(sk->port=%u)\n", inet_sk(sk)->inet_num);
147 BUG(); /* "Please do not press this button again." */
150 void ping_unhash(struct sock *sk)
152 struct inet_sock *isk = inet_sk(sk);
153 pr_debug("ping_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
155 write_lock_bh(&ping_table.lock);
156 hlist_nulls_del(&sk->sk_nulls_node);
160 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
161 write_unlock_bh(&ping_table.lock);
164 EXPORT_SYMBOL_GPL(ping_unhash);
166 static struct sock *ping_lookup(struct net *net, struct sk_buff *skb, u16 ident)
168 struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
169 struct sock *sk = NULL;
170 struct inet_sock *isk;
171 struct hlist_nulls_node *hnode;
172 int dif = skb->dev->ifindex;
174 if (skb->protocol == htons(ETH_P_IP)) {
175 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
176 (int)ident, &ip_hdr(skb)->daddr, dif);
177 #if IS_ENABLED(CONFIG_IPV6)
178 } else if (skb->protocol == htons(ETH_P_IPV6)) {
179 pr_debug("try to find: num = %d, daddr = %pI6c, dif = %d\n",
180 (int)ident, &ipv6_hdr(skb)->daddr, dif);
184 read_lock_bh(&ping_table.lock);
186 ping_portaddr_for_each_entry(sk, hnode, hslot) {
189 pr_debug("iterate\n");
190 if (isk->inet_num != ident)
193 if (skb->protocol == htons(ETH_P_IP) &&
194 sk->sk_family == AF_INET) {
195 pr_debug("found: %p: num=%d, daddr=%pI4, dif=%d\n", sk,
196 (int) isk->inet_num, &isk->inet_rcv_saddr,
197 sk->sk_bound_dev_if);
199 if (isk->inet_rcv_saddr &&
200 isk->inet_rcv_saddr != ip_hdr(skb)->daddr)
202 #if IS_ENABLED(CONFIG_IPV6)
203 } else if (skb->protocol == htons(ETH_P_IPV6) &&
204 sk->sk_family == AF_INET6) {
205 struct ipv6_pinfo *np = inet6_sk(sk);
207 pr_debug("found: %p: num=%d, daddr=%pI6c, dif=%d\n", sk,
209 &inet6_sk(sk)->rcv_saddr,
210 sk->sk_bound_dev_if);
212 if (!ipv6_addr_any(&np->rcv_saddr) &&
213 !ipv6_addr_equal(&np->rcv_saddr,
214 &ipv6_hdr(skb)->daddr))
221 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
230 read_unlock_bh(&ping_table.lock);
235 static void inet_get_ping_group_range_net(struct net *net, kgid_t *low,
238 kgid_t *data = net->ipv4.sysctl_ping_group_range;
242 seq = read_seqbegin(&sysctl_local_ports.lock);
246 } while (read_seqretry(&sysctl_local_ports.lock, seq));
250 int ping_init_sock(struct sock *sk)
252 struct net *net = sock_net(sk);
253 kgid_t group = current_egid();
254 struct group_info *group_info;
259 if (sk->sk_family == AF_INET6)
260 inet6_sk(sk)->ipv6only = 1;
262 inet_get_ping_group_range_net(net, &low, &high);
263 if (gid_lte(low, group) && gid_lte(group, high))
266 group_info = get_current_groups();
267 count = group_info->ngroups;
268 for (i = 0; i < group_info->nblocks; i++) {
269 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
270 for (j = 0; j < cp_count; j++) {
271 kgid_t gid = group_info->blocks[i][j];
272 if (gid_lte(low, gid) && gid_lte(gid, high))
273 goto out_release_group;
282 put_group_info(group_info);
285 EXPORT_SYMBOL_GPL(ping_init_sock);
287 void ping_close(struct sock *sk, long timeout)
289 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
290 inet_sk(sk), inet_sk(sk)->inet_num);
291 pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
293 sk_common_release(sk);
295 EXPORT_SYMBOL_GPL(ping_close);
297 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
298 int ping_check_bind_addr(struct sock *sk, struct inet_sock *isk,
299 struct sockaddr *uaddr, int addr_len) {
300 struct net *net = sock_net(sk);
301 if (sk->sk_family == AF_INET) {
302 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
305 if (addr_len < sizeof(*addr))
308 if (addr->sin_family != AF_INET &&
309 !(addr->sin_family == AF_UNSPEC &&
310 addr->sin_addr.s_addr == htonl(INADDR_ANY)))
311 return -EAFNOSUPPORT;
313 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI4,port=%d)\n",
314 sk, &addr->sin_addr.s_addr, ntohs(addr->sin_port));
316 chk_addr_ret = inet_addr_type(net, addr->sin_addr.s_addr);
318 if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
319 chk_addr_ret = RTN_LOCAL;
321 if ((sysctl_ip_nonlocal_bind == 0 &&
322 isk->freebind == 0 && isk->transparent == 0 &&
323 chk_addr_ret != RTN_LOCAL) ||
324 chk_addr_ret == RTN_MULTICAST ||
325 chk_addr_ret == RTN_BROADCAST)
326 return -EADDRNOTAVAIL;
328 #if IS_ENABLED(CONFIG_IPV6)
329 } else if (sk->sk_family == AF_INET6) {
330 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) uaddr;
331 int addr_type, scoped, has_addr;
332 struct net_device *dev = NULL;
334 if (addr_len < sizeof(*addr))
337 if (addr->sin6_family != AF_INET6)
338 return -EAFNOSUPPORT;
340 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI6c,port=%d)\n",
341 sk, addr->sin6_addr.s6_addr, ntohs(addr->sin6_port));
343 addr_type = ipv6_addr_type(&addr->sin6_addr);
344 scoped = __ipv6_addr_needs_scope_id(addr_type);
345 if ((addr_type != IPV6_ADDR_ANY &&
346 !(addr_type & IPV6_ADDR_UNICAST)) ||
347 (scoped && !addr->sin6_scope_id))
351 if (addr->sin6_scope_id) {
352 dev = dev_get_by_index_rcu(net, addr->sin6_scope_id);
358 has_addr = pingv6_ops.ipv6_chk_addr(net, &addr->sin6_addr, dev,
362 if (!(isk->freebind || isk->transparent || has_addr ||
363 addr_type == IPV6_ADDR_ANY))
364 return -EADDRNOTAVAIL;
367 sk->sk_bound_dev_if = addr->sin6_scope_id;
370 return -EAFNOSUPPORT;
375 void ping_set_saddr(struct sock *sk, struct sockaddr *saddr)
377 if (saddr->sa_family == AF_INET) {
378 struct inet_sock *isk = inet_sk(sk);
379 struct sockaddr_in *addr = (struct sockaddr_in *) saddr;
380 isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
381 #if IS_ENABLED(CONFIG_IPV6)
382 } else if (saddr->sa_family == AF_INET6) {
383 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) saddr;
384 struct ipv6_pinfo *np = inet6_sk(sk);
385 np->rcv_saddr = np->saddr = addr->sin6_addr;
390 void ping_clear_saddr(struct sock *sk, int dif)
392 sk->sk_bound_dev_if = dif;
393 if (sk->sk_family == AF_INET) {
394 struct inet_sock *isk = inet_sk(sk);
395 isk->inet_rcv_saddr = isk->inet_saddr = 0;
396 #if IS_ENABLED(CONFIG_IPV6)
397 } else if (sk->sk_family == AF_INET6) {
398 struct ipv6_pinfo *np = inet6_sk(sk);
399 memset(&np->rcv_saddr, 0, sizeof(np->rcv_saddr));
400 memset(&np->saddr, 0, sizeof(np->saddr));
405 * We need our own bind because there are no privileged id's == local ports.
406 * Moreover, we don't allow binding to multi- and broadcast addresses.
409 int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
411 struct inet_sock *isk = inet_sk(sk);
414 int dif = sk->sk_bound_dev_if;
416 err = ping_check_bind_addr(sk, isk, uaddr, addr_len);
423 if (isk->inet_num != 0)
427 ping_set_saddr(sk, uaddr);
428 snum = ntohs(((struct sockaddr_in *)uaddr)->sin_port);
429 if (ping_get_port(sk, snum) != 0) {
430 ping_clear_saddr(sk, dif);
434 pr_debug("after bind(): num = %d, dif = %d\n",
436 (int)sk->sk_bound_dev_if);
439 if ((sk->sk_family == AF_INET && isk->inet_rcv_saddr) ||
440 (sk->sk_family == AF_INET6 &&
441 !ipv6_addr_any(&inet6_sk(sk)->rcv_saddr)))
442 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
445 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
446 isk->inet_sport = htons(isk->inet_num);
450 #if IS_ENABLED(CONFIG_IPV6)
451 if (sk->sk_family == AF_INET6)
452 memset(&inet6_sk(sk)->daddr, 0, sizeof(inet6_sk(sk)->daddr));
458 pr_debug("ping_v4_bind -> %d\n", err);
461 EXPORT_SYMBOL_GPL(ping_bind);
464 * Is this a supported type of ICMP message?
467 static inline int ping_supported(int family, int type, int code)
469 return (family == AF_INET && type == ICMP_ECHO && code == 0) ||
470 (family == AF_INET6 && type == ICMPV6_ECHO_REQUEST && code == 0);
474 * This routine is called by the ICMP module when it gets some
475 * sort of error condition.
478 void ping_err(struct sk_buff *skb, int offset, u32 info)
481 struct icmphdr *icmph;
482 struct inet_sock *inet_sock;
485 struct net *net = dev_net(skb->dev);
490 if (skb->protocol == htons(ETH_P_IP)) {
492 type = icmp_hdr(skb)->type;
493 code = icmp_hdr(skb)->code;
494 icmph = (struct icmphdr *)(skb->data + offset);
495 } else if (skb->protocol == htons(ETH_P_IPV6)) {
497 type = icmp6_hdr(skb)->icmp6_type;
498 code = icmp6_hdr(skb)->icmp6_code;
499 icmph = (struct icmphdr *) (skb->data + offset);
504 /* We assume the packet has already been checked by icmp_unreach */
506 if (!ping_supported(family, icmph->type, icmph->code))
509 pr_debug("ping_err(proto=0x%x,type=%d,code=%d,id=%04x,seq=%04x)\n",
510 skb->protocol, type, code, ntohs(icmph->un.echo.id),
511 ntohs(icmph->un.echo.sequence));
513 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
515 pr_debug("no socket, dropping\n");
516 return; /* No socket for error */
518 pr_debug("err on socket %p\n", sk);
522 inet_sock = inet_sk(sk);
524 if (skb->protocol == htons(ETH_P_IP)) {
527 case ICMP_TIME_EXCEEDED:
530 case ICMP_SOURCE_QUENCH:
531 /* This is not a real error but ping wants to see it.
532 * Report it with some fake errno.
536 case ICMP_PARAMETERPROB:
540 case ICMP_DEST_UNREACH:
541 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
542 ipv4_sk_update_pmtu(skb, sk, info);
543 if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
551 if (code <= NR_ICMP_UNREACH) {
552 harderr = icmp_err_convert[code].fatal;
553 err = icmp_err_convert[code].errno;
557 /* See ICMP_SOURCE_QUENCH */
558 ipv4_sk_redirect(skb, sk);
562 #if IS_ENABLED(CONFIG_IPV6)
563 } else if (skb->protocol == htons(ETH_P_IPV6)) {
564 harderr = pingv6_ops.icmpv6_err_convert(type, code, &err);
569 * RFC1122: OK. Passes ICMP errors back to application, as per
572 if ((family == AF_INET && !inet_sock->recverr) ||
573 (family == AF_INET6 && !inet6_sk(sk)->recverr)) {
574 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
577 if (family == AF_INET) {
578 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
580 #if IS_ENABLED(CONFIG_IPV6)
581 } else if (family == AF_INET6) {
582 pingv6_ops.ipv6_icmp_error(sk, skb, err, 0,
588 sk->sk_error_report(sk);
592 EXPORT_SYMBOL_GPL(ping_err);
595 * Copy and checksum an ICMP Echo packet from user space into a buffer
596 * starting from the payload.
599 int ping_getfrag(void *from, char *to,
600 int offset, int fraglen, int odd, struct sk_buff *skb)
602 struct pingfakehdr *pfh = (struct pingfakehdr *)from;
605 if (fraglen < sizeof(struct icmphdr))
607 if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
608 pfh->iov, 0, fraglen - sizeof(struct icmphdr),
611 } else if (offset < sizeof(struct icmphdr)) {
614 if (csum_partial_copy_fromiovecend
615 (to, pfh->iov, offset - sizeof(struct icmphdr),
616 fraglen, &pfh->wcheck))
620 #if IS_ENABLED(CONFIG_IPV6)
621 /* For IPv6, checksum each skb as we go along, as expected by
622 * icmpv6_push_pending_frames. For IPv4, accumulate the checksum in
623 * wcheck, it will be finalized in ping_v4_push_pending_frames.
625 if (pfh->family == AF_INET6) {
626 skb->csum = pfh->wcheck;
627 skb->ip_summed = CHECKSUM_NONE;
634 EXPORT_SYMBOL_GPL(ping_getfrag);
636 static int ping_v4_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
639 struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
641 pfh->wcheck = csum_partial((char *)&pfh->icmph,
642 sizeof(struct icmphdr), pfh->wcheck);
643 pfh->icmph.checksum = csum_fold(pfh->wcheck);
644 memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
645 skb->ip_summed = CHECKSUM_NONE;
646 return ip_push_pending_frames(sk, fl4);
649 int ping_common_sendmsg(int family, struct msghdr *msg, size_t len,
650 void *user_icmph, size_t icmph_len) {
660 /* Mirror BSD error message compatibility */
661 if (msg->msg_flags & MSG_OOB)
665 * Fetch the ICMP header provided by the userland.
666 * iovec is modified! The ICMP header is consumed.
668 if (memcpy_fromiovec(user_icmph, msg->msg_iov, icmph_len))
671 if (family == AF_INET) {
672 type = ((struct icmphdr *) user_icmph)->type;
673 code = ((struct icmphdr *) user_icmph)->code;
674 #if IS_ENABLED(CONFIG_IPV6)
675 } else if (family == AF_INET6) {
676 type = ((struct icmp6hdr *) user_icmph)->icmp6_type;
677 code = ((struct icmp6hdr *) user_icmph)->icmp6_code;
683 if (!ping_supported(family, type, code))
688 EXPORT_SYMBOL_GPL(ping_common_sendmsg);
690 int ping_v4_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
693 struct net *net = sock_net(sk);
695 struct inet_sock *inet = inet_sk(sk);
696 struct ipcm_cookie ipc;
697 struct icmphdr user_icmph;
698 struct pingfakehdr pfh;
699 struct rtable *rt = NULL;
700 struct ip_options_data opt_copy;
702 __be32 saddr, daddr, faddr;
706 pr_debug("ping_v4_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
708 err = ping_common_sendmsg(AF_INET, msg, len, &user_icmph,
714 * Get and verify the address.
718 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
719 if (msg->msg_namelen < sizeof(*usin))
721 if (usin->sin_family != AF_INET)
722 return -EAFNOSUPPORT;
723 daddr = usin->sin_addr.s_addr;
726 if (sk->sk_state != TCP_ESTABLISHED)
727 return -EDESTADDRREQ;
728 daddr = inet->inet_daddr;
732 ipc.addr = inet->inet_saddr;
734 ipc.oif = sk->sk_bound_dev_if;
737 sock_tx_timestamp(sk, &ipc.tx_flags);
739 if (msg->msg_controllen) {
740 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
747 struct ip_options_rcu *inet_opt;
750 inet_opt = rcu_dereference(inet->inet_opt);
752 memcpy(&opt_copy, inet_opt,
753 sizeof(*inet_opt) + inet_opt->opt.optlen);
754 ipc.opt = &opt_copy.opt;
760 ipc.addr = faddr = daddr;
762 if (ipc.opt && ipc.opt->opt.srr) {
765 faddr = ipc.opt->opt.faddr;
767 tos = RT_TOS(inet->tos);
768 if (sock_flag(sk, SOCK_LOCALROUTE) ||
769 (msg->msg_flags & MSG_DONTROUTE) ||
770 (ipc.opt && ipc.opt->opt.is_strictroute)) {
774 if (ipv4_is_multicast(daddr)) {
776 ipc.oif = inet->mc_index;
778 saddr = inet->mc_addr;
780 ipc.oif = inet->uc_index;
782 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
783 RT_SCOPE_UNIVERSE, sk->sk_protocol,
784 inet_sk_flowi_flags(sk), faddr, saddr, 0, 0,
787 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
788 rt = ip_route_output_flow(net, &fl4, sk);
792 if (err == -ENETUNREACH)
793 IP_INC_STATS(net, IPSTATS_MIB_OUTNOROUTES);
798 if ((rt->rt_flags & RTCF_BROADCAST) &&
799 !sock_flag(sk, SOCK_BROADCAST))
802 if (msg->msg_flags & MSG_CONFIRM)
807 ipc.addr = fl4.daddr;
811 pfh.icmph.type = user_icmph.type; /* already checked */
812 pfh.icmph.code = user_icmph.code; /* ditto */
813 pfh.icmph.checksum = 0;
814 pfh.icmph.un.echo.id = inet->inet_sport;
815 pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
816 pfh.iov = msg->msg_iov;
818 pfh.family = AF_INET;
820 err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
821 0, &ipc, &rt, msg->msg_flags);
823 ip_flush_pending_frames(sk);
825 err = ping_v4_push_pending_frames(sk, &pfh, &fl4);
833 icmp_out_count(sock_net(sk), user_icmph.type);
839 dst_confirm(&rt->dst);
840 if (!(msg->msg_flags & MSG_PROBE) || len)
841 goto back_from_confirm;
846 int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
847 size_t len, int noblock, int flags, int *addr_len)
849 struct inet_sock *isk = inet_sk(sk);
850 int family = sk->sk_family;
851 struct sockaddr_in *sin;
852 struct sockaddr_in6 *sin6;
856 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
863 if (family == AF_INET)
864 *addr_len = sizeof(*sin);
865 else if (family == AF_INET6 && addr_len)
866 *addr_len = sizeof(*sin6);
869 if (flags & MSG_ERRQUEUE) {
870 if (family == AF_INET) {
871 return ip_recv_error(sk, msg, len, addr_len);
872 #if IS_ENABLED(CONFIG_IPV6)
873 } else if (family == AF_INET6) {
874 return pingv6_ops.ipv6_recv_error(sk, msg, len);
879 skb = skb_recv_datagram(sk, flags, noblock, &err);
885 msg->msg_flags |= MSG_TRUNC;
889 /* Don't bother checking the checksum */
890 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
894 sock_recv_timestamp(msg, sk, skb);
896 /* Copy the address and add cmsg data. */
897 if (family == AF_INET) {
898 sin = (struct sockaddr_in *) msg->msg_name;
900 sin->sin_family = AF_INET;
901 sin->sin_port = 0 /* skb->h.uh->source */;
902 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
903 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
907 ip_cmsg_recv(msg, skb);
909 #if IS_ENABLED(CONFIG_IPV6)
910 } else if (family == AF_INET6) {
911 struct ipv6_pinfo *np = inet6_sk(sk);
912 struct ipv6hdr *ip6 = ipv6_hdr(skb);
913 sin6 = (struct sockaddr_in6 *) msg->msg_name;
916 sin6->sin6_family = AF_INET6;
918 sin6->sin6_addr = ip6->saddr;
919 sin6->sin6_flowinfo = 0;
921 sin6->sin6_flowinfo = ip6_flowinfo(ip6);
922 sin6->sin6_scope_id =
923 ipv6_iface_scope_id(&sin6->sin6_addr,
927 if (inet6_sk(sk)->rxopt.all)
928 pingv6_ops.ip6_datagram_recv_ctl(sk, msg, skb);
937 skb_free_datagram(sk, skb);
939 pr_debug("ping_recvmsg -> %d\n", err);
942 EXPORT_SYMBOL_GPL(ping_recvmsg);
944 int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
946 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
947 inet_sk(sk), inet_sk(sk)->inet_num, skb);
948 if (sock_queue_rcv_skb(sk, skb) < 0) {
950 pr_debug("ping_queue_rcv_skb -> failed\n");
955 EXPORT_SYMBOL_GPL(ping_queue_rcv_skb);
959 * All we need to do is get the socket.
962 void ping_rcv(struct sk_buff *skb)
965 struct net *net = dev_net(skb->dev);
966 struct icmphdr *icmph = icmp_hdr(skb);
968 /* We assume the packet has already been checked by icmp_rcv */
970 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
971 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
973 /* Push ICMP header back */
974 skb_push(skb, skb->data - (u8 *)icmph);
976 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
978 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
980 pr_debug("rcv on socket %p\n", sk);
982 ping_queue_rcv_skb(sk, skb2);
986 pr_debug("no socket, dropping\n");
988 /* We're called from icmp_rcv(). kfree_skb() is done there. */
990 EXPORT_SYMBOL_GPL(ping_rcv);
992 struct proto ping_prot = {
994 .owner = THIS_MODULE,
995 .init = ping_init_sock,
997 .connect = ip4_datagram_connect,
998 .disconnect = udp_disconnect,
999 .setsockopt = ip_setsockopt,
1000 .getsockopt = ip_getsockopt,
1001 .sendmsg = ping_v4_sendmsg,
1002 .recvmsg = ping_recvmsg,
1004 .backlog_rcv = ping_queue_rcv_skb,
1005 .release_cb = ip4_datagram_release_cb,
1007 .unhash = ping_unhash,
1008 .get_port = ping_get_port,
1009 .obj_size = sizeof(struct inet_sock),
1011 EXPORT_SYMBOL(ping_prot);
1013 #ifdef CONFIG_PROC_FS
1015 static struct sock *ping_get_first(struct seq_file *seq, int start)
1018 struct ping_iter_state *state = seq->private;
1019 struct net *net = seq_file_net(seq);
1021 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
1023 struct hlist_nulls_node *node;
1024 struct hlist_nulls_head *hslot;
1026 hslot = &ping_table.hash[state->bucket];
1028 if (hlist_nulls_empty(hslot))
1031 sk_nulls_for_each(sk, node, hslot) {
1032 if (net_eq(sock_net(sk), net))
1041 static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
1043 struct ping_iter_state *state = seq->private;
1044 struct net *net = seq_file_net(seq);
1047 sk = sk_nulls_next(sk);
1048 } while (sk && (!net_eq(sock_net(sk), net)));
1051 return ping_get_first(seq, state->bucket + 1);
1055 static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
1057 struct sock *sk = ping_get_first(seq, 0);
1060 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
1062 return pos ? NULL : sk;
1065 static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
1067 struct ping_iter_state *state = seq->private;
1070 read_lock_bh(&ping_table.lock);
1072 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
1075 static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1079 if (v == SEQ_START_TOKEN)
1080 sk = ping_get_idx(seq, 0);
1082 sk = ping_get_next(seq, v);
1088 static void ping_seq_stop(struct seq_file *seq, void *v)
1090 read_unlock_bh(&ping_table.lock);
1093 static void ping_format_sock(struct sock *sp, struct seq_file *f,
1094 int bucket, int *len)
1096 struct inet_sock *inet = inet_sk(sp);
1097 __be32 dest = inet->inet_daddr;
1098 __be32 src = inet->inet_rcv_saddr;
1099 __u16 destp = ntohs(inet->inet_dport);
1100 __u16 srcp = ntohs(inet->inet_sport);
1102 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
1103 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
1104 bucket, src, srcp, dest, destp, sp->sk_state,
1105 sk_wmem_alloc_get(sp),
1106 sk_rmem_alloc_get(sp),
1108 from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
1110 atomic_read(&sp->sk_refcnt), sp,
1111 atomic_read(&sp->sk_drops), len);
1114 static int ping_seq_show(struct seq_file *seq, void *v)
1116 if (v == SEQ_START_TOKEN)
1117 seq_printf(seq, "%-127s\n",
1118 " sl local_address rem_address st tx_queue "
1119 "rx_queue tr tm->when retrnsmt uid timeout "
1120 "inode ref pointer drops");
1122 struct ping_iter_state *state = seq->private;
1125 ping_format_sock(v, seq, state->bucket, &len);
1126 seq_printf(seq, "%*s\n", 127 - len, "");
1131 static const struct seq_operations ping_seq_ops = {
1132 .show = ping_seq_show,
1133 .start = ping_seq_start,
1134 .next = ping_seq_next,
1135 .stop = ping_seq_stop,
1138 static int ping_seq_open(struct inode *inode, struct file *file)
1140 return seq_open_net(inode, file, &ping_seq_ops,
1141 sizeof(struct ping_iter_state));
1144 static const struct file_operations ping_seq_fops = {
1145 .open = ping_seq_open,
1147 .llseek = seq_lseek,
1148 .release = seq_release_net,
1151 static int ping_proc_register(struct net *net)
1153 struct proc_dir_entry *p;
1156 p = proc_create("icmp", S_IRUGO, net->proc_net, &ping_seq_fops);
1162 static void ping_proc_unregister(struct net *net)
1164 remove_proc_entry("icmp", net->proc_net);
1168 static int __net_init ping_proc_init_net(struct net *net)
1170 return ping_proc_register(net);
1173 static void __net_exit ping_proc_exit_net(struct net *net)
1175 ping_proc_unregister(net);
1178 static struct pernet_operations ping_net_ops = {
1179 .init = ping_proc_init_net,
1180 .exit = ping_proc_exit_net,
1183 int __init ping_proc_init(void)
1185 return register_pernet_subsys(&ping_net_ops);
1188 void ping_proc_exit(void)
1190 unregister_pernet_subsys(&ping_net_ops);
1195 void __init ping_init(void)
1199 for (i = 0; i < PING_HTABLE_SIZE; i++)
1200 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
1201 rwlock_init(&ping_table.lock);