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 int 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." */
156 void ping_unhash(struct sock *sk)
158 struct inet_sock *isk = inet_sk(sk);
160 pr_debug("ping_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
161 write_lock_bh(&ping_table.lock);
163 hlist_nulls_del(&sk->sk_nulls_node);
164 sk_nulls_node_init(&sk->sk_nulls_node);
168 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
170 write_unlock_bh(&ping_table.lock);
172 EXPORT_SYMBOL_GPL(ping_unhash);
174 static struct sock *ping_lookup(struct net *net, struct sk_buff *skb, u16 ident)
176 struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
177 struct sock *sk = NULL;
178 struct inet_sock *isk;
179 struct hlist_nulls_node *hnode;
182 if (skb->protocol == htons(ETH_P_IP)) {
184 sdif = inet_sdif(skb);
185 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
186 (int)ident, &ip_hdr(skb)->daddr, dif);
187 #if IS_ENABLED(CONFIG_IPV6)
188 } else if (skb->protocol == htons(ETH_P_IPV6)) {
189 dif = inet6_iif(skb);
190 sdif = inet6_sdif(skb);
191 pr_debug("try to find: num = %d, daddr = %pI6c, dif = %d\n",
192 (int)ident, &ipv6_hdr(skb)->daddr, dif);
198 read_lock_bh(&ping_table.lock);
200 ping_portaddr_for_each_entry(sk, hnode, hslot) {
203 pr_debug("iterate\n");
204 if (isk->inet_num != ident)
207 if (skb->protocol == htons(ETH_P_IP) &&
208 sk->sk_family == AF_INET) {
209 pr_debug("found: %p: num=%d, daddr=%pI4, dif=%d\n", sk,
210 (int) isk->inet_num, &isk->inet_rcv_saddr,
211 sk->sk_bound_dev_if);
213 if (isk->inet_rcv_saddr &&
214 isk->inet_rcv_saddr != ip_hdr(skb)->daddr)
216 #if IS_ENABLED(CONFIG_IPV6)
217 } else if (skb->protocol == htons(ETH_P_IPV6) &&
218 sk->sk_family == AF_INET6) {
220 pr_debug("found: %p: num=%d, daddr=%pI6c, dif=%d\n", sk,
222 &sk->sk_v6_rcv_saddr,
223 sk->sk_bound_dev_if);
225 if (!ipv6_addr_any(&sk->sk_v6_rcv_saddr) &&
226 !ipv6_addr_equal(&sk->sk_v6_rcv_saddr,
227 &ipv6_hdr(skb)->daddr))
234 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif &&
235 sk->sk_bound_dev_if != sdif)
244 read_unlock_bh(&ping_table.lock);
249 static void inet_get_ping_group_range_net(struct net *net, kgid_t *low,
252 kgid_t *data = net->ipv4.ping_group_range.range;
256 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
260 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
264 int ping_init_sock(struct sock *sk)
266 struct net *net = sock_net(sk);
267 kgid_t group = current_egid();
268 struct group_info *group_info;
273 if (sk->sk_family == AF_INET6)
276 inet_get_ping_group_range_net(net, &low, &high);
277 if (gid_lte(low, group) && gid_lte(group, high))
280 group_info = get_current_groups();
281 for (i = 0; i < group_info->ngroups; i++) {
282 kgid_t gid = group_info->gid[i];
284 if (gid_lte(low, gid) && gid_lte(gid, high))
285 goto out_release_group;
291 put_group_info(group_info);
294 EXPORT_SYMBOL_GPL(ping_init_sock);
296 void ping_close(struct sock *sk, long timeout)
298 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
299 inet_sk(sk), inet_sk(sk)->inet_num);
300 pr_debug("isk->refcnt = %d\n", refcount_read(&sk->sk_refcnt));
302 sk_common_release(sk);
304 EXPORT_SYMBOL_GPL(ping_close);
306 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
307 static int ping_check_bind_addr(struct sock *sk, struct inet_sock *isk,
308 struct sockaddr *uaddr, int addr_len) {
309 struct net *net = sock_net(sk);
310 if (sk->sk_family == AF_INET) {
311 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
312 u32 tb_id = RT_TABLE_LOCAL;
315 if (addr_len < sizeof(*addr))
318 if (addr->sin_family != AF_INET &&
319 !(addr->sin_family == AF_UNSPEC &&
320 addr->sin_addr.s_addr == htonl(INADDR_ANY)))
321 return -EAFNOSUPPORT;
323 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI4,port=%d)\n",
324 sk, &addr->sin_addr.s_addr, ntohs(addr->sin_port));
326 tb_id = l3mdev_fib_table_by_index(net, sk->sk_bound_dev_if) ? : tb_id;
327 chk_addr_ret = inet_addr_type_table(net, addr->sin_addr.s_addr, tb_id);
329 if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
330 chk_addr_ret = RTN_LOCAL;
332 if ((net->ipv4.sysctl_ip_nonlocal_bind == 0 &&
333 isk->freebind == 0 && isk->transparent == 0 &&
334 chk_addr_ret != RTN_LOCAL) ||
335 chk_addr_ret == RTN_MULTICAST ||
336 chk_addr_ret == RTN_BROADCAST)
337 return -EADDRNOTAVAIL;
339 #if IS_ENABLED(CONFIG_IPV6)
340 } else if (sk->sk_family == AF_INET6) {
341 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) uaddr;
342 int addr_type, scoped, has_addr;
343 struct net_device *dev = NULL;
345 if (addr_len < sizeof(*addr))
348 if (addr->sin6_family != AF_INET6)
349 return -EAFNOSUPPORT;
351 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI6c,port=%d)\n",
352 sk, addr->sin6_addr.s6_addr, ntohs(addr->sin6_port));
354 addr_type = ipv6_addr_type(&addr->sin6_addr);
355 scoped = __ipv6_addr_needs_scope_id(addr_type);
356 if ((addr_type != IPV6_ADDR_ANY &&
357 !(addr_type & IPV6_ADDR_UNICAST)) ||
358 (scoped && !addr->sin6_scope_id))
362 if (addr->sin6_scope_id) {
363 dev = dev_get_by_index_rcu(net, addr->sin6_scope_id);
370 if (!dev && sk->sk_bound_dev_if) {
371 dev = dev_get_by_index_rcu(net, sk->sk_bound_dev_if);
377 has_addr = pingv6_ops.ipv6_chk_addr(net, &addr->sin6_addr, dev,
381 if (!(net->ipv6.sysctl.ip_nonlocal_bind ||
382 isk->freebind || isk->transparent || has_addr ||
383 addr_type == IPV6_ADDR_ANY))
384 return -EADDRNOTAVAIL;
387 sk->sk_bound_dev_if = addr->sin6_scope_id;
390 return -EAFNOSUPPORT;
395 static void ping_set_saddr(struct sock *sk, struct sockaddr *saddr)
397 if (saddr->sa_family == AF_INET) {
398 struct inet_sock *isk = inet_sk(sk);
399 struct sockaddr_in *addr = (struct sockaddr_in *) saddr;
400 isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
401 #if IS_ENABLED(CONFIG_IPV6)
402 } else if (saddr->sa_family == AF_INET6) {
403 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) saddr;
404 struct ipv6_pinfo *np = inet6_sk(sk);
405 sk->sk_v6_rcv_saddr = np->saddr = addr->sin6_addr;
410 static void ping_clear_saddr(struct sock *sk, int dif)
412 sk->sk_bound_dev_if = dif;
413 if (sk->sk_family == AF_INET) {
414 struct inet_sock *isk = inet_sk(sk);
415 isk->inet_rcv_saddr = isk->inet_saddr = 0;
416 #if IS_ENABLED(CONFIG_IPV6)
417 } else if (sk->sk_family == AF_INET6) {
418 struct ipv6_pinfo *np = inet6_sk(sk);
419 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
420 memset(&np->saddr, 0, sizeof(np->saddr));
425 * We need our own bind because there are no privileged id's == local ports.
426 * Moreover, we don't allow binding to multi- and broadcast addresses.
429 int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
431 struct inet_sock *isk = inet_sk(sk);
434 int dif = sk->sk_bound_dev_if;
436 err = ping_check_bind_addr(sk, isk, uaddr, addr_len);
443 if (isk->inet_num != 0)
447 ping_set_saddr(sk, uaddr);
448 snum = ntohs(((struct sockaddr_in *)uaddr)->sin_port);
449 if (ping_get_port(sk, snum) != 0) {
450 ping_clear_saddr(sk, dif);
454 pr_debug("after bind(): num = %hu, dif = %d\n",
456 sk->sk_bound_dev_if);
459 if (sk->sk_family == AF_INET && isk->inet_rcv_saddr)
460 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
461 #if IS_ENABLED(CONFIG_IPV6)
462 if (sk->sk_family == AF_INET6 && !ipv6_addr_any(&sk->sk_v6_rcv_saddr))
463 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
467 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
468 isk->inet_sport = htons(isk->inet_num);
472 #if IS_ENABLED(CONFIG_IPV6)
473 if (sk->sk_family == AF_INET6)
474 memset(&sk->sk_v6_daddr, 0, sizeof(sk->sk_v6_daddr));
480 pr_debug("ping_v4_bind -> %d\n", err);
483 EXPORT_SYMBOL_GPL(ping_bind);
486 * Is this a supported type of ICMP message?
489 static inline int ping_supported(int family, int type, int code)
491 return (family == AF_INET && type == ICMP_ECHO && code == 0) ||
492 (family == AF_INET6 && type == ICMPV6_ECHO_REQUEST && code == 0);
496 * This routine is called by the ICMP module when it gets some
497 * sort of error condition.
500 void ping_err(struct sk_buff *skb, int offset, u32 info)
503 struct icmphdr *icmph;
504 struct inet_sock *inet_sock;
507 struct net *net = dev_net(skb->dev);
512 if (skb->protocol == htons(ETH_P_IP)) {
514 type = icmp_hdr(skb)->type;
515 code = icmp_hdr(skb)->code;
516 icmph = (struct icmphdr *)(skb->data + offset);
517 } else if (skb->protocol == htons(ETH_P_IPV6)) {
519 type = icmp6_hdr(skb)->icmp6_type;
520 code = icmp6_hdr(skb)->icmp6_code;
521 icmph = (struct icmphdr *) (skb->data + offset);
526 /* We assume the packet has already been checked by icmp_unreach */
528 if (!ping_supported(family, icmph->type, icmph->code))
531 pr_debug("ping_err(proto=0x%x,type=%d,code=%d,id=%04x,seq=%04x)\n",
532 skb->protocol, type, code, ntohs(icmph->un.echo.id),
533 ntohs(icmph->un.echo.sequence));
535 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
537 pr_debug("no socket, dropping\n");
538 return; /* No socket for error */
540 pr_debug("err on socket %p\n", sk);
544 inet_sock = inet_sk(sk);
546 if (skb->protocol == htons(ETH_P_IP)) {
549 case ICMP_TIME_EXCEEDED:
552 case ICMP_SOURCE_QUENCH:
553 /* This is not a real error but ping wants to see it.
554 * Report it with some fake errno.
558 case ICMP_PARAMETERPROB:
562 case ICMP_DEST_UNREACH:
563 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
564 ipv4_sk_update_pmtu(skb, sk, info);
565 if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
573 if (code <= NR_ICMP_UNREACH) {
574 harderr = icmp_err_convert[code].fatal;
575 err = icmp_err_convert[code].errno;
579 /* See ICMP_SOURCE_QUENCH */
580 ipv4_sk_redirect(skb, sk);
584 #if IS_ENABLED(CONFIG_IPV6)
585 } else if (skb->protocol == htons(ETH_P_IPV6)) {
586 harderr = pingv6_ops.icmpv6_err_convert(type, code, &err);
591 * RFC1122: OK. Passes ICMP errors back to application, as per
594 if ((family == AF_INET && !inet_sock->recverr) ||
595 (family == AF_INET6 && !inet6_sk(sk)->recverr)) {
596 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
599 if (family == AF_INET) {
600 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
602 #if IS_ENABLED(CONFIG_IPV6)
603 } else if (family == AF_INET6) {
604 pingv6_ops.ipv6_icmp_error(sk, skb, err, 0,
610 sk->sk_error_report(sk);
614 EXPORT_SYMBOL_GPL(ping_err);
617 * Copy and checksum an ICMP Echo packet from user space into a buffer
618 * starting from the payload.
621 int ping_getfrag(void *from, char *to,
622 int offset, int fraglen, int odd, struct sk_buff *skb)
624 struct pingfakehdr *pfh = (struct pingfakehdr *)from;
627 fraglen -= sizeof(struct icmphdr);
630 if (!csum_and_copy_from_iter_full(to + sizeof(struct icmphdr),
631 fraglen, &pfh->wcheck,
632 &pfh->msg->msg_iter))
634 } else if (offset < sizeof(struct icmphdr)) {
637 if (!csum_and_copy_from_iter_full(to, fraglen, &pfh->wcheck,
638 &pfh->msg->msg_iter))
642 #if IS_ENABLED(CONFIG_IPV6)
643 /* For IPv6, checksum each skb as we go along, as expected by
644 * icmpv6_push_pending_frames. For IPv4, accumulate the checksum in
645 * wcheck, it will be finalized in ping_v4_push_pending_frames.
647 if (pfh->family == AF_INET6) {
648 skb->csum = pfh->wcheck;
649 skb->ip_summed = CHECKSUM_NONE;
656 EXPORT_SYMBOL_GPL(ping_getfrag);
658 static int ping_v4_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
661 struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
665 pfh->wcheck = csum_partial((char *)&pfh->icmph,
666 sizeof(struct icmphdr), pfh->wcheck);
667 pfh->icmph.checksum = csum_fold(pfh->wcheck);
668 memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
669 skb->ip_summed = CHECKSUM_NONE;
670 return ip_push_pending_frames(sk, fl4);
673 int ping_common_sendmsg(int family, struct msghdr *msg, size_t len,
674 void *user_icmph, size_t icmph_len) {
680 /* Must have at least a full ICMP header. */
688 /* Mirror BSD error message compatibility */
689 if (msg->msg_flags & MSG_OOB)
693 * Fetch the ICMP header provided by the userland.
694 * iovec is modified! The ICMP header is consumed.
696 if (memcpy_from_msg(user_icmph, msg, icmph_len))
699 if (family == AF_INET) {
700 type = ((struct icmphdr *) user_icmph)->type;
701 code = ((struct icmphdr *) user_icmph)->code;
702 #if IS_ENABLED(CONFIG_IPV6)
703 } else if (family == AF_INET6) {
704 type = ((struct icmp6hdr *) user_icmph)->icmp6_type;
705 code = ((struct icmp6hdr *) user_icmph)->icmp6_code;
711 if (!ping_supported(family, type, code))
716 EXPORT_SYMBOL_GPL(ping_common_sendmsg);
718 static int ping_v4_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
720 struct net *net = sock_net(sk);
722 struct inet_sock *inet = inet_sk(sk);
723 struct ipcm_cookie ipc;
724 struct icmphdr user_icmph;
725 struct pingfakehdr pfh;
726 struct rtable *rt = NULL;
727 struct ip_options_data opt_copy;
729 __be32 saddr, daddr, faddr;
733 pr_debug("ping_v4_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
735 err = ping_common_sendmsg(AF_INET, msg, len, &user_icmph,
741 * Get and verify the address.
745 DECLARE_SOCKADDR(struct sockaddr_in *, usin, msg->msg_name);
746 if (msg->msg_namelen < sizeof(*usin))
748 if (usin->sin_family != AF_INET)
749 return -EAFNOSUPPORT;
750 daddr = usin->sin_addr.s_addr;
753 if (sk->sk_state != TCP_ESTABLISHED)
754 return -EDESTADDRREQ;
755 daddr = inet->inet_daddr;
759 ipc.sockc.tsflags = sk->sk_tsflags;
760 ipc.addr = inet->inet_saddr;
762 ipc.oif = sk->sk_bound_dev_if;
767 if (msg->msg_controllen) {
768 err = ip_cmsg_send(sk, msg, &ipc, false);
777 struct ip_options_rcu *inet_opt;
780 inet_opt = rcu_dereference(inet->inet_opt);
782 memcpy(&opt_copy, inet_opt,
783 sizeof(*inet_opt) + inet_opt->opt.optlen);
784 ipc.opt = &opt_copy.opt;
789 sock_tx_timestamp(sk, ipc.sockc.tsflags, &ipc.tx_flags);
792 ipc.addr = faddr = daddr;
794 if (ipc.opt && ipc.opt->opt.srr) {
799 faddr = ipc.opt->opt.faddr;
801 tos = get_rttos(&ipc, inet);
802 if (sock_flag(sk, SOCK_LOCALROUTE) ||
803 (msg->msg_flags & MSG_DONTROUTE) ||
804 (ipc.opt && ipc.opt->opt.is_strictroute)) {
808 if (ipv4_is_multicast(daddr)) {
810 ipc.oif = inet->mc_index;
812 saddr = inet->mc_addr;
814 ipc.oif = inet->uc_index;
816 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
817 RT_SCOPE_UNIVERSE, sk->sk_protocol,
818 inet_sk_flowi_flags(sk), faddr, saddr, 0, 0,
821 fl4.fl4_icmp_type = user_icmph.type;
822 fl4.fl4_icmp_code = user_icmph.code;
824 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
825 rt = ip_route_output_flow(net, &fl4, sk);
829 if (err == -ENETUNREACH)
830 IP_INC_STATS(net, IPSTATS_MIB_OUTNOROUTES);
835 if ((rt->rt_flags & RTCF_BROADCAST) &&
836 !sock_flag(sk, SOCK_BROADCAST))
839 if (msg->msg_flags & MSG_CONFIRM)
844 ipc.addr = fl4.daddr;
848 pfh.icmph.type = user_icmph.type; /* already checked */
849 pfh.icmph.code = user_icmph.code; /* ditto */
850 pfh.icmph.checksum = 0;
851 pfh.icmph.un.echo.id = inet->inet_sport;
852 pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
855 pfh.family = AF_INET;
857 err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
858 0, &ipc, &rt, msg->msg_flags);
860 ip_flush_pending_frames(sk);
862 err = ping_v4_push_pending_frames(sk, &pfh, &fl4);
871 icmp_out_count(sock_net(sk), user_icmph.type);
877 if (msg->msg_flags & MSG_PROBE)
878 dst_confirm_neigh(&rt->dst, &fl4.daddr);
879 if (!(msg->msg_flags & MSG_PROBE) || len)
880 goto back_from_confirm;
885 int ping_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int noblock,
886 int flags, int *addr_len)
888 struct inet_sock *isk = inet_sk(sk);
889 int family = sk->sk_family;
893 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
899 if (flags & MSG_ERRQUEUE)
900 return inet_recv_error(sk, msg, len, addr_len);
902 skb = skb_recv_datagram(sk, flags, noblock, &err);
908 msg->msg_flags |= MSG_TRUNC;
912 /* Don't bother checking the checksum */
913 err = skb_copy_datagram_msg(skb, 0, msg, copied);
917 sock_recv_timestamp(msg, sk, skb);
919 /* Copy the address and add cmsg data. */
920 if (family == AF_INET) {
921 DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name);
924 sin->sin_family = AF_INET;
925 sin->sin_port = 0 /* skb->h.uh->source */;
926 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
927 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
928 *addr_len = sizeof(*sin);
932 ip_cmsg_recv(msg, skb);
934 #if IS_ENABLED(CONFIG_IPV6)
935 } else if (family == AF_INET6) {
936 struct ipv6_pinfo *np = inet6_sk(sk);
937 struct ipv6hdr *ip6 = ipv6_hdr(skb);
938 DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
941 sin6->sin6_family = AF_INET6;
943 sin6->sin6_addr = ip6->saddr;
944 sin6->sin6_flowinfo = 0;
946 sin6->sin6_flowinfo = ip6_flowinfo(ip6);
947 sin6->sin6_scope_id =
948 ipv6_iface_scope_id(&sin6->sin6_addr,
950 *addr_len = sizeof(*sin6);
953 if (inet6_sk(sk)->rxopt.all)
954 pingv6_ops.ip6_datagram_recv_common_ctl(sk, msg, skb);
955 if (skb->protocol == htons(ETH_P_IPV6) &&
956 inet6_sk(sk)->rxopt.all)
957 pingv6_ops.ip6_datagram_recv_specific_ctl(sk, msg, skb);
958 else if (skb->protocol == htons(ETH_P_IP) && isk->cmsg_flags)
959 ip_cmsg_recv(msg, skb);
968 skb_free_datagram(sk, skb);
970 pr_debug("ping_recvmsg -> %d\n", err);
973 EXPORT_SYMBOL_GPL(ping_recvmsg);
975 int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
977 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
978 inet_sk(sk), inet_sk(sk)->inet_num, skb);
979 if (sock_queue_rcv_skb(sk, skb) < 0) {
981 pr_debug("ping_queue_rcv_skb -> failed\n");
986 EXPORT_SYMBOL_GPL(ping_queue_rcv_skb);
990 * All we need to do is get the socket.
993 bool ping_rcv(struct sk_buff *skb)
996 struct net *net = dev_net(skb->dev);
997 struct icmphdr *icmph = icmp_hdr(skb);
1000 /* We assume the packet has already been checked by icmp_rcv */
1002 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
1003 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
1005 /* Push ICMP header back */
1006 skb_push(skb, skb->data - (u8 *)icmph);
1008 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
1010 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
1012 pr_debug("rcv on socket %p\n", sk);
1013 if (skb2 && !ping_queue_rcv_skb(sk, skb2))
1019 pr_debug("no socket, dropping\n");
1023 EXPORT_SYMBOL_GPL(ping_rcv);
1025 struct proto ping_prot = {
1027 .owner = THIS_MODULE,
1028 .init = ping_init_sock,
1029 .close = ping_close,
1030 .connect = ip4_datagram_connect,
1031 .disconnect = __udp_disconnect,
1032 .setsockopt = ip_setsockopt,
1033 .getsockopt = ip_getsockopt,
1034 .sendmsg = ping_v4_sendmsg,
1035 .recvmsg = ping_recvmsg,
1037 .backlog_rcv = ping_queue_rcv_skb,
1038 .release_cb = ip4_datagram_release_cb,
1040 .unhash = ping_unhash,
1041 .get_port = ping_get_port,
1042 .obj_size = sizeof(struct inet_sock),
1044 EXPORT_SYMBOL(ping_prot);
1046 #ifdef CONFIG_PROC_FS
1048 static struct sock *ping_get_first(struct seq_file *seq, int start)
1051 struct ping_iter_state *state = seq->private;
1052 struct net *net = seq_file_net(seq);
1054 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
1056 struct hlist_nulls_node *node;
1057 struct hlist_nulls_head *hslot;
1059 hslot = &ping_table.hash[state->bucket];
1061 if (hlist_nulls_empty(hslot))
1064 sk_nulls_for_each(sk, node, hslot) {
1065 if (net_eq(sock_net(sk), net) &&
1066 sk->sk_family == state->family)
1075 static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
1077 struct ping_iter_state *state = seq->private;
1078 struct net *net = seq_file_net(seq);
1081 sk = sk_nulls_next(sk);
1082 } while (sk && (!net_eq(sock_net(sk), net)));
1085 return ping_get_first(seq, state->bucket + 1);
1089 static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
1091 struct sock *sk = ping_get_first(seq, 0);
1094 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
1096 return pos ? NULL : sk;
1099 void *ping_seq_start(struct seq_file *seq, loff_t *pos, sa_family_t family)
1100 __acquires(ping_table.lock)
1102 struct ping_iter_state *state = seq->private;
1104 state->family = family;
1106 read_lock_bh(&ping_table.lock);
1108 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
1110 EXPORT_SYMBOL_GPL(ping_seq_start);
1112 static void *ping_v4_seq_start(struct seq_file *seq, loff_t *pos)
1114 return ping_seq_start(seq, pos, AF_INET);
1117 void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1121 if (v == SEQ_START_TOKEN)
1122 sk = ping_get_idx(seq, 0);
1124 sk = ping_get_next(seq, v);
1129 EXPORT_SYMBOL_GPL(ping_seq_next);
1131 void ping_seq_stop(struct seq_file *seq, void *v)
1132 __releases(ping_table.lock)
1134 read_unlock_bh(&ping_table.lock);
1136 EXPORT_SYMBOL_GPL(ping_seq_stop);
1138 static void ping_v4_format_sock(struct sock *sp, struct seq_file *f,
1141 struct inet_sock *inet = inet_sk(sp);
1142 __be32 dest = inet->inet_daddr;
1143 __be32 src = inet->inet_rcv_saddr;
1144 __u16 destp = ntohs(inet->inet_dport);
1145 __u16 srcp = ntohs(inet->inet_sport);
1147 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
1148 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d",
1149 bucket, src, srcp, dest, destp, sp->sk_state,
1150 sk_wmem_alloc_get(sp),
1151 sk_rmem_alloc_get(sp),
1153 from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
1155 refcount_read(&sp->sk_refcnt), sp,
1156 atomic_read(&sp->sk_drops));
1159 static int ping_v4_seq_show(struct seq_file *seq, void *v)
1161 seq_setwidth(seq, 127);
1162 if (v == SEQ_START_TOKEN)
1163 seq_puts(seq, " sl local_address rem_address st tx_queue "
1164 "rx_queue tr tm->when retrnsmt uid timeout "
1165 "inode ref pointer drops");
1167 struct ping_iter_state *state = seq->private;
1169 ping_v4_format_sock(v, seq, state->bucket);
1175 static int ping_seq_open(struct inode *inode, struct file *file)
1177 struct ping_seq_afinfo *afinfo = PDE_DATA(inode);
1178 return seq_open_net(inode, file, &afinfo->seq_ops,
1179 sizeof(struct ping_iter_state));
1182 const struct file_operations ping_seq_fops = {
1183 .open = ping_seq_open,
1185 .llseek = seq_lseek,
1186 .release = seq_release_net,
1188 EXPORT_SYMBOL_GPL(ping_seq_fops);
1190 static struct ping_seq_afinfo ping_v4_seq_afinfo = {
1193 .seq_fops = &ping_seq_fops,
1195 .start = ping_v4_seq_start,
1196 .show = ping_v4_seq_show,
1197 .next = ping_seq_next,
1198 .stop = ping_seq_stop,
1202 int ping_proc_register(struct net *net, struct ping_seq_afinfo *afinfo)
1204 struct proc_dir_entry *p;
1205 p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
1206 afinfo->seq_fops, afinfo);
1211 EXPORT_SYMBOL_GPL(ping_proc_register);
1213 void ping_proc_unregister(struct net *net, struct ping_seq_afinfo *afinfo)
1215 remove_proc_entry(afinfo->name, net->proc_net);
1217 EXPORT_SYMBOL_GPL(ping_proc_unregister);
1219 static int __net_init ping_v4_proc_init_net(struct net *net)
1221 return ping_proc_register(net, &ping_v4_seq_afinfo);
1224 static void __net_exit ping_v4_proc_exit_net(struct net *net)
1226 ping_proc_unregister(net, &ping_v4_seq_afinfo);
1229 static struct pernet_operations ping_v4_net_ops = {
1230 .init = ping_v4_proc_init_net,
1231 .exit = ping_v4_proc_exit_net,
1234 int __init ping_proc_init(void)
1236 return register_pernet_subsys(&ping_v4_net_ops);
1239 void ping_proc_exit(void)
1241 unregister_pernet_subsys(&ping_v4_net_ops);
1246 void __init ping_init(void)
1250 for (i = 0; i < PING_HTABLE_SIZE; i++)
1251 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
1252 rwlock_init(&ping_table.lock);