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);
158 pr_debug("ping_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
159 write_lock_bh(&ping_table.lock);
161 hlist_nulls_del(&sk->sk_nulls_node);
162 sk_nulls_node_init(&sk->sk_nulls_node);
166 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
168 write_unlock_bh(&ping_table.lock);
170 EXPORT_SYMBOL_GPL(ping_unhash);
172 static struct sock *ping_lookup(struct net *net, struct sk_buff *skb, u16 ident)
174 struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
175 struct sock *sk = NULL;
176 struct inet_sock *isk;
177 struct hlist_nulls_node *hnode;
178 int dif = skb->dev->ifindex;
180 if (skb->protocol == htons(ETH_P_IP)) {
181 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
182 (int)ident, &ip_hdr(skb)->daddr, dif);
183 #if IS_ENABLED(CONFIG_IPV6)
184 } else if (skb->protocol == htons(ETH_P_IPV6)) {
185 pr_debug("try to find: num = %d, daddr = %pI6c, dif = %d\n",
186 (int)ident, &ipv6_hdr(skb)->daddr, dif);
190 read_lock_bh(&ping_table.lock);
192 ping_portaddr_for_each_entry(sk, hnode, hslot) {
195 pr_debug("iterate\n");
196 if (isk->inet_num != ident)
199 if (skb->protocol == htons(ETH_P_IP) &&
200 sk->sk_family == AF_INET) {
201 pr_debug("found: %p: num=%d, daddr=%pI4, dif=%d\n", sk,
202 (int) isk->inet_num, &isk->inet_rcv_saddr,
203 sk->sk_bound_dev_if);
205 if (isk->inet_rcv_saddr &&
206 isk->inet_rcv_saddr != ip_hdr(skb)->daddr)
208 #if IS_ENABLED(CONFIG_IPV6)
209 } else if (skb->protocol == htons(ETH_P_IPV6) &&
210 sk->sk_family == AF_INET6) {
212 pr_debug("found: %p: num=%d, daddr=%pI6c, dif=%d\n", sk,
214 &sk->sk_v6_rcv_saddr,
215 sk->sk_bound_dev_if);
217 if (!ipv6_addr_any(&sk->sk_v6_rcv_saddr) &&
218 !ipv6_addr_equal(&sk->sk_v6_rcv_saddr,
219 &ipv6_hdr(skb)->daddr))
226 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
235 read_unlock_bh(&ping_table.lock);
240 static void inet_get_ping_group_range_net(struct net *net, kgid_t *low,
243 kgid_t *data = net->ipv4.ping_group_range.range;
247 seq = read_seqbegin(&net->ipv4.ping_group_range.lock);
251 } while (read_seqretry(&net->ipv4.ping_group_range.lock, seq));
255 int ping_init_sock(struct sock *sk)
257 struct net *net = sock_net(sk);
258 kgid_t group = current_egid();
259 struct group_info *group_info;
264 if (sk->sk_family == AF_INET6)
267 inet_get_ping_group_range_net(net, &low, &high);
268 if (gid_lte(low, group) && gid_lte(group, high))
271 group_info = get_current_groups();
272 count = group_info->ngroups;
273 for (i = 0; i < group_info->nblocks; i++) {
274 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
275 for (j = 0; j < cp_count; j++) {
276 kgid_t gid = group_info->blocks[i][j];
277 if (gid_lte(low, gid) && gid_lte(gid, high))
278 goto out_release_group;
287 put_group_info(group_info);
290 EXPORT_SYMBOL_GPL(ping_init_sock);
292 void ping_close(struct sock *sk, long timeout)
294 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
295 inet_sk(sk), inet_sk(sk)->inet_num);
296 pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
298 sk_common_release(sk);
300 EXPORT_SYMBOL_GPL(ping_close);
302 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
303 static int ping_check_bind_addr(struct sock *sk, struct inet_sock *isk,
304 struct sockaddr *uaddr, int addr_len) {
305 struct net *net = sock_net(sk);
306 if (sk->sk_family == AF_INET) {
307 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
310 if (addr_len < sizeof(*addr))
313 if (addr->sin_family != AF_INET &&
314 !(addr->sin_family == AF_UNSPEC &&
315 addr->sin_addr.s_addr == htonl(INADDR_ANY)))
316 return -EAFNOSUPPORT;
318 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI4,port=%d)\n",
319 sk, &addr->sin_addr.s_addr, ntohs(addr->sin_port));
321 chk_addr_ret = inet_addr_type(net, addr->sin_addr.s_addr);
323 if (addr->sin_addr.s_addr == htonl(INADDR_ANY))
324 chk_addr_ret = RTN_LOCAL;
326 if ((net->ipv4.sysctl_ip_nonlocal_bind == 0 &&
327 isk->freebind == 0 && isk->transparent == 0 &&
328 chk_addr_ret != RTN_LOCAL) ||
329 chk_addr_ret == RTN_MULTICAST ||
330 chk_addr_ret == RTN_BROADCAST)
331 return -EADDRNOTAVAIL;
333 #if IS_ENABLED(CONFIG_IPV6)
334 } else if (sk->sk_family == AF_INET6) {
335 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) uaddr;
336 int addr_type, scoped, has_addr;
337 struct net_device *dev = NULL;
339 if (addr_len < sizeof(*addr))
342 if (addr->sin6_family != AF_INET6)
343 return -EAFNOSUPPORT;
345 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI6c,port=%d)\n",
346 sk, addr->sin6_addr.s6_addr, ntohs(addr->sin6_port));
348 addr_type = ipv6_addr_type(&addr->sin6_addr);
349 scoped = __ipv6_addr_needs_scope_id(addr_type);
350 if ((addr_type != IPV6_ADDR_ANY &&
351 !(addr_type & IPV6_ADDR_UNICAST)) ||
352 (scoped && !addr->sin6_scope_id))
356 if (addr->sin6_scope_id) {
357 dev = dev_get_by_index_rcu(net, addr->sin6_scope_id);
363 has_addr = pingv6_ops.ipv6_chk_addr(net, &addr->sin6_addr, dev,
367 if (!(net->ipv6.sysctl.ip_nonlocal_bind ||
368 isk->freebind || isk->transparent || has_addr ||
369 addr_type == IPV6_ADDR_ANY))
370 return -EADDRNOTAVAIL;
373 sk->sk_bound_dev_if = addr->sin6_scope_id;
376 return -EAFNOSUPPORT;
381 static void ping_set_saddr(struct sock *sk, struct sockaddr *saddr)
383 if (saddr->sa_family == AF_INET) {
384 struct inet_sock *isk = inet_sk(sk);
385 struct sockaddr_in *addr = (struct sockaddr_in *) saddr;
386 isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
387 #if IS_ENABLED(CONFIG_IPV6)
388 } else if (saddr->sa_family == AF_INET6) {
389 struct sockaddr_in6 *addr = (struct sockaddr_in6 *) saddr;
390 struct ipv6_pinfo *np = inet6_sk(sk);
391 sk->sk_v6_rcv_saddr = np->saddr = addr->sin6_addr;
396 static void ping_clear_saddr(struct sock *sk, int dif)
398 sk->sk_bound_dev_if = dif;
399 if (sk->sk_family == AF_INET) {
400 struct inet_sock *isk = inet_sk(sk);
401 isk->inet_rcv_saddr = isk->inet_saddr = 0;
402 #if IS_ENABLED(CONFIG_IPV6)
403 } else if (sk->sk_family == AF_INET6) {
404 struct ipv6_pinfo *np = inet6_sk(sk);
405 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
406 memset(&np->saddr, 0, sizeof(np->saddr));
411 * We need our own bind because there are no privileged id's == local ports.
412 * Moreover, we don't allow binding to multi- and broadcast addresses.
415 int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
417 struct inet_sock *isk = inet_sk(sk);
420 int dif = sk->sk_bound_dev_if;
422 err = ping_check_bind_addr(sk, isk, uaddr, addr_len);
429 if (isk->inet_num != 0)
433 ping_set_saddr(sk, uaddr);
434 snum = ntohs(((struct sockaddr_in *)uaddr)->sin_port);
435 if (ping_get_port(sk, snum) != 0) {
436 ping_clear_saddr(sk, dif);
440 pr_debug("after bind(): num = %d, dif = %d\n",
442 (int)sk->sk_bound_dev_if);
445 if (sk->sk_family == AF_INET && isk->inet_rcv_saddr)
446 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
447 #if IS_ENABLED(CONFIG_IPV6)
448 if (sk->sk_family == AF_INET6 && !ipv6_addr_any(&sk->sk_v6_rcv_saddr))
449 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
453 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
454 isk->inet_sport = htons(isk->inet_num);
458 #if IS_ENABLED(CONFIG_IPV6)
459 if (sk->sk_family == AF_INET6)
460 memset(&sk->sk_v6_daddr, 0, sizeof(sk->sk_v6_daddr));
466 pr_debug("ping_v4_bind -> %d\n", err);
469 EXPORT_SYMBOL_GPL(ping_bind);
472 * Is this a supported type of ICMP message?
475 static inline int ping_supported(int family, int type, int code)
477 return (family == AF_INET && type == ICMP_ECHO && code == 0) ||
478 (family == AF_INET6 && type == ICMPV6_ECHO_REQUEST && code == 0);
482 * This routine is called by the ICMP module when it gets some
483 * sort of error condition.
486 void ping_err(struct sk_buff *skb, int offset, u32 info)
489 struct icmphdr *icmph;
490 struct inet_sock *inet_sock;
493 struct net *net = dev_net(skb->dev);
498 if (skb->protocol == htons(ETH_P_IP)) {
500 type = icmp_hdr(skb)->type;
501 code = icmp_hdr(skb)->code;
502 icmph = (struct icmphdr *)(skb->data + offset);
503 } else if (skb->protocol == htons(ETH_P_IPV6)) {
505 type = icmp6_hdr(skb)->icmp6_type;
506 code = icmp6_hdr(skb)->icmp6_code;
507 icmph = (struct icmphdr *) (skb->data + offset);
512 /* We assume the packet has already been checked by icmp_unreach */
514 if (!ping_supported(family, icmph->type, icmph->code))
517 pr_debug("ping_err(proto=0x%x,type=%d,code=%d,id=%04x,seq=%04x)\n",
518 skb->protocol, type, code, ntohs(icmph->un.echo.id),
519 ntohs(icmph->un.echo.sequence));
521 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
523 pr_debug("no socket, dropping\n");
524 return; /* No socket for error */
526 pr_debug("err on socket %p\n", sk);
530 inet_sock = inet_sk(sk);
532 if (skb->protocol == htons(ETH_P_IP)) {
535 case ICMP_TIME_EXCEEDED:
538 case ICMP_SOURCE_QUENCH:
539 /* This is not a real error but ping wants to see it.
540 * Report it with some fake errno.
544 case ICMP_PARAMETERPROB:
548 case ICMP_DEST_UNREACH:
549 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
550 ipv4_sk_update_pmtu(skb, sk, info);
551 if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
559 if (code <= NR_ICMP_UNREACH) {
560 harderr = icmp_err_convert[code].fatal;
561 err = icmp_err_convert[code].errno;
565 /* See ICMP_SOURCE_QUENCH */
566 ipv4_sk_redirect(skb, sk);
570 #if IS_ENABLED(CONFIG_IPV6)
571 } else if (skb->protocol == htons(ETH_P_IPV6)) {
572 harderr = pingv6_ops.icmpv6_err_convert(type, code, &err);
577 * RFC1122: OK. Passes ICMP errors back to application, as per
580 if ((family == AF_INET && !inet_sock->recverr) ||
581 (family == AF_INET6 && !inet6_sk(sk)->recverr)) {
582 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
585 if (family == AF_INET) {
586 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
588 #if IS_ENABLED(CONFIG_IPV6)
589 } else if (family == AF_INET6) {
590 pingv6_ops.ipv6_icmp_error(sk, skb, err, 0,
596 sk->sk_error_report(sk);
600 EXPORT_SYMBOL_GPL(ping_err);
603 * Copy and checksum an ICMP Echo packet from user space into a buffer
604 * starting from the payload.
607 int ping_getfrag(void *from, char *to,
608 int offset, int fraglen, int odd, struct sk_buff *skb)
610 struct pingfakehdr *pfh = (struct pingfakehdr *)from;
613 fraglen -= sizeof(struct icmphdr);
616 if (csum_and_copy_from_iter(to + sizeof(struct icmphdr),
617 fraglen, &pfh->wcheck,
618 &pfh->msg->msg_iter) != fraglen)
620 } else if (offset < sizeof(struct icmphdr)) {
623 if (csum_and_copy_from_iter(to, fraglen, &pfh->wcheck,
624 &pfh->msg->msg_iter) != fraglen)
628 #if IS_ENABLED(CONFIG_IPV6)
629 /* For IPv6, checksum each skb as we go along, as expected by
630 * icmpv6_push_pending_frames. For IPv4, accumulate the checksum in
631 * wcheck, it will be finalized in ping_v4_push_pending_frames.
633 if (pfh->family == AF_INET6) {
634 skb->csum = pfh->wcheck;
635 skb->ip_summed = CHECKSUM_NONE;
642 EXPORT_SYMBOL_GPL(ping_getfrag);
644 static int ping_v4_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh,
647 struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
651 pfh->wcheck = csum_partial((char *)&pfh->icmph,
652 sizeof(struct icmphdr), pfh->wcheck);
653 pfh->icmph.checksum = csum_fold(pfh->wcheck);
654 memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
655 skb->ip_summed = CHECKSUM_NONE;
656 return ip_push_pending_frames(sk, fl4);
659 int ping_common_sendmsg(int family, struct msghdr *msg, size_t len,
660 void *user_icmph, size_t icmph_len) {
666 /* Must have at least a full ICMP header. */
674 /* Mirror BSD error message compatibility */
675 if (msg->msg_flags & MSG_OOB)
679 * Fetch the ICMP header provided by the userland.
680 * iovec is modified! The ICMP header is consumed.
682 if (memcpy_from_msg(user_icmph, msg, icmph_len))
685 if (family == AF_INET) {
686 type = ((struct icmphdr *) user_icmph)->type;
687 code = ((struct icmphdr *) user_icmph)->code;
688 #if IS_ENABLED(CONFIG_IPV6)
689 } else if (family == AF_INET6) {
690 type = ((struct icmp6hdr *) user_icmph)->icmp6_type;
691 code = ((struct icmp6hdr *) user_icmph)->icmp6_code;
697 if (!ping_supported(family, type, code))
702 EXPORT_SYMBOL_GPL(ping_common_sendmsg);
704 static int ping_v4_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
706 struct net *net = sock_net(sk);
708 struct inet_sock *inet = inet_sk(sk);
709 struct ipcm_cookie ipc;
710 struct icmphdr user_icmph;
711 struct pingfakehdr pfh;
712 struct rtable *rt = NULL;
713 struct ip_options_data opt_copy;
715 __be32 saddr, daddr, faddr;
719 pr_debug("ping_v4_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
721 err = ping_common_sendmsg(AF_INET, msg, len, &user_icmph,
727 * Get and verify the address.
731 DECLARE_SOCKADDR(struct sockaddr_in *, usin, msg->msg_name);
732 if (msg->msg_namelen < sizeof(*usin))
734 if (usin->sin_family != AF_INET)
735 return -EAFNOSUPPORT;
736 daddr = usin->sin_addr.s_addr;
739 if (sk->sk_state != TCP_ESTABLISHED)
740 return -EDESTADDRREQ;
741 daddr = inet->inet_daddr;
745 ipc.addr = inet->inet_saddr;
747 ipc.oif = sk->sk_bound_dev_if;
752 sock_tx_timestamp(sk, &ipc.tx_flags);
754 if (msg->msg_controllen) {
755 err = ip_cmsg_send(sock_net(sk), msg, &ipc, false);
764 struct ip_options_rcu *inet_opt;
767 inet_opt = rcu_dereference(inet->inet_opt);
769 memcpy(&opt_copy, inet_opt,
770 sizeof(*inet_opt) + inet_opt->opt.optlen);
771 ipc.opt = &opt_copy.opt;
777 ipc.addr = faddr = daddr;
779 if (ipc.opt && ipc.opt->opt.srr) {
784 faddr = ipc.opt->opt.faddr;
786 tos = get_rttos(&ipc, inet);
787 if (sock_flag(sk, SOCK_LOCALROUTE) ||
788 (msg->msg_flags & MSG_DONTROUTE) ||
789 (ipc.opt && ipc.opt->opt.is_strictroute)) {
793 if (ipv4_is_multicast(daddr)) {
795 ipc.oif = inet->mc_index;
797 saddr = inet->mc_addr;
799 ipc.oif = inet->uc_index;
801 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
802 RT_SCOPE_UNIVERSE, sk->sk_protocol,
803 inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
805 fl4.fl4_icmp_type = user_icmph.type;
806 fl4.fl4_icmp_code = user_icmph.code;
808 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
809 rt = ip_route_output_flow(net, &fl4, sk);
813 if (err == -ENETUNREACH)
814 IP_INC_STATS(net, IPSTATS_MIB_OUTNOROUTES);
819 if ((rt->rt_flags & RTCF_BROADCAST) &&
820 !sock_flag(sk, SOCK_BROADCAST))
823 if (msg->msg_flags & MSG_CONFIRM)
828 ipc.addr = fl4.daddr;
832 pfh.icmph.type = user_icmph.type; /* already checked */
833 pfh.icmph.code = user_icmph.code; /* ditto */
834 pfh.icmph.checksum = 0;
835 pfh.icmph.un.echo.id = inet->inet_sport;
836 pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
839 pfh.family = AF_INET;
841 err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
842 0, &ipc, &rt, msg->msg_flags);
844 ip_flush_pending_frames(sk);
846 err = ping_v4_push_pending_frames(sk, &pfh, &fl4);
855 icmp_out_count(sock_net(sk), user_icmph.type);
861 dst_confirm(&rt->dst);
862 if (!(msg->msg_flags & MSG_PROBE) || len)
863 goto back_from_confirm;
868 int ping_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int noblock,
869 int flags, int *addr_len)
871 struct inet_sock *isk = inet_sk(sk);
872 int family = sk->sk_family;
876 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
882 if (flags & MSG_ERRQUEUE)
883 return inet_recv_error(sk, msg, len, addr_len);
885 skb = skb_recv_datagram(sk, flags, noblock, &err);
891 msg->msg_flags |= MSG_TRUNC;
895 /* Don't bother checking the checksum */
896 err = skb_copy_datagram_msg(skb, 0, msg, copied);
900 sock_recv_timestamp(msg, sk, skb);
902 /* Copy the address and add cmsg data. */
903 if (family == AF_INET) {
904 DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name);
907 sin->sin_family = AF_INET;
908 sin->sin_port = 0 /* skb->h.uh->source */;
909 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
910 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
911 *addr_len = sizeof(*sin);
915 ip_cmsg_recv(msg, skb);
917 #if IS_ENABLED(CONFIG_IPV6)
918 } else if (family == AF_INET6) {
919 struct ipv6_pinfo *np = inet6_sk(sk);
920 struct ipv6hdr *ip6 = ipv6_hdr(skb);
921 DECLARE_SOCKADDR(struct sockaddr_in6 *, sin6, msg->msg_name);
924 sin6->sin6_family = AF_INET6;
926 sin6->sin6_addr = ip6->saddr;
927 sin6->sin6_flowinfo = 0;
929 sin6->sin6_flowinfo = ip6_flowinfo(ip6);
930 sin6->sin6_scope_id =
931 ipv6_iface_scope_id(&sin6->sin6_addr,
933 *addr_len = sizeof(*sin6);
936 if (inet6_sk(sk)->rxopt.all)
937 pingv6_ops.ip6_datagram_recv_common_ctl(sk, msg, skb);
938 if (skb->protocol == htons(ETH_P_IPV6) &&
939 inet6_sk(sk)->rxopt.all)
940 pingv6_ops.ip6_datagram_recv_specific_ctl(sk, msg, skb);
941 else if (skb->protocol == htons(ETH_P_IP) && isk->cmsg_flags)
942 ip_cmsg_recv(msg, skb);
951 skb_free_datagram(sk, skb);
953 pr_debug("ping_recvmsg -> %d\n", err);
956 EXPORT_SYMBOL_GPL(ping_recvmsg);
958 int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
960 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
961 inet_sk(sk), inet_sk(sk)->inet_num, skb);
962 if (sock_queue_rcv_skb(sk, skb) < 0) {
964 pr_debug("ping_queue_rcv_skb -> failed\n");
969 EXPORT_SYMBOL_GPL(ping_queue_rcv_skb);
973 * All we need to do is get the socket.
976 bool ping_rcv(struct sk_buff *skb)
979 struct net *net = dev_net(skb->dev);
980 struct icmphdr *icmph = icmp_hdr(skb);
983 /* We assume the packet has already been checked by icmp_rcv */
985 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
986 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
988 /* Push ICMP header back */
989 skb_push(skb, skb->data - (u8 *)icmph);
991 sk = ping_lookup(net, skb, ntohs(icmph->un.echo.id));
993 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
995 pr_debug("rcv on socket %p\n", sk);
996 if (skb2 && !ping_queue_rcv_skb(sk, skb2))
1002 pr_debug("no socket, dropping\n");
1006 EXPORT_SYMBOL_GPL(ping_rcv);
1008 struct proto ping_prot = {
1010 .owner = THIS_MODULE,
1011 .init = ping_init_sock,
1012 .close = ping_close,
1013 .connect = ip4_datagram_connect,
1014 .disconnect = udp_disconnect,
1015 .setsockopt = ip_setsockopt,
1016 .getsockopt = ip_getsockopt,
1017 .sendmsg = ping_v4_sendmsg,
1018 .recvmsg = ping_recvmsg,
1020 .backlog_rcv = ping_queue_rcv_skb,
1021 .release_cb = ip4_datagram_release_cb,
1023 .unhash = ping_unhash,
1024 .get_port = ping_get_port,
1025 .obj_size = sizeof(struct inet_sock),
1027 EXPORT_SYMBOL(ping_prot);
1029 #ifdef CONFIG_PROC_FS
1031 static struct sock *ping_get_first(struct seq_file *seq, int start)
1034 struct ping_iter_state *state = seq->private;
1035 struct net *net = seq_file_net(seq);
1037 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
1039 struct hlist_nulls_node *node;
1040 struct hlist_nulls_head *hslot;
1042 hslot = &ping_table.hash[state->bucket];
1044 if (hlist_nulls_empty(hslot))
1047 sk_nulls_for_each(sk, node, hslot) {
1048 if (net_eq(sock_net(sk), net) &&
1049 sk->sk_family == state->family)
1058 static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
1060 struct ping_iter_state *state = seq->private;
1061 struct net *net = seq_file_net(seq);
1064 sk = sk_nulls_next(sk);
1065 } while (sk && (!net_eq(sock_net(sk), net)));
1068 return ping_get_first(seq, state->bucket + 1);
1072 static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
1074 struct sock *sk = ping_get_first(seq, 0);
1077 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
1079 return pos ? NULL : sk;
1082 void *ping_seq_start(struct seq_file *seq, loff_t *pos, sa_family_t family)
1084 struct ping_iter_state *state = seq->private;
1086 state->family = family;
1088 read_lock_bh(&ping_table.lock);
1090 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
1092 EXPORT_SYMBOL_GPL(ping_seq_start);
1094 static void *ping_v4_seq_start(struct seq_file *seq, loff_t *pos)
1096 return ping_seq_start(seq, pos, AF_INET);
1099 void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
1103 if (v == SEQ_START_TOKEN)
1104 sk = ping_get_idx(seq, 0);
1106 sk = ping_get_next(seq, v);
1111 EXPORT_SYMBOL_GPL(ping_seq_next);
1113 void ping_seq_stop(struct seq_file *seq, void *v)
1115 read_unlock_bh(&ping_table.lock);
1117 EXPORT_SYMBOL_GPL(ping_seq_stop);
1119 static void ping_v4_format_sock(struct sock *sp, struct seq_file *f,
1122 struct inet_sock *inet = inet_sk(sp);
1123 __be32 dest = inet->inet_daddr;
1124 __be32 src = inet->inet_rcv_saddr;
1125 __u16 destp = ntohs(inet->inet_dport);
1126 __u16 srcp = ntohs(inet->inet_sport);
1128 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
1129 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d",
1130 bucket, src, srcp, dest, destp, sp->sk_state,
1131 sk_wmem_alloc_get(sp),
1132 sk_rmem_alloc_get(sp),
1134 from_kuid_munged(seq_user_ns(f), sock_i_uid(sp)),
1136 atomic_read(&sp->sk_refcnt), sp,
1137 atomic_read(&sp->sk_drops));
1140 static int ping_v4_seq_show(struct seq_file *seq, void *v)
1142 seq_setwidth(seq, 127);
1143 if (v == SEQ_START_TOKEN)
1144 seq_puts(seq, " sl local_address rem_address st tx_queue "
1145 "rx_queue tr tm->when retrnsmt uid timeout "
1146 "inode ref pointer drops");
1148 struct ping_iter_state *state = seq->private;
1150 ping_v4_format_sock(v, seq, state->bucket);
1156 static const struct seq_operations ping_v4_seq_ops = {
1157 .show = ping_v4_seq_show,
1158 .start = ping_v4_seq_start,
1159 .next = ping_seq_next,
1160 .stop = ping_seq_stop,
1163 static int ping_seq_open(struct inode *inode, struct file *file)
1165 struct ping_seq_afinfo *afinfo = PDE_DATA(inode);
1166 return seq_open_net(inode, file, &afinfo->seq_ops,
1167 sizeof(struct ping_iter_state));
1170 const struct file_operations ping_seq_fops = {
1171 .open = ping_seq_open,
1173 .llseek = seq_lseek,
1174 .release = seq_release_net,
1176 EXPORT_SYMBOL_GPL(ping_seq_fops);
1178 static struct ping_seq_afinfo ping_v4_seq_afinfo = {
1181 .seq_fops = &ping_seq_fops,
1183 .start = ping_v4_seq_start,
1184 .show = ping_v4_seq_show,
1185 .next = ping_seq_next,
1186 .stop = ping_seq_stop,
1190 int ping_proc_register(struct net *net, struct ping_seq_afinfo *afinfo)
1192 struct proc_dir_entry *p;
1193 p = proc_create_data(afinfo->name, S_IRUGO, net->proc_net,
1194 afinfo->seq_fops, afinfo);
1199 EXPORT_SYMBOL_GPL(ping_proc_register);
1201 void ping_proc_unregister(struct net *net, struct ping_seq_afinfo *afinfo)
1203 remove_proc_entry(afinfo->name, net->proc_net);
1205 EXPORT_SYMBOL_GPL(ping_proc_unregister);
1207 static int __net_init ping_v4_proc_init_net(struct net *net)
1209 return ping_proc_register(net, &ping_v4_seq_afinfo);
1212 static void __net_exit ping_v4_proc_exit_net(struct net *net)
1214 ping_proc_unregister(net, &ping_v4_seq_afinfo);
1217 static struct pernet_operations ping_v4_net_ops = {
1218 .init = ping_v4_proc_init_net,
1219 .exit = ping_v4_proc_exit_net,
1222 int __init ping_proc_init(void)
1224 return register_pernet_subsys(&ping_v4_net_ops);
1227 void ping_proc_exit(void)
1229 unregister_pernet_subsys(&ping_v4_net_ops);
1234 void __init ping_init(void)
1238 for (i = 0; i < PING_HTABLE_SIZE; i++)
1239 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
1240 rwlock_init(&ping_table.lock);