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
);
159 pr_debug("ping_unhash(isk=%p,isk->num=%u)\n", isk
, isk
->inet_num
);
161 write_lock_bh(&ping_table
.lock
);
162 hlist_nulls_del(&sk
->sk_nulls_node
);
163 sk_nulls_node_init(&sk
->sk_nulls_node
);
167 sock_prot_inuse_add(sock_net(sk
), sk
->sk_prot
, -1);
168 write_unlock_bh(&ping_table
.lock
);
171 EXPORT_SYMBOL_GPL(ping_unhash
);
173 static struct sock
*ping_lookup(struct net
*net
, struct sk_buff
*skb
, u16 ident
)
175 struct hlist_nulls_head
*hslot
= ping_hashslot(&ping_table
, net
, ident
);
176 struct sock
*sk
= NULL
;
177 struct inet_sock
*isk
;
178 struct hlist_nulls_node
*hnode
;
179 int dif
= skb
->dev
->ifindex
;
181 if (skb
->protocol
== htons(ETH_P_IP
)) {
182 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
183 (int)ident
, &ip_hdr(skb
)->daddr
, dif
);
184 #if IS_ENABLED(CONFIG_IPV6)
185 } else if (skb
->protocol
== htons(ETH_P_IPV6
)) {
186 pr_debug("try to find: num = %d, daddr = %pI6c, dif = %d\n",
187 (int)ident
, &ipv6_hdr(skb
)->daddr
, dif
);
191 read_lock_bh(&ping_table
.lock
);
193 ping_portaddr_for_each_entry(sk
, hnode
, hslot
) {
196 pr_debug("iterate\n");
197 if (isk
->inet_num
!= ident
)
200 if (skb
->protocol
== htons(ETH_P_IP
) &&
201 sk
->sk_family
== AF_INET
) {
202 pr_debug("found: %p: num=%d, daddr=%pI4, dif=%d\n", sk
,
203 (int) isk
->inet_num
, &isk
->inet_rcv_saddr
,
204 sk
->sk_bound_dev_if
);
206 if (isk
->inet_rcv_saddr
&&
207 isk
->inet_rcv_saddr
!= ip_hdr(skb
)->daddr
)
209 #if IS_ENABLED(CONFIG_IPV6)
210 } else if (skb
->protocol
== htons(ETH_P_IPV6
) &&
211 sk
->sk_family
== AF_INET6
) {
213 pr_debug("found: %p: num=%d, daddr=%pI6c, dif=%d\n", sk
,
215 &sk
->sk_v6_rcv_saddr
,
216 sk
->sk_bound_dev_if
);
218 if (!ipv6_addr_any(&sk
->sk_v6_rcv_saddr
) &&
219 !ipv6_addr_equal(&sk
->sk_v6_rcv_saddr
,
220 &ipv6_hdr(skb
)->daddr
))
227 if (sk
->sk_bound_dev_if
&& sk
->sk_bound_dev_if
!= dif
)
236 read_unlock_bh(&ping_table
.lock
);
241 static void inet_get_ping_group_range_net(struct net
*net
, kgid_t
*low
,
244 kgid_t
*data
= net
->ipv4
.ping_group_range
.range
;
248 seq
= read_seqbegin(&net
->ipv4
.ping_group_range
.lock
);
252 } while (read_seqretry(&net
->ipv4
.ping_group_range
.lock
, seq
));
256 int ping_init_sock(struct sock
*sk
)
258 struct net
*net
= sock_net(sk
);
259 kgid_t group
= current_egid();
260 struct group_info
*group_info
;
265 if (sk
->sk_family
== AF_INET6
)
268 inet_get_ping_group_range_net(net
, &low
, &high
);
269 if (gid_lte(low
, group
) && gid_lte(group
, high
))
272 group_info
= get_current_groups();
273 count
= group_info
->ngroups
;
274 for (i
= 0; i
< group_info
->nblocks
; i
++) {
275 int cp_count
= min_t(int, NGROUPS_PER_BLOCK
, count
);
276 for (j
= 0; j
< cp_count
; j
++) {
277 kgid_t gid
= group_info
->blocks
[i
][j
];
278 if (gid_lte(low
, gid
) && gid_lte(gid
, high
))
279 goto out_release_group
;
288 put_group_info(group_info
);
291 EXPORT_SYMBOL_GPL(ping_init_sock
);
293 void ping_close(struct sock
*sk
, long timeout
)
295 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
296 inet_sk(sk
), inet_sk(sk
)->inet_num
);
297 pr_debug("isk->refcnt = %d\n", sk
->sk_refcnt
.counter
);
299 sk_common_release(sk
);
301 EXPORT_SYMBOL_GPL(ping_close
);
303 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
304 static int ping_check_bind_addr(struct sock
*sk
, struct inet_sock
*isk
,
305 struct sockaddr
*uaddr
, int addr_len
) {
306 struct net
*net
= sock_net(sk
);
307 if (sk
->sk_family
== AF_INET
) {
308 struct sockaddr_in
*addr
= (struct sockaddr_in
*) uaddr
;
311 if (addr_len
< sizeof(*addr
))
314 if (addr
->sin_family
!= AF_INET
&&
315 !(addr
->sin_family
== AF_UNSPEC
&&
316 addr
->sin_addr
.s_addr
== htonl(INADDR_ANY
)))
317 return -EAFNOSUPPORT
;
319 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI4,port=%d)\n",
320 sk
, &addr
->sin_addr
.s_addr
, ntohs(addr
->sin_port
));
322 chk_addr_ret
= inet_addr_type(net
, addr
->sin_addr
.s_addr
);
324 if (addr
->sin_addr
.s_addr
== htonl(INADDR_ANY
))
325 chk_addr_ret
= RTN_LOCAL
;
327 if ((net
->ipv4
.sysctl_ip_nonlocal_bind
== 0 &&
328 isk
->freebind
== 0 && isk
->transparent
== 0 &&
329 chk_addr_ret
!= RTN_LOCAL
) ||
330 chk_addr_ret
== RTN_MULTICAST
||
331 chk_addr_ret
== RTN_BROADCAST
)
332 return -EADDRNOTAVAIL
;
334 #if IS_ENABLED(CONFIG_IPV6)
335 } else if (sk
->sk_family
== AF_INET6
) {
336 struct sockaddr_in6
*addr
= (struct sockaddr_in6
*) uaddr
;
337 int addr_type
, scoped
, has_addr
;
338 struct net_device
*dev
= NULL
;
340 if (addr_len
< sizeof(*addr
))
343 if (addr
->sin6_family
!= AF_INET6
)
344 return -EAFNOSUPPORT
;
346 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI6c,port=%d)\n",
347 sk
, addr
->sin6_addr
.s6_addr
, ntohs(addr
->sin6_port
));
349 addr_type
= ipv6_addr_type(&addr
->sin6_addr
);
350 scoped
= __ipv6_addr_needs_scope_id(addr_type
);
351 if ((addr_type
!= IPV6_ADDR_ANY
&&
352 !(addr_type
& IPV6_ADDR_UNICAST
)) ||
353 (scoped
&& !addr
->sin6_scope_id
))
357 if (addr
->sin6_scope_id
) {
358 dev
= dev_get_by_index_rcu(net
, addr
->sin6_scope_id
);
364 has_addr
= pingv6_ops
.ipv6_chk_addr(net
, &addr
->sin6_addr
, dev
,
368 if (!(net
->ipv6
.sysctl
.ip_nonlocal_bind
||
369 isk
->freebind
|| isk
->transparent
|| has_addr
||
370 addr_type
== IPV6_ADDR_ANY
))
371 return -EADDRNOTAVAIL
;
374 sk
->sk_bound_dev_if
= addr
->sin6_scope_id
;
377 return -EAFNOSUPPORT
;
382 static void ping_set_saddr(struct sock
*sk
, struct sockaddr
*saddr
)
384 if (saddr
->sa_family
== AF_INET
) {
385 struct inet_sock
*isk
= inet_sk(sk
);
386 struct sockaddr_in
*addr
= (struct sockaddr_in
*) saddr
;
387 isk
->inet_rcv_saddr
= isk
->inet_saddr
= addr
->sin_addr
.s_addr
;
388 #if IS_ENABLED(CONFIG_IPV6)
389 } else if (saddr
->sa_family
== AF_INET6
) {
390 struct sockaddr_in6
*addr
= (struct sockaddr_in6
*) saddr
;
391 struct ipv6_pinfo
*np
= inet6_sk(sk
);
392 sk
->sk_v6_rcv_saddr
= np
->saddr
= addr
->sin6_addr
;
397 static void ping_clear_saddr(struct sock
*sk
, int dif
)
399 sk
->sk_bound_dev_if
= dif
;
400 if (sk
->sk_family
== AF_INET
) {
401 struct inet_sock
*isk
= inet_sk(sk
);
402 isk
->inet_rcv_saddr
= isk
->inet_saddr
= 0;
403 #if IS_ENABLED(CONFIG_IPV6)
404 } else if (sk
->sk_family
== AF_INET6
) {
405 struct ipv6_pinfo
*np
= inet6_sk(sk
);
406 memset(&sk
->sk_v6_rcv_saddr
, 0, sizeof(sk
->sk_v6_rcv_saddr
));
407 memset(&np
->saddr
, 0, sizeof(np
->saddr
));
412 * We need our own bind because there are no privileged id's == local ports.
413 * Moreover, we don't allow binding to multi- and broadcast addresses.
416 int ping_bind(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
)
418 struct inet_sock
*isk
= inet_sk(sk
);
421 int dif
= sk
->sk_bound_dev_if
;
423 err
= ping_check_bind_addr(sk
, isk
, uaddr
, addr_len
);
430 if (isk
->inet_num
!= 0)
434 ping_set_saddr(sk
, uaddr
);
435 snum
= ntohs(((struct sockaddr_in
*)uaddr
)->sin_port
);
436 if (ping_get_port(sk
, snum
) != 0) {
437 ping_clear_saddr(sk
, dif
);
441 pr_debug("after bind(): num = %d, dif = %d\n",
443 (int)sk
->sk_bound_dev_if
);
446 if (sk
->sk_family
== AF_INET
&& isk
->inet_rcv_saddr
)
447 sk
->sk_userlocks
|= SOCK_BINDADDR_LOCK
;
448 #if IS_ENABLED(CONFIG_IPV6)
449 if (sk
->sk_family
== AF_INET6
&& !ipv6_addr_any(&sk
->sk_v6_rcv_saddr
))
450 sk
->sk_userlocks
|= SOCK_BINDADDR_LOCK
;
454 sk
->sk_userlocks
|= SOCK_BINDPORT_LOCK
;
455 isk
->inet_sport
= htons(isk
->inet_num
);
459 #if IS_ENABLED(CONFIG_IPV6)
460 if (sk
->sk_family
== AF_INET6
)
461 memset(&sk
->sk_v6_daddr
, 0, sizeof(sk
->sk_v6_daddr
));
467 pr_debug("ping_v4_bind -> %d\n", err
);
470 EXPORT_SYMBOL_GPL(ping_bind
);
473 * Is this a supported type of ICMP message?
476 static inline int ping_supported(int family
, int type
, int code
)
478 return (family
== AF_INET
&& type
== ICMP_ECHO
&& code
== 0) ||
479 (family
== AF_INET6
&& type
== ICMPV6_ECHO_REQUEST
&& code
== 0);
483 * This routine is called by the ICMP module when it gets some
484 * sort of error condition.
487 void ping_err(struct sk_buff
*skb
, int offset
, u32 info
)
490 struct icmphdr
*icmph
;
491 struct inet_sock
*inet_sock
;
494 struct net
*net
= dev_net(skb
->dev
);
499 if (skb
->protocol
== htons(ETH_P_IP
)) {
501 type
= icmp_hdr(skb
)->type
;
502 code
= icmp_hdr(skb
)->code
;
503 icmph
= (struct icmphdr
*)(skb
->data
+ offset
);
504 } else if (skb
->protocol
== htons(ETH_P_IPV6
)) {
506 type
= icmp6_hdr(skb
)->icmp6_type
;
507 code
= icmp6_hdr(skb
)->icmp6_code
;
508 icmph
= (struct icmphdr
*) (skb
->data
+ offset
);
513 /* We assume the packet has already been checked by icmp_unreach */
515 if (!ping_supported(family
, icmph
->type
, icmph
->code
))
518 pr_debug("ping_err(proto=0x%x,type=%d,code=%d,id=%04x,seq=%04x)\n",
519 skb
->protocol
, type
, code
, ntohs(icmph
->un
.echo
.id
),
520 ntohs(icmph
->un
.echo
.sequence
));
522 sk
= ping_lookup(net
, skb
, ntohs(icmph
->un
.echo
.id
));
524 pr_debug("no socket, dropping\n");
525 return; /* No socket for error */
527 pr_debug("err on socket %p\n", sk
);
531 inet_sock
= inet_sk(sk
);
533 if (skb
->protocol
== htons(ETH_P_IP
)) {
536 case ICMP_TIME_EXCEEDED
:
539 case ICMP_SOURCE_QUENCH
:
540 /* This is not a real error but ping wants to see it.
541 * Report it with some fake errno.
545 case ICMP_PARAMETERPROB
:
549 case ICMP_DEST_UNREACH
:
550 if (code
== ICMP_FRAG_NEEDED
) { /* Path MTU discovery */
551 ipv4_sk_update_pmtu(skb
, sk
, info
);
552 if (inet_sock
->pmtudisc
!= IP_PMTUDISC_DONT
) {
560 if (code
<= NR_ICMP_UNREACH
) {
561 harderr
= icmp_err_convert
[code
].fatal
;
562 err
= icmp_err_convert
[code
].errno
;
566 /* See ICMP_SOURCE_QUENCH */
567 ipv4_sk_redirect(skb
, sk
);
571 #if IS_ENABLED(CONFIG_IPV6)
572 } else if (skb
->protocol
== htons(ETH_P_IPV6
)) {
573 harderr
= pingv6_ops
.icmpv6_err_convert(type
, code
, &err
);
578 * RFC1122: OK. Passes ICMP errors back to application, as per
581 if ((family
== AF_INET
&& !inet_sock
->recverr
) ||
582 (family
== AF_INET6
&& !inet6_sk(sk
)->recverr
)) {
583 if (!harderr
|| sk
->sk_state
!= TCP_ESTABLISHED
)
586 if (family
== AF_INET
) {
587 ip_icmp_error(sk
, skb
, err
, 0 /* no remote port */,
589 #if IS_ENABLED(CONFIG_IPV6)
590 } else if (family
== AF_INET6
) {
591 pingv6_ops
.ipv6_icmp_error(sk
, skb
, err
, 0,
597 sk
->sk_error_report(sk
);
601 EXPORT_SYMBOL_GPL(ping_err
);
604 * Copy and checksum an ICMP Echo packet from user space into a buffer
605 * starting from the payload.
608 int ping_getfrag(void *from
, char *to
,
609 int offset
, int fraglen
, int odd
, struct sk_buff
*skb
)
611 struct pingfakehdr
*pfh
= (struct pingfakehdr
*)from
;
614 fraglen
-= sizeof(struct icmphdr
);
617 if (csum_and_copy_from_iter(to
+ sizeof(struct icmphdr
),
618 fraglen
, &pfh
->wcheck
,
619 &pfh
->msg
->msg_iter
) != fraglen
)
621 } else if (offset
< sizeof(struct icmphdr
)) {
624 if (csum_and_copy_from_iter(to
, fraglen
, &pfh
->wcheck
,
625 &pfh
->msg
->msg_iter
) != fraglen
)
629 #if IS_ENABLED(CONFIG_IPV6)
630 /* For IPv6, checksum each skb as we go along, as expected by
631 * icmpv6_push_pending_frames. For IPv4, accumulate the checksum in
632 * wcheck, it will be finalized in ping_v4_push_pending_frames.
634 if (pfh
->family
== AF_INET6
) {
635 skb
->csum
= pfh
->wcheck
;
636 skb
->ip_summed
= CHECKSUM_NONE
;
643 EXPORT_SYMBOL_GPL(ping_getfrag
);
645 static int ping_v4_push_pending_frames(struct sock
*sk
, struct pingfakehdr
*pfh
,
648 struct sk_buff
*skb
= skb_peek(&sk
->sk_write_queue
);
650 pfh
->wcheck
= csum_partial((char *)&pfh
->icmph
,
651 sizeof(struct icmphdr
), pfh
->wcheck
);
652 pfh
->icmph
.checksum
= csum_fold(pfh
->wcheck
);
653 memcpy(icmp_hdr(skb
), &pfh
->icmph
, sizeof(struct icmphdr
));
654 skb
->ip_summed
= CHECKSUM_NONE
;
655 return ip_push_pending_frames(sk
, fl4
);
658 int ping_common_sendmsg(int family
, struct msghdr
*msg
, size_t len
,
659 void *user_icmph
, size_t icmph_len
) {
669 /* Mirror BSD error message compatibility */
670 if (msg
->msg_flags
& MSG_OOB
)
674 * Fetch the ICMP header provided by the userland.
675 * iovec is modified! The ICMP header is consumed.
677 if (memcpy_from_msg(user_icmph
, msg
, icmph_len
))
680 if (family
== AF_INET
) {
681 type
= ((struct icmphdr
*) user_icmph
)->type
;
682 code
= ((struct icmphdr
*) user_icmph
)->code
;
683 #if IS_ENABLED(CONFIG_IPV6)
684 } else if (family
== AF_INET6
) {
685 type
= ((struct icmp6hdr
*) user_icmph
)->icmp6_type
;
686 code
= ((struct icmp6hdr
*) user_icmph
)->icmp6_code
;
692 if (!ping_supported(family
, type
, code
))
697 EXPORT_SYMBOL_GPL(ping_common_sendmsg
);
699 static int ping_v4_sendmsg(struct sock
*sk
, struct msghdr
*msg
, size_t len
)
701 struct net
*net
= sock_net(sk
);
703 struct inet_sock
*inet
= inet_sk(sk
);
704 struct ipcm_cookie ipc
;
705 struct icmphdr user_icmph
;
706 struct pingfakehdr pfh
;
707 struct rtable
*rt
= NULL
;
708 struct ip_options_data opt_copy
;
710 __be32 saddr
, daddr
, faddr
;
714 pr_debug("ping_v4_sendmsg(sk=%p,sk->num=%u)\n", inet
, inet
->inet_num
);
716 err
= ping_common_sendmsg(AF_INET
, msg
, len
, &user_icmph
,
722 * Get and verify the address.
726 DECLARE_SOCKADDR(struct sockaddr_in
*, usin
, msg
->msg_name
);
727 if (msg
->msg_namelen
< sizeof(*usin
))
729 if (usin
->sin_family
!= AF_INET
)
730 return -EAFNOSUPPORT
;
731 daddr
= usin
->sin_addr
.s_addr
;
734 if (sk
->sk_state
!= TCP_ESTABLISHED
)
735 return -EDESTADDRREQ
;
736 daddr
= inet
->inet_daddr
;
740 ipc
.addr
= inet
->inet_saddr
;
742 ipc
.oif
= sk
->sk_bound_dev_if
;
747 sock_tx_timestamp(sk
, &ipc
.tx_flags
);
749 if (msg
->msg_controllen
) {
750 err
= ip_cmsg_send(sock_net(sk
), msg
, &ipc
, false);
759 struct ip_options_rcu
*inet_opt
;
762 inet_opt
= rcu_dereference(inet
->inet_opt
);
764 memcpy(&opt_copy
, inet_opt
,
765 sizeof(*inet_opt
) + inet_opt
->opt
.optlen
);
766 ipc
.opt
= &opt_copy
.opt
;
772 ipc
.addr
= faddr
= daddr
;
774 if (ipc
.opt
&& ipc
.opt
->opt
.srr
) {
777 faddr
= ipc
.opt
->opt
.faddr
;
779 tos
= get_rttos(&ipc
, inet
);
780 if (sock_flag(sk
, SOCK_LOCALROUTE
) ||
781 (msg
->msg_flags
& MSG_DONTROUTE
) ||
782 (ipc
.opt
&& ipc
.opt
->opt
.is_strictroute
)) {
786 if (ipv4_is_multicast(daddr
)) {
788 ipc
.oif
= inet
->mc_index
;
790 saddr
= inet
->mc_addr
;
792 ipc
.oif
= inet
->uc_index
;
794 flowi4_init_output(&fl4
, ipc
.oif
, sk
->sk_mark
, tos
,
795 RT_SCOPE_UNIVERSE
, sk
->sk_protocol
,
796 inet_sk_flowi_flags(sk
), faddr
, saddr
, 0, 0);
798 security_sk_classify_flow(sk
, flowi4_to_flowi(&fl4
));
799 rt
= ip_route_output_flow(net
, &fl4
, sk
);
803 if (err
== -ENETUNREACH
)
804 IP_INC_STATS(net
, IPSTATS_MIB_OUTNOROUTES
);
809 if ((rt
->rt_flags
& RTCF_BROADCAST
) &&
810 !sock_flag(sk
, SOCK_BROADCAST
))
813 if (msg
->msg_flags
& MSG_CONFIRM
)
818 ipc
.addr
= fl4
.daddr
;
822 pfh
.icmph
.type
= user_icmph
.type
; /* already checked */
823 pfh
.icmph
.code
= user_icmph
.code
; /* ditto */
824 pfh
.icmph
.checksum
= 0;
825 pfh
.icmph
.un
.echo
.id
= inet
->inet_sport
;
826 pfh
.icmph
.un
.echo
.sequence
= user_icmph
.un
.echo
.sequence
;
829 pfh
.family
= AF_INET
;
831 err
= ip_append_data(sk
, &fl4
, ping_getfrag
, &pfh
, len
,
832 0, &ipc
, &rt
, msg
->msg_flags
);
834 ip_flush_pending_frames(sk
);
836 err
= ping_v4_push_pending_frames(sk
, &pfh
, &fl4
);
844 icmp_out_count(sock_net(sk
), user_icmph
.type
);
850 dst_confirm(&rt
->dst
);
851 if (!(msg
->msg_flags
& MSG_PROBE
) || len
)
852 goto back_from_confirm
;
857 int ping_recvmsg(struct sock
*sk
, struct msghdr
*msg
, size_t len
, int noblock
,
858 int flags
, int *addr_len
)
860 struct inet_sock
*isk
= inet_sk(sk
);
861 int family
= sk
->sk_family
;
865 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk
, isk
->inet_num
);
871 if (flags
& MSG_ERRQUEUE
)
872 return inet_recv_error(sk
, msg
, len
, addr_len
);
874 skb
= skb_recv_datagram(sk
, flags
, noblock
, &err
);
880 msg
->msg_flags
|= MSG_TRUNC
;
884 /* Don't bother checking the checksum */
885 err
= skb_copy_datagram_msg(skb
, 0, msg
, copied
);
889 sock_recv_timestamp(msg
, sk
, skb
);
891 /* Copy the address and add cmsg data. */
892 if (family
== AF_INET
) {
893 DECLARE_SOCKADDR(struct sockaddr_in
*, sin
, msg
->msg_name
);
896 sin
->sin_family
= AF_INET
;
897 sin
->sin_port
= 0 /* skb->h.uh->source */;
898 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
899 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
900 *addr_len
= sizeof(*sin
);
904 ip_cmsg_recv(msg
, skb
);
906 #if IS_ENABLED(CONFIG_IPV6)
907 } else if (family
== AF_INET6
) {
908 struct ipv6_pinfo
*np
= inet6_sk(sk
);
909 struct ipv6hdr
*ip6
= ipv6_hdr(skb
);
910 DECLARE_SOCKADDR(struct sockaddr_in6
*, sin6
, msg
->msg_name
);
913 sin6
->sin6_family
= AF_INET6
;
915 sin6
->sin6_addr
= ip6
->saddr
;
916 sin6
->sin6_flowinfo
= 0;
918 sin6
->sin6_flowinfo
= ip6_flowinfo(ip6
);
919 sin6
->sin6_scope_id
=
920 ipv6_iface_scope_id(&sin6
->sin6_addr
,
922 *addr_len
= sizeof(*sin6
);
925 if (inet6_sk(sk
)->rxopt
.all
)
926 pingv6_ops
.ip6_datagram_recv_common_ctl(sk
, msg
, skb
);
927 if (skb
->protocol
== htons(ETH_P_IPV6
) &&
928 inet6_sk(sk
)->rxopt
.all
)
929 pingv6_ops
.ip6_datagram_recv_specific_ctl(sk
, msg
, skb
);
930 else if (skb
->protocol
== htons(ETH_P_IP
) && isk
->cmsg_flags
)
931 ip_cmsg_recv(msg
, skb
);
940 skb_free_datagram(sk
, skb
);
942 pr_debug("ping_recvmsg -> %d\n", err
);
945 EXPORT_SYMBOL_GPL(ping_recvmsg
);
947 int ping_queue_rcv_skb(struct sock
*sk
, struct sk_buff
*skb
)
949 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
950 inet_sk(sk
), inet_sk(sk
)->inet_num
, skb
);
951 if (sock_queue_rcv_skb(sk
, skb
) < 0) {
953 pr_debug("ping_queue_rcv_skb -> failed\n");
958 EXPORT_SYMBOL_GPL(ping_queue_rcv_skb
);
962 * All we need to do is get the socket.
965 bool ping_rcv(struct sk_buff
*skb
)
968 struct net
*net
= dev_net(skb
->dev
);
969 struct icmphdr
*icmph
= icmp_hdr(skb
);
971 /* We assume the packet has already been checked by icmp_rcv */
973 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
974 skb
, ntohs(icmph
->un
.echo
.id
), ntohs(icmph
->un
.echo
.sequence
));
976 /* Push ICMP header back */
977 skb_push(skb
, skb
->data
- (u8
*)icmph
);
979 sk
= ping_lookup(net
, skb
, ntohs(icmph
->un
.echo
.id
));
981 struct sk_buff
*skb2
= skb_clone(skb
, GFP_ATOMIC
);
983 pr_debug("rcv on socket %p\n", sk
);
985 ping_queue_rcv_skb(sk
, skb2
);
989 pr_debug("no socket, dropping\n");
993 EXPORT_SYMBOL_GPL(ping_rcv
);
995 struct proto ping_prot
= {
997 .owner
= THIS_MODULE
,
998 .init
= ping_init_sock
,
1000 .connect
= ip4_datagram_connect
,
1001 .disconnect
= udp_disconnect
,
1002 .setsockopt
= ip_setsockopt
,
1003 .getsockopt
= ip_getsockopt
,
1004 .sendmsg
= ping_v4_sendmsg
,
1005 .recvmsg
= ping_recvmsg
,
1007 .backlog_rcv
= ping_queue_rcv_skb
,
1008 .release_cb
= ip4_datagram_release_cb
,
1010 .unhash
= ping_unhash
,
1011 .get_port
= ping_get_port
,
1012 .obj_size
= sizeof(struct inet_sock
),
1014 EXPORT_SYMBOL(ping_prot
);
1016 #ifdef CONFIG_PROC_FS
1018 static struct sock
*ping_get_first(struct seq_file
*seq
, int start
)
1021 struct ping_iter_state
*state
= seq
->private;
1022 struct net
*net
= seq_file_net(seq
);
1024 for (state
->bucket
= start
; state
->bucket
< PING_HTABLE_SIZE
;
1026 struct hlist_nulls_node
*node
;
1027 struct hlist_nulls_head
*hslot
;
1029 hslot
= &ping_table
.hash
[state
->bucket
];
1031 if (hlist_nulls_empty(hslot
))
1034 sk_nulls_for_each(sk
, node
, hslot
) {
1035 if (net_eq(sock_net(sk
), net
) &&
1036 sk
->sk_family
== state
->family
)
1045 static struct sock
*ping_get_next(struct seq_file
*seq
, struct sock
*sk
)
1047 struct ping_iter_state
*state
= seq
->private;
1048 struct net
*net
= seq_file_net(seq
);
1051 sk
= sk_nulls_next(sk
);
1052 } while (sk
&& (!net_eq(sock_net(sk
), net
)));
1055 return ping_get_first(seq
, state
->bucket
+ 1);
1059 static struct sock
*ping_get_idx(struct seq_file
*seq
, loff_t pos
)
1061 struct sock
*sk
= ping_get_first(seq
, 0);
1064 while (pos
&& (sk
= ping_get_next(seq
, sk
)) != NULL
)
1066 return pos
? NULL
: sk
;
1069 void *ping_seq_start(struct seq_file
*seq
, loff_t
*pos
, sa_family_t family
)
1070 __acquires(ping_table
.lock
)
1072 struct ping_iter_state
*state
= seq
->private;
1074 state
->family
= family
;
1076 read_lock_bh(&ping_table
.lock
);
1078 return *pos
? ping_get_idx(seq
, *pos
-1) : SEQ_START_TOKEN
;
1080 EXPORT_SYMBOL_GPL(ping_seq_start
);
1082 static void *ping_v4_seq_start(struct seq_file
*seq
, loff_t
*pos
)
1084 return ping_seq_start(seq
, pos
, AF_INET
);
1087 void *ping_seq_next(struct seq_file
*seq
, void *v
, loff_t
*pos
)
1091 if (v
== SEQ_START_TOKEN
)
1092 sk
= ping_get_idx(seq
, 0);
1094 sk
= ping_get_next(seq
, v
);
1099 EXPORT_SYMBOL_GPL(ping_seq_next
);
1101 void ping_seq_stop(struct seq_file
*seq
, void *v
)
1102 __releases(ping_table
.lock
)
1104 read_unlock_bh(&ping_table
.lock
);
1106 EXPORT_SYMBOL_GPL(ping_seq_stop
);
1108 static void ping_v4_format_sock(struct sock
*sp
, struct seq_file
*f
,
1111 struct inet_sock
*inet
= inet_sk(sp
);
1112 __be32 dest
= inet
->inet_daddr
;
1113 __be32 src
= inet
->inet_rcv_saddr
;
1114 __u16 destp
= ntohs(inet
->inet_dport
);
1115 __u16 srcp
= ntohs(inet
->inet_sport
);
1117 seq_printf(f
, "%5d: %08X:%04X %08X:%04X"
1118 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d",
1119 bucket
, src
, srcp
, dest
, destp
, sp
->sk_state
,
1120 sk_wmem_alloc_get(sp
),
1121 sk_rmem_alloc_get(sp
),
1123 from_kuid_munged(seq_user_ns(f
), sock_i_uid(sp
)),
1125 atomic_read(&sp
->sk_refcnt
), sp
,
1126 atomic_read(&sp
->sk_drops
));
1129 static int ping_v4_seq_show(struct seq_file
*seq
, void *v
)
1131 seq_setwidth(seq
, 127);
1132 if (v
== SEQ_START_TOKEN
)
1133 seq_puts(seq
, " sl local_address rem_address st tx_queue "
1134 "rx_queue tr tm->when retrnsmt uid timeout "
1135 "inode ref pointer drops");
1137 struct ping_iter_state
*state
= seq
->private;
1139 ping_v4_format_sock(v
, seq
, state
->bucket
);
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
);