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
) {
206 pr_debug("found: %p: num=%d, daddr=%pI6c, dif=%d\n", sk
,
208 &sk
->sk_v6_rcv_saddr
,
209 sk
->sk_bound_dev_if
);
211 if (!ipv6_addr_any(&sk
->sk_v6_rcv_saddr
) &&
212 !ipv6_addr_equal(&sk
->sk_v6_rcv_saddr
,
213 &ipv6_hdr(skb
)->daddr
))
218 if (sk
->sk_bound_dev_if
&& sk
->sk_bound_dev_if
!= dif
)
227 read_unlock_bh(&ping_table
.lock
);
232 static void inet_get_ping_group_range_net(struct net
*net
, kgid_t
*low
,
235 kgid_t
*data
= net
->ipv4
.sysctl_ping_group_range
;
239 seq
= read_seqbegin(&net
->ipv4
.sysctl_local_ports
.lock
);
243 } while (read_seqretry(&net
->ipv4
.sysctl_local_ports
.lock
, seq
));
247 int ping_init_sock(struct sock
*sk
)
249 struct net
*net
= sock_net(sk
);
250 kgid_t group
= current_egid();
251 struct group_info
*group_info
= get_current_groups();
252 int i
, j
, count
= group_info
->ngroups
;
255 inet_get_ping_group_range_net(net
, &low
, &high
);
256 if (gid_lte(low
, group
) && gid_lte(group
, high
))
259 for (i
= 0; i
< group_info
->nblocks
; i
++) {
260 int cp_count
= min_t(int, NGROUPS_PER_BLOCK
, count
);
261 for (j
= 0; j
< cp_count
; j
++) {
262 kgid_t gid
= group_info
->blocks
[i
][j
];
263 if (gid_lte(low
, gid
) && gid_lte(gid
, high
))
272 EXPORT_SYMBOL_GPL(ping_init_sock
);
274 void ping_close(struct sock
*sk
, long timeout
)
276 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
277 inet_sk(sk
), inet_sk(sk
)->inet_num
);
278 pr_debug("isk->refcnt = %d\n", sk
->sk_refcnt
.counter
);
280 sk_common_release(sk
);
282 EXPORT_SYMBOL_GPL(ping_close
);
284 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
285 static int ping_check_bind_addr(struct sock
*sk
, struct inet_sock
*isk
,
286 struct sockaddr
*uaddr
, int addr_len
) {
287 struct net
*net
= sock_net(sk
);
288 if (sk
->sk_family
== AF_INET
) {
289 struct sockaddr_in
*addr
= (struct sockaddr_in
*) uaddr
;
292 if (addr_len
< sizeof(*addr
))
295 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI4,port=%d)\n",
296 sk
, &addr
->sin_addr
.s_addr
, ntohs(addr
->sin_port
));
298 chk_addr_ret
= inet_addr_type(net
, addr
->sin_addr
.s_addr
);
300 if (addr
->sin_addr
.s_addr
== htonl(INADDR_ANY
))
301 chk_addr_ret
= RTN_LOCAL
;
303 if ((sysctl_ip_nonlocal_bind
== 0 &&
304 isk
->freebind
== 0 && isk
->transparent
== 0 &&
305 chk_addr_ret
!= RTN_LOCAL
) ||
306 chk_addr_ret
== RTN_MULTICAST
||
307 chk_addr_ret
== RTN_BROADCAST
)
308 return -EADDRNOTAVAIL
;
310 #if IS_ENABLED(CONFIG_IPV6)
311 } else if (sk
->sk_family
== AF_INET6
) {
312 struct sockaddr_in6
*addr
= (struct sockaddr_in6
*) uaddr
;
313 int addr_type
, scoped
, has_addr
;
314 struct net_device
*dev
= NULL
;
316 if (addr_len
< sizeof(*addr
))
319 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI6c,port=%d)\n",
320 sk
, addr
->sin6_addr
.s6_addr
, ntohs(addr
->sin6_port
));
322 addr_type
= ipv6_addr_type(&addr
->sin6_addr
);
323 scoped
= __ipv6_addr_needs_scope_id(addr_type
);
324 if ((addr_type
!= IPV6_ADDR_ANY
&&
325 !(addr_type
& IPV6_ADDR_UNICAST
)) ||
326 (scoped
&& !addr
->sin6_scope_id
))
330 if (addr
->sin6_scope_id
) {
331 dev
= dev_get_by_index_rcu(net
, addr
->sin6_scope_id
);
337 has_addr
= pingv6_ops
.ipv6_chk_addr(net
, &addr
->sin6_addr
, dev
,
341 if (!(isk
->freebind
|| isk
->transparent
|| has_addr
||
342 addr_type
== IPV6_ADDR_ANY
))
343 return -EADDRNOTAVAIL
;
346 sk
->sk_bound_dev_if
= addr
->sin6_scope_id
;
349 return -EAFNOSUPPORT
;
354 static void ping_set_saddr(struct sock
*sk
, struct sockaddr
*saddr
)
356 if (saddr
->sa_family
== AF_INET
) {
357 struct inet_sock
*isk
= inet_sk(sk
);
358 struct sockaddr_in
*addr
= (struct sockaddr_in
*) saddr
;
359 isk
->inet_rcv_saddr
= isk
->inet_saddr
= addr
->sin_addr
.s_addr
;
360 #if IS_ENABLED(CONFIG_IPV6)
361 } else if (saddr
->sa_family
== AF_INET6
) {
362 struct sockaddr_in6
*addr
= (struct sockaddr_in6
*) saddr
;
363 struct ipv6_pinfo
*np
= inet6_sk(sk
);
364 sk
->sk_v6_rcv_saddr
= np
->saddr
= addr
->sin6_addr
;
369 static void ping_clear_saddr(struct sock
*sk
, int dif
)
371 sk
->sk_bound_dev_if
= dif
;
372 if (sk
->sk_family
== AF_INET
) {
373 struct inet_sock
*isk
= inet_sk(sk
);
374 isk
->inet_rcv_saddr
= isk
->inet_saddr
= 0;
375 #if IS_ENABLED(CONFIG_IPV6)
376 } else if (sk
->sk_family
== AF_INET6
) {
377 struct ipv6_pinfo
*np
= inet6_sk(sk
);
378 memset(&sk
->sk_v6_rcv_saddr
, 0, sizeof(sk
->sk_v6_rcv_saddr
));
379 memset(&np
->saddr
, 0, sizeof(np
->saddr
));
384 * We need our own bind because there are no privileged id's == local ports.
385 * Moreover, we don't allow binding to multi- and broadcast addresses.
388 int ping_bind(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
)
390 struct inet_sock
*isk
= inet_sk(sk
);
393 int dif
= sk
->sk_bound_dev_if
;
395 err
= ping_check_bind_addr(sk
, isk
, uaddr
, addr_len
);
402 if (isk
->inet_num
!= 0)
406 ping_set_saddr(sk
, uaddr
);
407 snum
= ntohs(((struct sockaddr_in
*)uaddr
)->sin_port
);
408 if (ping_get_port(sk
, snum
) != 0) {
409 ping_clear_saddr(sk
, dif
);
413 pr_debug("after bind(): num = %d, dif = %d\n",
415 (int)sk
->sk_bound_dev_if
);
418 if (sk
->sk_family
== AF_INET
&& isk
->inet_rcv_saddr
)
419 sk
->sk_userlocks
|= SOCK_BINDADDR_LOCK
;
420 #if IS_ENABLED(CONFIG_IPV6)
421 if (sk
->sk_family
== AF_INET6
&& !ipv6_addr_any(&sk
->sk_v6_rcv_saddr
))
422 sk
->sk_userlocks
|= SOCK_BINDADDR_LOCK
;
426 sk
->sk_userlocks
|= SOCK_BINDPORT_LOCK
;
427 isk
->inet_sport
= htons(isk
->inet_num
);
431 #if IS_ENABLED(CONFIG_IPV6)
432 if (sk
->sk_family
== AF_INET6
)
433 memset(&sk
->sk_v6_daddr
, 0, sizeof(sk
->sk_v6_daddr
));
439 pr_debug("ping_v4_bind -> %d\n", err
);
442 EXPORT_SYMBOL_GPL(ping_bind
);
445 * Is this a supported type of ICMP message?
448 static inline int ping_supported(int family
, int type
, int code
)
450 return (family
== AF_INET
&& type
== ICMP_ECHO
&& code
== 0) ||
451 (family
== AF_INET6
&& type
== ICMPV6_ECHO_REQUEST
&& code
== 0);
455 * This routine is called by the ICMP module when it gets some
456 * sort of error condition.
459 void ping_err(struct sk_buff
*skb
, int offset
, u32 info
)
462 struct icmphdr
*icmph
;
463 struct inet_sock
*inet_sock
;
466 struct net
*net
= dev_net(skb
->dev
);
471 if (skb
->protocol
== htons(ETH_P_IP
)) {
473 type
= icmp_hdr(skb
)->type
;
474 code
= icmp_hdr(skb
)->code
;
475 icmph
= (struct icmphdr
*)(skb
->data
+ offset
);
476 } else if (skb
->protocol
== htons(ETH_P_IPV6
)) {
478 type
= icmp6_hdr(skb
)->icmp6_type
;
479 code
= icmp6_hdr(skb
)->icmp6_code
;
480 icmph
= (struct icmphdr
*) (skb
->data
+ offset
);
485 /* We assume the packet has already been checked by icmp_unreach */
487 if (!ping_supported(family
, icmph
->type
, icmph
->code
))
490 pr_debug("ping_err(proto=0x%x,type=%d,code=%d,id=%04x,seq=%04x)\n",
491 skb
->protocol
, type
, code
, ntohs(icmph
->un
.echo
.id
),
492 ntohs(icmph
->un
.echo
.sequence
));
494 sk
= ping_lookup(net
, skb
, ntohs(icmph
->un
.echo
.id
));
496 pr_debug("no socket, dropping\n");
497 return; /* No socket for error */
499 pr_debug("err on socket %p\n", sk
);
503 inet_sock
= inet_sk(sk
);
505 if (skb
->protocol
== htons(ETH_P_IP
)) {
508 case ICMP_TIME_EXCEEDED
:
511 case ICMP_SOURCE_QUENCH
:
512 /* This is not a real error but ping wants to see it.
513 * Report it with some fake errno.
517 case ICMP_PARAMETERPROB
:
521 case ICMP_DEST_UNREACH
:
522 if (code
== ICMP_FRAG_NEEDED
) { /* Path MTU discovery */
523 ipv4_sk_update_pmtu(skb
, sk
, info
);
524 if (inet_sock
->pmtudisc
!= IP_PMTUDISC_DONT
) {
532 if (code
<= NR_ICMP_UNREACH
) {
533 harderr
= icmp_err_convert
[code
].fatal
;
534 err
= icmp_err_convert
[code
].errno
;
538 /* See ICMP_SOURCE_QUENCH */
539 ipv4_sk_redirect(skb
, sk
);
543 #if IS_ENABLED(CONFIG_IPV6)
544 } else if (skb
->protocol
== htons(ETH_P_IPV6
)) {
545 harderr
= pingv6_ops
.icmpv6_err_convert(type
, code
, &err
);
550 * RFC1122: OK. Passes ICMP errors back to application, as per
553 if ((family
== AF_INET
&& !inet_sock
->recverr
) ||
554 (family
== AF_INET6
&& !inet6_sk(sk
)->recverr
)) {
555 if (!harderr
|| sk
->sk_state
!= TCP_ESTABLISHED
)
558 if (family
== AF_INET
) {
559 ip_icmp_error(sk
, skb
, err
, 0 /* no remote port */,
561 #if IS_ENABLED(CONFIG_IPV6)
562 } else if (family
== AF_INET6
) {
563 pingv6_ops
.ipv6_icmp_error(sk
, skb
, err
, 0,
569 sk
->sk_error_report(sk
);
573 EXPORT_SYMBOL_GPL(ping_err
);
576 * Copy and checksum an ICMP Echo packet from user space into a buffer
577 * starting from the payload.
580 int ping_getfrag(void *from
, char *to
,
581 int offset
, int fraglen
, int odd
, struct sk_buff
*skb
)
583 struct pingfakehdr
*pfh
= (struct pingfakehdr
*)from
;
586 if (fraglen
< sizeof(struct icmphdr
))
588 if (csum_partial_copy_fromiovecend(to
+ sizeof(struct icmphdr
),
589 pfh
->iov
, 0, fraglen
- sizeof(struct icmphdr
),
592 } else if (offset
< sizeof(struct icmphdr
)) {
595 if (csum_partial_copy_fromiovecend
596 (to
, pfh
->iov
, offset
- sizeof(struct icmphdr
),
597 fraglen
, &pfh
->wcheck
))
601 #if IS_ENABLED(CONFIG_IPV6)
602 /* For IPv6, checksum each skb as we go along, as expected by
603 * icmpv6_push_pending_frames. For IPv4, accumulate the checksum in
604 * wcheck, it will be finalized in ping_v4_push_pending_frames.
606 if (pfh
->family
== AF_INET6
) {
607 skb
->csum
= pfh
->wcheck
;
608 skb
->ip_summed
= CHECKSUM_NONE
;
615 EXPORT_SYMBOL_GPL(ping_getfrag
);
617 static int ping_v4_push_pending_frames(struct sock
*sk
, struct pingfakehdr
*pfh
,
620 struct sk_buff
*skb
= skb_peek(&sk
->sk_write_queue
);
622 pfh
->wcheck
= csum_partial((char *)&pfh
->icmph
,
623 sizeof(struct icmphdr
), pfh
->wcheck
);
624 pfh
->icmph
.checksum
= csum_fold(pfh
->wcheck
);
625 memcpy(icmp_hdr(skb
), &pfh
->icmph
, sizeof(struct icmphdr
));
626 skb
->ip_summed
= CHECKSUM_NONE
;
627 return ip_push_pending_frames(sk
, fl4
);
630 int ping_common_sendmsg(int family
, struct msghdr
*msg
, size_t len
,
631 void *user_icmph
, size_t icmph_len
) {
641 /* Mirror BSD error message compatibility */
642 if (msg
->msg_flags
& MSG_OOB
)
646 * Fetch the ICMP header provided by the userland.
647 * iovec is modified! The ICMP header is consumed.
649 if (memcpy_fromiovec(user_icmph
, msg
->msg_iov
, icmph_len
))
652 if (family
== AF_INET
) {
653 type
= ((struct icmphdr
*) user_icmph
)->type
;
654 code
= ((struct icmphdr
*) user_icmph
)->code
;
655 #if IS_ENABLED(CONFIG_IPV6)
656 } else if (family
== AF_INET6
) {
657 type
= ((struct icmp6hdr
*) user_icmph
)->icmp6_type
;
658 code
= ((struct icmp6hdr
*) user_icmph
)->icmp6_code
;
664 if (!ping_supported(family
, type
, code
))
669 EXPORT_SYMBOL_GPL(ping_common_sendmsg
);
671 int ping_v4_sendmsg(struct kiocb
*iocb
, struct sock
*sk
, struct msghdr
*msg
,
674 struct net
*net
= sock_net(sk
);
676 struct inet_sock
*inet
= inet_sk(sk
);
677 struct ipcm_cookie ipc
;
678 struct icmphdr user_icmph
;
679 struct pingfakehdr pfh
;
680 struct rtable
*rt
= NULL
;
681 struct ip_options_data opt_copy
;
683 __be32 saddr
, daddr
, faddr
;
687 pr_debug("ping_v4_sendmsg(sk=%p,sk->num=%u)\n", inet
, inet
->inet_num
);
689 err
= ping_common_sendmsg(AF_INET
, msg
, len
, &user_icmph
,
695 * Get and verify the address.
699 struct sockaddr_in
*usin
= (struct sockaddr_in
*)msg
->msg_name
;
700 if (msg
->msg_namelen
< sizeof(*usin
))
702 if (usin
->sin_family
!= AF_INET
)
704 daddr
= usin
->sin_addr
.s_addr
;
707 if (sk
->sk_state
!= TCP_ESTABLISHED
)
708 return -EDESTADDRREQ
;
709 daddr
= inet
->inet_daddr
;
713 ipc
.addr
= inet
->inet_saddr
;
715 ipc
.oif
= sk
->sk_bound_dev_if
;
720 sock_tx_timestamp(sk
, &ipc
.tx_flags
);
722 if (msg
->msg_controllen
) {
723 err
= ip_cmsg_send(sock_net(sk
), msg
, &ipc
);
730 struct ip_options_rcu
*inet_opt
;
733 inet_opt
= rcu_dereference(inet
->inet_opt
);
735 memcpy(&opt_copy
, inet_opt
,
736 sizeof(*inet_opt
) + inet_opt
->opt
.optlen
);
737 ipc
.opt
= &opt_copy
.opt
;
743 ipc
.addr
= faddr
= daddr
;
745 if (ipc
.opt
&& ipc
.opt
->opt
.srr
) {
748 faddr
= ipc
.opt
->opt
.faddr
;
750 tos
= get_rttos(&ipc
, inet
);
751 if (sock_flag(sk
, SOCK_LOCALROUTE
) ||
752 (msg
->msg_flags
& MSG_DONTROUTE
) ||
753 (ipc
.opt
&& ipc
.opt
->opt
.is_strictroute
)) {
757 if (ipv4_is_multicast(daddr
)) {
759 ipc
.oif
= inet
->mc_index
;
761 saddr
= inet
->mc_addr
;
763 ipc
.oif
= inet
->uc_index
;
765 flowi4_init_output(&fl4
, ipc
.oif
, sk
->sk_mark
, tos
,
766 RT_SCOPE_UNIVERSE
, sk
->sk_protocol
,
767 inet_sk_flowi_flags(sk
), faddr
, saddr
, 0, 0);
769 security_sk_classify_flow(sk
, flowi4_to_flowi(&fl4
));
770 rt
= ip_route_output_flow(net
, &fl4
, sk
);
774 if (err
== -ENETUNREACH
)
775 IP_INC_STATS_BH(net
, IPSTATS_MIB_OUTNOROUTES
);
780 if ((rt
->rt_flags
& RTCF_BROADCAST
) &&
781 !sock_flag(sk
, SOCK_BROADCAST
))
784 if (msg
->msg_flags
& MSG_CONFIRM
)
789 ipc
.addr
= fl4
.daddr
;
793 pfh
.icmph
.type
= user_icmph
.type
; /* already checked */
794 pfh
.icmph
.code
= user_icmph
.code
; /* ditto */
795 pfh
.icmph
.checksum
= 0;
796 pfh
.icmph
.un
.echo
.id
= inet
->inet_sport
;
797 pfh
.icmph
.un
.echo
.sequence
= user_icmph
.un
.echo
.sequence
;
798 pfh
.iov
= msg
->msg_iov
;
800 pfh
.family
= AF_INET
;
802 err
= ip_append_data(sk
, &fl4
, ping_getfrag
, &pfh
, len
,
803 0, &ipc
, &rt
, msg
->msg_flags
);
805 ip_flush_pending_frames(sk
);
807 err
= ping_v4_push_pending_frames(sk
, &pfh
, &fl4
);
815 icmp_out_count(sock_net(sk
), user_icmph
.type
);
821 dst_confirm(&rt
->dst
);
822 if (!(msg
->msg_flags
& MSG_PROBE
) || len
)
823 goto back_from_confirm
;
828 int ping_recvmsg(struct kiocb
*iocb
, struct sock
*sk
, struct msghdr
*msg
,
829 size_t len
, int noblock
, int flags
, int *addr_len
)
831 struct inet_sock
*isk
= inet_sk(sk
);
832 int family
= sk
->sk_family
;
833 struct sockaddr_in
*sin
;
834 struct sockaddr_in6
*sin6
;
838 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk
, isk
->inet_num
);
845 if (family
== AF_INET
)
846 *addr_len
= sizeof(*sin
);
847 else if (family
== AF_INET6
&& addr_len
)
848 *addr_len
= sizeof(*sin6
);
851 if (flags
& MSG_ERRQUEUE
) {
852 if (family
== AF_INET
) {
853 return ip_recv_error(sk
, msg
, len
);
854 #if IS_ENABLED(CONFIG_IPV6)
855 } else if (family
== AF_INET6
) {
856 return pingv6_ops
.ipv6_recv_error(sk
, msg
, len
);
861 skb
= skb_recv_datagram(sk
, flags
, noblock
, &err
);
867 msg
->msg_flags
|= MSG_TRUNC
;
871 /* Don't bother checking the checksum */
872 err
= skb_copy_datagram_iovec(skb
, 0, msg
->msg_iov
, copied
);
876 sock_recv_timestamp(msg
, sk
, skb
);
878 /* Copy the address and add cmsg data. */
879 if (family
== AF_INET
) {
880 sin
= (struct sockaddr_in
*) msg
->msg_name
;
881 sin
->sin_family
= AF_INET
;
882 sin
->sin_port
= 0 /* skb->h.uh->source */;
883 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
884 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
887 ip_cmsg_recv(msg
, skb
);
889 #if IS_ENABLED(CONFIG_IPV6)
890 } else if (family
== AF_INET6
) {
891 struct ipv6_pinfo
*np
= inet6_sk(sk
);
892 struct ipv6hdr
*ip6
= ipv6_hdr(skb
);
893 sin6
= (struct sockaddr_in6
*) msg
->msg_name
;
894 sin6
->sin6_family
= AF_INET6
;
896 sin6
->sin6_addr
= ip6
->saddr
;
898 sin6
->sin6_flowinfo
= 0;
900 sin6
->sin6_flowinfo
= ip6_flowinfo(ip6
);
902 sin6
->sin6_scope_id
= ipv6_iface_scope_id(&sin6
->sin6_addr
,
905 if (inet6_sk(sk
)->rxopt
.all
)
906 pingv6_ops
.ip6_datagram_recv_ctl(sk
, msg
, skb
);
915 skb_free_datagram(sk
, skb
);
917 pr_debug("ping_recvmsg -> %d\n", err
);
920 EXPORT_SYMBOL_GPL(ping_recvmsg
);
922 int ping_queue_rcv_skb(struct sock
*sk
, struct sk_buff
*skb
)
924 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
925 inet_sk(sk
), inet_sk(sk
)->inet_num
, skb
);
926 if (sock_queue_rcv_skb(sk
, skb
) < 0) {
928 pr_debug("ping_queue_rcv_skb -> failed\n");
933 EXPORT_SYMBOL_GPL(ping_queue_rcv_skb
);
937 * All we need to do is get the socket.
940 void ping_rcv(struct sk_buff
*skb
)
943 struct net
*net
= dev_net(skb
->dev
);
944 struct icmphdr
*icmph
= icmp_hdr(skb
);
946 /* We assume the packet has already been checked by icmp_rcv */
948 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
949 skb
, ntohs(icmph
->un
.echo
.id
), ntohs(icmph
->un
.echo
.sequence
));
951 /* Push ICMP header back */
952 skb_push(skb
, skb
->data
- (u8
*)icmph
);
954 sk
= ping_lookup(net
, skb
, ntohs(icmph
->un
.echo
.id
));
956 pr_debug("rcv on socket %p\n", sk
);
957 ping_queue_rcv_skb(sk
, skb_get(skb
));
961 pr_debug("no socket, dropping\n");
963 /* We're called from icmp_rcv(). kfree_skb() is done there. */
965 EXPORT_SYMBOL_GPL(ping_rcv
);
967 struct proto ping_prot
= {
969 .owner
= THIS_MODULE
,
970 .init
= ping_init_sock
,
972 .connect
= ip4_datagram_connect
,
973 .disconnect
= udp_disconnect
,
974 .setsockopt
= ip_setsockopt
,
975 .getsockopt
= ip_getsockopt
,
976 .sendmsg
= ping_v4_sendmsg
,
977 .recvmsg
= ping_recvmsg
,
979 .backlog_rcv
= ping_queue_rcv_skb
,
980 .release_cb
= ip4_datagram_release_cb
,
982 .unhash
= ping_unhash
,
983 .get_port
= ping_get_port
,
984 .obj_size
= sizeof(struct inet_sock
),
986 EXPORT_SYMBOL(ping_prot
);
988 #ifdef CONFIG_PROC_FS
990 static struct sock
*ping_get_first(struct seq_file
*seq
, int start
)
993 struct ping_iter_state
*state
= seq
->private;
994 struct net
*net
= seq_file_net(seq
);
996 for (state
->bucket
= start
; state
->bucket
< PING_HTABLE_SIZE
;
998 struct hlist_nulls_node
*node
;
999 struct hlist_nulls_head
*hslot
;
1001 hslot
= &ping_table
.hash
[state
->bucket
];
1003 if (hlist_nulls_empty(hslot
))
1006 sk_nulls_for_each(sk
, node
, hslot
) {
1007 if (net_eq(sock_net(sk
), net
) &&
1008 sk
->sk_family
== state
->family
)
1017 static struct sock
*ping_get_next(struct seq_file
*seq
, struct sock
*sk
)
1019 struct ping_iter_state
*state
= seq
->private;
1020 struct net
*net
= seq_file_net(seq
);
1023 sk
= sk_nulls_next(sk
);
1024 } while (sk
&& (!net_eq(sock_net(sk
), net
)));
1027 return ping_get_first(seq
, state
->bucket
+ 1);
1031 static struct sock
*ping_get_idx(struct seq_file
*seq
, loff_t pos
)
1033 struct sock
*sk
= ping_get_first(seq
, 0);
1036 while (pos
&& (sk
= ping_get_next(seq
, sk
)) != NULL
)
1038 return pos
? NULL
: sk
;
1041 void *ping_seq_start(struct seq_file
*seq
, loff_t
*pos
, sa_family_t family
)
1043 struct ping_iter_state
*state
= seq
->private;
1045 state
->family
= family
;
1047 read_lock_bh(&ping_table
.lock
);
1049 return *pos
? ping_get_idx(seq
, *pos
-1) : SEQ_START_TOKEN
;
1051 EXPORT_SYMBOL_GPL(ping_seq_start
);
1053 static void *ping_v4_seq_start(struct seq_file
*seq
, loff_t
*pos
)
1055 return ping_seq_start(seq
, pos
, AF_INET
);
1058 void *ping_seq_next(struct seq_file
*seq
, void *v
, loff_t
*pos
)
1062 if (v
== SEQ_START_TOKEN
)
1063 sk
= ping_get_idx(seq
, 0);
1065 sk
= ping_get_next(seq
, v
);
1070 EXPORT_SYMBOL_GPL(ping_seq_next
);
1072 void ping_seq_stop(struct seq_file
*seq
, void *v
)
1074 read_unlock_bh(&ping_table
.lock
);
1076 EXPORT_SYMBOL_GPL(ping_seq_stop
);
1078 static void ping_v4_format_sock(struct sock
*sp
, struct seq_file
*f
,
1079 int bucket
, int *len
)
1081 struct inet_sock
*inet
= inet_sk(sp
);
1082 __be32 dest
= inet
->inet_daddr
;
1083 __be32 src
= inet
->inet_rcv_saddr
;
1084 __u16 destp
= ntohs(inet
->inet_dport
);
1085 __u16 srcp
= ntohs(inet
->inet_sport
);
1087 seq_printf(f
, "%5d: %08X:%04X %08X:%04X"
1088 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d%n",
1089 bucket
, src
, srcp
, dest
, destp
, sp
->sk_state
,
1090 sk_wmem_alloc_get(sp
),
1091 sk_rmem_alloc_get(sp
),
1093 from_kuid_munged(seq_user_ns(f
), sock_i_uid(sp
)),
1095 atomic_read(&sp
->sk_refcnt
), sp
,
1096 atomic_read(&sp
->sk_drops
), len
);
1099 static int ping_v4_seq_show(struct seq_file
*seq
, void *v
)
1101 if (v
== SEQ_START_TOKEN
)
1102 seq_printf(seq
, "%-127s\n",
1103 " sl local_address rem_address st tx_queue "
1104 "rx_queue tr tm->when retrnsmt uid timeout "
1105 "inode ref pointer drops");
1107 struct ping_iter_state
*state
= seq
->private;
1110 ping_v4_format_sock(v
, seq
, state
->bucket
, &len
);
1111 seq_printf(seq
, "%*s\n", 127 - len
, "");
1116 static const struct seq_operations ping_v4_seq_ops
= {
1117 .show
= ping_v4_seq_show
,
1118 .start
= ping_v4_seq_start
,
1119 .next
= ping_seq_next
,
1120 .stop
= ping_seq_stop
,
1123 static int ping_seq_open(struct inode
*inode
, struct file
*file
)
1125 struct ping_seq_afinfo
*afinfo
= PDE_DATA(inode
);
1126 return seq_open_net(inode
, file
, &afinfo
->seq_ops
,
1127 sizeof(struct ping_iter_state
));
1130 const struct file_operations ping_seq_fops
= {
1131 .open
= ping_seq_open
,
1133 .llseek
= seq_lseek
,
1134 .release
= seq_release_net
,
1136 EXPORT_SYMBOL_GPL(ping_seq_fops
);
1138 static struct ping_seq_afinfo ping_v4_seq_afinfo
= {
1141 .seq_fops
= &ping_seq_fops
,
1143 .start
= ping_v4_seq_start
,
1144 .show
= ping_v4_seq_show
,
1145 .next
= ping_seq_next
,
1146 .stop
= ping_seq_stop
,
1150 int ping_proc_register(struct net
*net
, struct ping_seq_afinfo
*afinfo
)
1152 struct proc_dir_entry
*p
;
1153 p
= proc_create_data(afinfo
->name
, S_IRUGO
, net
->proc_net
,
1154 afinfo
->seq_fops
, afinfo
);
1159 EXPORT_SYMBOL_GPL(ping_proc_register
);
1161 void ping_proc_unregister(struct net
*net
, struct ping_seq_afinfo
*afinfo
)
1163 remove_proc_entry(afinfo
->name
, net
->proc_net
);
1165 EXPORT_SYMBOL_GPL(ping_proc_unregister
);
1167 static int __net_init
ping_v4_proc_init_net(struct net
*net
)
1169 return ping_proc_register(net
, &ping_v4_seq_afinfo
);
1172 static void __net_exit
ping_v4_proc_exit_net(struct net
*net
)
1174 ping_proc_unregister(net
, &ping_v4_seq_afinfo
);
1177 static struct pernet_operations ping_v4_net_ops
= {
1178 .init
= ping_v4_proc_init_net
,
1179 .exit
= ping_v4_proc_exit_net
,
1182 int __init
ping_proc_init(void)
1184 return register_pernet_subsys(&ping_v4_net_ops
);
1187 void ping_proc_exit(void)
1189 unregister_pernet_subsys(&ping_v4_net_ops
);
1194 void __init
ping_init(void)
1198 for (i
= 0; i
< PING_HTABLE_SIZE
; i
++)
1199 INIT_HLIST_NULLS_HEAD(&ping_table
.hash
[i
], i
);
1200 rwlock_init(&ping_table
.lock
);