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
);
157 sk_nulls_node_init(&sk
->sk_nulls_node
);
161 sock_prot_inuse_add(sock_net(sk
), sk
->sk_prot
, -1);
162 write_unlock_bh(&ping_table
.lock
);
165 EXPORT_SYMBOL_GPL(ping_unhash
);
167 static struct sock
*ping_lookup(struct net
*net
, struct sk_buff
*skb
, u16 ident
)
169 struct hlist_nulls_head
*hslot
= ping_hashslot(&ping_table
, net
, ident
);
170 struct sock
*sk
= NULL
;
171 struct inet_sock
*isk
;
172 struct hlist_nulls_node
*hnode
;
173 int dif
= skb
->dev
->ifindex
;
175 if (skb
->protocol
== htons(ETH_P_IP
)) {
176 pr_debug("try to find: num = %d, daddr = %pI4, dif = %d\n",
177 (int)ident
, &ip_hdr(skb
)->daddr
, dif
);
178 #if IS_ENABLED(CONFIG_IPV6)
179 } else if (skb
->protocol
== htons(ETH_P_IPV6
)) {
180 pr_debug("try to find: num = %d, daddr = %pI6c, dif = %d\n",
181 (int)ident
, &ipv6_hdr(skb
)->daddr
, dif
);
185 read_lock_bh(&ping_table
.lock
);
187 ping_portaddr_for_each_entry(sk
, hnode
, hslot
) {
190 pr_debug("iterate\n");
191 if (isk
->inet_num
!= ident
)
194 if (skb
->protocol
== htons(ETH_P_IP
) &&
195 sk
->sk_family
== AF_INET
) {
196 pr_debug("found: %p: num=%d, daddr=%pI4, dif=%d\n", sk
,
197 (int) isk
->inet_num
, &isk
->inet_rcv_saddr
,
198 sk
->sk_bound_dev_if
);
200 if (isk
->inet_rcv_saddr
&&
201 isk
->inet_rcv_saddr
!= ip_hdr(skb
)->daddr
)
203 #if IS_ENABLED(CONFIG_IPV6)
204 } else if (skb
->protocol
== htons(ETH_P_IPV6
) &&
205 sk
->sk_family
== AF_INET6
) {
206 struct ipv6_pinfo
*np
= inet6_sk(sk
);
208 pr_debug("found: %p: num=%d, daddr=%pI6c, dif=%d\n", sk
,
210 &inet6_sk(sk
)->rcv_saddr
,
211 sk
->sk_bound_dev_if
);
213 if (!ipv6_addr_any(&np
->rcv_saddr
) &&
214 !ipv6_addr_equal(&np
->rcv_saddr
,
215 &ipv6_hdr(skb
)->daddr
))
222 if (sk
->sk_bound_dev_if
&& sk
->sk_bound_dev_if
!= dif
)
231 read_unlock_bh(&ping_table
.lock
);
236 static void inet_get_ping_group_range_net(struct net
*net
, kgid_t
*low
,
239 kgid_t
*data
= net
->ipv4
.sysctl_ping_group_range
;
243 seq
= read_seqbegin(&sysctl_local_ports
.lock
);
247 } while (read_seqretry(&sysctl_local_ports
.lock
, seq
));
251 int ping_init_sock(struct sock
*sk
)
253 struct net
*net
= sock_net(sk
);
254 kgid_t group
= current_egid();
255 struct group_info
*group_info
;
260 #if IS_ENABLED(CONFIG_IPV6)
261 if (sk
->sk_family
== AF_INET6
)
262 inet6_sk(sk
)->ipv6only
= 1;
264 inet_get_ping_group_range_net(net
, &low
, &high
);
265 if (gid_lte(low
, group
) && gid_lte(group
, high
))
268 group_info
= get_current_groups();
269 count
= group_info
->ngroups
;
270 for (i
= 0; i
< group_info
->nblocks
; i
++) {
271 int cp_count
= min_t(int, NGROUPS_PER_BLOCK
, count
);
272 for (j
= 0; j
< cp_count
; j
++) {
273 kgid_t gid
= group_info
->blocks
[i
][j
];
274 if (gid_lte(low
, gid
) && gid_lte(gid
, high
))
275 goto out_release_group
;
284 put_group_info(group_info
);
287 EXPORT_SYMBOL_GPL(ping_init_sock
);
289 void ping_close(struct sock
*sk
, long timeout
)
291 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
292 inet_sk(sk
), inet_sk(sk
)->inet_num
);
293 pr_debug("isk->refcnt = %d\n", sk
->sk_refcnt
.counter
);
295 sk_common_release(sk
);
297 EXPORT_SYMBOL_GPL(ping_close
);
299 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
300 static int ping_check_bind_addr(struct sock
*sk
, struct inet_sock
*isk
,
301 struct sockaddr
*uaddr
, int addr_len
) {
302 struct net
*net
= sock_net(sk
);
303 if (sk
->sk_family
== AF_INET
) {
304 struct sockaddr_in
*addr
= (struct sockaddr_in
*) uaddr
;
307 if (addr_len
< sizeof(*addr
))
310 if (addr
->sin_family
!= AF_INET
&&
311 !(addr
->sin_family
== AF_UNSPEC
&&
312 addr
->sin_addr
.s_addr
== htonl(INADDR_ANY
)))
313 return -EAFNOSUPPORT
;
315 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI4,port=%d)\n",
316 sk
, &addr
->sin_addr
.s_addr
, ntohs(addr
->sin_port
));
318 chk_addr_ret
= inet_addr_type(net
, addr
->sin_addr
.s_addr
);
320 if (addr
->sin_addr
.s_addr
== htonl(INADDR_ANY
))
321 chk_addr_ret
= RTN_LOCAL
;
323 if ((sysctl_ip_nonlocal_bind
== 0 &&
324 isk
->freebind
== 0 && isk
->transparent
== 0 &&
325 chk_addr_ret
!= RTN_LOCAL
) ||
326 chk_addr_ret
== RTN_MULTICAST
||
327 chk_addr_ret
== RTN_BROADCAST
)
328 return -EADDRNOTAVAIL
;
330 #if IS_ENABLED(CONFIG_IPV6)
331 } else if (sk
->sk_family
== AF_INET6
) {
332 struct sockaddr_in6
*addr
= (struct sockaddr_in6
*) uaddr
;
333 int addr_type
, scoped
, has_addr
;
334 struct net_device
*dev
= NULL
;
336 if (addr_len
< sizeof(*addr
))
339 if (addr
->sin6_family
!= AF_INET6
)
340 return -EAFNOSUPPORT
;
342 pr_debug("ping_check_bind_addr(sk=%p,addr=%pI6c,port=%d)\n",
343 sk
, addr
->sin6_addr
.s6_addr
, ntohs(addr
->sin6_port
));
345 addr_type
= ipv6_addr_type(&addr
->sin6_addr
);
346 scoped
= __ipv6_addr_needs_scope_id(addr_type
);
347 if ((addr_type
!= IPV6_ADDR_ANY
&&
348 !(addr_type
& IPV6_ADDR_UNICAST
)) ||
349 (scoped
&& !addr
->sin6_scope_id
))
353 if (addr
->sin6_scope_id
) {
354 dev
= dev_get_by_index_rcu(net
, addr
->sin6_scope_id
);
360 has_addr
= pingv6_ops
.ipv6_chk_addr(net
, &addr
->sin6_addr
, dev
,
364 if (!(isk
->freebind
|| isk
->transparent
|| has_addr
||
365 addr_type
== IPV6_ADDR_ANY
))
366 return -EADDRNOTAVAIL
;
369 sk
->sk_bound_dev_if
= addr
->sin6_scope_id
;
372 return -EAFNOSUPPORT
;
377 static void ping_set_saddr(struct sock
*sk
, struct sockaddr
*saddr
)
379 if (saddr
->sa_family
== AF_INET
) {
380 struct inet_sock
*isk
= inet_sk(sk
);
381 struct sockaddr_in
*addr
= (struct sockaddr_in
*) saddr
;
382 isk
->inet_rcv_saddr
= isk
->inet_saddr
= addr
->sin_addr
.s_addr
;
383 #if IS_ENABLED(CONFIG_IPV6)
384 } else if (saddr
->sa_family
== AF_INET6
) {
385 struct sockaddr_in6
*addr
= (struct sockaddr_in6
*) saddr
;
386 struct ipv6_pinfo
*np
= inet6_sk(sk
);
387 np
->rcv_saddr
= np
->saddr
= addr
->sin6_addr
;
392 static void ping_clear_saddr(struct sock
*sk
, int dif
)
394 sk
->sk_bound_dev_if
= dif
;
395 if (sk
->sk_family
== AF_INET
) {
396 struct inet_sock
*isk
= inet_sk(sk
);
397 isk
->inet_rcv_saddr
= isk
->inet_saddr
= 0;
398 #if IS_ENABLED(CONFIG_IPV6)
399 } else if (sk
->sk_family
== AF_INET6
) {
400 struct ipv6_pinfo
*np
= inet6_sk(sk
);
401 memset(&np
->rcv_saddr
, 0, sizeof(np
->rcv_saddr
));
402 memset(&np
->saddr
, 0, sizeof(np
->saddr
));
407 * We need our own bind because there are no privileged id's == local ports.
408 * Moreover, we don't allow binding to multi- and broadcast addresses.
411 int ping_bind(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
)
413 struct inet_sock
*isk
= inet_sk(sk
);
416 int dif
= sk
->sk_bound_dev_if
;
418 err
= ping_check_bind_addr(sk
, isk
, uaddr
, addr_len
);
425 if (isk
->inet_num
!= 0)
429 ping_set_saddr(sk
, uaddr
);
430 snum
= ntohs(((struct sockaddr_in
*)uaddr
)->sin_port
);
431 if (ping_get_port(sk
, snum
) != 0) {
432 ping_clear_saddr(sk
, dif
);
436 pr_debug("after bind(): num = %d, dif = %d\n",
438 (int)sk
->sk_bound_dev_if
);
441 if ((sk
->sk_family
== AF_INET
&& isk
->inet_rcv_saddr
) ||
442 (sk
->sk_family
== AF_INET6
&&
443 !ipv6_addr_any(&inet6_sk(sk
)->rcv_saddr
)))
444 sk
->sk_userlocks
|= SOCK_BINDADDR_LOCK
;
447 sk
->sk_userlocks
|= SOCK_BINDPORT_LOCK
;
448 isk
->inet_sport
= htons(isk
->inet_num
);
452 #if IS_ENABLED(CONFIG_IPV6)
453 if (sk
->sk_family
== AF_INET6
)
454 memset(&inet6_sk(sk
)->daddr
, 0, sizeof(inet6_sk(sk
)->daddr
));
460 pr_debug("ping_v4_bind -> %d\n", err
);
463 EXPORT_SYMBOL_GPL(ping_bind
);
466 * Is this a supported type of ICMP message?
469 static inline int ping_supported(int family
, int type
, int code
)
471 return (family
== AF_INET
&& type
== ICMP_ECHO
&& code
== 0) ||
472 (family
== AF_INET6
&& type
== ICMPV6_ECHO_REQUEST
&& code
== 0);
476 * This routine is called by the ICMP module when it gets some
477 * sort of error condition.
480 void ping_err(struct sk_buff
*skb
, int offset
, u32 info
)
483 struct icmphdr
*icmph
;
484 struct inet_sock
*inet_sock
;
487 struct net
*net
= dev_net(skb
->dev
);
492 if (skb
->protocol
== htons(ETH_P_IP
)) {
494 type
= icmp_hdr(skb
)->type
;
495 code
= icmp_hdr(skb
)->code
;
496 icmph
= (struct icmphdr
*)(skb
->data
+ offset
);
497 } else if (skb
->protocol
== htons(ETH_P_IPV6
)) {
499 type
= icmp6_hdr(skb
)->icmp6_type
;
500 code
= icmp6_hdr(skb
)->icmp6_code
;
501 icmph
= (struct icmphdr
*) (skb
->data
+ offset
);
506 /* We assume the packet has already been checked by icmp_unreach */
508 if (!ping_supported(family
, icmph
->type
, icmph
->code
))
511 pr_debug("ping_err(proto=0x%x,type=%d,code=%d,id=%04x,seq=%04x)\n",
512 skb
->protocol
, type
, code
, ntohs(icmph
->un
.echo
.id
),
513 ntohs(icmph
->un
.echo
.sequence
));
515 sk
= ping_lookup(net
, skb
, ntohs(icmph
->un
.echo
.id
));
517 pr_debug("no socket, dropping\n");
518 return; /* No socket for error */
520 pr_debug("err on socket %p\n", sk
);
524 inet_sock
= inet_sk(sk
);
526 if (skb
->protocol
== htons(ETH_P_IP
)) {
529 case ICMP_TIME_EXCEEDED
:
532 case ICMP_SOURCE_QUENCH
:
533 /* This is not a real error but ping wants to see it.
534 * Report it with some fake errno.
538 case ICMP_PARAMETERPROB
:
542 case ICMP_DEST_UNREACH
:
543 if (code
== ICMP_FRAG_NEEDED
) { /* Path MTU discovery */
544 ipv4_sk_update_pmtu(skb
, sk
, info
);
545 if (inet_sock
->pmtudisc
!= IP_PMTUDISC_DONT
) {
553 if (code
<= NR_ICMP_UNREACH
) {
554 harderr
= icmp_err_convert
[code
].fatal
;
555 err
= icmp_err_convert
[code
].errno
;
559 /* See ICMP_SOURCE_QUENCH */
560 ipv4_sk_redirect(skb
, sk
);
564 #if IS_ENABLED(CONFIG_IPV6)
565 } else if (skb
->protocol
== htons(ETH_P_IPV6
)) {
566 harderr
= pingv6_ops
.icmpv6_err_convert(type
, code
, &err
);
571 * RFC1122: OK. Passes ICMP errors back to application, as per
574 if ((family
== AF_INET
&& !inet_sock
->recverr
) ||
575 (family
== AF_INET6
&& !inet6_sk(sk
)->recverr
)) {
576 if (!harderr
|| sk
->sk_state
!= TCP_ESTABLISHED
)
579 if (family
== AF_INET
) {
580 ip_icmp_error(sk
, skb
, err
, 0 /* no remote port */,
582 #if IS_ENABLED(CONFIG_IPV6)
583 } else if (family
== AF_INET6
) {
584 pingv6_ops
.ipv6_icmp_error(sk
, skb
, err
, 0,
590 sk
->sk_error_report(sk
);
594 EXPORT_SYMBOL_GPL(ping_err
);
597 * Copy and checksum an ICMP Echo packet from user space into a buffer
598 * starting from the payload.
601 int ping_getfrag(void *from
, char *to
,
602 int offset
, int fraglen
, int odd
, struct sk_buff
*skb
)
604 struct pingfakehdr
*pfh
= (struct pingfakehdr
*)from
;
607 if (fraglen
< sizeof(struct icmphdr
))
609 if (csum_partial_copy_fromiovecend(to
+ sizeof(struct icmphdr
),
610 pfh
->iov
, 0, fraglen
- sizeof(struct icmphdr
),
613 } else if (offset
< sizeof(struct icmphdr
)) {
616 if (csum_partial_copy_fromiovecend
617 (to
, pfh
->iov
, offset
- sizeof(struct icmphdr
),
618 fraglen
, &pfh
->wcheck
))
622 #if IS_ENABLED(CONFIG_IPV6)
623 /* For IPv6, checksum each skb as we go along, as expected by
624 * icmpv6_push_pending_frames. For IPv4, accumulate the checksum in
625 * wcheck, it will be finalized in ping_v4_push_pending_frames.
627 if (pfh
->family
== AF_INET6
) {
628 skb
->csum
= pfh
->wcheck
;
629 skb
->ip_summed
= CHECKSUM_NONE
;
636 EXPORT_SYMBOL_GPL(ping_getfrag
);
638 static int ping_v4_push_pending_frames(struct sock
*sk
, struct pingfakehdr
*pfh
,
641 struct sk_buff
*skb
= skb_peek(&sk
->sk_write_queue
);
643 pfh
->wcheck
= csum_partial((char *)&pfh
->icmph
,
644 sizeof(struct icmphdr
), pfh
->wcheck
);
645 pfh
->icmph
.checksum
= csum_fold(pfh
->wcheck
);
646 memcpy(icmp_hdr(skb
), &pfh
->icmph
, sizeof(struct icmphdr
));
647 skb
->ip_summed
= CHECKSUM_NONE
;
648 return ip_push_pending_frames(sk
, fl4
);
651 int ping_common_sendmsg(int family
, struct msghdr
*msg
, size_t len
,
652 void *user_icmph
, size_t icmph_len
) {
662 /* Mirror BSD error message compatibility */
663 if (msg
->msg_flags
& MSG_OOB
)
667 * Fetch the ICMP header provided by the userland.
668 * iovec is modified! The ICMP header is consumed.
670 if (memcpy_fromiovec(user_icmph
, msg
->msg_iov
, icmph_len
))
673 if (family
== AF_INET
) {
674 type
= ((struct icmphdr
*) user_icmph
)->type
;
675 code
= ((struct icmphdr
*) user_icmph
)->code
;
676 #if IS_ENABLED(CONFIG_IPV6)
677 } else if (family
== AF_INET6
) {
678 type
= ((struct icmp6hdr
*) user_icmph
)->icmp6_type
;
679 code
= ((struct icmp6hdr
*) user_icmph
)->icmp6_code
;
685 if (!ping_supported(family
, type
, code
))
690 EXPORT_SYMBOL_GPL(ping_common_sendmsg
);
692 int ping_v4_sendmsg(struct kiocb
*iocb
, struct sock
*sk
, struct msghdr
*msg
,
695 struct net
*net
= sock_net(sk
);
697 struct inet_sock
*inet
= inet_sk(sk
);
698 struct ipcm_cookie ipc
;
699 struct icmphdr user_icmph
;
700 struct pingfakehdr pfh
;
701 struct rtable
*rt
= NULL
;
702 struct ip_options_data opt_copy
;
704 __be32 saddr
, daddr
, faddr
;
708 pr_debug("ping_v4_sendmsg(sk=%p,sk->num=%u)\n", inet
, inet
->inet_num
);
710 err
= ping_common_sendmsg(AF_INET
, msg
, len
, &user_icmph
,
716 * Get and verify the address.
720 struct sockaddr_in
*usin
= (struct sockaddr_in
*)msg
->msg_name
;
721 if (msg
->msg_namelen
< sizeof(*usin
))
723 if (usin
->sin_family
!= AF_INET
)
724 return -EAFNOSUPPORT
;
725 daddr
= usin
->sin_addr
.s_addr
;
728 if (sk
->sk_state
!= TCP_ESTABLISHED
)
729 return -EDESTADDRREQ
;
730 daddr
= inet
->inet_daddr
;
734 ipc
.addr
= inet
->inet_saddr
;
736 ipc
.oif
= sk
->sk_bound_dev_if
;
739 sock_tx_timestamp(sk
, &ipc
.tx_flags
);
741 if (msg
->msg_controllen
) {
742 err
= ip_cmsg_send(sock_net(sk
), msg
, &ipc
);
751 struct ip_options_rcu
*inet_opt
;
754 inet_opt
= rcu_dereference(inet
->inet_opt
);
756 memcpy(&opt_copy
, inet_opt
,
757 sizeof(*inet_opt
) + inet_opt
->opt
.optlen
);
758 ipc
.opt
= &opt_copy
.opt
;
764 ipc
.addr
= faddr
= daddr
;
766 if (ipc
.opt
&& ipc
.opt
->opt
.srr
) {
769 faddr
= ipc
.opt
->opt
.faddr
;
771 tos
= RT_TOS(inet
->tos
);
772 if (sock_flag(sk
, SOCK_LOCALROUTE
) ||
773 (msg
->msg_flags
& MSG_DONTROUTE
) ||
774 (ipc
.opt
&& ipc
.opt
->opt
.is_strictroute
)) {
778 if (ipv4_is_multicast(daddr
)) {
780 ipc
.oif
= inet
->mc_index
;
782 saddr
= inet
->mc_addr
;
784 ipc
.oif
= inet
->uc_index
;
786 flowi4_init_output(&fl4
, ipc
.oif
, sk
->sk_mark
, tos
,
787 RT_SCOPE_UNIVERSE
, sk
->sk_protocol
,
788 inet_sk_flowi_flags(sk
), faddr
, saddr
, 0, 0);
790 security_sk_classify_flow(sk
, flowi4_to_flowi(&fl4
));
791 rt
= ip_route_output_flow(net
, &fl4
, sk
);
795 if (err
== -ENETUNREACH
)
796 IP_INC_STATS(net
, IPSTATS_MIB_OUTNOROUTES
);
801 if ((rt
->rt_flags
& RTCF_BROADCAST
) &&
802 !sock_flag(sk
, SOCK_BROADCAST
))
805 if (msg
->msg_flags
& MSG_CONFIRM
)
810 ipc
.addr
= fl4
.daddr
;
814 pfh
.icmph
.type
= user_icmph
.type
; /* already checked */
815 pfh
.icmph
.code
= user_icmph
.code
; /* ditto */
816 pfh
.icmph
.checksum
= 0;
817 pfh
.icmph
.un
.echo
.id
= inet
->inet_sport
;
818 pfh
.icmph
.un
.echo
.sequence
= user_icmph
.un
.echo
.sequence
;
819 pfh
.iov
= msg
->msg_iov
;
821 pfh
.family
= AF_INET
;
823 err
= ip_append_data(sk
, &fl4
, ping_getfrag
, &pfh
, len
,
824 0, &ipc
, &rt
, msg
->msg_flags
);
826 ip_flush_pending_frames(sk
);
828 err
= ping_v4_push_pending_frames(sk
, &pfh
, &fl4
);
836 icmp_out_count(sock_net(sk
), user_icmph
.type
);
842 dst_confirm(&rt
->dst
);
843 if (!(msg
->msg_flags
& MSG_PROBE
) || len
)
844 goto back_from_confirm
;
849 int ping_recvmsg(struct kiocb
*iocb
, struct sock
*sk
, struct msghdr
*msg
,
850 size_t len
, int noblock
, int flags
, int *addr_len
)
852 struct inet_sock
*isk
= inet_sk(sk
);
853 int family
= sk
->sk_family
;
857 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk
, isk
->inet_num
);
863 if (flags
& MSG_ERRQUEUE
) {
864 if (family
== AF_INET
) {
865 return ip_recv_error(sk
, msg
, len
, addr_len
);
866 #if IS_ENABLED(CONFIG_IPV6)
867 } else if (family
== AF_INET6
) {
868 return pingv6_ops
.ipv6_recv_error(sk
, msg
, 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_iovec(skb
, 0, msg
->msg_iov
, copied
);
889 sock_recv_timestamp(msg
, sk
, skb
);
891 /* Copy the address and add cmsg data. */
892 if (family
== AF_INET
) {
893 struct sockaddr_in
*sin
= (struct sockaddr_in
*)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 struct sockaddr_in6
*sin6
=
911 (struct sockaddr_in6
*)msg
->msg_name
;
914 sin6
->sin6_family
= AF_INET6
;
916 sin6
->sin6_addr
= ip6
->saddr
;
917 sin6
->sin6_flowinfo
= 0;
919 sin6
->sin6_flowinfo
= ip6_flowinfo(ip6
);
920 sin6
->sin6_scope_id
=
921 ipv6_iface_scope_id(&sin6
->sin6_addr
,
923 *addr_len
= sizeof(*sin6
);
926 if (inet6_sk(sk
)->rxopt
.all
)
927 pingv6_ops
.ip6_datagram_recv_ctl(sk
, msg
, skb
);
936 skb_free_datagram(sk
, skb
);
938 pr_debug("ping_recvmsg -> %d\n", err
);
941 EXPORT_SYMBOL_GPL(ping_recvmsg
);
943 int ping_queue_rcv_skb(struct sock
*sk
, struct sk_buff
*skb
)
945 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
946 inet_sk(sk
), inet_sk(sk
)->inet_num
, skb
);
947 if (sock_queue_rcv_skb(sk
, skb
) < 0) {
949 pr_debug("ping_queue_rcv_skb -> failed\n");
954 EXPORT_SYMBOL_GPL(ping_queue_rcv_skb
);
958 * All we need to do is get the socket.
961 void ping_rcv(struct sk_buff
*skb
)
964 struct net
*net
= dev_net(skb
->dev
);
965 struct icmphdr
*icmph
= icmp_hdr(skb
);
967 /* We assume the packet has already been checked by icmp_rcv */
969 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
970 skb
, ntohs(icmph
->un
.echo
.id
), ntohs(icmph
->un
.echo
.sequence
));
972 /* Push ICMP header back */
973 skb_push(skb
, skb
->data
- (u8
*)icmph
);
975 sk
= ping_lookup(net
, skb
, ntohs(icmph
->un
.echo
.id
));
977 struct sk_buff
*skb2
= skb_clone(skb
, GFP_ATOMIC
);
979 pr_debug("rcv on socket %p\n", sk
);
981 ping_queue_rcv_skb(sk
, skb2
);
985 pr_debug("no socket, dropping\n");
987 /* We're called from icmp_rcv(). kfree_skb() is done there. */
989 EXPORT_SYMBOL_GPL(ping_rcv
);
991 struct proto ping_prot
= {
993 .owner
= THIS_MODULE
,
994 .init
= ping_init_sock
,
996 .connect
= ip4_datagram_connect
,
997 .disconnect
= udp_disconnect
,
998 .setsockopt
= ip_setsockopt
,
999 .getsockopt
= ip_getsockopt
,
1000 .sendmsg
= ping_v4_sendmsg
,
1001 .recvmsg
= ping_recvmsg
,
1003 .backlog_rcv
= ping_queue_rcv_skb
,
1004 .release_cb
= ip4_datagram_release_cb
,
1006 .unhash
= ping_unhash
,
1007 .get_port
= ping_get_port
,
1008 .obj_size
= sizeof(struct inet_sock
),
1010 EXPORT_SYMBOL(ping_prot
);
1012 #ifdef CONFIG_PROC_FS
1014 static struct sock
*ping_get_first(struct seq_file
*seq
, int start
)
1017 struct ping_iter_state
*state
= seq
->private;
1018 struct net
*net
= seq_file_net(seq
);
1020 for (state
->bucket
= start
; state
->bucket
< PING_HTABLE_SIZE
;
1022 struct hlist_nulls_node
*node
;
1023 struct hlist_nulls_head
*hslot
;
1025 hslot
= &ping_table
.hash
[state
->bucket
];
1027 if (hlist_nulls_empty(hslot
))
1030 sk_nulls_for_each(sk
, node
, hslot
) {
1031 if (net_eq(sock_net(sk
), net
) &&
1032 sk
->sk_family
== state
->family
)
1041 static struct sock
*ping_get_next(struct seq_file
*seq
, struct sock
*sk
)
1043 struct ping_iter_state
*state
= seq
->private;
1044 struct net
*net
= seq_file_net(seq
);
1047 sk
= sk_nulls_next(sk
);
1048 } while (sk
&& (!net_eq(sock_net(sk
), net
)));
1051 return ping_get_first(seq
, state
->bucket
+ 1);
1055 static struct sock
*ping_get_idx(struct seq_file
*seq
, loff_t pos
)
1057 struct sock
*sk
= ping_get_first(seq
, 0);
1060 while (pos
&& (sk
= ping_get_next(seq
, sk
)) != NULL
)
1062 return pos
? NULL
: sk
;
1065 void *ping_seq_start(struct seq_file
*seq
, loff_t
*pos
, sa_family_t family
)
1067 struct ping_iter_state
*state
= seq
->private;
1069 state
->family
= family
;
1071 read_lock_bh(&ping_table
.lock
);
1073 return *pos
? ping_get_idx(seq
, *pos
-1) : SEQ_START_TOKEN
;
1075 EXPORT_SYMBOL_GPL(ping_seq_start
);
1077 static void *ping_v4_seq_start(struct seq_file
*seq
, loff_t
*pos
)
1079 return ping_seq_start(seq
, pos
, AF_INET
);
1082 void *ping_seq_next(struct seq_file
*seq
, void *v
, loff_t
*pos
)
1086 if (v
== SEQ_START_TOKEN
)
1087 sk
= ping_get_idx(seq
, 0);
1089 sk
= ping_get_next(seq
, v
);
1094 EXPORT_SYMBOL_GPL(ping_seq_next
);
1096 void ping_seq_stop(struct seq_file
*seq
, void *v
)
1098 read_unlock_bh(&ping_table
.lock
);
1100 EXPORT_SYMBOL_GPL(ping_seq_stop
);
1102 static void ping_v4_format_sock(struct sock
*sp
, struct seq_file
*f
,
1103 int bucket
, int *len
)
1105 struct inet_sock
*inet
= inet_sk(sp
);
1106 __be32 dest
= inet
->inet_daddr
;
1107 __be32 src
= inet
->inet_rcv_saddr
;
1108 __u16 destp
= ntohs(inet
->inet_dport
);
1109 __u16 srcp
= ntohs(inet
->inet_sport
);
1111 seq_printf(f
, "%5d: %08X:%04X %08X:%04X"
1112 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d%n",
1113 bucket
, src
, srcp
, dest
, destp
, sp
->sk_state
,
1114 sk_wmem_alloc_get(sp
),
1115 sk_rmem_alloc_get(sp
),
1117 from_kuid_munged(seq_user_ns(f
), sock_i_uid(sp
)),
1119 atomic_read(&sp
->sk_refcnt
), sp
,
1120 atomic_read(&sp
->sk_drops
), len
);
1123 static int ping_v4_seq_show(struct seq_file
*seq
, void *v
)
1125 if (v
== SEQ_START_TOKEN
)
1126 seq_printf(seq
, "%-127s\n",
1127 " sl local_address rem_address st tx_queue "
1128 "rx_queue tr tm->when retrnsmt uid timeout "
1129 "inode ref pointer drops");
1131 struct ping_iter_state
*state
= seq
->private;
1134 ping_v4_format_sock(v
, seq
, state
->bucket
, &len
);
1135 seq_printf(seq
, "%*s\n", 127 - len
, "");
1140 static const struct seq_operations ping_v4_seq_ops
= {
1141 .show
= ping_v4_seq_show
,
1142 .start
= ping_v4_seq_start
,
1143 .next
= ping_seq_next
,
1144 .stop
= ping_seq_stop
,
1147 static int ping_seq_open(struct inode
*inode
, struct file
*file
)
1149 struct ping_seq_afinfo
*afinfo
= PDE_DATA(inode
);
1150 return seq_open_net(inode
, file
, &afinfo
->seq_ops
,
1151 sizeof(struct ping_iter_state
));
1154 const struct file_operations ping_seq_fops
= {
1155 .open
= ping_seq_open
,
1157 .llseek
= seq_lseek
,
1158 .release
= seq_release_net
,
1160 EXPORT_SYMBOL_GPL(ping_seq_fops
);
1162 static struct ping_seq_afinfo ping_v4_seq_afinfo
= {
1165 .seq_fops
= &ping_seq_fops
,
1167 .start
= ping_v4_seq_start
,
1168 .show
= ping_v4_seq_show
,
1169 .next
= ping_seq_next
,
1170 .stop
= ping_seq_stop
,
1174 int ping_proc_register(struct net
*net
, struct ping_seq_afinfo
*afinfo
)
1176 struct proc_dir_entry
*p
;
1177 p
= proc_create_data(afinfo
->name
, S_IRUGO
, net
->proc_net
,
1178 afinfo
->seq_fops
, afinfo
);
1183 EXPORT_SYMBOL_GPL(ping_proc_register
);
1185 void ping_proc_unregister(struct net
*net
, struct ping_seq_afinfo
*afinfo
)
1187 remove_proc_entry(afinfo
->name
, net
->proc_net
);
1189 EXPORT_SYMBOL_GPL(ping_proc_unregister
);
1191 static int __net_init
ping_v4_proc_init_net(struct net
*net
)
1193 return ping_proc_register(net
, &ping_v4_seq_afinfo
);
1196 static void __net_exit
ping_v4_proc_exit_net(struct net
*net
)
1198 ping_proc_unregister(net
, &ping_v4_seq_afinfo
);
1201 static struct pernet_operations ping_v4_net_ops
= {
1202 .init
= ping_v4_proc_init_net
,
1203 .exit
= ping_v4_proc_exit_net
,
1206 int __init
ping_proc_init(void)
1208 return register_pernet_subsys(&ping_v4_net_ops
);
1211 void ping_proc_exit(void)
1213 unregister_pernet_subsys(&ping_v4_net_ops
);
1218 void __init
ping_init(void)
1222 for (i
= 0; i
< PING_HTABLE_SIZE
; i
++)
1223 INIT_HLIST_NULLS_HEAD(&ping_table
.hash
[i
], i
);
1224 rwlock_init(&ping_table
.lock
);