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.
6 * RAW - implementation of IP "raw" sockets.
9 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12 * Alan Cox : verify_area() fixed up
13 * Alan Cox : ICMP error handling
14 * Alan Cox : EMSGSIZE if you send too big a packet
15 * Alan Cox : Now uses generic datagrams and shared
16 * skbuff library. No more peek crashes,
18 * Alan Cox : Checks sk->broadcast.
19 * Alan Cox : Uses skb_free_datagram/skb_copy_datagram
20 * Alan Cox : Raw passes ip options too
21 * Alan Cox : Setsocketopt added
22 * Alan Cox : Fixed error return for broadcasts
23 * Alan Cox : Removed wake_up calls
24 * Alan Cox : Use ttl/tos
25 * Alan Cox : Cleaned up old debugging
26 * Alan Cox : Use new kernel side addresses
27 * Arnt Gulbrandsen : Fixed MSG_DONTROUTE in raw sockets.
28 * Alan Cox : BSD style RAW socket demultiplexing.
29 * Alan Cox : Beginnings of mrouted support.
30 * Alan Cox : Added IP_HDRINCL option.
31 * Alan Cox : Skip broadcast check if BSDism set.
32 * David S. Miller : New socket lookup architecture.
34 * This program is free software; you can redistribute it and/or
35 * modify it under the terms of the GNU General Public License
36 * as published by the Free Software Foundation; either version
37 * 2 of the License, or (at your option) any later version.
40 #include <linux/types.h>
41 #include <linux/atomic.h>
42 #include <asm/byteorder.h>
43 #include <asm/current.h>
44 #include <asm/uaccess.h>
45 #include <asm/ioctls.h>
46 #include <linux/stddef.h>
47 #include <linux/slab.h>
48 #include <linux/errno.h>
49 #include <linux/aio.h>
50 #include <linux/kernel.h>
51 #include <linux/export.h>
52 #include <linux/spinlock.h>
53 #include <linux/sockios.h>
54 #include <linux/socket.h>
56 #include <linux/mroute.h>
57 #include <linux/netdevice.h>
58 #include <linux/in_route.h>
59 #include <linux/route.h>
60 #include <linux/skbuff.h>
61 #include <linux/igmp.h>
62 #include <net/net_namespace.h>
66 #include <linux/net.h>
72 #include <net/tcp_states.h>
73 #include <net/inet_common.h>
74 #include <net/checksum.h>
76 #include <linux/rtnetlink.h>
77 #include <linux/proc_fs.h>
78 #include <linux/seq_file.h>
79 #include <linux/netfilter.h>
80 #include <linux/netfilter_ipv4.h>
81 #include <linux/compat.h>
83 static struct raw_hashinfo raw_v4_hashinfo
= {
84 .lock
= __RW_LOCK_UNLOCKED(raw_v4_hashinfo
.lock
),
87 void raw_hash_sk(struct sock
*sk
)
89 struct raw_hashinfo
*h
= sk
->sk_prot
->h
.raw_hash
;
90 struct hlist_head
*head
;
92 head
= &h
->ht
[inet_sk(sk
)->inet_num
& (RAW_HTABLE_SIZE
- 1)];
94 write_lock_bh(&h
->lock
);
95 sk_add_node(sk
, head
);
96 sock_prot_inuse_add(sock_net(sk
), sk
->sk_prot
, 1);
97 write_unlock_bh(&h
->lock
);
99 EXPORT_SYMBOL_GPL(raw_hash_sk
);
101 void raw_unhash_sk(struct sock
*sk
)
103 struct raw_hashinfo
*h
= sk
->sk_prot
->h
.raw_hash
;
105 write_lock_bh(&h
->lock
);
106 if (sk_del_node_init(sk
))
107 sock_prot_inuse_add(sock_net(sk
), sk
->sk_prot
, -1);
108 write_unlock_bh(&h
->lock
);
110 EXPORT_SYMBOL_GPL(raw_unhash_sk
);
112 static struct sock
*__raw_v4_lookup(struct net
*net
, struct sock
*sk
,
113 unsigned short num
, __be32 raddr
, __be32 laddr
, int dif
)
115 sk_for_each_from(sk
) {
116 struct inet_sock
*inet
= inet_sk(sk
);
118 if (net_eq(sock_net(sk
), net
) && inet
->inet_num
== num
&&
119 !(inet
->inet_daddr
&& inet
->inet_daddr
!= raddr
) &&
120 !(inet
->inet_rcv_saddr
&& inet
->inet_rcv_saddr
!= laddr
) &&
121 !(sk
->sk_bound_dev_if
&& sk
->sk_bound_dev_if
!= dif
))
122 goto found
; /* gotcha */
133 static int icmp_filter(const struct sock
*sk
, const struct sk_buff
*skb
)
136 const struct icmphdr
*hdr
;
138 hdr
= skb_header_pointer(skb
, skb_transport_offset(skb
),
139 sizeof(_hdr
), &_hdr
);
143 if (hdr
->type
< 32) {
144 __u32 data
= raw_sk(sk
)->filter
.data
;
146 return ((1U << hdr
->type
) & data
) != 0;
149 /* Do not block unknown ICMP types */
153 /* IP input processing comes here for RAW socket delivery.
154 * Caller owns SKB, so we must make clones.
156 * RFC 1122: SHOULD pass TOS value up to the transport layer.
157 * -> It does. And not only TOS, but all IP header.
159 static int raw_v4_input(struct sk_buff
*skb
, const struct iphdr
*iph
, int hash
)
162 struct hlist_head
*head
;
166 read_lock(&raw_v4_hashinfo
.lock
);
167 head
= &raw_v4_hashinfo
.ht
[hash
];
168 if (hlist_empty(head
))
171 net
= dev_net(skb
->dev
);
172 sk
= __raw_v4_lookup(net
, __sk_head(head
), iph
->protocol
,
173 iph
->saddr
, iph
->daddr
,
178 if ((iph
->protocol
!= IPPROTO_ICMP
|| !icmp_filter(sk
, skb
)) &&
179 ip_mc_sf_allow(sk
, iph
->daddr
, iph
->saddr
,
180 skb
->dev
->ifindex
)) {
181 struct sk_buff
*clone
= skb_clone(skb
, GFP_ATOMIC
);
183 /* Not releasing hash table! */
187 sk
= __raw_v4_lookup(net
, sk_next(sk
), iph
->protocol
,
188 iph
->saddr
, iph
->daddr
,
192 read_unlock(&raw_v4_hashinfo
.lock
);
196 int raw_local_deliver(struct sk_buff
*skb
, int protocol
)
201 hash
= protocol
& (RAW_HTABLE_SIZE
- 1);
202 raw_sk
= sk_head(&raw_v4_hashinfo
.ht
[hash
]);
204 /* If there maybe a raw socket we must check - if not we
207 if (raw_sk
&& !raw_v4_input(skb
, ip_hdr(skb
), hash
))
210 return raw_sk
!= NULL
;
214 static void raw_err(struct sock
*sk
, struct sk_buff
*skb
, u32 info
)
216 struct inet_sock
*inet
= inet_sk(sk
);
217 const int type
= icmp_hdr(skb
)->type
;
218 const int code
= icmp_hdr(skb
)->code
;
222 if (type
== ICMP_DEST_UNREACH
&& code
== ICMP_FRAG_NEEDED
)
223 ipv4_sk_update_pmtu(skb
, sk
, info
);
224 else if (type
== ICMP_REDIRECT
) {
225 ipv4_sk_redirect(skb
, sk
);
229 /* Report error on raw socket, if:
230 1. User requested ip_recverr.
231 2. Socket is connected (otherwise the error indication
232 is useless without ip_recverr and error is hard.
234 if (!inet
->recverr
&& sk
->sk_state
!= TCP_ESTABLISHED
)
239 case ICMP_TIME_EXCEEDED
:
242 case ICMP_SOURCE_QUENCH
:
244 case ICMP_PARAMETERPROB
:
248 case ICMP_DEST_UNREACH
:
250 if (code
> NR_ICMP_UNREACH
)
252 err
= icmp_err_convert
[code
].errno
;
253 harderr
= icmp_err_convert
[code
].fatal
;
254 if (code
== ICMP_FRAG_NEEDED
) {
255 harderr
= inet
->pmtudisc
!= IP_PMTUDISC_DONT
;
261 const struct iphdr
*iph
= (const struct iphdr
*)skb
->data
;
262 u8
*payload
= skb
->data
+ (iph
->ihl
<< 2);
266 ip_icmp_error(sk
, skb
, err
, 0, info
, payload
);
269 if (inet
->recverr
|| harderr
) {
271 sk
->sk_error_report(sk
);
275 void raw_icmp_error(struct sk_buff
*skb
, int protocol
, u32 info
)
279 const struct iphdr
*iph
;
282 hash
= protocol
& (RAW_HTABLE_SIZE
- 1);
284 read_lock(&raw_v4_hashinfo
.lock
);
285 raw_sk
= sk_head(&raw_v4_hashinfo
.ht
[hash
]);
286 if (raw_sk
!= NULL
) {
287 iph
= (const struct iphdr
*)skb
->data
;
288 net
= dev_net(skb
->dev
);
290 while ((raw_sk
= __raw_v4_lookup(net
, raw_sk
, protocol
,
291 iph
->daddr
, iph
->saddr
,
292 skb
->dev
->ifindex
)) != NULL
) {
293 raw_err(raw_sk
, skb
, info
);
294 raw_sk
= sk_next(raw_sk
);
295 iph
= (const struct iphdr
*)skb
->data
;
298 read_unlock(&raw_v4_hashinfo
.lock
);
301 static int raw_rcv_skb(struct sock
*sk
, struct sk_buff
*skb
)
303 /* Charge it to the socket. */
305 ipv4_pktinfo_prepare(sk
, skb
);
306 if (sock_queue_rcv_skb(sk
, skb
) < 0) {
311 return NET_RX_SUCCESS
;
314 int raw_rcv(struct sock
*sk
, struct sk_buff
*skb
)
316 if (!xfrm4_policy_check(sk
, XFRM_POLICY_IN
, skb
)) {
317 atomic_inc(&sk
->sk_drops
);
323 skb_push(skb
, skb
->data
- skb_network_header(skb
));
325 raw_rcv_skb(sk
, skb
);
329 static int raw_send_hdrinc(struct sock
*sk
, struct flowi4
*fl4
,
330 void *from
, size_t length
,
334 struct inet_sock
*inet
= inet_sk(sk
);
335 struct net
*net
= sock_net(sk
);
340 struct rtable
*rt
= *rtp
;
343 if (length
> rt
->dst
.dev
->mtu
) {
344 ip_local_error(sk
, EMSGSIZE
, fl4
->daddr
, inet
->inet_dport
,
351 hlen
= LL_RESERVED_SPACE(rt
->dst
.dev
);
352 tlen
= rt
->dst
.dev
->needed_tailroom
;
353 skb
= sock_alloc_send_skb(sk
,
354 length
+ hlen
+ tlen
+ 15,
355 flags
& MSG_DONTWAIT
, &err
);
358 skb_reserve(skb
, hlen
);
360 skb
->priority
= sk
->sk_priority
;
361 skb
->mark
= sk
->sk_mark
;
362 skb_dst_set(skb
, &rt
->dst
);
365 skb_reset_network_header(skb
);
367 skb_put(skb
, length
);
369 skb
->ip_summed
= CHECKSUM_NONE
;
371 sock_tx_timestamp(sk
, &skb_shinfo(skb
)->tx_flags
);
373 skb
->transport_header
= skb
->network_header
;
375 if (memcpy_fromiovecend((void *)iph
, from
, 0, length
))
378 iphlen
= iph
->ihl
* 4;
381 * We don't want to modify the ip header, but we do need to
382 * be sure that it won't cause problems later along the network
383 * stack. Specifically we want to make sure that iph->ihl is a
384 * sane value. If ihl points beyond the length of the buffer passed
385 * in, reject the frame as invalid
391 if (iphlen
>= sizeof(*iph
)) {
393 iph
->saddr
= fl4
->saddr
;
395 iph
->tot_len
= htons(length
);
397 ip_select_ident(skb
, NULL
);
399 iph
->check
= ip_fast_csum((unsigned char *)iph
, iph
->ihl
);
401 if (iph
->protocol
== IPPROTO_ICMP
)
402 icmp_out_count(net
, ((struct icmphdr
*)
403 skb_transport_header(skb
))->type
);
405 err
= NF_HOOK(NFPROTO_IPV4
, NF_INET_LOCAL_OUT
, skb
, NULL
,
406 rt
->dst
.dev
, dst_output
);
408 err
= net_xmit_errno(err
);
417 IP_INC_STATS(net
, IPSTATS_MIB_OUTDISCARDS
);
418 if (err
== -ENOBUFS
&& !inet
->recverr
)
423 static int raw_probe_proto_opt(struct flowi4
*fl4
, struct msghdr
*msg
)
426 u8 __user
*type
= NULL
;
427 u8 __user
*code
= NULL
;
434 for (i
= 0; i
< msg
->msg_iovlen
; i
++) {
435 iov
= &msg
->msg_iov
[i
];
439 switch (fl4
->flowi4_proto
) {
441 /* check if one-byte field is readable or not. */
442 if (iov
->iov_base
&& iov
->iov_len
< 1)
446 type
= iov
->iov_base
;
447 /* check if code field is readable or not. */
448 if (iov
->iov_len
> 1)
451 code
= iov
->iov_base
;
454 if (get_user(fl4
->fl4_icmp_type
, type
) ||
455 get_user(fl4
->fl4_icmp_code
, code
))
470 static int raw_sendmsg(struct kiocb
*iocb
, struct sock
*sk
, struct msghdr
*msg
,
473 struct inet_sock
*inet
= inet_sk(sk
);
474 struct ipcm_cookie ipc
;
475 struct rtable
*rt
= NULL
;
482 struct ip_options_data opt_copy
;
493 if (msg
->msg_flags
& MSG_OOB
) /* Mirror BSD error message */
494 goto out
; /* compatibility */
497 * Get and verify the address.
500 if (msg
->msg_namelen
) {
501 DECLARE_SOCKADDR(struct sockaddr_in
*, usin
, msg
->msg_name
);
503 if (msg
->msg_namelen
< sizeof(*usin
))
505 if (usin
->sin_family
!= AF_INET
) {
506 pr_info_once("%s: %s forgot to set AF_INET. Fix it!\n",
507 __func__
, current
->comm
);
509 if (usin
->sin_family
)
512 daddr
= usin
->sin_addr
.s_addr
;
513 /* ANK: I did not forget to get protocol from port field.
514 * I just do not know, who uses this weirdness.
515 * IP_HDRINCL is much more convenient.
519 if (sk
->sk_state
!= TCP_ESTABLISHED
)
521 daddr
= inet
->inet_daddr
;
524 ipc
.addr
= inet
->inet_saddr
;
529 ipc
.oif
= sk
->sk_bound_dev_if
;
531 if (msg
->msg_controllen
) {
532 err
= ip_cmsg_send(sock_net(sk
), msg
, &ipc
, false);
543 struct ip_options_rcu
*inet_opt
;
546 inet_opt
= rcu_dereference(inet
->inet_opt
);
548 memcpy(&opt_copy
, inet_opt
,
549 sizeof(*inet_opt
) + inet_opt
->opt
.optlen
);
550 ipc
.opt
= &opt_copy
.opt
;
557 /* Linux does not mangle headers on raw sockets,
558 * so that IP options + IP_HDRINCL is non-sense.
562 if (ipc
.opt
->opt
.srr
) {
565 daddr
= ipc
.opt
->opt
.faddr
;
568 tos
= get_rtconn_flags(&ipc
, sk
);
569 if (msg
->msg_flags
& MSG_DONTROUTE
)
572 if (ipv4_is_multicast(daddr
)) {
574 ipc
.oif
= inet
->mc_index
;
576 saddr
= inet
->mc_addr
;
578 ipc
.oif
= inet
->uc_index
;
580 flowi4_init_output(&fl4
, ipc
.oif
, sk
->sk_mark
, tos
,
582 inet
->hdrincl
? IPPROTO_RAW
: sk
->sk_protocol
,
583 inet_sk_flowi_flags(sk
) |
584 (inet
->hdrincl
? FLOWI_FLAG_KNOWN_NH
: 0),
587 if (!inet
->hdrincl
) {
588 err
= raw_probe_proto_opt(&fl4
, msg
);
593 security_sk_classify_flow(sk
, flowi4_to_flowi(&fl4
));
594 rt
= ip_route_output_flow(sock_net(sk
), &fl4
, sk
);
602 if (rt
->rt_flags
& RTCF_BROADCAST
&& !sock_flag(sk
, SOCK_BROADCAST
))
605 if (msg
->msg_flags
& MSG_CONFIRM
)
610 err
= raw_send_hdrinc(sk
, &fl4
, msg
->msg_iov
, len
,
611 &rt
, msg
->msg_flags
);
614 sock_tx_timestamp(sk
, &ipc
.tx_flags
);
617 ipc
.addr
= fl4
.daddr
;
619 err
= ip_append_data(sk
, &fl4
, ip_generic_getfrag
,
620 msg
->msg_iov
, len
, 0,
621 &ipc
, &rt
, msg
->msg_flags
);
623 ip_flush_pending_frames(sk
);
624 else if (!(msg
->msg_flags
& MSG_MORE
)) {
625 err
= ip_push_pending_frames(sk
, &fl4
);
626 if (err
== -ENOBUFS
&& !inet
->recverr
)
642 dst_confirm(&rt
->dst
);
643 if (!(msg
->msg_flags
& MSG_PROBE
) || len
)
644 goto back_from_confirm
;
649 static void raw_close(struct sock
*sk
, long timeout
)
652 * Raw sockets may have direct kernel references. Kill them.
654 ip_ra_control(sk
, 0, NULL
);
656 sk_common_release(sk
);
659 static void raw_destroy(struct sock
*sk
)
662 ip_flush_pending_frames(sk
);
666 /* This gets rid of all the nasties in af_inet. -DaveM */
667 static int raw_bind(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
)
669 struct inet_sock
*inet
= inet_sk(sk
);
670 struct sockaddr_in
*addr
= (struct sockaddr_in
*) uaddr
;
674 if (sk
->sk_state
!= TCP_CLOSE
|| addr_len
< sizeof(struct sockaddr_in
))
676 chk_addr_ret
= inet_addr_type(sock_net(sk
), addr
->sin_addr
.s_addr
);
677 ret
= -EADDRNOTAVAIL
;
678 if (addr
->sin_addr
.s_addr
&& chk_addr_ret
!= RTN_LOCAL
&&
679 chk_addr_ret
!= RTN_MULTICAST
&& chk_addr_ret
!= RTN_BROADCAST
)
681 inet
->inet_rcv_saddr
= inet
->inet_saddr
= addr
->sin_addr
.s_addr
;
682 if (chk_addr_ret
== RTN_MULTICAST
|| chk_addr_ret
== RTN_BROADCAST
)
683 inet
->inet_saddr
= 0; /* Use device */
690 * This should be easy, if there is something there
691 * we return it, otherwise we block.
694 static int raw_recvmsg(struct kiocb
*iocb
, struct sock
*sk
, struct msghdr
*msg
,
695 size_t len
, int noblock
, int flags
, int *addr_len
)
697 struct inet_sock
*inet
= inet_sk(sk
);
699 int err
= -EOPNOTSUPP
;
700 DECLARE_SOCKADDR(struct sockaddr_in
*, sin
, msg
->msg_name
);
706 if (flags
& MSG_ERRQUEUE
) {
707 err
= ip_recv_error(sk
, msg
, len
, addr_len
);
711 skb
= skb_recv_datagram(sk
, flags
, noblock
, &err
);
717 msg
->msg_flags
|= MSG_TRUNC
;
721 err
= skb_copy_datagram_iovec(skb
, 0, msg
->msg_iov
, copied
);
725 sock_recv_ts_and_drops(msg
, sk
, skb
);
727 /* Copy the address. */
729 sin
->sin_family
= AF_INET
;
730 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
732 memset(&sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
733 *addr_len
= sizeof(*sin
);
735 if (inet
->cmsg_flags
)
736 ip_cmsg_recv(msg
, skb
);
737 if (flags
& MSG_TRUNC
)
740 skb_free_datagram(sk
, skb
);
747 static int raw_init(struct sock
*sk
)
749 struct raw_sock
*rp
= raw_sk(sk
);
751 if (inet_sk(sk
)->inet_num
== IPPROTO_ICMP
)
752 memset(&rp
->filter
, 0, sizeof(rp
->filter
));
756 static int raw_seticmpfilter(struct sock
*sk
, char __user
*optval
, int optlen
)
758 if (optlen
> sizeof(struct icmp_filter
))
759 optlen
= sizeof(struct icmp_filter
);
760 if (copy_from_user(&raw_sk(sk
)->filter
, optval
, optlen
))
765 static int raw_geticmpfilter(struct sock
*sk
, char __user
*optval
, int __user
*optlen
)
767 int len
, ret
= -EFAULT
;
769 if (get_user(len
, optlen
))
774 if (len
> sizeof(struct icmp_filter
))
775 len
= sizeof(struct icmp_filter
);
777 if (put_user(len
, optlen
) ||
778 copy_to_user(optval
, &raw_sk(sk
)->filter
, len
))
784 static int do_raw_setsockopt(struct sock
*sk
, int level
, int optname
,
785 char __user
*optval
, unsigned int optlen
)
787 if (optname
== ICMP_FILTER
) {
788 if (inet_sk(sk
)->inet_num
!= IPPROTO_ICMP
)
791 return raw_seticmpfilter(sk
, optval
, optlen
);
796 static int raw_setsockopt(struct sock
*sk
, int level
, int optname
,
797 char __user
*optval
, unsigned int optlen
)
799 if (level
!= SOL_RAW
)
800 return ip_setsockopt(sk
, level
, optname
, optval
, optlen
);
801 return do_raw_setsockopt(sk
, level
, optname
, optval
, optlen
);
805 static int compat_raw_setsockopt(struct sock
*sk
, int level
, int optname
,
806 char __user
*optval
, unsigned int optlen
)
808 if (level
!= SOL_RAW
)
809 return compat_ip_setsockopt(sk
, level
, optname
, optval
, optlen
);
810 return do_raw_setsockopt(sk
, level
, optname
, optval
, optlen
);
814 static int do_raw_getsockopt(struct sock
*sk
, int level
, int optname
,
815 char __user
*optval
, int __user
*optlen
)
817 if (optname
== ICMP_FILTER
) {
818 if (inet_sk(sk
)->inet_num
!= IPPROTO_ICMP
)
821 return raw_geticmpfilter(sk
, optval
, optlen
);
826 static int raw_getsockopt(struct sock
*sk
, int level
, int optname
,
827 char __user
*optval
, int __user
*optlen
)
829 if (level
!= SOL_RAW
)
830 return ip_getsockopt(sk
, level
, optname
, optval
, optlen
);
831 return do_raw_getsockopt(sk
, level
, optname
, optval
, optlen
);
835 static int compat_raw_getsockopt(struct sock
*sk
, int level
, int optname
,
836 char __user
*optval
, int __user
*optlen
)
838 if (level
!= SOL_RAW
)
839 return compat_ip_getsockopt(sk
, level
, optname
, optval
, optlen
);
840 return do_raw_getsockopt(sk
, level
, optname
, optval
, optlen
);
844 static int raw_ioctl(struct sock
*sk
, int cmd
, unsigned long arg
)
848 int amount
= sk_wmem_alloc_get(sk
);
850 return put_user(amount
, (int __user
*)arg
);
856 spin_lock_bh(&sk
->sk_receive_queue
.lock
);
857 skb
= skb_peek(&sk
->sk_receive_queue
);
860 spin_unlock_bh(&sk
->sk_receive_queue
.lock
);
861 return put_user(amount
, (int __user
*)arg
);
865 #ifdef CONFIG_IP_MROUTE
866 return ipmr_ioctl(sk
, cmd
, (void __user
*)arg
);
874 static int compat_raw_ioctl(struct sock
*sk
, unsigned int cmd
, unsigned long arg
)
881 #ifdef CONFIG_IP_MROUTE
882 return ipmr_compat_ioctl(sk
, cmd
, compat_ptr(arg
));
890 struct proto raw_prot
= {
892 .owner
= THIS_MODULE
,
894 .destroy
= raw_destroy
,
895 .connect
= ip4_datagram_connect
,
896 .disconnect
= udp_disconnect
,
899 .setsockopt
= raw_setsockopt
,
900 .getsockopt
= raw_getsockopt
,
901 .sendmsg
= raw_sendmsg
,
902 .recvmsg
= raw_recvmsg
,
904 .backlog_rcv
= raw_rcv_skb
,
905 .release_cb
= ip4_datagram_release_cb
,
907 .unhash
= raw_unhash_sk
,
908 .obj_size
= sizeof(struct raw_sock
),
909 .h
.raw_hash
= &raw_v4_hashinfo
,
911 .compat_setsockopt
= compat_raw_setsockopt
,
912 .compat_getsockopt
= compat_raw_getsockopt
,
913 .compat_ioctl
= compat_raw_ioctl
,
917 #ifdef CONFIG_PROC_FS
918 static struct sock
*raw_get_first(struct seq_file
*seq
)
921 struct raw_iter_state
*state
= raw_seq_private(seq
);
923 for (state
->bucket
= 0; state
->bucket
< RAW_HTABLE_SIZE
;
925 sk_for_each(sk
, &state
->h
->ht
[state
->bucket
])
926 if (sock_net(sk
) == seq_file_net(seq
))
934 static struct sock
*raw_get_next(struct seq_file
*seq
, struct sock
*sk
)
936 struct raw_iter_state
*state
= raw_seq_private(seq
);
942 } while (sk
&& sock_net(sk
) != seq_file_net(seq
));
944 if (!sk
&& ++state
->bucket
< RAW_HTABLE_SIZE
) {
945 sk
= sk_head(&state
->h
->ht
[state
->bucket
]);
951 static struct sock
*raw_get_idx(struct seq_file
*seq
, loff_t pos
)
953 struct sock
*sk
= raw_get_first(seq
);
956 while (pos
&& (sk
= raw_get_next(seq
, sk
)) != NULL
)
958 return pos
? NULL
: sk
;
961 void *raw_seq_start(struct seq_file
*seq
, loff_t
*pos
)
963 struct raw_iter_state
*state
= raw_seq_private(seq
);
965 read_lock(&state
->h
->lock
);
966 return *pos
? raw_get_idx(seq
, *pos
- 1) : SEQ_START_TOKEN
;
968 EXPORT_SYMBOL_GPL(raw_seq_start
);
970 void *raw_seq_next(struct seq_file
*seq
, void *v
, loff_t
*pos
)
974 if (v
== SEQ_START_TOKEN
)
975 sk
= raw_get_first(seq
);
977 sk
= raw_get_next(seq
, v
);
981 EXPORT_SYMBOL_GPL(raw_seq_next
);
983 void raw_seq_stop(struct seq_file
*seq
, void *v
)
985 struct raw_iter_state
*state
= raw_seq_private(seq
);
987 read_unlock(&state
->h
->lock
);
989 EXPORT_SYMBOL_GPL(raw_seq_stop
);
991 static void raw_sock_seq_show(struct seq_file
*seq
, struct sock
*sp
, int i
)
993 struct inet_sock
*inet
= inet_sk(sp
);
994 __be32 dest
= inet
->inet_daddr
,
995 src
= inet
->inet_rcv_saddr
;
997 srcp
= inet
->inet_num
;
999 seq_printf(seq
, "%4d: %08X:%04X %08X:%04X"
1000 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d\n",
1001 i
, src
, srcp
, dest
, destp
, sp
->sk_state
,
1002 sk_wmem_alloc_get(sp
),
1003 sk_rmem_alloc_get(sp
),
1005 from_kuid_munged(seq_user_ns(seq
), sock_i_uid(sp
)),
1007 atomic_read(&sp
->sk_refcnt
), sp
, atomic_read(&sp
->sk_drops
));
1010 static int raw_seq_show(struct seq_file
*seq
, void *v
)
1012 if (v
== SEQ_START_TOKEN
)
1013 seq_printf(seq
, " sl local_address rem_address st tx_queue "
1014 "rx_queue tr tm->when retrnsmt uid timeout "
1015 "inode ref pointer drops\n");
1017 raw_sock_seq_show(seq
, v
, raw_seq_private(seq
)->bucket
);
1021 static const struct seq_operations raw_seq_ops
= {
1022 .start
= raw_seq_start
,
1023 .next
= raw_seq_next
,
1024 .stop
= raw_seq_stop
,
1025 .show
= raw_seq_show
,
1028 int raw_seq_open(struct inode
*ino
, struct file
*file
,
1029 struct raw_hashinfo
*h
, const struct seq_operations
*ops
)
1032 struct raw_iter_state
*i
;
1034 err
= seq_open_net(ino
, file
, ops
, sizeof(struct raw_iter_state
));
1038 i
= raw_seq_private((struct seq_file
*)file
->private_data
);
1042 EXPORT_SYMBOL_GPL(raw_seq_open
);
1044 static int raw_v4_seq_open(struct inode
*inode
, struct file
*file
)
1046 return raw_seq_open(inode
, file
, &raw_v4_hashinfo
, &raw_seq_ops
);
1049 static const struct file_operations raw_seq_fops
= {
1050 .owner
= THIS_MODULE
,
1051 .open
= raw_v4_seq_open
,
1053 .llseek
= seq_lseek
,
1054 .release
= seq_release_net
,
1057 static __net_init
int raw_init_net(struct net
*net
)
1059 if (!proc_create("raw", S_IRUGO
, net
->proc_net
, &raw_seq_fops
))
1065 static __net_exit
void raw_exit_net(struct net
*net
)
1067 remove_proc_entry("raw", net
->proc_net
);
1070 static __net_initdata
struct pernet_operations raw_net_ops
= {
1071 .init
= raw_init_net
,
1072 .exit
= raw_exit_net
,
1075 int __init
raw_proc_init(void)
1077 return register_pernet_subsys(&raw_net_ops
);
1080 void __init
raw_proc_exit(void)
1082 unregister_pernet_subsys(&raw_net_ops
);
1084 #endif /* CONFIG_PROC_FS */