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/kernel.h>
50 #include <linux/export.h>
51 #include <linux/spinlock.h>
52 #include <linux/sockios.h>
53 #include <linux/socket.h>
55 #include <linux/mroute.h>
56 #include <linux/netdevice.h>
57 #include <linux/in_route.h>
58 #include <linux/route.h>
59 #include <linux/skbuff.h>
60 #include <linux/igmp.h>
61 #include <net/net_namespace.h>
65 #include <linux/net.h>
71 #include <net/tcp_states.h>
72 #include <net/inet_common.h>
73 #include <net/checksum.h>
75 #include <linux/rtnetlink.h>
76 #include <linux/proc_fs.h>
77 #include <linux/seq_file.h>
78 #include <linux/netfilter.h>
79 #include <linux/netfilter_ipv4.h>
80 #include <linux/compat.h>
81 #include <linux/uio.h>
92 static struct raw_hashinfo raw_v4_hashinfo
= {
93 .lock
= __RW_LOCK_UNLOCKED(raw_v4_hashinfo
.lock
),
96 void raw_hash_sk(struct sock
*sk
)
98 struct raw_hashinfo
*h
= sk
->sk_prot
->h
.raw_hash
;
99 struct hlist_head
*head
;
101 head
= &h
->ht
[inet_sk(sk
)->inet_num
& (RAW_HTABLE_SIZE
- 1)];
103 write_lock_bh(&h
->lock
);
104 sk_add_node(sk
, head
);
105 sock_prot_inuse_add(sock_net(sk
), sk
->sk_prot
, 1);
106 write_unlock_bh(&h
->lock
);
108 EXPORT_SYMBOL_GPL(raw_hash_sk
);
110 void raw_unhash_sk(struct sock
*sk
)
112 struct raw_hashinfo
*h
= sk
->sk_prot
->h
.raw_hash
;
114 write_lock_bh(&h
->lock
);
115 if (sk_del_node_init(sk
))
116 sock_prot_inuse_add(sock_net(sk
), sk
->sk_prot
, -1);
117 write_unlock_bh(&h
->lock
);
119 EXPORT_SYMBOL_GPL(raw_unhash_sk
);
121 static struct sock
*__raw_v4_lookup(struct net
*net
, struct sock
*sk
,
122 unsigned short num
, __be32 raddr
, __be32 laddr
, int dif
)
124 sk_for_each_from(sk
) {
125 struct inet_sock
*inet
= inet_sk(sk
);
127 if (net_eq(sock_net(sk
), net
) && inet
->inet_num
== num
&&
128 !(inet
->inet_daddr
&& inet
->inet_daddr
!= raddr
) &&
129 !(inet
->inet_rcv_saddr
&& inet
->inet_rcv_saddr
!= laddr
) &&
130 !(sk
->sk_bound_dev_if
&& sk
->sk_bound_dev_if
!= dif
))
131 goto found
; /* gotcha */
142 static int icmp_filter(const struct sock
*sk
, const struct sk_buff
*skb
)
145 const struct icmphdr
*hdr
;
147 hdr
= skb_header_pointer(skb
, skb_transport_offset(skb
),
148 sizeof(_hdr
), &_hdr
);
152 if (hdr
->type
< 32) {
153 __u32 data
= raw_sk(sk
)->filter
.data
;
155 return ((1U << hdr
->type
) & data
) != 0;
158 /* Do not block unknown ICMP types */
162 /* IP input processing comes here for RAW socket delivery.
163 * Caller owns SKB, so we must make clones.
165 * RFC 1122: SHOULD pass TOS value up to the transport layer.
166 * -> It does. And not only TOS, but all IP header.
168 static int raw_v4_input(struct sk_buff
*skb
, const struct iphdr
*iph
, int hash
)
171 struct hlist_head
*head
;
175 read_lock(&raw_v4_hashinfo
.lock
);
176 head
= &raw_v4_hashinfo
.ht
[hash
];
177 if (hlist_empty(head
))
180 net
= dev_net(skb
->dev
);
181 sk
= __raw_v4_lookup(net
, __sk_head(head
), iph
->protocol
,
182 iph
->saddr
, iph
->daddr
,
187 if ((iph
->protocol
!= IPPROTO_ICMP
|| !icmp_filter(sk
, skb
)) &&
188 ip_mc_sf_allow(sk
, iph
->daddr
, iph
->saddr
,
189 skb
->dev
->ifindex
)) {
190 struct sk_buff
*clone
= skb_clone(skb
, GFP_ATOMIC
);
192 /* Not releasing hash table! */
196 sk
= __raw_v4_lookup(net
, sk_next(sk
), iph
->protocol
,
197 iph
->saddr
, iph
->daddr
,
201 read_unlock(&raw_v4_hashinfo
.lock
);
205 int raw_local_deliver(struct sk_buff
*skb
, int protocol
)
210 hash
= protocol
& (RAW_HTABLE_SIZE
- 1);
211 raw_sk
= sk_head(&raw_v4_hashinfo
.ht
[hash
]);
213 /* If there maybe a raw socket we must check - if not we
216 if (raw_sk
&& !raw_v4_input(skb
, ip_hdr(skb
), hash
))
219 return raw_sk
!= NULL
;
223 static void raw_err(struct sock
*sk
, struct sk_buff
*skb
, u32 info
)
225 struct inet_sock
*inet
= inet_sk(sk
);
226 const int type
= icmp_hdr(skb
)->type
;
227 const int code
= icmp_hdr(skb
)->code
;
231 if (type
== ICMP_DEST_UNREACH
&& code
== ICMP_FRAG_NEEDED
)
232 ipv4_sk_update_pmtu(skb
, sk
, info
);
233 else if (type
== ICMP_REDIRECT
) {
234 ipv4_sk_redirect(skb
, sk
);
238 /* Report error on raw socket, if:
239 1. User requested ip_recverr.
240 2. Socket is connected (otherwise the error indication
241 is useless without ip_recverr and error is hard.
243 if (!inet
->recverr
&& sk
->sk_state
!= TCP_ESTABLISHED
)
248 case ICMP_TIME_EXCEEDED
:
251 case ICMP_SOURCE_QUENCH
:
253 case ICMP_PARAMETERPROB
:
257 case ICMP_DEST_UNREACH
:
259 if (code
> NR_ICMP_UNREACH
)
261 err
= icmp_err_convert
[code
].errno
;
262 harderr
= icmp_err_convert
[code
].fatal
;
263 if (code
== ICMP_FRAG_NEEDED
) {
264 harderr
= inet
->pmtudisc
!= IP_PMTUDISC_DONT
;
270 const struct iphdr
*iph
= (const struct iphdr
*)skb
->data
;
271 u8
*payload
= skb
->data
+ (iph
->ihl
<< 2);
275 ip_icmp_error(sk
, skb
, err
, 0, info
, payload
);
278 if (inet
->recverr
|| harderr
) {
280 sk
->sk_error_report(sk
);
284 void raw_icmp_error(struct sk_buff
*skb
, int protocol
, u32 info
)
288 const struct iphdr
*iph
;
291 hash
= protocol
& (RAW_HTABLE_SIZE
- 1);
293 read_lock(&raw_v4_hashinfo
.lock
);
294 raw_sk
= sk_head(&raw_v4_hashinfo
.ht
[hash
]);
296 iph
= (const struct iphdr
*)skb
->data
;
297 net
= dev_net(skb
->dev
);
299 while ((raw_sk
= __raw_v4_lookup(net
, raw_sk
, protocol
,
300 iph
->daddr
, iph
->saddr
,
301 skb
->dev
->ifindex
)) != NULL
) {
302 raw_err(raw_sk
, skb
, info
);
303 raw_sk
= sk_next(raw_sk
);
304 iph
= (const struct iphdr
*)skb
->data
;
307 read_unlock(&raw_v4_hashinfo
.lock
);
310 static int raw_rcv_skb(struct sock
*sk
, struct sk_buff
*skb
)
312 /* Charge it to the socket. */
314 ipv4_pktinfo_prepare(sk
, skb
);
315 if (sock_queue_rcv_skb(sk
, skb
) < 0) {
320 return NET_RX_SUCCESS
;
323 int raw_rcv(struct sock
*sk
, struct sk_buff
*skb
)
325 if (!xfrm4_policy_check(sk
, XFRM_POLICY_IN
, skb
)) {
326 atomic_inc(&sk
->sk_drops
);
332 skb_push(skb
, skb
->data
- skb_network_header(skb
));
334 raw_rcv_skb(sk
, skb
);
338 static int raw_send_hdrinc(struct sock
*sk
, struct flowi4
*fl4
,
339 struct msghdr
*msg
, size_t length
,
343 struct inet_sock
*inet
= inet_sk(sk
);
344 struct net
*net
= sock_net(sk
);
349 struct rtable
*rt
= *rtp
;
352 if (length
> rt
->dst
.dev
->mtu
) {
353 ip_local_error(sk
, EMSGSIZE
, fl4
->daddr
, inet
->inet_dport
,
360 hlen
= LL_RESERVED_SPACE(rt
->dst
.dev
);
361 tlen
= rt
->dst
.dev
->needed_tailroom
;
362 skb
= sock_alloc_send_skb(sk
,
363 length
+ hlen
+ tlen
+ 15,
364 flags
& MSG_DONTWAIT
, &err
);
367 skb_reserve(skb
, hlen
);
369 skb
->priority
= sk
->sk_priority
;
370 skb
->mark
= sk
->sk_mark
;
371 skb_dst_set(skb
, &rt
->dst
);
374 skb_reset_network_header(skb
);
376 skb_put(skb
, length
);
378 skb
->ip_summed
= CHECKSUM_NONE
;
380 sock_tx_timestamp(sk
, &skb_shinfo(skb
)->tx_flags
);
382 skb
->transport_header
= skb
->network_header
;
384 if (memcpy_from_msg(iph
, msg
, length
))
387 iphlen
= iph
->ihl
* 4;
390 * We don't want to modify the ip header, but we do need to
391 * be sure that it won't cause problems later along the network
392 * stack. Specifically we want to make sure that iph->ihl is a
393 * sane value. If ihl points beyond the length of the buffer passed
394 * in, reject the frame as invalid
400 if (iphlen
>= sizeof(*iph
)) {
402 iph
->saddr
= fl4
->saddr
;
404 iph
->tot_len
= htons(length
);
406 ip_select_ident(net
, skb
, NULL
);
408 iph
->check
= ip_fast_csum((unsigned char *)iph
, iph
->ihl
);
410 if (iph
->protocol
== IPPROTO_ICMP
)
411 icmp_out_count(net
, ((struct icmphdr
*)
412 skb_transport_header(skb
))->type
);
414 err
= NF_HOOK(NFPROTO_IPV4
, NF_INET_LOCAL_OUT
, sk
, skb
,
415 NULL
, rt
->dst
.dev
, dst_output_sk
);
417 err
= net_xmit_errno(err
);
426 IP_INC_STATS(net
, IPSTATS_MIB_OUTDISCARDS
);
427 if (err
== -ENOBUFS
&& !inet
->recverr
)
432 static int raw_probe_proto_opt(struct raw_frag_vec
*rfv
, struct flowi4
*fl4
)
436 if (fl4
->flowi4_proto
!= IPPROTO_ICMP
)
439 /* We only need the first two bytes. */
442 err
= memcpy_from_msg(rfv
->hdr
.c
, rfv
->msg
, rfv
->hlen
);
446 fl4
->fl4_icmp_type
= rfv
->hdr
.icmph
.type
;
447 fl4
->fl4_icmp_code
= rfv
->hdr
.icmph
.code
;
452 static int raw_getfrag(void *from
, char *to
, int offset
, int len
, int odd
,
455 struct raw_frag_vec
*rfv
= from
;
457 if (offset
< rfv
->hlen
) {
458 int copy
= min(rfv
->hlen
- offset
, len
);
460 if (skb
->ip_summed
== CHECKSUM_PARTIAL
)
461 memcpy(to
, rfv
->hdr
.c
+ offset
, copy
);
463 skb
->csum
= csum_block_add(
465 csum_partial_copy_nocheck(rfv
->hdr
.c
+ offset
,
480 return ip_generic_getfrag(rfv
->msg
, to
, offset
, len
, odd
, skb
);
483 static int raw_sendmsg(struct sock
*sk
, struct msghdr
*msg
, size_t len
)
485 struct inet_sock
*inet
= inet_sk(sk
);
486 struct ipcm_cookie ipc
;
487 struct rtable
*rt
= NULL
;
494 struct ip_options_data opt_copy
;
495 struct raw_frag_vec rfv
;
506 if (msg
->msg_flags
& MSG_OOB
) /* Mirror BSD error message */
507 goto out
; /* compatibility */
510 * Get and verify the address.
513 if (msg
->msg_namelen
) {
514 DECLARE_SOCKADDR(struct sockaddr_in
*, usin
, msg
->msg_name
);
516 if (msg
->msg_namelen
< sizeof(*usin
))
518 if (usin
->sin_family
!= AF_INET
) {
519 pr_info_once("%s: %s forgot to set AF_INET. Fix it!\n",
520 __func__
, current
->comm
);
522 if (usin
->sin_family
)
525 daddr
= usin
->sin_addr
.s_addr
;
526 /* ANK: I did not forget to get protocol from port field.
527 * I just do not know, who uses this weirdness.
528 * IP_HDRINCL is much more convenient.
532 if (sk
->sk_state
!= TCP_ESTABLISHED
)
534 daddr
= inet
->inet_daddr
;
537 ipc
.addr
= inet
->inet_saddr
;
542 ipc
.oif
= sk
->sk_bound_dev_if
;
544 if (msg
->msg_controllen
) {
545 err
= ip_cmsg_send(sock_net(sk
), msg
, &ipc
, false);
556 struct ip_options_rcu
*inet_opt
;
559 inet_opt
= rcu_dereference(inet
->inet_opt
);
561 memcpy(&opt_copy
, inet_opt
,
562 sizeof(*inet_opt
) + inet_opt
->opt
.optlen
);
563 ipc
.opt
= &opt_copy
.opt
;
570 /* Linux does not mangle headers on raw sockets,
571 * so that IP options + IP_HDRINCL is non-sense.
575 if (ipc
.opt
->opt
.srr
) {
578 daddr
= ipc
.opt
->opt
.faddr
;
581 tos
= get_rtconn_flags(&ipc
, sk
);
582 if (msg
->msg_flags
& MSG_DONTROUTE
)
585 if (ipv4_is_multicast(daddr
)) {
587 ipc
.oif
= inet
->mc_index
;
589 saddr
= inet
->mc_addr
;
591 ipc
.oif
= inet
->uc_index
;
593 flowi4_init_output(&fl4
, ipc
.oif
, sk
->sk_mark
, tos
,
595 inet
->hdrincl
? IPPROTO_RAW
: sk
->sk_protocol
,
596 inet_sk_flowi_flags(sk
) |
597 (inet
->hdrincl
? FLOWI_FLAG_KNOWN_NH
: 0),
600 if (!inet
->hdrincl
) {
604 err
= raw_probe_proto_opt(&rfv
, &fl4
);
609 security_sk_classify_flow(sk
, flowi4_to_flowi(&fl4
));
610 rt
= ip_route_output_flow(sock_net(sk
), &fl4
, sk
);
618 if (rt
->rt_flags
& RTCF_BROADCAST
&& !sock_flag(sk
, SOCK_BROADCAST
))
621 if (msg
->msg_flags
& MSG_CONFIRM
)
626 err
= raw_send_hdrinc(sk
, &fl4
, msg
, len
,
627 &rt
, msg
->msg_flags
);
630 sock_tx_timestamp(sk
, &ipc
.tx_flags
);
633 ipc
.addr
= fl4
.daddr
;
635 err
= ip_append_data(sk
, &fl4
, raw_getfrag
,
637 &ipc
, &rt
, msg
->msg_flags
);
639 ip_flush_pending_frames(sk
);
640 else if (!(msg
->msg_flags
& MSG_MORE
)) {
641 err
= ip_push_pending_frames(sk
, &fl4
);
642 if (err
== -ENOBUFS
&& !inet
->recverr
)
658 dst_confirm(&rt
->dst
);
659 if (!(msg
->msg_flags
& MSG_PROBE
) || len
)
660 goto back_from_confirm
;
665 static void raw_close(struct sock
*sk
, long timeout
)
668 * Raw sockets may have direct kernel references. Kill them.
670 ip_ra_control(sk
, 0, NULL
);
672 sk_common_release(sk
);
675 static void raw_destroy(struct sock
*sk
)
678 ip_flush_pending_frames(sk
);
682 /* This gets rid of all the nasties in af_inet. -DaveM */
683 static int raw_bind(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
)
685 struct inet_sock
*inet
= inet_sk(sk
);
686 struct sockaddr_in
*addr
= (struct sockaddr_in
*) uaddr
;
690 if (sk
->sk_state
!= TCP_CLOSE
|| addr_len
< sizeof(struct sockaddr_in
))
692 chk_addr_ret
= inet_addr_type(sock_net(sk
), addr
->sin_addr
.s_addr
);
693 ret
= -EADDRNOTAVAIL
;
694 if (addr
->sin_addr
.s_addr
&& chk_addr_ret
!= RTN_LOCAL
&&
695 chk_addr_ret
!= RTN_MULTICAST
&& chk_addr_ret
!= RTN_BROADCAST
)
697 inet
->inet_rcv_saddr
= inet
->inet_saddr
= addr
->sin_addr
.s_addr
;
698 if (chk_addr_ret
== RTN_MULTICAST
|| chk_addr_ret
== RTN_BROADCAST
)
699 inet
->inet_saddr
= 0; /* Use device */
706 * This should be easy, if there is something there
707 * we return it, otherwise we block.
710 static int raw_recvmsg(struct sock
*sk
, struct msghdr
*msg
, size_t len
,
711 int noblock
, int flags
, int *addr_len
)
713 struct inet_sock
*inet
= inet_sk(sk
);
715 int err
= -EOPNOTSUPP
;
716 DECLARE_SOCKADDR(struct sockaddr_in
*, sin
, msg
->msg_name
);
722 if (flags
& MSG_ERRQUEUE
) {
723 err
= ip_recv_error(sk
, msg
, len
, addr_len
);
727 skb
= skb_recv_datagram(sk
, flags
, noblock
, &err
);
733 msg
->msg_flags
|= MSG_TRUNC
;
737 err
= skb_copy_datagram_msg(skb
, 0, msg
, copied
);
741 sock_recv_ts_and_drops(msg
, sk
, skb
);
743 /* Copy the address. */
745 sin
->sin_family
= AF_INET
;
746 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
748 memset(&sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
749 *addr_len
= sizeof(*sin
);
751 if (inet
->cmsg_flags
)
752 ip_cmsg_recv(msg
, skb
);
753 if (flags
& MSG_TRUNC
)
756 skb_free_datagram(sk
, skb
);
763 static int raw_init(struct sock
*sk
)
765 struct raw_sock
*rp
= raw_sk(sk
);
767 if (inet_sk(sk
)->inet_num
== IPPROTO_ICMP
)
768 memset(&rp
->filter
, 0, sizeof(rp
->filter
));
772 static int raw_seticmpfilter(struct sock
*sk
, char __user
*optval
, int optlen
)
774 if (optlen
> sizeof(struct icmp_filter
))
775 optlen
= sizeof(struct icmp_filter
);
776 if (copy_from_user(&raw_sk(sk
)->filter
, optval
, optlen
))
781 static int raw_geticmpfilter(struct sock
*sk
, char __user
*optval
, int __user
*optlen
)
783 int len
, ret
= -EFAULT
;
785 if (get_user(len
, optlen
))
790 if (len
> sizeof(struct icmp_filter
))
791 len
= sizeof(struct icmp_filter
);
793 if (put_user(len
, optlen
) ||
794 copy_to_user(optval
, &raw_sk(sk
)->filter
, len
))
800 static int do_raw_setsockopt(struct sock
*sk
, int level
, int optname
,
801 char __user
*optval
, unsigned int optlen
)
803 if (optname
== ICMP_FILTER
) {
804 if (inet_sk(sk
)->inet_num
!= IPPROTO_ICMP
)
807 return raw_seticmpfilter(sk
, optval
, optlen
);
812 static int raw_setsockopt(struct sock
*sk
, int level
, int optname
,
813 char __user
*optval
, unsigned int optlen
)
815 if (level
!= SOL_RAW
)
816 return ip_setsockopt(sk
, level
, optname
, optval
, optlen
);
817 return do_raw_setsockopt(sk
, level
, optname
, optval
, optlen
);
821 static int compat_raw_setsockopt(struct sock
*sk
, int level
, int optname
,
822 char __user
*optval
, unsigned int optlen
)
824 if (level
!= SOL_RAW
)
825 return compat_ip_setsockopt(sk
, level
, optname
, optval
, optlen
);
826 return do_raw_setsockopt(sk
, level
, optname
, optval
, optlen
);
830 static int do_raw_getsockopt(struct sock
*sk
, int level
, int optname
,
831 char __user
*optval
, int __user
*optlen
)
833 if (optname
== ICMP_FILTER
) {
834 if (inet_sk(sk
)->inet_num
!= IPPROTO_ICMP
)
837 return raw_geticmpfilter(sk
, optval
, optlen
);
842 static int raw_getsockopt(struct sock
*sk
, int level
, int optname
,
843 char __user
*optval
, int __user
*optlen
)
845 if (level
!= SOL_RAW
)
846 return ip_getsockopt(sk
, level
, optname
, optval
, optlen
);
847 return do_raw_getsockopt(sk
, level
, optname
, optval
, optlen
);
851 static int compat_raw_getsockopt(struct sock
*sk
, int level
, int optname
,
852 char __user
*optval
, int __user
*optlen
)
854 if (level
!= SOL_RAW
)
855 return compat_ip_getsockopt(sk
, level
, optname
, optval
, optlen
);
856 return do_raw_getsockopt(sk
, level
, optname
, optval
, optlen
);
860 static int raw_ioctl(struct sock
*sk
, int cmd
, unsigned long arg
)
864 int amount
= sk_wmem_alloc_get(sk
);
866 return put_user(amount
, (int __user
*)arg
);
872 spin_lock_bh(&sk
->sk_receive_queue
.lock
);
873 skb
= skb_peek(&sk
->sk_receive_queue
);
876 spin_unlock_bh(&sk
->sk_receive_queue
.lock
);
877 return put_user(amount
, (int __user
*)arg
);
881 #ifdef CONFIG_IP_MROUTE
882 return ipmr_ioctl(sk
, cmd
, (void __user
*)arg
);
890 static int compat_raw_ioctl(struct sock
*sk
, unsigned int cmd
, unsigned long arg
)
897 #ifdef CONFIG_IP_MROUTE
898 return ipmr_compat_ioctl(sk
, cmd
, compat_ptr(arg
));
906 struct proto raw_prot
= {
908 .owner
= THIS_MODULE
,
910 .destroy
= raw_destroy
,
911 .connect
= ip4_datagram_connect
,
912 .disconnect
= udp_disconnect
,
915 .setsockopt
= raw_setsockopt
,
916 .getsockopt
= raw_getsockopt
,
917 .sendmsg
= raw_sendmsg
,
918 .recvmsg
= raw_recvmsg
,
920 .backlog_rcv
= raw_rcv_skb
,
921 .release_cb
= ip4_datagram_release_cb
,
923 .unhash
= raw_unhash_sk
,
924 .obj_size
= sizeof(struct raw_sock
),
925 .h
.raw_hash
= &raw_v4_hashinfo
,
927 .compat_setsockopt
= compat_raw_setsockopt
,
928 .compat_getsockopt
= compat_raw_getsockopt
,
929 .compat_ioctl
= compat_raw_ioctl
,
933 #ifdef CONFIG_PROC_FS
934 static struct sock
*raw_get_first(struct seq_file
*seq
)
937 struct raw_iter_state
*state
= raw_seq_private(seq
);
939 for (state
->bucket
= 0; state
->bucket
< RAW_HTABLE_SIZE
;
941 sk_for_each(sk
, &state
->h
->ht
[state
->bucket
])
942 if (sock_net(sk
) == seq_file_net(seq
))
950 static struct sock
*raw_get_next(struct seq_file
*seq
, struct sock
*sk
)
952 struct raw_iter_state
*state
= raw_seq_private(seq
);
958 } while (sk
&& sock_net(sk
) != seq_file_net(seq
));
960 if (!sk
&& ++state
->bucket
< RAW_HTABLE_SIZE
) {
961 sk
= sk_head(&state
->h
->ht
[state
->bucket
]);
967 static struct sock
*raw_get_idx(struct seq_file
*seq
, loff_t pos
)
969 struct sock
*sk
= raw_get_first(seq
);
972 while (pos
&& (sk
= raw_get_next(seq
, sk
)) != NULL
)
974 return pos
? NULL
: sk
;
977 void *raw_seq_start(struct seq_file
*seq
, loff_t
*pos
)
979 struct raw_iter_state
*state
= raw_seq_private(seq
);
981 read_lock(&state
->h
->lock
);
982 return *pos
? raw_get_idx(seq
, *pos
- 1) : SEQ_START_TOKEN
;
984 EXPORT_SYMBOL_GPL(raw_seq_start
);
986 void *raw_seq_next(struct seq_file
*seq
, void *v
, loff_t
*pos
)
990 if (v
== SEQ_START_TOKEN
)
991 sk
= raw_get_first(seq
);
993 sk
= raw_get_next(seq
, v
);
997 EXPORT_SYMBOL_GPL(raw_seq_next
);
999 void raw_seq_stop(struct seq_file
*seq
, void *v
)
1001 struct raw_iter_state
*state
= raw_seq_private(seq
);
1003 read_unlock(&state
->h
->lock
);
1005 EXPORT_SYMBOL_GPL(raw_seq_stop
);
1007 static void raw_sock_seq_show(struct seq_file
*seq
, struct sock
*sp
, int i
)
1009 struct inet_sock
*inet
= inet_sk(sp
);
1010 __be32 dest
= inet
->inet_daddr
,
1011 src
= inet
->inet_rcv_saddr
;
1013 srcp
= inet
->inet_num
;
1015 seq_printf(seq
, "%4d: %08X:%04X %08X:%04X"
1016 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d\n",
1017 i
, src
, srcp
, dest
, destp
, sp
->sk_state
,
1018 sk_wmem_alloc_get(sp
),
1019 sk_rmem_alloc_get(sp
),
1021 from_kuid_munged(seq_user_ns(seq
), sock_i_uid(sp
)),
1023 atomic_read(&sp
->sk_refcnt
), sp
, atomic_read(&sp
->sk_drops
));
1026 static int raw_seq_show(struct seq_file
*seq
, void *v
)
1028 if (v
== SEQ_START_TOKEN
)
1029 seq_printf(seq
, " sl local_address rem_address st tx_queue "
1030 "rx_queue tr tm->when retrnsmt uid timeout "
1031 "inode ref pointer drops\n");
1033 raw_sock_seq_show(seq
, v
, raw_seq_private(seq
)->bucket
);
1037 static const struct seq_operations raw_seq_ops
= {
1038 .start
= raw_seq_start
,
1039 .next
= raw_seq_next
,
1040 .stop
= raw_seq_stop
,
1041 .show
= raw_seq_show
,
1044 int raw_seq_open(struct inode
*ino
, struct file
*file
,
1045 struct raw_hashinfo
*h
, const struct seq_operations
*ops
)
1048 struct raw_iter_state
*i
;
1050 err
= seq_open_net(ino
, file
, ops
, sizeof(struct raw_iter_state
));
1054 i
= raw_seq_private((struct seq_file
*)file
->private_data
);
1058 EXPORT_SYMBOL_GPL(raw_seq_open
);
1060 static int raw_v4_seq_open(struct inode
*inode
, struct file
*file
)
1062 return raw_seq_open(inode
, file
, &raw_v4_hashinfo
, &raw_seq_ops
);
1065 static const struct file_operations raw_seq_fops
= {
1066 .owner
= THIS_MODULE
,
1067 .open
= raw_v4_seq_open
,
1069 .llseek
= seq_lseek
,
1070 .release
= seq_release_net
,
1073 static __net_init
int raw_init_net(struct net
*net
)
1075 if (!proc_create("raw", S_IRUGO
, net
->proc_net
, &raw_seq_fops
))
1081 static __net_exit
void raw_exit_net(struct net
*net
)
1083 remove_proc_entry("raw", net
->proc_net
);
1086 static __net_initdata
struct pernet_operations raw_net_ops
= {
1087 .init
= raw_init_net
,
1088 .exit
= raw_exit_net
,
1091 int __init
raw_proc_init(void)
1093 return register_pernet_subsys(&raw_net_ops
);
1096 void __init
raw_proc_exit(void)
1098 unregister_pernet_subsys(&raw_net_ops
);
1100 #endif /* CONFIG_PROC_FS */