1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
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>
55#include <linux/in.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 <net/net_namespace.h>
62#include <net/dst.h>
63#include <net/sock.h>
64#include <linux/ip.h>
65#include <linux/net.h>
66#include <net/ip.h>
67#include <net/icmp.h>
68#include <net/udp.h>
69#include <net/raw.h>
70#include <net/snmp.h>
71#include <net/tcp_states.h>
72#include <net/inet_common.h>
73#include <net/checksum.h>
74#include <net/xfrm.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
82static struct raw_hashinfo raw_v4_hashinfo = {
83 .lock = __RW_LOCK_UNLOCKED(raw_v4_hashinfo.lock),
84};
85
86void raw_hash_sk(struct sock *sk)
87{
88 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
89 struct hlist_head *head;
90
91 head = &h->ht[inet_sk(sk)->inet_num & (RAW_HTABLE_SIZE - 1)];
92
93 write_lock_bh(&h->lock);
94 sk_add_node(sk, head);
95 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
96 write_unlock_bh(&h->lock);
97}
98EXPORT_SYMBOL_GPL(raw_hash_sk);
99
100void raw_unhash_sk(struct sock *sk)
101{
102 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
103
104 write_lock_bh(&h->lock);
105 if (sk_del_node_init(sk))
106 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
107 write_unlock_bh(&h->lock);
108}
109EXPORT_SYMBOL_GPL(raw_unhash_sk);
110
111static struct sock *__raw_v4_lookup(struct net *net, struct sock *sk,
112 unsigned short num, __be32 raddr, __be32 laddr, int dif)
113{
114 sk_for_each_from(sk) {
115 struct inet_sock *inet = inet_sk(sk);
116
117 if (net_eq(sock_net(sk), net) && inet->inet_num == num &&
118 !(inet->inet_daddr && inet->inet_daddr != raddr) &&
119 !(inet->inet_rcv_saddr && inet->inet_rcv_saddr != laddr) &&
120 !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
121 goto found;
122 }
123 sk = NULL;
124found:
125 return sk;
126}
127
128
129
130
131
132static int icmp_filter(const struct sock *sk, const struct sk_buff *skb)
133{
134 struct icmphdr _hdr;
135 const struct icmphdr *hdr;
136
137 hdr = skb_header_pointer(skb, skb_transport_offset(skb),
138 sizeof(_hdr), &_hdr);
139 if (!hdr)
140 return 1;
141
142 if (hdr->type < 32) {
143 __u32 data = raw_sk(sk)->filter.data;
144
145 return ((1U << hdr->type) & data) != 0;
146 }
147
148
149 return 0;
150}
151
152
153
154
155
156
157
158static int raw_v4_input(struct sk_buff *skb, const struct iphdr *iph, int hash)
159{
160 struct sock *sk;
161 struct hlist_head *head;
162 int delivered = 0;
163 struct net *net;
164
165 read_lock(&raw_v4_hashinfo.lock);
166 head = &raw_v4_hashinfo.ht[hash];
167 if (hlist_empty(head))
168 goto out;
169
170 net = dev_net(skb->dev);
171 sk = __raw_v4_lookup(net, __sk_head(head), iph->protocol,
172 iph->saddr, iph->daddr,
173 skb->dev->ifindex);
174
175 while (sk) {
176 delivered = 1;
177 if (iph->protocol != IPPROTO_ICMP || !icmp_filter(sk, skb)) {
178 struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
179
180
181 if (clone)
182 raw_rcv(sk, clone);
183 }
184 sk = __raw_v4_lookup(net, sk_next(sk), iph->protocol,
185 iph->saddr, iph->daddr,
186 skb->dev->ifindex);
187 }
188out:
189 read_unlock(&raw_v4_hashinfo.lock);
190 return delivered;
191}
192
193int raw_local_deliver(struct sk_buff *skb, int protocol)
194{
195 int hash;
196 struct sock *raw_sk;
197
198 hash = protocol & (RAW_HTABLE_SIZE - 1);
199 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
200
201
202
203
204 if (raw_sk && !raw_v4_input(skb, ip_hdr(skb), hash))
205 raw_sk = NULL;
206
207 return raw_sk != NULL;
208
209}
210
211static void raw_err(struct sock *sk, struct sk_buff *skb, u32 info)
212{
213 struct inet_sock *inet = inet_sk(sk);
214 const int type = icmp_hdr(skb)->type;
215 const int code = icmp_hdr(skb)->code;
216 int err = 0;
217 int harderr = 0;
218
219 if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED)
220 ipv4_sk_update_pmtu(skb, sk, info);
221 else if (type == ICMP_REDIRECT)
222 ipv4_sk_redirect(skb, sk);
223
224
225
226
227
228
229 if (!inet->recverr && sk->sk_state != TCP_ESTABLISHED)
230 return;
231
232 switch (type) {
233 default:
234 case ICMP_TIME_EXCEEDED:
235 err = EHOSTUNREACH;
236 break;
237 case ICMP_SOURCE_QUENCH:
238 return;
239 case ICMP_PARAMETERPROB:
240 err = EPROTO;
241 harderr = 1;
242 break;
243 case ICMP_DEST_UNREACH:
244 err = EHOSTUNREACH;
245 if (code > NR_ICMP_UNREACH)
246 break;
247 err = icmp_err_convert[code].errno;
248 harderr = icmp_err_convert[code].fatal;
249 if (code == ICMP_FRAG_NEEDED) {
250 harderr = inet->pmtudisc != IP_PMTUDISC_DONT;
251 err = EMSGSIZE;
252 }
253 }
254
255 if (inet->recverr) {
256 const struct iphdr *iph = (const struct iphdr *)skb->data;
257 u8 *payload = skb->data + (iph->ihl << 2);
258
259 if (inet->hdrincl)
260 payload = skb->data;
261 ip_icmp_error(sk, skb, err, 0, info, payload);
262 }
263
264 if (inet->recverr || harderr) {
265 sk->sk_err = err;
266 sk->sk_error_report(sk);
267 }
268}
269
270void raw_icmp_error(struct sk_buff *skb, int protocol, u32 info)
271{
272 int hash;
273 struct sock *raw_sk;
274 const struct iphdr *iph;
275 struct net *net;
276
277 hash = protocol & (RAW_HTABLE_SIZE - 1);
278
279 read_lock(&raw_v4_hashinfo.lock);
280 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
281 if (raw_sk != NULL) {
282 iph = (const struct iphdr *)skb->data;
283 net = dev_net(skb->dev);
284
285 while ((raw_sk = __raw_v4_lookup(net, raw_sk, protocol,
286 iph->daddr, iph->saddr,
287 skb->dev->ifindex)) != NULL) {
288 raw_err(raw_sk, skb, info);
289 raw_sk = sk_next(raw_sk);
290 iph = (const struct iphdr *)skb->data;
291 }
292 }
293 read_unlock(&raw_v4_hashinfo.lock);
294}
295
296static int raw_rcv_skb(struct sock *sk, struct sk_buff *skb)
297{
298
299
300 ipv4_pktinfo_prepare(sk, skb);
301 if (sock_queue_rcv_skb(sk, skb) < 0) {
302 kfree_skb(skb);
303 return NET_RX_DROP;
304 }
305
306 return NET_RX_SUCCESS;
307}
308
309int raw_rcv(struct sock *sk, struct sk_buff *skb)
310{
311 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
312 atomic_inc(&sk->sk_drops);
313 kfree_skb(skb);
314 return NET_RX_DROP;
315 }
316 nf_reset(skb);
317
318 skb_push(skb, skb->data - skb_network_header(skb));
319
320 raw_rcv_skb(sk, skb);
321 return 0;
322}
323
324static int raw_send_hdrinc(struct sock *sk, struct flowi4 *fl4,
325 void *from, size_t length,
326 struct rtable **rtp,
327 unsigned int flags)
328{
329 struct inet_sock *inet = inet_sk(sk);
330 struct net *net = sock_net(sk);
331 struct iphdr *iph;
332 struct sk_buff *skb;
333 unsigned int iphlen;
334 int err;
335 struct rtable *rt = *rtp;
336 int hlen, tlen;
337
338 if (length > rt->dst.dev->mtu) {
339 ip_local_error(sk, EMSGSIZE, fl4->daddr, inet->inet_dport,
340 rt->dst.dev->mtu);
341 return -EMSGSIZE;
342 }
343 if (flags&MSG_PROBE)
344 goto out;
345
346 hlen = LL_RESERVED_SPACE(rt->dst.dev);
347 tlen = rt->dst.dev->needed_tailroom;
348 skb = sock_alloc_send_skb(sk,
349 length + hlen + tlen + 15,
350 flags & MSG_DONTWAIT, &err);
351 if (skb == NULL)
352 goto error;
353 skb_reserve(skb, hlen);
354
355 skb->priority = sk->sk_priority;
356 skb->mark = sk->sk_mark;
357 skb_dst_set(skb, &rt->dst);
358 *rtp = NULL;
359
360 skb_reset_network_header(skb);
361 iph = ip_hdr(skb);
362 skb_put(skb, length);
363
364 skb->ip_summed = CHECKSUM_NONE;
365
366 if (flags & MSG_CONFIRM)
367 skb_set_dst_pending_confirm(skb, 1);
368
369 skb->transport_header = skb->network_header;
370 err = -EFAULT;
371 if (memcpy_fromiovecend((void *)iph, from, 0, length))
372 goto error_free;
373
374 iphlen = iph->ihl * 4;
375
376
377
378
379
380
381
382
383 err = -EINVAL;
384 if (iphlen > length)
385 goto error_free;
386
387 if (iphlen >= sizeof(*iph)) {
388 if (!iph->saddr)
389 iph->saddr = fl4->saddr;
390 iph->check = 0;
391 iph->tot_len = htons(length);
392 if (!iph->id)
393 ip_select_ident(net, skb, NULL);
394
395 iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
396 }
397 if (iph->protocol == IPPROTO_ICMP)
398 icmp_out_count(net, ((struct icmphdr *)
399 skb_transport_header(skb))->type);
400
401 err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT, sk, skb,
402 NULL, rt->dst.dev, dst_output_sk);
403 if (err > 0)
404 err = net_xmit_errno(err);
405 if (err)
406 goto error;
407out:
408 return 0;
409
410error_free:
411 kfree_skb(skb);
412error:
413 IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
414 if (err == -ENOBUFS && !inet->recverr)
415 err = 0;
416 return err;
417}
418
419static int raw_probe_proto_opt(struct flowi4 *fl4, struct msghdr *msg)
420{
421 struct iovec *iov;
422 u8 __user *type = NULL;
423 u8 __user *code = NULL;
424 int probed = 0;
425 unsigned int i;
426
427 if (!msg->msg_iov)
428 return 0;
429
430 for (i = 0; i < msg->msg_iovlen; i++) {
431 iov = &msg->msg_iov[i];
432 if (!iov)
433 continue;
434
435 switch (fl4->flowi4_proto) {
436 case IPPROTO_ICMP:
437
438 if (iov->iov_base && iov->iov_len < 1)
439 break;
440
441 if (!type) {
442 type = iov->iov_base;
443
444 if (iov->iov_len > 1)
445 code = type + 1;
446 } else if (!code)
447 code = iov->iov_base;
448
449 if (type && code) {
450 if (get_user(fl4->fl4_icmp_type, type) ||
451 get_user(fl4->fl4_icmp_code, code))
452 return -EFAULT;
453 probed = 1;
454 }
455 break;
456 default:
457 probed = 1;
458 break;
459 }
460 if (probed)
461 break;
462 }
463 return 0;
464}
465
466static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
467 size_t len)
468{
469 struct inet_sock *inet = inet_sk(sk);
470 struct ipcm_cookie ipc;
471 struct rtable *rt = NULL;
472 struct flowi4 fl4;
473 int free = 0;
474 __be32 daddr;
475 __be32 saddr;
476 u8 tos;
477 int err;
478 struct ip_options_data opt_copy;
479
480 err = -EMSGSIZE;
481 if (len > 0xFFFF)
482 goto out;
483
484
485
486
487
488 err = -EOPNOTSUPP;
489 if (msg->msg_flags & MSG_OOB)
490 goto out;
491
492
493
494
495
496 if (msg->msg_namelen) {
497 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
498 err = -EINVAL;
499 if (msg->msg_namelen < sizeof(*usin))
500 goto out;
501 if (usin->sin_family != AF_INET) {
502 pr_info_once("%s: %s forgot to set AF_INET. Fix it!\n",
503 __func__, current->comm);
504 err = -EAFNOSUPPORT;
505 if (usin->sin_family)
506 goto out;
507 }
508 daddr = usin->sin_addr.s_addr;
509
510
511
512
513 } else {
514 err = -EDESTADDRREQ;
515 if (sk->sk_state != TCP_ESTABLISHED)
516 goto out;
517 daddr = inet->inet_daddr;
518 }
519
520 ipc.addr = inet->inet_saddr;
521 ipc.opt = NULL;
522 ipc.tx_flags = 0;
523 ipc.ttl = 0;
524 ipc.tos = -1;
525 ipc.oif = sk->sk_bound_dev_if;
526
527 if (msg->msg_controllen) {
528 err = ip_cmsg_send(sock_net(sk), msg, &ipc, false);
529 if (unlikely(err)) {
530 kfree(ipc.opt);
531 goto out;
532 }
533 if (ipc.opt)
534 free = 1;
535 }
536
537 saddr = ipc.addr;
538 ipc.addr = daddr;
539
540 if (!ipc.opt) {
541 struct ip_options_rcu *inet_opt;
542
543 rcu_read_lock();
544 inet_opt = rcu_dereference(inet->inet_opt);
545 if (inet_opt) {
546 memcpy(&opt_copy, inet_opt,
547 sizeof(*inet_opt) + inet_opt->opt.optlen);
548 ipc.opt = &opt_copy.opt;
549 }
550 rcu_read_unlock();
551 }
552
553 if (ipc.opt) {
554 err = -EINVAL;
555
556
557
558 if (inet->hdrincl)
559 goto done;
560 if (ipc.opt->opt.srr) {
561 if (!daddr)
562 goto done;
563 daddr = ipc.opt->opt.faddr;
564 }
565 }
566 tos = get_rtconn_flags(&ipc, sk);
567 if (msg->msg_flags & MSG_DONTROUTE)
568 tos |= RTO_ONLINK;
569
570 if (ipv4_is_multicast(daddr)) {
571 if (!ipc.oif)
572 ipc.oif = inet->mc_index;
573 if (!saddr)
574 saddr = inet->mc_addr;
575 } else if (!ipc.oif)
576 ipc.oif = inet->uc_index;
577
578 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
579 RT_SCOPE_UNIVERSE,
580 inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol,
581 inet_sk_flowi_flags(sk) |
582 (inet->hdrincl ? FLOWI_FLAG_KNOWN_NH : 0),
583 daddr, saddr, 0, 0);
584
585 if (!inet->hdrincl) {
586 err = raw_probe_proto_opt(&fl4, msg);
587 if (err)
588 goto done;
589 }
590
591 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
592 rt = ip_route_output_flow(sock_net(sk), &fl4, sk);
593 if (IS_ERR(rt)) {
594 err = PTR_ERR(rt);
595 rt = NULL;
596 goto done;
597 }
598
599 err = -EACCES;
600 if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST))
601 goto done;
602
603 if (msg->msg_flags & MSG_CONFIRM)
604 goto do_confirm;
605back_from_confirm:
606
607 if (inet->hdrincl)
608 err = raw_send_hdrinc(sk, &fl4, msg->msg_iov, len,
609 &rt, msg->msg_flags);
610
611 else {
612 if (!ipc.addr)
613 ipc.addr = fl4.daddr;
614 lock_sock(sk);
615 err = ip_append_data(sk, &fl4, ip_generic_getfrag,
616 msg->msg_iov, len, 0,
617 &ipc, &rt, msg->msg_flags);
618 if (err)
619 ip_flush_pending_frames(sk);
620 else if (!(msg->msg_flags & MSG_MORE)) {
621 err = ip_push_pending_frames(sk, &fl4);
622 if (err == -ENOBUFS && !inet->recverr)
623 err = 0;
624 }
625 release_sock(sk);
626 }
627done:
628 if (free)
629 kfree(ipc.opt);
630 ip_rt_put(rt);
631
632out:
633 if (err < 0)
634 return err;
635 return len;
636
637do_confirm:
638 if (msg->msg_flags & MSG_PROBE)
639 dst_confirm_neigh(&rt->dst, &fl4.daddr);
640 if (!(msg->msg_flags & MSG_PROBE) || len)
641 goto back_from_confirm;
642 err = 0;
643 goto done;
644}
645
646static void raw_close(struct sock *sk, long timeout)
647{
648
649
650
651 ip_ra_control(sk, 0, NULL);
652
653 sk_common_release(sk);
654}
655
656static void raw_destroy(struct sock *sk)
657{
658 lock_sock(sk);
659 ip_flush_pending_frames(sk);
660 release_sock(sk);
661}
662
663
664static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
665{
666 struct inet_sock *inet = inet_sk(sk);
667 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
668 int ret = -EINVAL;
669 int chk_addr_ret;
670
671 if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
672 goto out;
673 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
674 ret = -EADDRNOTAVAIL;
675 if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
676 chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
677 goto out;
678 inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
679 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
680 inet->inet_saddr = 0;
681 sk_dst_reset(sk);
682 ret = 0;
683out: return ret;
684}
685
686
687
688
689
690
691static int raw_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
692 size_t len, int noblock, int flags, int *addr_len)
693{
694 struct inet_sock *inet = inet_sk(sk);
695 size_t copied = 0;
696 int err = -EOPNOTSUPP;
697 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
698 struct sk_buff *skb;
699
700 if (flags & MSG_OOB)
701 goto out;
702
703 if (flags & MSG_ERRQUEUE) {
704 err = ip_recv_error(sk, msg, len, addr_len);
705 goto out;
706 }
707
708 skb = skb_recv_datagram(sk, flags, noblock, &err);
709 if (!skb)
710 goto out;
711
712 copied = skb->len;
713 if (len < copied) {
714 msg->msg_flags |= MSG_TRUNC;
715 copied = len;
716 }
717
718 err = skb_copy_datagram_msg(skb, 0, msg, copied);
719 if (err)
720 goto done;
721
722 sock_recv_ts_and_drops(msg, sk, skb);
723
724
725 if (sin) {
726 sin->sin_family = AF_INET;
727 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
728 sin->sin_port = 0;
729 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
730 *addr_len = sizeof(*sin);
731 }
732 if (inet->cmsg_flags)
733 ip_cmsg_recv(msg, skb);
734 if (flags & MSG_TRUNC)
735 copied = skb->len;
736done:
737 skb_free_datagram(sk, skb);
738out:
739 if (err)
740 return err;
741 return copied;
742}
743
744static int raw_init(struct sock *sk)
745{
746 struct raw_sock *rp = raw_sk(sk);
747
748 if (inet_sk(sk)->inet_num == IPPROTO_ICMP)
749 memset(&rp->filter, 0, sizeof(rp->filter));
750 return 0;
751}
752
753static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen)
754{
755 if (optlen > sizeof(struct icmp_filter))
756 optlen = sizeof(struct icmp_filter);
757 if (copy_from_user(&raw_sk(sk)->filter, optval, optlen))
758 return -EFAULT;
759 return 0;
760}
761
762static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen)
763{
764 int len, ret = -EFAULT;
765
766 if (get_user(len, optlen))
767 goto out;
768 ret = -EINVAL;
769 if (len < 0)
770 goto out;
771 if (len > sizeof(struct icmp_filter))
772 len = sizeof(struct icmp_filter);
773 ret = -EFAULT;
774 if (put_user(len, optlen) ||
775 copy_to_user(optval, &raw_sk(sk)->filter, len))
776 goto out;
777 ret = 0;
778out: return ret;
779}
780
781static int do_raw_setsockopt(struct sock *sk, int level, int optname,
782 char __user *optval, unsigned int optlen)
783{
784 if (optname == ICMP_FILTER) {
785 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
786 return -EOPNOTSUPP;
787 else
788 return raw_seticmpfilter(sk, optval, optlen);
789 }
790 return -ENOPROTOOPT;
791}
792
793static int raw_setsockopt(struct sock *sk, int level, int optname,
794 char __user *optval, unsigned int optlen)
795{
796 if (level != SOL_RAW)
797 return ip_setsockopt(sk, level, optname, optval, optlen);
798 return do_raw_setsockopt(sk, level, optname, optval, optlen);
799}
800
801#ifdef CONFIG_COMPAT
802static int compat_raw_setsockopt(struct sock *sk, int level, int optname,
803 char __user *optval, unsigned int optlen)
804{
805 if (level != SOL_RAW)
806 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
807 return do_raw_setsockopt(sk, level, optname, optval, optlen);
808}
809#endif
810
811static int do_raw_getsockopt(struct sock *sk, int level, int optname,
812 char __user *optval, int __user *optlen)
813{
814 if (optname == ICMP_FILTER) {
815 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
816 return -EOPNOTSUPP;
817 else
818 return raw_geticmpfilter(sk, optval, optlen);
819 }
820 return -ENOPROTOOPT;
821}
822
823static int raw_getsockopt(struct sock *sk, int level, int optname,
824 char __user *optval, int __user *optlen)
825{
826 if (level != SOL_RAW)
827 return ip_getsockopt(sk, level, optname, optval, optlen);
828 return do_raw_getsockopt(sk, level, optname, optval, optlen);
829}
830
831#ifdef CONFIG_COMPAT
832static int compat_raw_getsockopt(struct sock *sk, int level, int optname,
833 char __user *optval, int __user *optlen)
834{
835 if (level != SOL_RAW)
836 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
837 return do_raw_getsockopt(sk, level, optname, optval, optlen);
838}
839#endif
840
841static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg)
842{
843 switch (cmd) {
844 case SIOCOUTQ: {
845 int amount = sk_wmem_alloc_get(sk);
846
847 return put_user(amount, (int __user *)arg);
848 }
849 case SIOCINQ: {
850 struct sk_buff *skb;
851 int amount = 0;
852
853 spin_lock_bh(&sk->sk_receive_queue.lock);
854 skb = skb_peek(&sk->sk_receive_queue);
855 if (skb != NULL)
856 amount = skb->len;
857 spin_unlock_bh(&sk->sk_receive_queue.lock);
858 return put_user(amount, (int __user *)arg);
859 }
860
861 default:
862#ifdef CONFIG_IP_MROUTE
863 return ipmr_ioctl(sk, cmd, (void __user *)arg);
864#else
865 return -ENOIOCTLCMD;
866#endif
867 }
868}
869
870#ifdef CONFIG_COMPAT
871static int compat_raw_ioctl(struct sock *sk, unsigned int cmd, unsigned long arg)
872{
873 switch (cmd) {
874 case SIOCOUTQ:
875 case SIOCINQ:
876 return -ENOIOCTLCMD;
877 default:
878#ifdef CONFIG_IP_MROUTE
879 return ipmr_compat_ioctl(sk, cmd, compat_ptr(arg));
880#else
881 return -ENOIOCTLCMD;
882#endif
883 }
884}
885#endif
886
887struct proto raw_prot = {
888 .name = "RAW",
889 .owner = THIS_MODULE,
890 .close = raw_close,
891 .destroy = raw_destroy,
892 .connect = ip4_datagram_connect,
893 .disconnect = udp_disconnect,
894 .ioctl = raw_ioctl,
895 .init = raw_init,
896 .setsockopt = raw_setsockopt,
897 .getsockopt = raw_getsockopt,
898 .sendmsg = raw_sendmsg,
899 .recvmsg = raw_recvmsg,
900 .bind = raw_bind,
901 .backlog_rcv = raw_rcv_skb,
902 .release_cb = ip4_datagram_release_cb,
903 .hash = raw_hash_sk,
904 .unhash = raw_unhash_sk,
905 .obj_size = sizeof(struct raw_sock),
906 .h.raw_hash = &raw_v4_hashinfo,
907#ifdef CONFIG_COMPAT
908 .compat_setsockopt = compat_raw_setsockopt,
909 .compat_getsockopt = compat_raw_getsockopt,
910 .compat_ioctl = compat_raw_ioctl,
911#endif
912};
913
914#ifdef CONFIG_PROC_FS
915static struct sock *raw_get_first(struct seq_file *seq)
916{
917 struct sock *sk;
918 struct raw_iter_state *state = raw_seq_private(seq);
919
920 for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE;
921 ++state->bucket) {
922 sk_for_each(sk, &state->h->ht[state->bucket])
923 if (sock_net(sk) == seq_file_net(seq))
924 goto found;
925 }
926 sk = NULL;
927found:
928 return sk;
929}
930
931static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk)
932{
933 struct raw_iter_state *state = raw_seq_private(seq);
934
935 do {
936 sk = sk_next(sk);
937try_again:
938 ;
939 } while (sk && sock_net(sk) != seq_file_net(seq));
940
941 if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
942 sk = sk_head(&state->h->ht[state->bucket]);
943 goto try_again;
944 }
945 return sk;
946}
947
948static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos)
949{
950 struct sock *sk = raw_get_first(seq);
951
952 if (sk)
953 while (pos && (sk = raw_get_next(seq, sk)) != NULL)
954 --pos;
955 return pos ? NULL : sk;
956}
957
958void *raw_seq_start(struct seq_file *seq, loff_t *pos)
959{
960 struct raw_iter_state *state = raw_seq_private(seq);
961
962 read_lock(&state->h->lock);
963 return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
964}
965EXPORT_SYMBOL_GPL(raw_seq_start);
966
967void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos)
968{
969 struct sock *sk;
970
971 if (v == SEQ_START_TOKEN)
972 sk = raw_get_first(seq);
973 else
974 sk = raw_get_next(seq, v);
975 ++*pos;
976 return sk;
977}
978EXPORT_SYMBOL_GPL(raw_seq_next);
979
980void raw_seq_stop(struct seq_file *seq, void *v)
981{
982 struct raw_iter_state *state = raw_seq_private(seq);
983
984 read_unlock(&state->h->lock);
985}
986EXPORT_SYMBOL_GPL(raw_seq_stop);
987
988static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
989{
990 struct inet_sock *inet = inet_sk(sp);
991 __be32 dest = inet->inet_daddr,
992 src = inet->inet_rcv_saddr;
993 __u16 destp = 0,
994 srcp = inet->inet_num;
995
996 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
997 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d\n",
998 i, src, srcp, dest, destp, sp->sk_state,
999 sk_wmem_alloc_get(sp),
1000 sk_rmem_alloc_get(sp),
1001 0, 0L, 0,
1002 from_kuid_munged(seq_user_ns(seq), sock_i_uid(sp)),
1003 0, sock_i_ino(sp),
1004 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
1005}
1006
1007static int raw_seq_show(struct seq_file *seq, void *v)
1008{
1009 if (v == SEQ_START_TOKEN)
1010 seq_printf(seq, " sl local_address rem_address st tx_queue "
1011 "rx_queue tr tm->when retrnsmt uid timeout "
1012 "inode ref pointer drops\n");
1013 else
1014 raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
1015 return 0;
1016}
1017
1018static const struct seq_operations raw_seq_ops = {
1019 .start = raw_seq_start,
1020 .next = raw_seq_next,
1021 .stop = raw_seq_stop,
1022 .show = raw_seq_show,
1023};
1024
1025int raw_seq_open(struct inode *ino, struct file *file,
1026 struct raw_hashinfo *h, const struct seq_operations *ops)
1027{
1028 int err;
1029 struct raw_iter_state *i;
1030
1031 err = seq_open_net(ino, file, ops, sizeof(struct raw_iter_state));
1032 if (err < 0)
1033 return err;
1034
1035 i = raw_seq_private((struct seq_file *)file->private_data);
1036 i->h = h;
1037 return 0;
1038}
1039EXPORT_SYMBOL_GPL(raw_seq_open);
1040
1041static int raw_v4_seq_open(struct inode *inode, struct file *file)
1042{
1043 return raw_seq_open(inode, file, &raw_v4_hashinfo, &raw_seq_ops);
1044}
1045
1046static const struct file_operations raw_seq_fops = {
1047 .owner = THIS_MODULE,
1048 .open = raw_v4_seq_open,
1049 .read = seq_read,
1050 .llseek = seq_lseek,
1051 .release = seq_release_net,
1052};
1053
1054static __net_init int raw_init_net(struct net *net)
1055{
1056 if (!proc_create("raw", S_IRUGO, net->proc_net, &raw_seq_fops))
1057 return -ENOMEM;
1058
1059 return 0;
1060}
1061
1062static __net_exit void raw_exit_net(struct net *net)
1063{
1064 remove_proc_entry("raw", net->proc_net);
1065}
1066
1067static __net_initdata struct pernet_operations raw_net_ops = {
1068 .init = raw_init_net,
1069 .exit = raw_exit_net,
1070};
1071
1072int __init raw_proc_init(void)
1073{
1074 return register_pernet_subsys(&raw_net_ops);
1075}
1076
1077void __init raw_proc_exit(void)
1078{
1079 unregister_pernet_subsys(&raw_net_ops);
1080}
1081#endif
1082
