?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
netfilter: xtables: replace XT_ENTRY_ITERATE macro
[linux-2.6/x86.git] / net / ipv4 / ip_sockglue.c
blob644dc43a55dec954dc7232b3edb58c3412951540
1 /*
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.
5 *
6 * The IP to API glue.
7 *
8 * Authors: see ip.c
9 *
10 * Fixes:
11 * Many : Split from ip.c , see ip.c for history.
12 * Martin Mares : TOS setting fixed.
13 * Alan Cox : Fixed a couple of oopses in Martin's
14 * TOS tweaks.
15 * Mike McLagan : Routing by source
16 */
17
18 #include <linux/module.h>
19 #include <linux/types.h>
20 #include <linux/mm.h>
21 #include <linux/skbuff.h>
22 #include <linux/ip.h>
23 #include <linux/icmp.h>
24 #include <linux/inetdevice.h>
25 #include <linux/netdevice.h>
26 #include <net/sock.h>
27 #include <net/ip.h>
28 #include <net/icmp.h>
29 #include <net/tcp_states.h>
30 #include <linux/udp.h>
31 #include <linux/igmp.h>
32 #include <linux/netfilter.h>
33 #include <linux/route.h>
34 #include <linux/mroute.h>
35 #include <net/route.h>
36 #include <net/xfrm.h>
37 #include <net/compat.h>
38 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
39 #include <net/transp_v6.h>
40 #endif
41
42 #include <linux/errqueue.h>
43 #include <asm/uaccess.h>
44
45 #define IP_CMSG_PKTINFO 1
46 #define IP_CMSG_TTL 2
47 #define IP_CMSG_TOS 4
48 #define IP_CMSG_RECVOPTS 8
49 #define IP_CMSG_RETOPTS 16
50 #define IP_CMSG_PASSSEC 32
51 #define IP_CMSG_ORIGDSTADDR 64
52
53 /*
54 * SOL_IP control messages.
55 */
56
57 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
58 {
59 struct in_pktinfo info;
60 struct rtable *rt = skb_rtable(skb);
61
62 info.ipi_addr.s_addr = ip_hdr(skb)->daddr;
63 if (rt) {
64 info.ipi_ifindex = rt->rt_iif;
65 info.ipi_spec_dst.s_addr = rt->rt_spec_dst;
66 } else {
67 info.ipi_ifindex = 0;
68 info.ipi_spec_dst.s_addr = 0;
69 }
70
71 put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
72 }
73
74 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb)
75 {
76 int ttl = ip_hdr(skb)->ttl;
77 put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl);
78 }
79
80 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb)
81 {
82 put_cmsg(msg, SOL_IP, IP_TOS, 1, &ip_hdr(skb)->tos);
83 }
84
85 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb)
86 {
87 if (IPCB(skb)->opt.optlen == 0)
88 return;
89
90 put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen,
91 ip_hdr(skb) + 1);
92 }
93
94
95 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb)
96 {
97 unsigned char optbuf[sizeof(struct ip_options) + 40];
98 struct ip_options * opt = (struct ip_options *)optbuf;
99
100 if (IPCB(skb)->opt.optlen == 0)
101 return;
102
103 if (ip_options_echo(opt, skb)) {
104 msg->msg_flags |= MSG_CTRUNC;
105 return;
106 }
107 ip_options_undo(opt);
108
109 put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data);
110 }
111
112 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb)
113 {
114 char *secdata;
115 u32 seclen, secid;
116 int err;
117
118 err = security_socket_getpeersec_dgram(NULL, skb, &secid);
119 if (err)
120 return;
121
122 err = security_secid_to_secctx(secid, &secdata, &seclen);
123 if (err)
124 return;
125
126 put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata);
127 security_release_secctx(secdata, seclen);
128 }
129
130 static void ip_cmsg_recv_dstaddr(struct msghdr *msg, struct sk_buff *skb)
131 {
132 struct sockaddr_in sin;
133 struct iphdr *iph = ip_hdr(skb);
134 __be16 *ports = (__be16 *)skb_transport_header(skb);
135
136 if (skb_transport_offset(skb) + 4 > skb->len)
137 return;
138
139 /* All current transport protocols have the port numbers in the
140 * first four bytes of the transport header and this function is
141 * written with this assumption in mind.
142 */
143
144 sin.sin_family = AF_INET;
145 sin.sin_addr.s_addr = iph->daddr;
146 sin.sin_port = ports[1];
147 memset(sin.sin_zero, 0, sizeof(sin.sin_zero));
148
149 put_cmsg(msg, SOL_IP, IP_ORIGDSTADDR, sizeof(sin), &sin);
150 }
151
152 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
153 {
154 struct inet_sock *inet = inet_sk(skb->sk);
155 unsigned flags = inet->cmsg_flags;
156
157 /* Ordered by supposed usage frequency */
158 if (flags & 1)
159 ip_cmsg_recv_pktinfo(msg, skb);
160 if ((flags >>= 1) == 0)
161 return;
162
163 if (flags & 1)
164 ip_cmsg_recv_ttl(msg, skb);
165 if ((flags >>= 1) == 0)
166 return;
167
168 if (flags & 1)
169 ip_cmsg_recv_tos(msg, skb);
170 if ((flags >>= 1) == 0)
171 return;
172
173 if (flags & 1)
174 ip_cmsg_recv_opts(msg, skb);
175 if ((flags >>= 1) == 0)
176 return;
177
178 if (flags & 1)
179 ip_cmsg_recv_retopts(msg, skb);
180 if ((flags >>= 1) == 0)
181 return;
182
183 if (flags & 1)
184 ip_cmsg_recv_security(msg, skb);
185
186 if ((flags >>= 1) == 0)
187 return;
188 if (flags & 1)
189 ip_cmsg_recv_dstaddr(msg, skb);
190
191 }
192 EXPORT_SYMBOL(ip_cmsg_recv);
193
194 int ip_cmsg_send(struct net *net, struct msghdr *msg, struct ipcm_cookie *ipc)
195 {
196 int err;
197 struct cmsghdr *cmsg;
198
199 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
200 if (!CMSG_OK(msg, cmsg))
201 return -EINVAL;
202 if (cmsg->cmsg_level != SOL_IP)
203 continue;
204 switch (cmsg->cmsg_type) {
205 case IP_RETOPTS:
206 err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr));
207 err = ip_options_get(net, &ipc->opt, CMSG_DATA(cmsg),
208 err < 40 ? err : 40);
209 if (err)
210 return err;
211 break;
212 case IP_PKTINFO:
213 {
214 struct in_pktinfo *info;
215 if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo)))
216 return -EINVAL;
217 info = (struct in_pktinfo *)CMSG_DATA(cmsg);
218 ipc->oif = info->ipi_ifindex;
219 ipc->addr = info->ipi_spec_dst.s_addr;
220 break;
221 }
222 default:
223 return -EINVAL;
224 }
225 }
226 return 0;
227 }
228
229
230 /* Special input handler for packets caught by router alert option.
231 They are selected only by protocol field, and then processed likely
232 local ones; but only if someone wants them! Otherwise, router
233 not running rsvpd will kill RSVP.
234
235 It is user level problem, what it will make with them.
236 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
237 but receiver should be enough clever f.e. to forward mtrace requests,
238 sent to multicast group to reach destination designated router.
239 */
240 struct ip_ra_chain *ip_ra_chain;
241 DEFINE_RWLOCK(ip_ra_lock);
242
243 int ip_ra_control(struct sock *sk, unsigned char on,
244 void (*destructor)(struct sock *))
245 {
246 struct ip_ra_chain *ra, *new_ra, **rap;
247
248 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->inet_num == IPPROTO_RAW)
249 return -EINVAL;
250
251 new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL;
252
253 write_lock_bh(&ip_ra_lock);
254 for (rap = &ip_ra_chain; (ra = *rap) != NULL; rap = &ra->next) {
255 if (ra->sk == sk) {
256 if (on) {
257 write_unlock_bh(&ip_ra_lock);
258 kfree(new_ra);
259 return -EADDRINUSE;
260 }
261 *rap = ra->next;
262 write_unlock_bh(&ip_ra_lock);
263
264 if (ra->destructor)
265 ra->destructor(sk);
266 sock_put(sk);
267 kfree(ra);
268 return 0;
269 }
270 }
271 if (new_ra == NULL) {
272 write_unlock_bh(&ip_ra_lock);
273 return -ENOBUFS;
274 }
275 new_ra->sk = sk;
276 new_ra->destructor = destructor;
277
278 new_ra->next = ra;
279 *rap = new_ra;
280 sock_hold(sk);
281 write_unlock_bh(&ip_ra_lock);
282
283 return 0;
284 }
285
286 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
287 __be16 port, u32 info, u8 *payload)
288 {
289 struct inet_sock *inet = inet_sk(sk);
290 struct sock_exterr_skb *serr;
291
292 if (!inet->recverr)
293 return;
294
295 skb = skb_clone(skb, GFP_ATOMIC);
296 if (!skb)
297 return;
298
299 serr = SKB_EXT_ERR(skb);
300 serr->ee.ee_errno = err;
301 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP;
302 serr->ee.ee_type = icmp_hdr(skb)->type;
303 serr->ee.ee_code = icmp_hdr(skb)->code;
304 serr->ee.ee_pad = 0;
305 serr->ee.ee_info = info;
306 serr->ee.ee_data = 0;
307 serr->addr_offset = (u8 *)&(((struct iphdr *)(icmp_hdr(skb) + 1))->daddr) -
308 skb_network_header(skb);
309 serr->port = port;
310
311 if (skb_pull(skb, payload - skb->data) != NULL) {
312 skb_reset_transport_header(skb);
313 if (sock_queue_err_skb(sk, skb) == 0)
314 return;
315 }
316 kfree_skb(skb);
317 }
318
319 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info)
320 {
321 struct inet_sock *inet = inet_sk(sk);
322 struct sock_exterr_skb *serr;
323 struct iphdr *iph;
324 struct sk_buff *skb;
325
326 if (!inet->recverr)
327 return;
328
329 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC);
330 if (!skb)
331 return;
332
333 skb_put(skb, sizeof(struct iphdr));
334 skb_reset_network_header(skb);
335 iph = ip_hdr(skb);
336 iph->daddr = daddr;
337
338 serr = SKB_EXT_ERR(skb);
339 serr->ee.ee_errno = err;
340 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
341 serr->ee.ee_type = 0;
342 serr->ee.ee_code = 0;
343 serr->ee.ee_pad = 0;
344 serr->ee.ee_info = info;
345 serr->ee.ee_data = 0;
346 serr->addr_offset = (u8 *)&iph->daddr - skb_network_header(skb);
347 serr->port = port;
348
349 __skb_pull(skb, skb_tail_pointer(skb) - skb->data);
350 skb_reset_transport_header(skb);
351
352 if (sock_queue_err_skb(sk, skb))
353 kfree_skb(skb);
354 }
355
356 /*
357 * Handle MSG_ERRQUEUE
358 */
359 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len)
360 {
361 struct sock_exterr_skb *serr;
362 struct sk_buff *skb, *skb2;
363 struct sockaddr_in *sin;
364 struct {
365 struct sock_extended_err ee;
366 struct sockaddr_in offender;
367 } errhdr;
368 int err;
369 int copied;
370
371 err = -EAGAIN;
372 skb = skb_dequeue(&sk->sk_error_queue);
373 if (skb == NULL)
374 goto out;
375
376 copied = skb->len;
377 if (copied > len) {
378 msg->msg_flags |= MSG_TRUNC;
379 copied = len;
380 }
381 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
382 if (err)
383 goto out_free_skb;
384
385 sock_recv_timestamp(msg, sk, skb);
386
387 serr = SKB_EXT_ERR(skb);
388
389 sin = (struct sockaddr_in *)msg->msg_name;
390 if (sin) {
391 sin->sin_family = AF_INET;
392 sin->sin_addr.s_addr = *(__be32 *)(skb_network_header(skb) +
393 serr->addr_offset);
394 sin->sin_port = serr->port;
395 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
396 }
397
398 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err));
399 sin = &errhdr.offender;
400 sin->sin_family = AF_UNSPEC;
401 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) {
402 struct inet_sock *inet = inet_sk(sk);
403
404 sin->sin_family = AF_INET;
405 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
406 sin->sin_port = 0;
407 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
408 if (inet->cmsg_flags)
409 ip_cmsg_recv(msg, skb);
410 }
411
412 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr);
413
414 /* Now we could try to dump offended packet options */
415
416 msg->msg_flags |= MSG_ERRQUEUE;
417 err = copied;
418
419 /* Reset and regenerate socket error */
420 spin_lock_bh(&sk->sk_error_queue.lock);
421 sk->sk_err = 0;
422 skb2 = skb_peek(&sk->sk_error_queue);
423 if (skb2 != NULL) {
424 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno;
425 spin_unlock_bh(&sk->sk_error_queue.lock);
426 sk->sk_error_report(sk);
427 } else
428 spin_unlock_bh(&sk->sk_error_queue.lock);
429
430 out_free_skb:
431 kfree_skb(skb);
432 out:
433 return err;
434 }
435
436
437 /*
438 * Socket option code for IP. This is the end of the line after any
439 * TCP,UDP etc options on an IP socket.
440 */
441
442 static int do_ip_setsockopt(struct sock *sk, int level,
443 int optname, char __user *optval, unsigned int optlen)
444 {
445 struct inet_sock *inet = inet_sk(sk);
446 int val = 0, err;
447
448 if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) |
449 (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) |
450 (1<<IP_RETOPTS) | (1<<IP_TOS) |
451 (1<<IP_TTL) | (1<<IP_HDRINCL) |
452 (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) |
453 (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND) |
454 (1<<IP_PASSSEC) | (1<<IP_TRANSPARENT) |
455 (1<<IP_MINTTL))) ||
456 optname == IP_MULTICAST_TTL ||
457 optname == IP_MULTICAST_ALL ||
458 optname == IP_MULTICAST_LOOP ||
459 optname == IP_RECVORIGDSTADDR) {
460 if (optlen >= sizeof(int)) {
461 if (get_user(val, (int __user *) optval))
462 return -EFAULT;
463 } else if (optlen >= sizeof(char)) {
464 unsigned char ucval;
465
466 if (get_user(ucval, (unsigned char __user *) optval))
467 return -EFAULT;
468 val = (int) ucval;
469 }
470 }
471
472 /* If optlen==0, it is equivalent to val == 0 */
473
474 if (ip_mroute_opt(optname))
475 return ip_mroute_setsockopt(sk, optname, optval, optlen);
476
477 err = 0;
478 lock_sock(sk);
479
480 switch (optname) {
481 case IP_OPTIONS:
482 {
483 struct ip_options *opt = NULL;
484 if (optlen > 40)
485 goto e_inval;
486 err = ip_options_get_from_user(sock_net(sk), &opt,
487 optval, optlen);
488 if (err)
489 break;
490 if (inet->is_icsk) {
491 struct inet_connection_sock *icsk = inet_csk(sk);
492 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
493 if (sk->sk_family == PF_INET ||
494 (!((1 << sk->sk_state) &
495 (TCPF_LISTEN | TCPF_CLOSE)) &&
496 inet->inet_daddr != LOOPBACK4_IPV6)) {
497 #endif
498 if (inet->opt)
499 icsk->icsk_ext_hdr_len -= inet->opt->optlen;
500 if (opt)
501 icsk->icsk_ext_hdr_len += opt->optlen;
502 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie);
503 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
504 }
505 #endif
506 }
507 opt = xchg(&inet->opt, opt);
508 kfree(opt);
509 break;
510 }
511 case IP_PKTINFO:
512 if (val)
513 inet->cmsg_flags |= IP_CMSG_PKTINFO;
514 else
515 inet->cmsg_flags &= ~IP_CMSG_PKTINFO;
516 break;
517 case IP_RECVTTL:
518 if (val)
519 inet->cmsg_flags |= IP_CMSG_TTL;
520 else
521 inet->cmsg_flags &= ~IP_CMSG_TTL;
522 break;
523 case IP_RECVTOS:
524 if (val)
525 inet->cmsg_flags |= IP_CMSG_TOS;
526 else
527 inet->cmsg_flags &= ~IP_CMSG_TOS;
528 break;
529 case IP_RECVOPTS:
530 if (val)
531 inet->cmsg_flags |= IP_CMSG_RECVOPTS;
532 else
533 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS;
534 break;
535 case IP_RETOPTS:
536 if (val)
537 inet->cmsg_flags |= IP_CMSG_RETOPTS;
538 else
539 inet->cmsg_flags &= ~IP_CMSG_RETOPTS;
540 break;
541 case IP_PASSSEC:
542 if (val)
543 inet->cmsg_flags |= IP_CMSG_PASSSEC;
544 else
545 inet->cmsg_flags &= ~IP_CMSG_PASSSEC;
546 break;
547 case IP_RECVORIGDSTADDR:
548 if (val)
549 inet->cmsg_flags |= IP_CMSG_ORIGDSTADDR;
550 else
551 inet->cmsg_flags &= ~IP_CMSG_ORIGDSTADDR;
552 break;
553 case IP_TOS: /* This sets both TOS and Precedence */
554 if (sk->sk_type == SOCK_STREAM) {
555 val &= ~3;
556 val |= inet->tos & 3;
557 }
558 if (inet->tos != val) {
559 inet->tos = val;
560 sk->sk_priority = rt_tos2priority(val);
561 sk_dst_reset(sk);
562 }
563 break;
564 case IP_TTL:
565 if (optlen < 1)
566 goto e_inval;
567 if (val != -1 && (val < 0 || val > 255))
568 goto e_inval;
569 inet->uc_ttl = val;
570 break;
571 case IP_HDRINCL:
572 if (sk->sk_type != SOCK_RAW) {
573 err = -ENOPROTOOPT;
574 break;
575 }
576 inet->hdrincl = val ? 1 : 0;
577 break;
578 case IP_MTU_DISCOVER:
579 if (val < IP_PMTUDISC_DONT || val > IP_PMTUDISC_PROBE)
580 goto e_inval;
581 inet->pmtudisc = val;
582 break;
583 case IP_RECVERR:
584 inet->recverr = !!val;
585 if (!val)
586 skb_queue_purge(&sk->sk_error_queue);
587 break;
588 case IP_MULTICAST_TTL:
589 if (sk->sk_type == SOCK_STREAM)
590 goto e_inval;
591 if (optlen < 1)
592 goto e_inval;
593 if (val == -1)
594 val = 1;
595 if (val < 0 || val > 255)
596 goto e_inval;
597 inet->mc_ttl = val;
598 break;
599 case IP_MULTICAST_LOOP:
600 if (optlen < 1)
601 goto e_inval;
602 inet->mc_loop = !!val;
603 break;
604 case IP_MULTICAST_IF:
605 {
606 struct ip_mreqn mreq;
607 struct net_device *dev = NULL;
608
609 if (sk->sk_type == SOCK_STREAM)
610 goto e_inval;
611 /*
612 * Check the arguments are allowable
613 */
614
615 if (optlen < sizeof(struct in_addr))
616 goto e_inval;
617
618 err = -EFAULT;
619 if (optlen >= sizeof(struct ip_mreqn)) {
620 if (copy_from_user(&mreq, optval, sizeof(mreq)))
621 break;
622 } else {
623 memset(&mreq, 0, sizeof(mreq));
624 if (optlen >= sizeof(struct in_addr) &&
625 copy_from_user(&mreq.imr_address, optval,
626 sizeof(struct in_addr)))
627 break;
628 }
629
630 if (!mreq.imr_ifindex) {
631 if (mreq.imr_address.s_addr == htonl(INADDR_ANY)) {
632 inet->mc_index = 0;
633 inet->mc_addr = 0;
634 err = 0;
635 break;
636 }
637 dev = ip_dev_find(sock_net(sk), mreq.imr_address.s_addr);
638 if (dev)
639 mreq.imr_ifindex = dev->ifindex;
640 } else
641 dev = dev_get_by_index(sock_net(sk), mreq.imr_ifindex);
642
643
644 err = -EADDRNOTAVAIL;
645 if (!dev)
646 break;
647 dev_put(dev);
648
649 err = -EINVAL;
650 if (sk->sk_bound_dev_if &&
651 mreq.imr_ifindex != sk->sk_bound_dev_if)
652 break;
653
654 inet->mc_index = mreq.imr_ifindex;
655 inet->mc_addr = mreq.imr_address.s_addr;
656 err = 0;
657 break;
658 }
659
660 case IP_ADD_MEMBERSHIP:
661 case IP_DROP_MEMBERSHIP:
662 {
663 struct ip_mreqn mreq;
664
665 err = -EPROTO;
666 if (inet_sk(sk)->is_icsk)
667 break;
668
669 if (optlen < sizeof(struct ip_mreq))
670 goto e_inval;
671 err = -EFAULT;
672 if (optlen >= sizeof(struct ip_mreqn)) {
673 if (copy_from_user(&mreq, optval, sizeof(mreq)))
674 break;
675 } else {
676 memset(&mreq, 0, sizeof(mreq));
677 if (copy_from_user(&mreq, optval, sizeof(struct ip_mreq)))
678 break;
679 }
680
681 if (optname == IP_ADD_MEMBERSHIP)
682 err = ip_mc_join_group(sk, &mreq);
683 else
684 err = ip_mc_leave_group(sk, &mreq);
685 break;
686 }
687 case IP_MSFILTER:
688 {
689 struct ip_msfilter *msf;
690
691 if (optlen < IP_MSFILTER_SIZE(0))
692 goto e_inval;
693 if (optlen > sysctl_optmem_max) {
694 err = -ENOBUFS;
695 break;
696 }
697 msf = kmalloc(optlen, GFP_KERNEL);
698 if (!msf) {
699 err = -ENOBUFS;
700 break;
701 }
702 err = -EFAULT;
703 if (copy_from_user(msf, optval, optlen)) {
704 kfree(msf);
705 break;
706 }
707 /* numsrc >= (1G-4) overflow in 32 bits */
708 if (msf->imsf_numsrc >= 0x3ffffffcU ||
709 msf->imsf_numsrc > sysctl_igmp_max_msf) {
710 kfree(msf);
711 err = -ENOBUFS;
712 break;
713 }
714 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) {
715 kfree(msf);
716 err = -EINVAL;
717 break;
718 }
719 err = ip_mc_msfilter(sk, msf, 0);
720 kfree(msf);
721 break;
722 }
723 case IP_BLOCK_SOURCE:
724 case IP_UNBLOCK_SOURCE:
725 case IP_ADD_SOURCE_MEMBERSHIP:
726 case IP_DROP_SOURCE_MEMBERSHIP:
727 {
728 struct ip_mreq_source mreqs;
729 int omode, add;
730
731 if (optlen != sizeof(struct ip_mreq_source))
732 goto e_inval;
733 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) {
734 err = -EFAULT;
735 break;
736 }
737 if (optname == IP_BLOCK_SOURCE) {
738 omode = MCAST_EXCLUDE;
739 add = 1;
740 } else if (optname == IP_UNBLOCK_SOURCE) {
741 omode = MCAST_EXCLUDE;
742 add = 0;
743 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) {
744 struct ip_mreqn mreq;
745
746 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr;
747 mreq.imr_address.s_addr = mreqs.imr_interface;
748 mreq.imr_ifindex = 0;
749 err = ip_mc_join_group(sk, &mreq);
750 if (err && err != -EADDRINUSE)
751 break;
752 omode = MCAST_INCLUDE;
753 add = 1;
754 } else /* IP_DROP_SOURCE_MEMBERSHIP */ {
755 omode = MCAST_INCLUDE;
756 add = 0;
757 }
758 err = ip_mc_source(add, omode, sk, &mreqs, 0);
759 break;
760 }
761 case MCAST_JOIN_GROUP:
762 case MCAST_LEAVE_GROUP:
763 {
764 struct group_req greq;
765 struct sockaddr_in *psin;
766 struct ip_mreqn mreq;
767
768 if (optlen < sizeof(struct group_req))
769 goto e_inval;
770 err = -EFAULT;
771 if (copy_from_user(&greq, optval, sizeof(greq)))
772 break;
773 psin = (struct sockaddr_in *)&greq.gr_group;
774 if (psin->sin_family != AF_INET)
775 goto e_inval;
776 memset(&mreq, 0, sizeof(mreq));
777 mreq.imr_multiaddr = psin->sin_addr;
778 mreq.imr_ifindex = greq.gr_interface;
779
780 if (optname == MCAST_JOIN_GROUP)
781 err = ip_mc_join_group(sk, &mreq);
782 else
783 err = ip_mc_leave_group(sk, &mreq);
784 break;
785 }
786 case MCAST_JOIN_SOURCE_GROUP:
787 case MCAST_LEAVE_SOURCE_GROUP:
788 case MCAST_BLOCK_SOURCE:
789 case MCAST_UNBLOCK_SOURCE:
790 {
791 struct group_source_req greqs;
792 struct ip_mreq_source mreqs;
793 struct sockaddr_in *psin;
794 int omode, add;
795
796 if (optlen != sizeof(struct group_source_req))
797 goto e_inval;
798 if (copy_from_user(&greqs, optval, sizeof(greqs))) {
799 err = -EFAULT;
800 break;
801 }
802 if (greqs.gsr_group.ss_family != AF_INET ||
803 greqs.gsr_source.ss_family != AF_INET) {
804 err = -EADDRNOTAVAIL;
805 break;
806 }
807 psin = (struct sockaddr_in *)&greqs.gsr_group;
808 mreqs.imr_multiaddr = psin->sin_addr.s_addr;
809 psin = (struct sockaddr_in *)&greqs.gsr_source;
810 mreqs.imr_sourceaddr = psin->sin_addr.s_addr;
811 mreqs.imr_interface = 0; /* use index for mc_source */
812
813 if (optname == MCAST_BLOCK_SOURCE) {
814 omode = MCAST_EXCLUDE;
815 add = 1;
816 } else if (optname == MCAST_UNBLOCK_SOURCE) {
817 omode = MCAST_EXCLUDE;
818 add = 0;
819 } else if (optname == MCAST_JOIN_SOURCE_GROUP) {
820 struct ip_mreqn mreq;
821
822 psin = (struct sockaddr_in *)&greqs.gsr_group;
823 mreq.imr_multiaddr = psin->sin_addr;
824 mreq.imr_address.s_addr = 0;
825 mreq.imr_ifindex = greqs.gsr_interface;
826 err = ip_mc_join_group(sk, &mreq);
827 if (err && err != -EADDRINUSE)
828 break;
829 greqs.gsr_interface = mreq.imr_ifindex;
830 omode = MCAST_INCLUDE;
831 add = 1;
832 } else /* MCAST_LEAVE_SOURCE_GROUP */ {
833 omode = MCAST_INCLUDE;
834 add = 0;
835 }
836 err = ip_mc_source(add, omode, sk, &mreqs,
837 greqs.gsr_interface);
838 break;
839 }
840 case MCAST_MSFILTER:
841 {
842 struct sockaddr_in *psin;
843 struct ip_msfilter *msf = NULL;
844 struct group_filter *gsf = NULL;
845 int msize, i, ifindex;
846
847 if (optlen < GROUP_FILTER_SIZE(0))
848 goto e_inval;
849 if (optlen > sysctl_optmem_max) {
850 err = -ENOBUFS;
851 break;
852 }
853 gsf = kmalloc(optlen, GFP_KERNEL);
854 if (!gsf) {
855 err = -ENOBUFS;
856 break;
857 }
858 err = -EFAULT;
859 if (copy_from_user(gsf, optval, optlen))
860 goto mc_msf_out;
861
862 /* numsrc >= (4G-140)/128 overflow in 32 bits */
863 if (gsf->gf_numsrc >= 0x1ffffff ||
864 gsf->gf_numsrc > sysctl_igmp_max_msf) {
865 err = -ENOBUFS;
866 goto mc_msf_out;
867 }
868 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) {
869 err = -EINVAL;
870 goto mc_msf_out;
871 }
872 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc);
873 msf = kmalloc(msize, GFP_KERNEL);
874 if (!msf) {
875 err = -ENOBUFS;
876 goto mc_msf_out;
877 }
878 ifindex = gsf->gf_interface;
879 psin = (struct sockaddr_in *)&gsf->gf_group;
880 if (psin->sin_family != AF_INET) {
881 err = -EADDRNOTAVAIL;
882 goto mc_msf_out;
883 }
884 msf->imsf_multiaddr = psin->sin_addr.s_addr;
885 msf->imsf_interface = 0;
886 msf->imsf_fmode = gsf->gf_fmode;
887 msf->imsf_numsrc = gsf->gf_numsrc;
888 err = -EADDRNOTAVAIL;
889 for (i = 0; i < gsf->gf_numsrc; ++i) {
890 psin = (struct sockaddr_in *)&gsf->gf_slist[i];
891
892 if (psin->sin_family != AF_INET)
893 goto mc_msf_out;
894 msf->imsf_slist[i] = psin->sin_addr.s_addr;
895 }
896 kfree(gsf);
897 gsf = NULL;
898
899 err = ip_mc_msfilter(sk, msf, ifindex);
900 mc_msf_out:
901 kfree(msf);
902 kfree(gsf);
903 break;
904 }
905 case IP_MULTICAST_ALL:
906 if (optlen < 1)
907 goto e_inval;
908 if (val != 0 && val != 1)
909 goto e_inval;
910 inet->mc_all = val;
911 break;
912 case IP_ROUTER_ALERT:
913 err = ip_ra_control(sk, val ? 1 : 0, NULL);
914 break;
915
916 case IP_FREEBIND:
917 if (optlen < 1)
918 goto e_inval;
919 inet->freebind = !!val;
920 break;
921
922 case IP_IPSEC_POLICY:
923 case IP_XFRM_POLICY:
924 err = -EPERM;
925 if (!capable(CAP_NET_ADMIN))
926 break;
927 err = xfrm_user_policy(sk, optname, optval, optlen);
928 break;
929
930 case IP_TRANSPARENT:
931 if (!capable(CAP_NET_ADMIN)) {
932 err = -EPERM;
933 break;
934 }
935 if (optlen < 1)
936 goto e_inval;
937 inet->transparent = !!val;
938 break;
939
940 case IP_MINTTL:
941 if (optlen < 1)
942 goto e_inval;
943 if (val < 0 || val > 255)
944 goto e_inval;
945 inet->min_ttl = val;
946 break;
947
948 default:
949 err = -ENOPROTOOPT;
950 break;
951 }
952 release_sock(sk);
953 return err;
954
955 e_inval:
956 release_sock(sk);
957 return -EINVAL;
958 }
959
960 int ip_setsockopt(struct sock *sk, int level,
961 int optname, char __user *optval, unsigned int optlen)
962 {
963 int err;
964
965 if (level != SOL_IP)
966 return -ENOPROTOOPT;
967
968 err = do_ip_setsockopt(sk, level, optname, optval, optlen);
969 #ifdef CONFIG_NETFILTER
970 /* we need to exclude all possible ENOPROTOOPTs except default case */
971 if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
972 optname != IP_IPSEC_POLICY &&
973 optname != IP_XFRM_POLICY &&
974 !ip_mroute_opt(optname)) {
975 lock_sock(sk);
976 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen);
977 release_sock(sk);
978 }
979 #endif
980 return err;
981 }
982 EXPORT_SYMBOL(ip_setsockopt);
983
984 #ifdef CONFIG_COMPAT
985 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
986 char __user *optval, unsigned int optlen)
987 {
988 int err;
989
990 if (level != SOL_IP)
991 return -ENOPROTOOPT;
992
993 if (optname >= MCAST_JOIN_GROUP && optname <= MCAST_MSFILTER)
994 return compat_mc_setsockopt(sk, level, optname, optval, optlen,
995 ip_setsockopt);
996
997 err = do_ip_setsockopt(sk, level, optname, optval, optlen);
998 #ifdef CONFIG_NETFILTER
999 /* we need to exclude all possible ENOPROTOOPTs except default case */
1000 if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
1001 optname != IP_IPSEC_POLICY &&
1002 optname != IP_XFRM_POLICY &&
1003 !ip_mroute_opt(optname)) {
1004 lock_sock(sk);
1005 err = compat_nf_setsockopt(sk, PF_INET, optname,
1006 optval, optlen);
1007 release_sock(sk);
1008 }
1009 #endif
1010 return err;
1011 }
1012 EXPORT_SYMBOL(compat_ip_setsockopt);
1013 #endif
1014
1015 /*
1016 * Get the options. Note for future reference. The GET of IP options gets
1017 * the _received_ ones. The set sets the _sent_ ones.
1018 */
1019
1020 static int do_ip_getsockopt(struct sock *sk, int level, int optname,
1021 char __user *optval, int __user *optlen)
1022 {
1023 struct inet_sock *inet = inet_sk(sk);
1024 int val;
1025 int len;
1026
1027 if (level != SOL_IP)
1028 return -EOPNOTSUPP;
1029
1030 if (ip_mroute_opt(optname))
1031 return ip_mroute_getsockopt(sk, optname, optval, optlen);
1032
1033 if (get_user(len, optlen))
1034 return -EFAULT;
1035 if (len < 0)
1036 return -EINVAL;
1037
1038 lock_sock(sk);
1039
1040 switch (optname) {
1041 case IP_OPTIONS:
1042 {
1043 unsigned char optbuf[sizeof(struct ip_options)+40];
1044 struct ip_options * opt = (struct ip_options *)optbuf;
1045 opt->optlen = 0;
1046 if (inet->opt)
1047 memcpy(optbuf, inet->opt,
1048 sizeof(struct ip_options)+
1049 inet->opt->optlen);
1050 release_sock(sk);
1051
1052 if (opt->optlen == 0)
1053 return put_user(0, optlen);
1054
1055 ip_options_undo(opt);
1056
1057 len = min_t(unsigned int, len, opt->optlen);
1058 if (put_user(len, optlen))
1059 return -EFAULT;
1060 if (copy_to_user(optval, opt->__data, len))
1061 return -EFAULT;
1062 return 0;
1063 }
1064 case IP_PKTINFO:
1065 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0;
1066 break;
1067 case IP_RECVTTL:
1068 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0;
1069 break;
1070 case IP_RECVTOS:
1071 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0;
1072 break;
1073 case IP_RECVOPTS:
1074 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0;
1075 break;
1076 case IP_RETOPTS:
1077 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0;
1078 break;
1079 case IP_PASSSEC:
1080 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0;
1081 break;
1082 case IP_RECVORIGDSTADDR:
1083 val = (inet->cmsg_flags & IP_CMSG_ORIGDSTADDR) != 0;
1084 break;
1085 case IP_TOS:
1086 val = inet->tos;
1087 break;
1088 case IP_TTL:
1089 val = (inet->uc_ttl == -1 ?
1090 sysctl_ip_default_ttl :
1091 inet->uc_ttl);
1092 break;
1093 case IP_HDRINCL:
1094 val = inet->hdrincl;
1095 break;
1096 case IP_MTU_DISCOVER:
1097 val = inet->pmtudisc;
1098 break;
1099 case IP_MTU:
1100 {
1101 struct dst_entry *dst;
1102 val = 0;
1103 dst = sk_dst_get(sk);
1104 if (dst) {
1105 val = dst_mtu(dst);
1106 dst_release(dst);
1107 }
1108 if (!val) {
1109 release_sock(sk);
1110 return -ENOTCONN;
1111 }
1112 break;
1113 }
1114 case IP_RECVERR:
1115 val = inet->recverr;
1116 break;
1117 case IP_MULTICAST_TTL:
1118 val = inet->mc_ttl;
1119 break;
1120 case IP_MULTICAST_LOOP:
1121 val = inet->mc_loop;
1122 break;
1123 case IP_MULTICAST_IF:
1124 {
1125 struct in_addr addr;
1126 len = min_t(unsigned int, len, sizeof(struct in_addr));
1127 addr.s_addr = inet->mc_addr;
1128 release_sock(sk);
1129
1130 if (put_user(len, optlen))
1131 return -EFAULT;
1132 if (copy_to_user(optval, &addr, len))
1133 return -EFAULT;
1134 return 0;
1135 }
1136 case IP_MSFILTER:
1137 {
1138 struct ip_msfilter msf;
1139 int err;
1140
1141 if (len < IP_MSFILTER_SIZE(0)) {
1142 release_sock(sk);
1143 return -EINVAL;
1144 }
1145 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) {
1146 release_sock(sk);
1147 return -EFAULT;
1148 }
1149 err = ip_mc_msfget(sk, &msf,
1150 (struct ip_msfilter __user *)optval, optlen);
1151 release_sock(sk);
1152 return err;
1153 }
1154 case MCAST_MSFILTER:
1155 {
1156 struct group_filter gsf;
1157 int err;
1158
1159 if (len < GROUP_FILTER_SIZE(0)) {
1160 release_sock(sk);
1161 return -EINVAL;
1162 }
1163 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) {
1164 release_sock(sk);
1165 return -EFAULT;
1166 }
1167 err = ip_mc_gsfget(sk, &gsf,
1168 (struct group_filter __user *)optval,
1169 optlen);
1170 release_sock(sk);
1171 return err;
1172 }
1173 case IP_MULTICAST_ALL:
1174 val = inet->mc_all;
1175 break;
1176 case IP_PKTOPTIONS:
1177 {
1178 struct msghdr msg;
1179
1180 release_sock(sk);
1181
1182 if (sk->sk_type != SOCK_STREAM)
1183 return -ENOPROTOOPT;
1184
1185 msg.msg_control = optval;
1186 msg.msg_controllen = len;
1187 msg.msg_flags = 0;
1188
1189 if (inet->cmsg_flags & IP_CMSG_PKTINFO) {
1190 struct in_pktinfo info;
1191
1192 info.ipi_addr.s_addr = inet->inet_rcv_saddr;
1193 info.ipi_spec_dst.s_addr = inet->inet_rcv_saddr;
1194 info.ipi_ifindex = inet->mc_index;
1195 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
1196 }
1197 if (inet->cmsg_flags & IP_CMSG_TTL) {
1198 int hlim = inet->mc_ttl;
1199 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim);
1200 }
1201 len -= msg.msg_controllen;
1202 return put_user(len, optlen);
1203 }
1204 case IP_FREEBIND:
1205 val = inet->freebind;
1206 break;
1207 case IP_TRANSPARENT:
1208 val = inet->transparent;
1209 break;
1210 case IP_MINTTL:
1211 val = inet->min_ttl;
1212 break;
1213 default:
1214 release_sock(sk);
1215 return -ENOPROTOOPT;
1216 }
1217 release_sock(sk);
1218
1219 if (len < sizeof(int) && len > 0 && val >= 0 && val <= 255) {
1220 unsigned char ucval = (unsigned char)val;
1221 len = 1;
1222 if (put_user(len, optlen))
1223 return -EFAULT;
1224 if (copy_to_user(optval, &ucval, 1))
1225 return -EFAULT;
1226 } else {
1227 len = min_t(unsigned int, sizeof(int), len);
1228 if (put_user(len, optlen))
1229 return -EFAULT;
1230 if (copy_to_user(optval, &val, len))
1231 return -EFAULT;
1232 }
1233 return 0;
1234 }
1235
1236 int ip_getsockopt(struct sock *sk, int level,
1237 int optname, char __user *optval, int __user *optlen)
1238 {
1239 int err;
1240
1241 err = do_ip_getsockopt(sk, level, optname, optval, optlen);
1242 #ifdef CONFIG_NETFILTER
1243 /* we need to exclude all possible ENOPROTOOPTs except default case */
1244 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1245 !ip_mroute_opt(optname)) {
1246 int len;
1247
1248 if (get_user(len, optlen))
1249 return -EFAULT;
1250
1251 lock_sock(sk);
1252 err = nf_getsockopt(sk, PF_INET, optname, optval,
1253 &len);
1254 release_sock(sk);
1255 if (err >= 0)
1256 err = put_user(len, optlen);
1257 return err;
1258 }
1259 #endif
1260 return err;
1261 }
1262 EXPORT_SYMBOL(ip_getsockopt);
1263
1264 #ifdef CONFIG_COMPAT
1265 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
1266 char __user *optval, int __user *optlen)
1267 {
1268 int err;
1269
1270 if (optname == MCAST_MSFILTER)
1271 return compat_mc_getsockopt(sk, level, optname, optval, optlen,
1272 ip_getsockopt);
1273
1274 err = do_ip_getsockopt(sk, level, optname, optval, optlen);
1275
1276 #ifdef CONFIG_NETFILTER
1277 /* we need to exclude all possible ENOPROTOOPTs except default case */
1278 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1279 !ip_mroute_opt(optname)) {
1280 int len;
1281
1282 if (get_user(len, optlen))
1283 return -EFAULT;
1284
1285 lock_sock(sk);
1286 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len);
1287 release_sock(sk);
1288 if (err >= 0)
1289 err = put_user(len, optlen);
1290 return err;
1291 }
1292 #endif
1293 return err;
1294 }
1295 EXPORT_SYMBOL(compat_ip_getsockopt);
1296 #endif
x86 "subsystem" repository