Linux 2.6.17.7
[linux/fpc-iii.git] / net / ipv4 / ip_sockglue.c
blob12e0bf19f24ad7c109382bbbfcc0a0ac572feccf
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.
6 * The IP to API glue.
7 *
8 * Version: $Id: ip_sockglue.c,v 1.62 2002/02/01 22:01:04 davem Exp $
10 * Authors: see ip.c
12 * Fixes:
13 * Many : Split from ip.c , see ip.c for history.
14 * Martin Mares : TOS setting fixed.
15 * Alan Cox : Fixed a couple of oopses in Martin's
16 * TOS tweaks.
17 * Mike McLagan : Routing by source
20 #include <linux/config.h>
21 #include <linux/module.h>
22 #include <linux/types.h>
23 #include <linux/mm.h>
24 #include <linux/sched.h>
25 #include <linux/skbuff.h>
26 #include <linux/ip.h>
27 #include <linux/icmp.h>
28 #include <linux/inetdevice.h>
29 #include <linux/netdevice.h>
30 #include <net/sock.h>
31 #include <net/ip.h>
32 #include <net/icmp.h>
33 #include <net/tcp_states.h>
34 #include <linux/udp.h>
35 #include <linux/igmp.h>
36 #include <linux/netfilter.h>
37 #include <linux/route.h>
38 #include <linux/mroute.h>
39 #include <net/route.h>
40 #include <net/xfrm.h>
41 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
42 #include <net/transp_v6.h>
43 #endif
45 #include <linux/errqueue.h>
46 #include <asm/uaccess.h>
48 #define IP_CMSG_PKTINFO 1
49 #define IP_CMSG_TTL 2
50 #define IP_CMSG_TOS 4
51 #define IP_CMSG_RECVOPTS 8
52 #define IP_CMSG_RETOPTS 16
53 #define IP_CMSG_PASSSEC 32
56 * SOL_IP control messages.
59 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
61 struct in_pktinfo info;
62 struct rtable *rt = (struct rtable *)skb->dst;
64 info.ipi_addr.s_addr = skb->nh.iph->daddr;
65 if (rt) {
66 info.ipi_ifindex = rt->rt_iif;
67 info.ipi_spec_dst.s_addr = rt->rt_spec_dst;
68 } else {
69 info.ipi_ifindex = 0;
70 info.ipi_spec_dst.s_addr = 0;
73 put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
76 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb)
78 int ttl = skb->nh.iph->ttl;
79 put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl);
82 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb)
84 put_cmsg(msg, SOL_IP, IP_TOS, 1, &skb->nh.iph->tos);
87 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb)
89 if (IPCB(skb)->opt.optlen == 0)
90 return;
92 put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen, skb->nh.iph+1);
96 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb)
98 unsigned char optbuf[sizeof(struct ip_options) + 40];
99 struct ip_options * opt = (struct ip_options*)optbuf;
101 if (IPCB(skb)->opt.optlen == 0)
102 return;
104 if (ip_options_echo(opt, skb)) {
105 msg->msg_flags |= MSG_CTRUNC;
106 return;
108 ip_options_undo(opt);
110 put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data);
113 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb)
115 char *secdata;
116 u32 seclen;
117 int err;
119 err = security_socket_getpeersec_dgram(skb, &secdata, &seclen);
120 if (err)
121 return;
123 put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata);
127 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
129 struct inet_sock *inet = inet_sk(skb->sk);
130 unsigned flags = inet->cmsg_flags;
132 /* Ordered by supposed usage frequency */
133 if (flags & 1)
134 ip_cmsg_recv_pktinfo(msg, skb);
135 if ((flags>>=1) == 0)
136 return;
138 if (flags & 1)
139 ip_cmsg_recv_ttl(msg, skb);
140 if ((flags>>=1) == 0)
141 return;
143 if (flags & 1)
144 ip_cmsg_recv_tos(msg, skb);
145 if ((flags>>=1) == 0)
146 return;
148 if (flags & 1)
149 ip_cmsg_recv_opts(msg, skb);
150 if ((flags>>=1) == 0)
151 return;
153 if (flags & 1)
154 ip_cmsg_recv_retopts(msg, skb);
155 if ((flags>>=1) == 0)
156 return;
158 if (flags & 1)
159 ip_cmsg_recv_security(msg, skb);
162 int ip_cmsg_send(struct msghdr *msg, struct ipcm_cookie *ipc)
164 int err;
165 struct cmsghdr *cmsg;
167 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
168 if (!CMSG_OK(msg, cmsg))
169 return -EINVAL;
170 if (cmsg->cmsg_level != SOL_IP)
171 continue;
172 switch (cmsg->cmsg_type) {
173 case IP_RETOPTS:
174 err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr));
175 err = ip_options_get(&ipc->opt, CMSG_DATA(cmsg), err < 40 ? err : 40);
176 if (err)
177 return err;
178 break;
179 case IP_PKTINFO:
181 struct in_pktinfo *info;
182 if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo)))
183 return -EINVAL;
184 info = (struct in_pktinfo *)CMSG_DATA(cmsg);
185 ipc->oif = info->ipi_ifindex;
186 ipc->addr = info->ipi_spec_dst.s_addr;
187 break;
189 default:
190 return -EINVAL;
193 return 0;
197 /* Special input handler for packets caught by router alert option.
198 They are selected only by protocol field, and then processed likely
199 local ones; but only if someone wants them! Otherwise, router
200 not running rsvpd will kill RSVP.
202 It is user level problem, what it will make with them.
203 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
204 but receiver should be enough clever f.e. to forward mtrace requests,
205 sent to multicast group to reach destination designated router.
207 struct ip_ra_chain *ip_ra_chain;
208 DEFINE_RWLOCK(ip_ra_lock);
210 int ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *))
212 struct ip_ra_chain *ra, *new_ra, **rap;
214 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->num == IPPROTO_RAW)
215 return -EINVAL;
217 new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL;
219 write_lock_bh(&ip_ra_lock);
220 for (rap = &ip_ra_chain; (ra=*rap) != NULL; rap = &ra->next) {
221 if (ra->sk == sk) {
222 if (on) {
223 write_unlock_bh(&ip_ra_lock);
224 kfree(new_ra);
225 return -EADDRINUSE;
227 *rap = ra->next;
228 write_unlock_bh(&ip_ra_lock);
230 if (ra->destructor)
231 ra->destructor(sk);
232 sock_put(sk);
233 kfree(ra);
234 return 0;
237 if (new_ra == NULL) {
238 write_unlock_bh(&ip_ra_lock);
239 return -ENOBUFS;
241 new_ra->sk = sk;
242 new_ra->destructor = destructor;
244 new_ra->next = ra;
245 *rap = new_ra;
246 sock_hold(sk);
247 write_unlock_bh(&ip_ra_lock);
249 return 0;
252 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
253 u16 port, u32 info, u8 *payload)
255 struct inet_sock *inet = inet_sk(sk);
256 struct sock_exterr_skb *serr;
258 if (!inet->recverr)
259 return;
261 skb = skb_clone(skb, GFP_ATOMIC);
262 if (!skb)
263 return;
265 serr = SKB_EXT_ERR(skb);
266 serr->ee.ee_errno = err;
267 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP;
268 serr->ee.ee_type = skb->h.icmph->type;
269 serr->ee.ee_code = skb->h.icmph->code;
270 serr->ee.ee_pad = 0;
271 serr->ee.ee_info = info;
272 serr->ee.ee_data = 0;
273 serr->addr_offset = (u8*)&(((struct iphdr*)(skb->h.icmph+1))->daddr) - skb->nh.raw;
274 serr->port = port;
276 skb->h.raw = payload;
277 if (!skb_pull(skb, payload - skb->data) ||
278 sock_queue_err_skb(sk, skb))
279 kfree_skb(skb);
282 void ip_local_error(struct sock *sk, int err, u32 daddr, u16 port, u32 info)
284 struct inet_sock *inet = inet_sk(sk);
285 struct sock_exterr_skb *serr;
286 struct iphdr *iph;
287 struct sk_buff *skb;
289 if (!inet->recverr)
290 return;
292 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC);
293 if (!skb)
294 return;
296 iph = (struct iphdr*)skb_put(skb, sizeof(struct iphdr));
297 skb->nh.iph = iph;
298 iph->daddr = daddr;
300 serr = SKB_EXT_ERR(skb);
301 serr->ee.ee_errno = err;
302 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
303 serr->ee.ee_type = 0;
304 serr->ee.ee_code = 0;
305 serr->ee.ee_pad = 0;
306 serr->ee.ee_info = info;
307 serr->ee.ee_data = 0;
308 serr->addr_offset = (u8*)&iph->daddr - skb->nh.raw;
309 serr->port = port;
311 skb->h.raw = skb->tail;
312 __skb_pull(skb, skb->tail - skb->data);
314 if (sock_queue_err_skb(sk, skb))
315 kfree_skb(skb);
319 * Handle MSG_ERRQUEUE
321 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len)
323 struct sock_exterr_skb *serr;
324 struct sk_buff *skb, *skb2;
325 struct sockaddr_in *sin;
326 struct {
327 struct sock_extended_err ee;
328 struct sockaddr_in offender;
329 } errhdr;
330 int err;
331 int copied;
333 err = -EAGAIN;
334 skb = skb_dequeue(&sk->sk_error_queue);
335 if (skb == NULL)
336 goto out;
338 copied = skb->len;
339 if (copied > len) {
340 msg->msg_flags |= MSG_TRUNC;
341 copied = len;
343 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
344 if (err)
345 goto out_free_skb;
347 sock_recv_timestamp(msg, sk, skb);
349 serr = SKB_EXT_ERR(skb);
351 sin = (struct sockaddr_in *)msg->msg_name;
352 if (sin) {
353 sin->sin_family = AF_INET;
354 sin->sin_addr.s_addr = *(u32*)(skb->nh.raw + serr->addr_offset);
355 sin->sin_port = serr->port;
356 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
359 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err));
360 sin = &errhdr.offender;
361 sin->sin_family = AF_UNSPEC;
362 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) {
363 struct inet_sock *inet = inet_sk(sk);
365 sin->sin_family = AF_INET;
366 sin->sin_addr.s_addr = skb->nh.iph->saddr;
367 sin->sin_port = 0;
368 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
369 if (inet->cmsg_flags)
370 ip_cmsg_recv(msg, skb);
373 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr);
375 /* Now we could try to dump offended packet options */
377 msg->msg_flags |= MSG_ERRQUEUE;
378 err = copied;
380 /* Reset and regenerate socket error */
381 spin_lock_bh(&sk->sk_error_queue.lock);
382 sk->sk_err = 0;
383 if ((skb2 = skb_peek(&sk->sk_error_queue)) != NULL) {
384 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno;
385 spin_unlock_bh(&sk->sk_error_queue.lock);
386 sk->sk_error_report(sk);
387 } else
388 spin_unlock_bh(&sk->sk_error_queue.lock);
390 out_free_skb:
391 kfree_skb(skb);
392 out:
393 return err;
398 * Socket option code for IP. This is the end of the line after any TCP,UDP etc options on
399 * an IP socket.
402 static int do_ip_setsockopt(struct sock *sk, int level,
403 int optname, char __user *optval, int optlen)
405 struct inet_sock *inet = inet_sk(sk);
406 int val=0,err;
408 if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) |
409 (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) |
410 (1<<IP_RETOPTS) | (1<<IP_TOS) |
411 (1<<IP_TTL) | (1<<IP_HDRINCL) |
412 (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) |
413 (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND) |
414 (1<<IP_PASSSEC))) ||
415 optname == IP_MULTICAST_TTL ||
416 optname == IP_MULTICAST_LOOP) {
417 if (optlen >= sizeof(int)) {
418 if (get_user(val, (int __user *) optval))
419 return -EFAULT;
420 } else if (optlen >= sizeof(char)) {
421 unsigned char ucval;
423 if (get_user(ucval, (unsigned char __user *) optval))
424 return -EFAULT;
425 val = (int) ucval;
429 /* If optlen==0, it is equivalent to val == 0 */
431 #ifdef CONFIG_IP_MROUTE
432 if (optname >= MRT_BASE && optname <= (MRT_BASE + 10))
433 return ip_mroute_setsockopt(sk,optname,optval,optlen);
434 #endif
436 err = 0;
437 lock_sock(sk);
439 switch (optname) {
440 case IP_OPTIONS:
442 struct ip_options * opt = NULL;
443 if (optlen > 40 || optlen < 0)
444 goto e_inval;
445 err = ip_options_get_from_user(&opt, optval, optlen);
446 if (err)
447 break;
448 if (inet->is_icsk) {
449 struct inet_connection_sock *icsk = inet_csk(sk);
450 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
451 if (sk->sk_family == PF_INET ||
452 (!((1 << sk->sk_state) &
453 (TCPF_LISTEN | TCPF_CLOSE)) &&
454 inet->daddr != LOOPBACK4_IPV6)) {
455 #endif
456 if (inet->opt)
457 icsk->icsk_ext_hdr_len -= inet->opt->optlen;
458 if (opt)
459 icsk->icsk_ext_hdr_len += opt->optlen;
460 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie);
461 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
463 #endif
465 opt = xchg(&inet->opt, opt);
466 kfree(opt);
467 break;
469 case IP_PKTINFO:
470 if (val)
471 inet->cmsg_flags |= IP_CMSG_PKTINFO;
472 else
473 inet->cmsg_flags &= ~IP_CMSG_PKTINFO;
474 break;
475 case IP_RECVTTL:
476 if (val)
477 inet->cmsg_flags |= IP_CMSG_TTL;
478 else
479 inet->cmsg_flags &= ~IP_CMSG_TTL;
480 break;
481 case IP_RECVTOS:
482 if (val)
483 inet->cmsg_flags |= IP_CMSG_TOS;
484 else
485 inet->cmsg_flags &= ~IP_CMSG_TOS;
486 break;
487 case IP_RECVOPTS:
488 if (val)
489 inet->cmsg_flags |= IP_CMSG_RECVOPTS;
490 else
491 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS;
492 break;
493 case IP_RETOPTS:
494 if (val)
495 inet->cmsg_flags |= IP_CMSG_RETOPTS;
496 else
497 inet->cmsg_flags &= ~IP_CMSG_RETOPTS;
498 break;
499 case IP_PASSSEC:
500 if (val)
501 inet->cmsg_flags |= IP_CMSG_PASSSEC;
502 else
503 inet->cmsg_flags &= ~IP_CMSG_PASSSEC;
504 break;
505 case IP_TOS: /* This sets both TOS and Precedence */
506 if (sk->sk_type == SOCK_STREAM) {
507 val &= ~3;
508 val |= inet->tos & 3;
510 if (IPTOS_PREC(val) >= IPTOS_PREC_CRITIC_ECP &&
511 !capable(CAP_NET_ADMIN)) {
512 err = -EPERM;
513 break;
515 if (inet->tos != val) {
516 inet->tos = val;
517 sk->sk_priority = rt_tos2priority(val);
518 sk_dst_reset(sk);
520 break;
521 case IP_TTL:
522 if (optlen<1)
523 goto e_inval;
524 if (val != -1 && (val < 1 || val>255))
525 goto e_inval;
526 inet->uc_ttl = val;
527 break;
528 case IP_HDRINCL:
529 if (sk->sk_type != SOCK_RAW) {
530 err = -ENOPROTOOPT;
531 break;
533 inet->hdrincl = val ? 1 : 0;
534 break;
535 case IP_MTU_DISCOVER:
536 if (val<0 || val>2)
537 goto e_inval;
538 inet->pmtudisc = val;
539 break;
540 case IP_RECVERR:
541 inet->recverr = !!val;
542 if (!val)
543 skb_queue_purge(&sk->sk_error_queue);
544 break;
545 case IP_MULTICAST_TTL:
546 if (sk->sk_type == SOCK_STREAM)
547 goto e_inval;
548 if (optlen<1)
549 goto e_inval;
550 if (val==-1)
551 val = 1;
552 if (val < 0 || val > 255)
553 goto e_inval;
554 inet->mc_ttl = val;
555 break;
556 case IP_MULTICAST_LOOP:
557 if (optlen<1)
558 goto e_inval;
559 inet->mc_loop = !!val;
560 break;
561 case IP_MULTICAST_IF:
563 struct ip_mreqn mreq;
564 struct net_device *dev = NULL;
566 if (sk->sk_type == SOCK_STREAM)
567 goto e_inval;
569 * Check the arguments are allowable
572 err = -EFAULT;
573 if (optlen >= sizeof(struct ip_mreqn)) {
574 if (copy_from_user(&mreq,optval,sizeof(mreq)))
575 break;
576 } else {
577 memset(&mreq, 0, sizeof(mreq));
578 if (optlen >= sizeof(struct in_addr) &&
579 copy_from_user(&mreq.imr_address,optval,sizeof(struct in_addr)))
580 break;
583 if (!mreq.imr_ifindex) {
584 if (mreq.imr_address.s_addr == INADDR_ANY) {
585 inet->mc_index = 0;
586 inet->mc_addr = 0;
587 err = 0;
588 break;
590 dev = ip_dev_find(mreq.imr_address.s_addr);
591 if (dev) {
592 mreq.imr_ifindex = dev->ifindex;
593 dev_put(dev);
595 } else
596 dev = __dev_get_by_index(mreq.imr_ifindex);
599 err = -EADDRNOTAVAIL;
600 if (!dev)
601 break;
603 err = -EINVAL;
604 if (sk->sk_bound_dev_if &&
605 mreq.imr_ifindex != sk->sk_bound_dev_if)
606 break;
608 inet->mc_index = mreq.imr_ifindex;
609 inet->mc_addr = mreq.imr_address.s_addr;
610 err = 0;
611 break;
614 case IP_ADD_MEMBERSHIP:
615 case IP_DROP_MEMBERSHIP:
617 struct ip_mreqn mreq;
619 if (optlen < sizeof(struct ip_mreq))
620 goto e_inval;
621 err = -EFAULT;
622 if (optlen >= sizeof(struct ip_mreqn)) {
623 if(copy_from_user(&mreq,optval,sizeof(mreq)))
624 break;
625 } else {
626 memset(&mreq, 0, sizeof(mreq));
627 if (copy_from_user(&mreq,optval,sizeof(struct ip_mreq)))
628 break;
631 if (optname == IP_ADD_MEMBERSHIP)
632 err = ip_mc_join_group(sk, &mreq);
633 else
634 err = ip_mc_leave_group(sk, &mreq);
635 break;
637 case IP_MSFILTER:
639 extern int sysctl_igmp_max_msf;
640 struct ip_msfilter *msf;
642 if (optlen < IP_MSFILTER_SIZE(0))
643 goto e_inval;
644 if (optlen > sysctl_optmem_max) {
645 err = -ENOBUFS;
646 break;
648 msf = kmalloc(optlen, GFP_KERNEL);
649 if (msf == 0) {
650 err = -ENOBUFS;
651 break;
653 err = -EFAULT;
654 if (copy_from_user(msf, optval, optlen)) {
655 kfree(msf);
656 break;
658 /* numsrc >= (1G-4) overflow in 32 bits */
659 if (msf->imsf_numsrc >= 0x3ffffffcU ||
660 msf->imsf_numsrc > sysctl_igmp_max_msf) {
661 kfree(msf);
662 err = -ENOBUFS;
663 break;
665 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) {
666 kfree(msf);
667 err = -EINVAL;
668 break;
670 err = ip_mc_msfilter(sk, msf, 0);
671 kfree(msf);
672 break;
674 case IP_BLOCK_SOURCE:
675 case IP_UNBLOCK_SOURCE:
676 case IP_ADD_SOURCE_MEMBERSHIP:
677 case IP_DROP_SOURCE_MEMBERSHIP:
679 struct ip_mreq_source mreqs;
680 int omode, add;
682 if (optlen != sizeof(struct ip_mreq_source))
683 goto e_inval;
684 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) {
685 err = -EFAULT;
686 break;
688 if (optname == IP_BLOCK_SOURCE) {
689 omode = MCAST_EXCLUDE;
690 add = 1;
691 } else if (optname == IP_UNBLOCK_SOURCE) {
692 omode = MCAST_EXCLUDE;
693 add = 0;
694 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) {
695 struct ip_mreqn mreq;
697 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr;
698 mreq.imr_address.s_addr = mreqs.imr_interface;
699 mreq.imr_ifindex = 0;
700 err = ip_mc_join_group(sk, &mreq);
701 if (err && err != -EADDRINUSE)
702 break;
703 omode = MCAST_INCLUDE;
704 add = 1;
705 } else /* IP_DROP_SOURCE_MEMBERSHIP */ {
706 omode = MCAST_INCLUDE;
707 add = 0;
709 err = ip_mc_source(add, omode, sk, &mreqs, 0);
710 break;
712 case MCAST_JOIN_GROUP:
713 case MCAST_LEAVE_GROUP:
715 struct group_req greq;
716 struct sockaddr_in *psin;
717 struct ip_mreqn mreq;
719 if (optlen < sizeof(struct group_req))
720 goto e_inval;
721 err = -EFAULT;
722 if(copy_from_user(&greq, optval, sizeof(greq)))
723 break;
724 psin = (struct sockaddr_in *)&greq.gr_group;
725 if (psin->sin_family != AF_INET)
726 goto e_inval;
727 memset(&mreq, 0, sizeof(mreq));
728 mreq.imr_multiaddr = psin->sin_addr;
729 mreq.imr_ifindex = greq.gr_interface;
731 if (optname == MCAST_JOIN_GROUP)
732 err = ip_mc_join_group(sk, &mreq);
733 else
734 err = ip_mc_leave_group(sk, &mreq);
735 break;
737 case MCAST_JOIN_SOURCE_GROUP:
738 case MCAST_LEAVE_SOURCE_GROUP:
739 case MCAST_BLOCK_SOURCE:
740 case MCAST_UNBLOCK_SOURCE:
742 struct group_source_req greqs;
743 struct ip_mreq_source mreqs;
744 struct sockaddr_in *psin;
745 int omode, add;
747 if (optlen != sizeof(struct group_source_req))
748 goto e_inval;
749 if (copy_from_user(&greqs, optval, sizeof(greqs))) {
750 err = -EFAULT;
751 break;
753 if (greqs.gsr_group.ss_family != AF_INET ||
754 greqs.gsr_source.ss_family != AF_INET) {
755 err = -EADDRNOTAVAIL;
756 break;
758 psin = (struct sockaddr_in *)&greqs.gsr_group;
759 mreqs.imr_multiaddr = psin->sin_addr.s_addr;
760 psin = (struct sockaddr_in *)&greqs.gsr_source;
761 mreqs.imr_sourceaddr = psin->sin_addr.s_addr;
762 mreqs.imr_interface = 0; /* use index for mc_source */
764 if (optname == MCAST_BLOCK_SOURCE) {
765 omode = MCAST_EXCLUDE;
766 add = 1;
767 } else if (optname == MCAST_UNBLOCK_SOURCE) {
768 omode = MCAST_EXCLUDE;
769 add = 0;
770 } else if (optname == MCAST_JOIN_SOURCE_GROUP) {
771 struct ip_mreqn mreq;
773 psin = (struct sockaddr_in *)&greqs.gsr_group;
774 mreq.imr_multiaddr = psin->sin_addr;
775 mreq.imr_address.s_addr = 0;
776 mreq.imr_ifindex = greqs.gsr_interface;
777 err = ip_mc_join_group(sk, &mreq);
778 if (err && err != -EADDRINUSE)
779 break;
780 greqs.gsr_interface = mreq.imr_ifindex;
781 omode = MCAST_INCLUDE;
782 add = 1;
783 } else /* MCAST_LEAVE_SOURCE_GROUP */ {
784 omode = MCAST_INCLUDE;
785 add = 0;
787 err = ip_mc_source(add, omode, sk, &mreqs,
788 greqs.gsr_interface);
789 break;
791 case MCAST_MSFILTER:
793 extern int sysctl_igmp_max_msf;
794 struct sockaddr_in *psin;
795 struct ip_msfilter *msf = NULL;
796 struct group_filter *gsf = NULL;
797 int msize, i, ifindex;
799 if (optlen < GROUP_FILTER_SIZE(0))
800 goto e_inval;
801 if (optlen > sysctl_optmem_max) {
802 err = -ENOBUFS;
803 break;
805 gsf = kmalloc(optlen,GFP_KERNEL);
806 if (gsf == 0) {
807 err = -ENOBUFS;
808 break;
810 err = -EFAULT;
811 if (copy_from_user(gsf, optval, optlen)) {
812 goto mc_msf_out;
814 /* numsrc >= (4G-140)/128 overflow in 32 bits */
815 if (gsf->gf_numsrc >= 0x1ffffff ||
816 gsf->gf_numsrc > sysctl_igmp_max_msf) {
817 err = -ENOBUFS;
818 goto mc_msf_out;
820 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) {
821 err = -EINVAL;
822 goto mc_msf_out;
824 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc);
825 msf = kmalloc(msize,GFP_KERNEL);
826 if (msf == 0) {
827 err = -ENOBUFS;
828 goto mc_msf_out;
830 ifindex = gsf->gf_interface;
831 psin = (struct sockaddr_in *)&gsf->gf_group;
832 if (psin->sin_family != AF_INET) {
833 err = -EADDRNOTAVAIL;
834 goto mc_msf_out;
836 msf->imsf_multiaddr = psin->sin_addr.s_addr;
837 msf->imsf_interface = 0;
838 msf->imsf_fmode = gsf->gf_fmode;
839 msf->imsf_numsrc = gsf->gf_numsrc;
840 err = -EADDRNOTAVAIL;
841 for (i=0; i<gsf->gf_numsrc; ++i) {
842 psin = (struct sockaddr_in *)&gsf->gf_slist[i];
844 if (psin->sin_family != AF_INET)
845 goto mc_msf_out;
846 msf->imsf_slist[i] = psin->sin_addr.s_addr;
848 kfree(gsf);
849 gsf = NULL;
851 err = ip_mc_msfilter(sk, msf, ifindex);
852 mc_msf_out:
853 kfree(msf);
854 kfree(gsf);
855 break;
857 case IP_ROUTER_ALERT:
858 err = ip_ra_control(sk, val ? 1 : 0, NULL);
859 break;
861 case IP_FREEBIND:
862 if (optlen<1)
863 goto e_inval;
864 inet->freebind = !!val;
865 break;
867 case IP_IPSEC_POLICY:
868 case IP_XFRM_POLICY:
869 err = -EPERM;
870 if (!capable(CAP_NET_ADMIN))
871 break;
872 err = xfrm_user_policy(sk, optname, optval, optlen);
873 break;
875 default:
876 err = -ENOPROTOOPT;
877 break;
879 release_sock(sk);
880 return err;
882 e_inval:
883 release_sock(sk);
884 return -EINVAL;
887 int ip_setsockopt(struct sock *sk, int level,
888 int optname, char __user *optval, int optlen)
890 int err;
892 if (level != SOL_IP)
893 return -ENOPROTOOPT;
895 err = do_ip_setsockopt(sk, level, optname, optval, optlen);
896 #ifdef CONFIG_NETFILTER
897 /* we need to exclude all possible ENOPROTOOPTs except default case */
898 if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
899 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY
900 #ifdef CONFIG_IP_MROUTE
901 && (optname < MRT_BASE || optname > (MRT_BASE + 10))
902 #endif
904 lock_sock(sk);
905 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen);
906 release_sock(sk);
908 #endif
909 return err;
912 #ifdef CONFIG_COMPAT
913 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
914 char __user *optval, int optlen)
916 int err;
918 if (level != SOL_IP)
919 return -ENOPROTOOPT;
921 err = do_ip_setsockopt(sk, level, optname, optval, optlen);
922 #ifdef CONFIG_NETFILTER
923 /* we need to exclude all possible ENOPROTOOPTs except default case */
924 if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
925 optname != IP_IPSEC_POLICY && optname != IP_XFRM_POLICY
926 #ifdef CONFIG_IP_MROUTE
927 && (optname < MRT_BASE || optname > (MRT_BASE + 10))
928 #endif
930 lock_sock(sk);
931 err = compat_nf_setsockopt(sk, PF_INET, optname,
932 optval, optlen);
933 release_sock(sk);
935 #endif
936 return err;
939 EXPORT_SYMBOL(compat_ip_setsockopt);
940 #endif
943 * Get the options. Note for future reference. The GET of IP options gets the
944 * _received_ ones. The set sets the _sent_ ones.
947 static int do_ip_getsockopt(struct sock *sk, int level, int optname,
948 char __user *optval, int __user *optlen)
950 struct inet_sock *inet = inet_sk(sk);
951 int val;
952 int len;
954 if(level!=SOL_IP)
955 return -EOPNOTSUPP;
957 #ifdef CONFIG_IP_MROUTE
958 if(optname>=MRT_BASE && optname <=MRT_BASE+10)
960 return ip_mroute_getsockopt(sk,optname,optval,optlen);
962 #endif
964 if(get_user(len,optlen))
965 return -EFAULT;
966 if(len < 0)
967 return -EINVAL;
969 lock_sock(sk);
971 switch(optname) {
972 case IP_OPTIONS:
974 unsigned char optbuf[sizeof(struct ip_options)+40];
975 struct ip_options * opt = (struct ip_options*)optbuf;
976 opt->optlen = 0;
977 if (inet->opt)
978 memcpy(optbuf, inet->opt,
979 sizeof(struct ip_options)+
980 inet->opt->optlen);
981 release_sock(sk);
983 if (opt->optlen == 0)
984 return put_user(0, optlen);
986 ip_options_undo(opt);
988 len = min_t(unsigned int, len, opt->optlen);
989 if(put_user(len, optlen))
990 return -EFAULT;
991 if(copy_to_user(optval, opt->__data, len))
992 return -EFAULT;
993 return 0;
995 case IP_PKTINFO:
996 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0;
997 break;
998 case IP_RECVTTL:
999 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0;
1000 break;
1001 case IP_RECVTOS:
1002 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0;
1003 break;
1004 case IP_RECVOPTS:
1005 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0;
1006 break;
1007 case IP_RETOPTS:
1008 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0;
1009 break;
1010 case IP_PASSSEC:
1011 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0;
1012 break;
1013 case IP_TOS:
1014 val = inet->tos;
1015 break;
1016 case IP_TTL:
1017 val = (inet->uc_ttl == -1 ?
1018 sysctl_ip_default_ttl :
1019 inet->uc_ttl);
1020 break;
1021 case IP_HDRINCL:
1022 val = inet->hdrincl;
1023 break;
1024 case IP_MTU_DISCOVER:
1025 val = inet->pmtudisc;
1026 break;
1027 case IP_MTU:
1029 struct dst_entry *dst;
1030 val = 0;
1031 dst = sk_dst_get(sk);
1032 if (dst) {
1033 val = dst_mtu(dst);
1034 dst_release(dst);
1036 if (!val) {
1037 release_sock(sk);
1038 return -ENOTCONN;
1040 break;
1042 case IP_RECVERR:
1043 val = inet->recverr;
1044 break;
1045 case IP_MULTICAST_TTL:
1046 val = inet->mc_ttl;
1047 break;
1048 case IP_MULTICAST_LOOP:
1049 val = inet->mc_loop;
1050 break;
1051 case IP_MULTICAST_IF:
1053 struct in_addr addr;
1054 len = min_t(unsigned int, len, sizeof(struct in_addr));
1055 addr.s_addr = inet->mc_addr;
1056 release_sock(sk);
1058 if(put_user(len, optlen))
1059 return -EFAULT;
1060 if(copy_to_user(optval, &addr, len))
1061 return -EFAULT;
1062 return 0;
1064 case IP_MSFILTER:
1066 struct ip_msfilter msf;
1067 int err;
1069 if (len < IP_MSFILTER_SIZE(0)) {
1070 release_sock(sk);
1071 return -EINVAL;
1073 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) {
1074 release_sock(sk);
1075 return -EFAULT;
1077 err = ip_mc_msfget(sk, &msf,
1078 (struct ip_msfilter __user *)optval, optlen);
1079 release_sock(sk);
1080 return err;
1082 case MCAST_MSFILTER:
1084 struct group_filter gsf;
1085 int err;
1087 if (len < GROUP_FILTER_SIZE(0)) {
1088 release_sock(sk);
1089 return -EINVAL;
1091 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) {
1092 release_sock(sk);
1093 return -EFAULT;
1095 err = ip_mc_gsfget(sk, &gsf,
1096 (struct group_filter __user *)optval, optlen);
1097 release_sock(sk);
1098 return err;
1100 case IP_PKTOPTIONS:
1102 struct msghdr msg;
1104 release_sock(sk);
1106 if (sk->sk_type != SOCK_STREAM)
1107 return -ENOPROTOOPT;
1109 msg.msg_control = optval;
1110 msg.msg_controllen = len;
1111 msg.msg_flags = 0;
1113 if (inet->cmsg_flags & IP_CMSG_PKTINFO) {
1114 struct in_pktinfo info;
1116 info.ipi_addr.s_addr = inet->rcv_saddr;
1117 info.ipi_spec_dst.s_addr = inet->rcv_saddr;
1118 info.ipi_ifindex = inet->mc_index;
1119 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
1121 if (inet->cmsg_flags & IP_CMSG_TTL) {
1122 int hlim = inet->mc_ttl;
1123 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim);
1125 len -= msg.msg_controllen;
1126 return put_user(len, optlen);
1128 case IP_FREEBIND:
1129 val = inet->freebind;
1130 break;
1131 default:
1132 release_sock(sk);
1133 return -ENOPROTOOPT;
1135 release_sock(sk);
1137 if (len < sizeof(int) && len > 0 && val>=0 && val<255) {
1138 unsigned char ucval = (unsigned char)val;
1139 len = 1;
1140 if(put_user(len, optlen))
1141 return -EFAULT;
1142 if(copy_to_user(optval,&ucval,1))
1143 return -EFAULT;
1144 } else {
1145 len = min_t(unsigned int, sizeof(int), len);
1146 if(put_user(len, optlen))
1147 return -EFAULT;
1148 if(copy_to_user(optval,&val,len))
1149 return -EFAULT;
1151 return 0;
1154 int ip_getsockopt(struct sock *sk, int level,
1155 int optname, char __user *optval, int __user *optlen)
1157 int err;
1159 err = do_ip_getsockopt(sk, level, optname, optval, optlen);
1160 #ifdef CONFIG_NETFILTER
1161 /* we need to exclude all possible ENOPROTOOPTs except default case */
1162 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS
1163 #ifdef CONFIG_IP_MROUTE
1164 && (optname < MRT_BASE || optname > MRT_BASE+10)
1165 #endif
1167 int len;
1169 if(get_user(len,optlen))
1170 return -EFAULT;
1172 lock_sock(sk);
1173 err = nf_getsockopt(sk, PF_INET, optname, optval,
1174 &len);
1175 release_sock(sk);
1176 if (err >= 0)
1177 err = put_user(len, optlen);
1178 return err;
1180 #endif
1181 return err;
1184 #ifdef CONFIG_COMPAT
1185 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
1186 char __user *optval, int __user *optlen)
1188 int err = do_ip_getsockopt(sk, level, optname, optval, optlen);
1189 #ifdef CONFIG_NETFILTER
1190 /* we need to exclude all possible ENOPROTOOPTs except default case */
1191 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS
1192 #ifdef CONFIG_IP_MROUTE
1193 && (optname < MRT_BASE || optname > MRT_BASE+10)
1194 #endif
1196 int len;
1198 if (get_user(len, optlen))
1199 return -EFAULT;
1201 lock_sock(sk);
1202 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len);
1203 release_sock(sk);
1204 if (err >= 0)
1205 err = put_user(len, optlen);
1206 return err;
1208 #endif
1209 return err;
1212 EXPORT_SYMBOL(compat_ip_getsockopt);
1213 #endif
1215 EXPORT_SYMBOL(ip_cmsg_recv);
1217 EXPORT_SYMBOL(ip_getsockopt);
1218 EXPORT_SYMBOL(ip_setsockopt);