mfd: wm8350-i2c: Make sure the i2c regmap functions are compiled
[linux/fpc-iii.git] / net / ipv4 / ip_sockglue.c
blob9e4f832aaf1369d5a436c8530b77de4e2774d410
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.
8 * Authors: see ip.c
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
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 <linux/slab.h>
27 #include <net/sock.h>
28 #include <net/ip.h>
29 #include <net/icmp.h>
30 #include <net/tcp_states.h>
31 #include <linux/udp.h>
32 #include <linux/igmp.h>
33 #include <linux/netfilter.h>
34 #include <linux/route.h>
35 #include <linux/mroute.h>
36 #include <net/inet_ecn.h>
37 #include <net/route.h>
38 #include <net/xfrm.h>
39 #include <net/compat.h>
40 #if IS_ENABLED(CONFIG_IPV6)
41 #include <net/transp_v6.h>
42 #endif
43 #include <net/ip_fib.h>
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
54 #define IP_CMSG_ORIGDSTADDR 64
57 * SOL_IP control messages.
59 #define PKTINFO_SKB_CB(__skb) ((struct in_pktinfo *)((__skb)->cb))
61 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
63 struct in_pktinfo info = *PKTINFO_SKB_CB(skb);
65 info.ipi_addr.s_addr = ip_hdr(skb)->daddr;
67 put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
70 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb)
72 int ttl = ip_hdr(skb)->ttl;
73 put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl);
76 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb)
78 put_cmsg(msg, SOL_IP, IP_TOS, 1, &ip_hdr(skb)->tos);
81 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb)
83 if (IPCB(skb)->opt.optlen == 0)
84 return;
86 put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen,
87 ip_hdr(skb) + 1);
91 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb)
93 unsigned char optbuf[sizeof(struct ip_options) + 40];
94 struct ip_options *opt = (struct ip_options *)optbuf;
96 if (IPCB(skb)->opt.optlen == 0)
97 return;
99 if (ip_options_echo(opt, skb)) {
100 msg->msg_flags |= MSG_CTRUNC;
101 return;
103 ip_options_undo(opt);
105 put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data);
108 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb)
110 char *secdata;
111 u32 seclen, secid;
112 int err;
114 err = security_socket_getpeersec_dgram(NULL, skb, &secid);
115 if (err)
116 return;
118 err = security_secid_to_secctx(secid, &secdata, &seclen);
119 if (err)
120 return;
122 put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata);
123 security_release_secctx(secdata, seclen);
126 static void ip_cmsg_recv_dstaddr(struct msghdr *msg, struct sk_buff *skb)
128 struct sockaddr_in sin;
129 const struct iphdr *iph = ip_hdr(skb);
130 __be16 *ports = (__be16 *)skb_transport_header(skb);
132 if (skb_transport_offset(skb) + 4 > skb->len)
133 return;
135 /* All current transport protocols have the port numbers in the
136 * first four bytes of the transport header and this function is
137 * written with this assumption in mind.
140 sin.sin_family = AF_INET;
141 sin.sin_addr.s_addr = iph->daddr;
142 sin.sin_port = ports[1];
143 memset(sin.sin_zero, 0, sizeof(sin.sin_zero));
145 put_cmsg(msg, SOL_IP, IP_ORIGDSTADDR, sizeof(sin), &sin);
148 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
150 struct inet_sock *inet = inet_sk(skb->sk);
151 unsigned int flags = inet->cmsg_flags;
153 /* Ordered by supposed usage frequency */
154 if (flags & 1)
155 ip_cmsg_recv_pktinfo(msg, skb);
156 if ((flags >>= 1) == 0)
157 return;
159 if (flags & 1)
160 ip_cmsg_recv_ttl(msg, skb);
161 if ((flags >>= 1) == 0)
162 return;
164 if (flags & 1)
165 ip_cmsg_recv_tos(msg, skb);
166 if ((flags >>= 1) == 0)
167 return;
169 if (flags & 1)
170 ip_cmsg_recv_opts(msg, skb);
171 if ((flags >>= 1) == 0)
172 return;
174 if (flags & 1)
175 ip_cmsg_recv_retopts(msg, skb);
176 if ((flags >>= 1) == 0)
177 return;
179 if (flags & 1)
180 ip_cmsg_recv_security(msg, skb);
182 if ((flags >>= 1) == 0)
183 return;
184 if (flags & 1)
185 ip_cmsg_recv_dstaddr(msg, skb);
188 EXPORT_SYMBOL(ip_cmsg_recv);
190 int ip_cmsg_send(struct net *net, struct msghdr *msg, struct ipcm_cookie *ipc)
192 int err;
193 struct cmsghdr *cmsg;
195 for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
196 if (!CMSG_OK(msg, cmsg))
197 return -EINVAL;
198 if (cmsg->cmsg_level != SOL_IP)
199 continue;
200 switch (cmsg->cmsg_type) {
201 case IP_RETOPTS:
202 err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr));
204 /* Our caller is responsible for freeing ipc->opt */
205 err = ip_options_get(net, &ipc->opt, CMSG_DATA(cmsg),
206 err < 40 ? err : 40);
207 if (err)
208 return err;
209 break;
210 case IP_PKTINFO:
212 struct in_pktinfo *info;
213 if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo)))
214 return -EINVAL;
215 info = (struct in_pktinfo *)CMSG_DATA(cmsg);
216 ipc->oif = info->ipi_ifindex;
217 ipc->addr = info->ipi_spec_dst.s_addr;
218 break;
220 default:
221 return -EINVAL;
224 return 0;
228 /* Special input handler for packets caught by router alert option.
229 They are selected only by protocol field, and then processed likely
230 local ones; but only if someone wants them! Otherwise, router
231 not running rsvpd will kill RSVP.
233 It is user level problem, what it will make with them.
234 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
235 but receiver should be enough clever f.e. to forward mtrace requests,
236 sent to multicast group to reach destination designated router.
238 struct ip_ra_chain __rcu *ip_ra_chain;
239 static DEFINE_SPINLOCK(ip_ra_lock);
242 static void ip_ra_destroy_rcu(struct rcu_head *head)
244 struct ip_ra_chain *ra = container_of(head, struct ip_ra_chain, rcu);
246 sock_put(ra->saved_sk);
247 kfree(ra);
250 int ip_ra_control(struct sock *sk, unsigned char on,
251 void (*destructor)(struct sock *))
253 struct ip_ra_chain *ra, *new_ra;
254 struct ip_ra_chain __rcu **rap;
256 if (sk->sk_type != SOCK_RAW || inet_sk(sk)->inet_num == IPPROTO_RAW)
257 return -EINVAL;
259 new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL;
261 spin_lock_bh(&ip_ra_lock);
262 for (rap = &ip_ra_chain;
263 (ra = rcu_dereference_protected(*rap,
264 lockdep_is_held(&ip_ra_lock))) != NULL;
265 rap = &ra->next) {
266 if (ra->sk == sk) {
267 if (on) {
268 spin_unlock_bh(&ip_ra_lock);
269 kfree(new_ra);
270 return -EADDRINUSE;
272 /* dont let ip_call_ra_chain() use sk again */
273 ra->sk = NULL;
274 rcu_assign_pointer(*rap, ra->next);
275 spin_unlock_bh(&ip_ra_lock);
277 if (ra->destructor)
278 ra->destructor(sk);
280 * Delay sock_put(sk) and kfree(ra) after one rcu grace
281 * period. This guarantee ip_call_ra_chain() dont need
282 * to mess with socket refcounts.
284 ra->saved_sk = sk;
285 call_rcu(&ra->rcu, ip_ra_destroy_rcu);
286 return 0;
289 if (new_ra == NULL) {
290 spin_unlock_bh(&ip_ra_lock);
291 return -ENOBUFS;
293 new_ra->sk = sk;
294 new_ra->destructor = destructor;
296 new_ra->next = ra;
297 rcu_assign_pointer(*rap, new_ra);
298 sock_hold(sk);
299 spin_unlock_bh(&ip_ra_lock);
301 return 0;
304 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
305 __be16 port, u32 info, u8 *payload)
307 struct sock_exterr_skb *serr;
309 skb = skb_clone(skb, GFP_ATOMIC);
310 if (!skb)
311 return;
313 serr = SKB_EXT_ERR(skb);
314 serr->ee.ee_errno = err;
315 serr->ee.ee_origin = SO_EE_ORIGIN_ICMP;
316 serr->ee.ee_type = icmp_hdr(skb)->type;
317 serr->ee.ee_code = icmp_hdr(skb)->code;
318 serr->ee.ee_pad = 0;
319 serr->ee.ee_info = info;
320 serr->ee.ee_data = 0;
321 serr->addr_offset = (u8 *)&(((struct iphdr *)(icmp_hdr(skb) + 1))->daddr) -
322 skb_network_header(skb);
323 serr->port = port;
325 if (skb_pull(skb, payload - skb->data) != NULL) {
326 skb_reset_transport_header(skb);
327 if (sock_queue_err_skb(sk, skb) == 0)
328 return;
330 kfree_skb(skb);
333 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info)
335 struct inet_sock *inet = inet_sk(sk);
336 struct sock_exterr_skb *serr;
337 struct iphdr *iph;
338 struct sk_buff *skb;
340 if (!inet->recverr)
341 return;
343 skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC);
344 if (!skb)
345 return;
347 skb_put(skb, sizeof(struct iphdr));
348 skb_reset_network_header(skb);
349 iph = ip_hdr(skb);
350 iph->daddr = daddr;
352 serr = SKB_EXT_ERR(skb);
353 serr->ee.ee_errno = err;
354 serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
355 serr->ee.ee_type = 0;
356 serr->ee.ee_code = 0;
357 serr->ee.ee_pad = 0;
358 serr->ee.ee_info = info;
359 serr->ee.ee_data = 0;
360 serr->addr_offset = (u8 *)&iph->daddr - skb_network_header(skb);
361 serr->port = port;
363 __skb_pull(skb, skb_tail_pointer(skb) - skb->data);
364 skb_reset_transport_header(skb);
366 if (sock_queue_err_skb(sk, skb))
367 kfree_skb(skb);
371 * Handle MSG_ERRQUEUE
373 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len)
375 struct sock_exterr_skb *serr;
376 struct sk_buff *skb, *skb2;
377 struct sockaddr_in *sin;
378 struct {
379 struct sock_extended_err ee;
380 struct sockaddr_in offender;
381 } errhdr;
382 int err;
383 int copied;
385 err = -EAGAIN;
386 skb = skb_dequeue(&sk->sk_error_queue);
387 if (skb == NULL)
388 goto out;
390 copied = skb->len;
391 if (copied > len) {
392 msg->msg_flags |= MSG_TRUNC;
393 copied = len;
395 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
396 if (err)
397 goto out_free_skb;
399 sock_recv_timestamp(msg, sk, skb);
401 serr = SKB_EXT_ERR(skb);
403 sin = (struct sockaddr_in *)msg->msg_name;
404 if (sin) {
405 sin->sin_family = AF_INET;
406 sin->sin_addr.s_addr = *(__be32 *)(skb_network_header(skb) +
407 serr->addr_offset);
408 sin->sin_port = serr->port;
409 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
410 *addr_len = sizeof(*sin);
413 memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err));
414 sin = &errhdr.offender;
415 memset(sin, 0, sizeof(*sin));
416 if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) {
417 sin->sin_family = AF_INET;
418 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
419 if (inet_sk(sk)->cmsg_flags)
420 ip_cmsg_recv(msg, skb);
423 put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr);
425 /* Now we could try to dump offended packet options */
427 msg->msg_flags |= MSG_ERRQUEUE;
428 err = copied;
430 /* Reset and regenerate socket error */
431 spin_lock_bh(&sk->sk_error_queue.lock);
432 sk->sk_err = 0;
433 skb2 = skb_peek(&sk->sk_error_queue);
434 if (skb2 != NULL) {
435 sk->sk_err = SKB_EXT_ERR(skb2)->ee.ee_errno;
436 spin_unlock_bh(&sk->sk_error_queue.lock);
437 sk->sk_error_report(sk);
438 } else
439 spin_unlock_bh(&sk->sk_error_queue.lock);
441 out_free_skb:
442 kfree_skb(skb);
443 out:
444 return err;
449 * Socket option code for IP. This is the end of the line after any
450 * TCP,UDP etc options on an IP socket.
453 static int do_ip_setsockopt(struct sock *sk, int level,
454 int optname, char __user *optval, unsigned int optlen)
456 struct inet_sock *inet = inet_sk(sk);
457 int val = 0, err;
459 switch (optname) {
460 case IP_PKTINFO:
461 case IP_RECVTTL:
462 case IP_RECVOPTS:
463 case IP_RECVTOS:
464 case IP_RETOPTS:
465 case IP_TOS:
466 case IP_TTL:
467 case IP_HDRINCL:
468 case IP_MTU_DISCOVER:
469 case IP_RECVERR:
470 case IP_ROUTER_ALERT:
471 case IP_FREEBIND:
472 case IP_PASSSEC:
473 case IP_TRANSPARENT:
474 case IP_MINTTL:
475 case IP_NODEFRAG:
476 case IP_UNICAST_IF:
477 case IP_MULTICAST_TTL:
478 case IP_MULTICAST_ALL:
479 case IP_MULTICAST_LOOP:
480 case IP_RECVORIGDSTADDR:
481 if (optlen >= sizeof(int)) {
482 if (get_user(val, (int __user *) optval))
483 return -EFAULT;
484 } else if (optlen >= sizeof(char)) {
485 unsigned char ucval;
487 if (get_user(ucval, (unsigned char __user *) optval))
488 return -EFAULT;
489 val = (int) ucval;
493 /* If optlen==0, it is equivalent to val == 0 */
495 if (ip_mroute_opt(optname))
496 return ip_mroute_setsockopt(sk, optname, optval, optlen);
498 err = 0;
499 lock_sock(sk);
501 switch (optname) {
502 case IP_OPTIONS:
504 struct ip_options_rcu *old, *opt = NULL;
506 if (optlen > 40)
507 goto e_inval;
508 err = ip_options_get_from_user(sock_net(sk), &opt,
509 optval, optlen);
510 if (err)
511 break;
512 old = rcu_dereference_protected(inet->inet_opt,
513 sock_owned_by_user(sk));
514 if (inet->is_icsk) {
515 struct inet_connection_sock *icsk = inet_csk(sk);
516 #if IS_ENABLED(CONFIG_IPV6)
517 if (sk->sk_family == PF_INET ||
518 (!((1 << sk->sk_state) &
519 (TCPF_LISTEN | TCPF_CLOSE)) &&
520 inet->inet_daddr != LOOPBACK4_IPV6)) {
521 #endif
522 if (old)
523 icsk->icsk_ext_hdr_len -= old->opt.optlen;
524 if (opt)
525 icsk->icsk_ext_hdr_len += opt->opt.optlen;
526 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie);
527 #if IS_ENABLED(CONFIG_IPV6)
529 #endif
531 rcu_assign_pointer(inet->inet_opt, opt);
532 if (old)
533 kfree_rcu(old, rcu);
534 break;
536 case IP_PKTINFO:
537 if (val)
538 inet->cmsg_flags |= IP_CMSG_PKTINFO;
539 else
540 inet->cmsg_flags &= ~IP_CMSG_PKTINFO;
541 break;
542 case IP_RECVTTL:
543 if (val)
544 inet->cmsg_flags |= IP_CMSG_TTL;
545 else
546 inet->cmsg_flags &= ~IP_CMSG_TTL;
547 break;
548 case IP_RECVTOS:
549 if (val)
550 inet->cmsg_flags |= IP_CMSG_TOS;
551 else
552 inet->cmsg_flags &= ~IP_CMSG_TOS;
553 break;
554 case IP_RECVOPTS:
555 if (val)
556 inet->cmsg_flags |= IP_CMSG_RECVOPTS;
557 else
558 inet->cmsg_flags &= ~IP_CMSG_RECVOPTS;
559 break;
560 case IP_RETOPTS:
561 if (val)
562 inet->cmsg_flags |= IP_CMSG_RETOPTS;
563 else
564 inet->cmsg_flags &= ~IP_CMSG_RETOPTS;
565 break;
566 case IP_PASSSEC:
567 if (val)
568 inet->cmsg_flags |= IP_CMSG_PASSSEC;
569 else
570 inet->cmsg_flags &= ~IP_CMSG_PASSSEC;
571 break;
572 case IP_RECVORIGDSTADDR:
573 if (val)
574 inet->cmsg_flags |= IP_CMSG_ORIGDSTADDR;
575 else
576 inet->cmsg_flags &= ~IP_CMSG_ORIGDSTADDR;
577 break;
578 case IP_TOS: /* This sets both TOS and Precedence */
579 if (sk->sk_type == SOCK_STREAM) {
580 val &= ~INET_ECN_MASK;
581 val |= inet->tos & INET_ECN_MASK;
583 if (inet->tos != val) {
584 inet->tos = val;
585 sk->sk_priority = rt_tos2priority(val);
586 sk_dst_reset(sk);
588 break;
589 case IP_TTL:
590 if (optlen < 1)
591 goto e_inval;
592 if (val != -1 && (val < 1 || val > 255))
593 goto e_inval;
594 inet->uc_ttl = val;
595 break;
596 case IP_HDRINCL:
597 if (sk->sk_type != SOCK_RAW) {
598 err = -ENOPROTOOPT;
599 break;
601 inet->hdrincl = val ? 1 : 0;
602 break;
603 case IP_NODEFRAG:
604 if (sk->sk_type != SOCK_RAW) {
605 err = -ENOPROTOOPT;
606 break;
608 inet->nodefrag = val ? 1 : 0;
609 break;
610 case IP_MTU_DISCOVER:
611 if (val < IP_PMTUDISC_DONT || val > IP_PMTUDISC_PROBE)
612 goto e_inval;
613 inet->pmtudisc = val;
614 break;
615 case IP_RECVERR:
616 inet->recverr = !!val;
617 if (!val)
618 skb_queue_purge(&sk->sk_error_queue);
619 break;
620 case IP_MULTICAST_TTL:
621 if (sk->sk_type == SOCK_STREAM)
622 goto e_inval;
623 if (optlen < 1)
624 goto e_inval;
625 if (val == -1)
626 val = 1;
627 if (val < 0 || val > 255)
628 goto e_inval;
629 inet->mc_ttl = val;
630 break;
631 case IP_MULTICAST_LOOP:
632 if (optlen < 1)
633 goto e_inval;
634 inet->mc_loop = !!val;
635 break;
636 case IP_UNICAST_IF:
638 struct net_device *dev = NULL;
639 int ifindex;
641 if (optlen != sizeof(int))
642 goto e_inval;
644 ifindex = (__force int)ntohl((__force __be32)val);
645 if (ifindex == 0) {
646 inet->uc_index = 0;
647 err = 0;
648 break;
651 dev = dev_get_by_index(sock_net(sk), ifindex);
652 err = -EADDRNOTAVAIL;
653 if (!dev)
654 break;
655 dev_put(dev);
657 err = -EINVAL;
658 if (sk->sk_bound_dev_if)
659 break;
661 inet->uc_index = ifindex;
662 err = 0;
663 break;
665 case IP_MULTICAST_IF:
667 struct ip_mreqn mreq;
668 struct net_device *dev = NULL;
670 if (sk->sk_type == SOCK_STREAM)
671 goto e_inval;
673 * Check the arguments are allowable
676 if (optlen < sizeof(struct in_addr))
677 goto e_inval;
679 err = -EFAULT;
680 if (optlen >= sizeof(struct ip_mreqn)) {
681 if (copy_from_user(&mreq, optval, sizeof(mreq)))
682 break;
683 } else {
684 memset(&mreq, 0, sizeof(mreq));
685 if (optlen >= sizeof(struct ip_mreq)) {
686 if (copy_from_user(&mreq, optval,
687 sizeof(struct ip_mreq)))
688 break;
689 } else if (optlen >= sizeof(struct in_addr)) {
690 if (copy_from_user(&mreq.imr_address, optval,
691 sizeof(struct in_addr)))
692 break;
696 if (!mreq.imr_ifindex) {
697 if (mreq.imr_address.s_addr == htonl(INADDR_ANY)) {
698 inet->mc_index = 0;
699 inet->mc_addr = 0;
700 err = 0;
701 break;
703 dev = ip_dev_find(sock_net(sk), mreq.imr_address.s_addr);
704 if (dev)
705 mreq.imr_ifindex = dev->ifindex;
706 } else
707 dev = dev_get_by_index(sock_net(sk), mreq.imr_ifindex);
710 err = -EADDRNOTAVAIL;
711 if (!dev)
712 break;
713 dev_put(dev);
715 err = -EINVAL;
716 if (sk->sk_bound_dev_if &&
717 mreq.imr_ifindex != sk->sk_bound_dev_if)
718 break;
720 inet->mc_index = mreq.imr_ifindex;
721 inet->mc_addr = mreq.imr_address.s_addr;
722 err = 0;
723 break;
726 case IP_ADD_MEMBERSHIP:
727 case IP_DROP_MEMBERSHIP:
729 struct ip_mreqn mreq;
731 err = -EPROTO;
732 if (inet_sk(sk)->is_icsk)
733 break;
735 if (optlen < sizeof(struct ip_mreq))
736 goto e_inval;
737 err = -EFAULT;
738 if (optlen >= sizeof(struct ip_mreqn)) {
739 if (copy_from_user(&mreq, optval, sizeof(mreq)))
740 break;
741 } else {
742 memset(&mreq, 0, sizeof(mreq));
743 if (copy_from_user(&mreq, optval, sizeof(struct ip_mreq)))
744 break;
747 if (optname == IP_ADD_MEMBERSHIP)
748 err = ip_mc_join_group(sk, &mreq);
749 else
750 err = ip_mc_leave_group(sk, &mreq);
751 break;
753 case IP_MSFILTER:
755 struct ip_msfilter *msf;
757 if (optlen < IP_MSFILTER_SIZE(0))
758 goto e_inval;
759 if (optlen > sysctl_optmem_max) {
760 err = -ENOBUFS;
761 break;
763 msf = kmalloc(optlen, GFP_KERNEL);
764 if (!msf) {
765 err = -ENOBUFS;
766 break;
768 err = -EFAULT;
769 if (copy_from_user(msf, optval, optlen)) {
770 kfree(msf);
771 break;
773 /* numsrc >= (1G-4) overflow in 32 bits */
774 if (msf->imsf_numsrc >= 0x3ffffffcU ||
775 msf->imsf_numsrc > sysctl_igmp_max_msf) {
776 kfree(msf);
777 err = -ENOBUFS;
778 break;
780 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) {
781 kfree(msf);
782 err = -EINVAL;
783 break;
785 err = ip_mc_msfilter(sk, msf, 0);
786 kfree(msf);
787 break;
789 case IP_BLOCK_SOURCE:
790 case IP_UNBLOCK_SOURCE:
791 case IP_ADD_SOURCE_MEMBERSHIP:
792 case IP_DROP_SOURCE_MEMBERSHIP:
794 struct ip_mreq_source mreqs;
795 int omode, add;
797 if (optlen != sizeof(struct ip_mreq_source))
798 goto e_inval;
799 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) {
800 err = -EFAULT;
801 break;
803 if (optname == IP_BLOCK_SOURCE) {
804 omode = MCAST_EXCLUDE;
805 add = 1;
806 } else if (optname == IP_UNBLOCK_SOURCE) {
807 omode = MCAST_EXCLUDE;
808 add = 0;
809 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) {
810 struct ip_mreqn mreq;
812 mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr;
813 mreq.imr_address.s_addr = mreqs.imr_interface;
814 mreq.imr_ifindex = 0;
815 err = ip_mc_join_group(sk, &mreq);
816 if (err && err != -EADDRINUSE)
817 break;
818 omode = MCAST_INCLUDE;
819 add = 1;
820 } else /* IP_DROP_SOURCE_MEMBERSHIP */ {
821 omode = MCAST_INCLUDE;
822 add = 0;
824 err = ip_mc_source(add, omode, sk, &mreqs, 0);
825 break;
827 case MCAST_JOIN_GROUP:
828 case MCAST_LEAVE_GROUP:
830 struct group_req greq;
831 struct sockaddr_in *psin;
832 struct ip_mreqn mreq;
834 if (optlen < sizeof(struct group_req))
835 goto e_inval;
836 err = -EFAULT;
837 if (copy_from_user(&greq, optval, sizeof(greq)))
838 break;
839 psin = (struct sockaddr_in *)&greq.gr_group;
840 if (psin->sin_family != AF_INET)
841 goto e_inval;
842 memset(&mreq, 0, sizeof(mreq));
843 mreq.imr_multiaddr = psin->sin_addr;
844 mreq.imr_ifindex = greq.gr_interface;
846 if (optname == MCAST_JOIN_GROUP)
847 err = ip_mc_join_group(sk, &mreq);
848 else
849 err = ip_mc_leave_group(sk, &mreq);
850 break;
852 case MCAST_JOIN_SOURCE_GROUP:
853 case MCAST_LEAVE_SOURCE_GROUP:
854 case MCAST_BLOCK_SOURCE:
855 case MCAST_UNBLOCK_SOURCE:
857 struct group_source_req greqs;
858 struct ip_mreq_source mreqs;
859 struct sockaddr_in *psin;
860 int omode, add;
862 if (optlen != sizeof(struct group_source_req))
863 goto e_inval;
864 if (copy_from_user(&greqs, optval, sizeof(greqs))) {
865 err = -EFAULT;
866 break;
868 if (greqs.gsr_group.ss_family != AF_INET ||
869 greqs.gsr_source.ss_family != AF_INET) {
870 err = -EADDRNOTAVAIL;
871 break;
873 psin = (struct sockaddr_in *)&greqs.gsr_group;
874 mreqs.imr_multiaddr = psin->sin_addr.s_addr;
875 psin = (struct sockaddr_in *)&greqs.gsr_source;
876 mreqs.imr_sourceaddr = psin->sin_addr.s_addr;
877 mreqs.imr_interface = 0; /* use index for mc_source */
879 if (optname == MCAST_BLOCK_SOURCE) {
880 omode = MCAST_EXCLUDE;
881 add = 1;
882 } else if (optname == MCAST_UNBLOCK_SOURCE) {
883 omode = MCAST_EXCLUDE;
884 add = 0;
885 } else if (optname == MCAST_JOIN_SOURCE_GROUP) {
886 struct ip_mreqn mreq;
888 psin = (struct sockaddr_in *)&greqs.gsr_group;
889 mreq.imr_multiaddr = psin->sin_addr;
890 mreq.imr_address.s_addr = 0;
891 mreq.imr_ifindex = greqs.gsr_interface;
892 err = ip_mc_join_group(sk, &mreq);
893 if (err && err != -EADDRINUSE)
894 break;
895 greqs.gsr_interface = mreq.imr_ifindex;
896 omode = MCAST_INCLUDE;
897 add = 1;
898 } else /* MCAST_LEAVE_SOURCE_GROUP */ {
899 omode = MCAST_INCLUDE;
900 add = 0;
902 err = ip_mc_source(add, omode, sk, &mreqs,
903 greqs.gsr_interface);
904 break;
906 case MCAST_MSFILTER:
908 struct sockaddr_in *psin;
909 struct ip_msfilter *msf = NULL;
910 struct group_filter *gsf = NULL;
911 int msize, i, ifindex;
913 if (optlen < GROUP_FILTER_SIZE(0))
914 goto e_inval;
915 if (optlen > sysctl_optmem_max) {
916 err = -ENOBUFS;
917 break;
919 gsf = kmalloc(optlen, GFP_KERNEL);
920 if (!gsf) {
921 err = -ENOBUFS;
922 break;
924 err = -EFAULT;
925 if (copy_from_user(gsf, optval, optlen))
926 goto mc_msf_out;
928 /* numsrc >= (4G-140)/128 overflow in 32 bits */
929 if (gsf->gf_numsrc >= 0x1ffffff ||
930 gsf->gf_numsrc > sysctl_igmp_max_msf) {
931 err = -ENOBUFS;
932 goto mc_msf_out;
934 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) {
935 err = -EINVAL;
936 goto mc_msf_out;
938 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc);
939 msf = kmalloc(msize, GFP_KERNEL);
940 if (!msf) {
941 err = -ENOBUFS;
942 goto mc_msf_out;
944 ifindex = gsf->gf_interface;
945 psin = (struct sockaddr_in *)&gsf->gf_group;
946 if (psin->sin_family != AF_INET) {
947 err = -EADDRNOTAVAIL;
948 goto mc_msf_out;
950 msf->imsf_multiaddr = psin->sin_addr.s_addr;
951 msf->imsf_interface = 0;
952 msf->imsf_fmode = gsf->gf_fmode;
953 msf->imsf_numsrc = gsf->gf_numsrc;
954 err = -EADDRNOTAVAIL;
955 for (i = 0; i < gsf->gf_numsrc; ++i) {
956 psin = (struct sockaddr_in *)&gsf->gf_slist[i];
958 if (psin->sin_family != AF_INET)
959 goto mc_msf_out;
960 msf->imsf_slist[i] = psin->sin_addr.s_addr;
962 kfree(gsf);
963 gsf = NULL;
965 err = ip_mc_msfilter(sk, msf, ifindex);
966 mc_msf_out:
967 kfree(msf);
968 kfree(gsf);
969 break;
971 case IP_MULTICAST_ALL:
972 if (optlen < 1)
973 goto e_inval;
974 if (val != 0 && val != 1)
975 goto e_inval;
976 inet->mc_all = val;
977 break;
978 case IP_ROUTER_ALERT:
979 err = ip_ra_control(sk, val ? 1 : 0, NULL);
980 break;
982 case IP_FREEBIND:
983 if (optlen < 1)
984 goto e_inval;
985 inet->freebind = !!val;
986 break;
988 case IP_IPSEC_POLICY:
989 case IP_XFRM_POLICY:
990 err = -EPERM;
991 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
992 break;
993 err = xfrm_user_policy(sk, optname, optval, optlen);
994 break;
996 case IP_TRANSPARENT:
997 if (!!val && !ns_capable(sock_net(sk)->user_ns, CAP_NET_RAW) &&
998 !ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)) {
999 err = -EPERM;
1000 break;
1002 if (optlen < 1)
1003 goto e_inval;
1004 inet->transparent = !!val;
1005 break;
1007 case IP_MINTTL:
1008 if (optlen < 1)
1009 goto e_inval;
1010 if (val < 0 || val > 255)
1011 goto e_inval;
1012 inet->min_ttl = val;
1013 break;
1015 default:
1016 err = -ENOPROTOOPT;
1017 break;
1019 release_sock(sk);
1020 return err;
1022 e_inval:
1023 release_sock(sk);
1024 return -EINVAL;
1028 * ipv4_pktinfo_prepare - transfert some info from rtable to skb
1029 * @sk: socket
1030 * @skb: buffer
1032 * To support IP_CMSG_PKTINFO option, we store rt_iif and specific
1033 * destination in skb->cb[] before dst drop.
1034 * This way, receiver doesnt make cache line misses to read rtable.
1036 void ipv4_pktinfo_prepare(struct sk_buff *skb)
1038 struct in_pktinfo *pktinfo = PKTINFO_SKB_CB(skb);
1040 if (skb_rtable(skb)) {
1041 pktinfo->ipi_ifindex = inet_iif(skb);
1042 pktinfo->ipi_spec_dst.s_addr = fib_compute_spec_dst(skb);
1043 } else {
1044 pktinfo->ipi_ifindex = 0;
1045 pktinfo->ipi_spec_dst.s_addr = 0;
1047 skb_dst_drop(skb);
1050 int ip_setsockopt(struct sock *sk, int level,
1051 int optname, char __user *optval, unsigned int optlen)
1053 int err;
1055 if (level != SOL_IP)
1056 return -ENOPROTOOPT;
1058 err = do_ip_setsockopt(sk, level, optname, optval, optlen);
1059 #ifdef CONFIG_NETFILTER
1060 /* we need to exclude all possible ENOPROTOOPTs except default case */
1061 if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
1062 optname != IP_IPSEC_POLICY &&
1063 optname != IP_XFRM_POLICY &&
1064 !ip_mroute_opt(optname)) {
1065 lock_sock(sk);
1066 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen);
1067 release_sock(sk);
1069 #endif
1070 return err;
1072 EXPORT_SYMBOL(ip_setsockopt);
1074 #ifdef CONFIG_COMPAT
1075 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
1076 char __user *optval, unsigned int optlen)
1078 int err;
1080 if (level != SOL_IP)
1081 return -ENOPROTOOPT;
1083 if (optname >= MCAST_JOIN_GROUP && optname <= MCAST_MSFILTER)
1084 return compat_mc_setsockopt(sk, level, optname, optval, optlen,
1085 ip_setsockopt);
1087 err = do_ip_setsockopt(sk, level, optname, optval, optlen);
1088 #ifdef CONFIG_NETFILTER
1089 /* we need to exclude all possible ENOPROTOOPTs except default case */
1090 if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
1091 optname != IP_IPSEC_POLICY &&
1092 optname != IP_XFRM_POLICY &&
1093 !ip_mroute_opt(optname)) {
1094 lock_sock(sk);
1095 err = compat_nf_setsockopt(sk, PF_INET, optname,
1096 optval, optlen);
1097 release_sock(sk);
1099 #endif
1100 return err;
1102 EXPORT_SYMBOL(compat_ip_setsockopt);
1103 #endif
1106 * Get the options. Note for future reference. The GET of IP options gets
1107 * the _received_ ones. The set sets the _sent_ ones.
1110 static int do_ip_getsockopt(struct sock *sk, int level, int optname,
1111 char __user *optval, int __user *optlen, unsigned int flags)
1113 struct inet_sock *inet = inet_sk(sk);
1114 int val;
1115 int len;
1117 if (level != SOL_IP)
1118 return -EOPNOTSUPP;
1120 if (ip_mroute_opt(optname))
1121 return ip_mroute_getsockopt(sk, optname, optval, optlen);
1123 if (get_user(len, optlen))
1124 return -EFAULT;
1125 if (len < 0)
1126 return -EINVAL;
1128 lock_sock(sk);
1130 switch (optname) {
1131 case IP_OPTIONS:
1133 unsigned char optbuf[sizeof(struct ip_options)+40];
1134 struct ip_options *opt = (struct ip_options *)optbuf;
1135 struct ip_options_rcu *inet_opt;
1137 inet_opt = rcu_dereference_protected(inet->inet_opt,
1138 sock_owned_by_user(sk));
1139 opt->optlen = 0;
1140 if (inet_opt)
1141 memcpy(optbuf, &inet_opt->opt,
1142 sizeof(struct ip_options) +
1143 inet_opt->opt.optlen);
1144 release_sock(sk);
1146 if (opt->optlen == 0)
1147 return put_user(0, optlen);
1149 ip_options_undo(opt);
1151 len = min_t(unsigned int, len, opt->optlen);
1152 if (put_user(len, optlen))
1153 return -EFAULT;
1154 if (copy_to_user(optval, opt->__data, len))
1155 return -EFAULT;
1156 return 0;
1158 case IP_PKTINFO:
1159 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0;
1160 break;
1161 case IP_RECVTTL:
1162 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0;
1163 break;
1164 case IP_RECVTOS:
1165 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0;
1166 break;
1167 case IP_RECVOPTS:
1168 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0;
1169 break;
1170 case IP_RETOPTS:
1171 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0;
1172 break;
1173 case IP_PASSSEC:
1174 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0;
1175 break;
1176 case IP_RECVORIGDSTADDR:
1177 val = (inet->cmsg_flags & IP_CMSG_ORIGDSTADDR) != 0;
1178 break;
1179 case IP_TOS:
1180 val = inet->tos;
1181 break;
1182 case IP_TTL:
1183 val = (inet->uc_ttl == -1 ?
1184 sysctl_ip_default_ttl :
1185 inet->uc_ttl);
1186 break;
1187 case IP_HDRINCL:
1188 val = inet->hdrincl;
1189 break;
1190 case IP_NODEFRAG:
1191 val = inet->nodefrag;
1192 break;
1193 case IP_MTU_DISCOVER:
1194 val = inet->pmtudisc;
1195 break;
1196 case IP_MTU:
1198 struct dst_entry *dst;
1199 val = 0;
1200 dst = sk_dst_get(sk);
1201 if (dst) {
1202 val = dst_mtu(dst);
1203 dst_release(dst);
1205 if (!val) {
1206 release_sock(sk);
1207 return -ENOTCONN;
1209 break;
1211 case IP_RECVERR:
1212 val = inet->recverr;
1213 break;
1214 case IP_MULTICAST_TTL:
1215 val = inet->mc_ttl;
1216 break;
1217 case IP_MULTICAST_LOOP:
1218 val = inet->mc_loop;
1219 break;
1220 case IP_UNICAST_IF:
1221 val = (__force int)htonl((__u32) inet->uc_index);
1222 break;
1223 case IP_MULTICAST_IF:
1225 struct in_addr addr;
1226 len = min_t(unsigned int, len, sizeof(struct in_addr));
1227 addr.s_addr = inet->mc_addr;
1228 release_sock(sk);
1230 if (put_user(len, optlen))
1231 return -EFAULT;
1232 if (copy_to_user(optval, &addr, len))
1233 return -EFAULT;
1234 return 0;
1236 case IP_MSFILTER:
1238 struct ip_msfilter msf;
1239 int err;
1241 if (len < IP_MSFILTER_SIZE(0)) {
1242 release_sock(sk);
1243 return -EINVAL;
1245 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) {
1246 release_sock(sk);
1247 return -EFAULT;
1249 err = ip_mc_msfget(sk, &msf,
1250 (struct ip_msfilter __user *)optval, optlen);
1251 release_sock(sk);
1252 return err;
1254 case MCAST_MSFILTER:
1256 struct group_filter gsf;
1257 int err;
1259 if (len < GROUP_FILTER_SIZE(0)) {
1260 release_sock(sk);
1261 return -EINVAL;
1263 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) {
1264 release_sock(sk);
1265 return -EFAULT;
1267 err = ip_mc_gsfget(sk, &gsf,
1268 (struct group_filter __user *)optval,
1269 optlen);
1270 release_sock(sk);
1271 return err;
1273 case IP_MULTICAST_ALL:
1274 val = inet->mc_all;
1275 break;
1276 case IP_PKTOPTIONS:
1278 struct msghdr msg;
1280 release_sock(sk);
1282 if (sk->sk_type != SOCK_STREAM)
1283 return -ENOPROTOOPT;
1285 msg.msg_control = optval;
1286 msg.msg_controllen = len;
1287 msg.msg_flags = flags;
1289 if (inet->cmsg_flags & IP_CMSG_PKTINFO) {
1290 struct in_pktinfo info;
1292 info.ipi_addr.s_addr = inet->inet_rcv_saddr;
1293 info.ipi_spec_dst.s_addr = inet->inet_rcv_saddr;
1294 info.ipi_ifindex = inet->mc_index;
1295 put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
1297 if (inet->cmsg_flags & IP_CMSG_TTL) {
1298 int hlim = inet->mc_ttl;
1299 put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim);
1301 if (inet->cmsg_flags & IP_CMSG_TOS) {
1302 int tos = inet->rcv_tos;
1303 put_cmsg(&msg, SOL_IP, IP_TOS, sizeof(tos), &tos);
1305 len -= msg.msg_controllen;
1306 return put_user(len, optlen);
1308 case IP_FREEBIND:
1309 val = inet->freebind;
1310 break;
1311 case IP_TRANSPARENT:
1312 val = inet->transparent;
1313 break;
1314 case IP_MINTTL:
1315 val = inet->min_ttl;
1316 break;
1317 default:
1318 release_sock(sk);
1319 return -ENOPROTOOPT;
1321 release_sock(sk);
1323 if (len < sizeof(int) && len > 0 && val >= 0 && val <= 255) {
1324 unsigned char ucval = (unsigned char)val;
1325 len = 1;
1326 if (put_user(len, optlen))
1327 return -EFAULT;
1328 if (copy_to_user(optval, &ucval, 1))
1329 return -EFAULT;
1330 } else {
1331 len = min_t(unsigned int, sizeof(int), len);
1332 if (put_user(len, optlen))
1333 return -EFAULT;
1334 if (copy_to_user(optval, &val, len))
1335 return -EFAULT;
1337 return 0;
1340 int ip_getsockopt(struct sock *sk, int level,
1341 int optname, char __user *optval, int __user *optlen)
1343 int err;
1345 err = do_ip_getsockopt(sk, level, optname, optval, optlen, 0);
1346 #ifdef CONFIG_NETFILTER
1347 /* we need to exclude all possible ENOPROTOOPTs except default case */
1348 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1349 !ip_mroute_opt(optname)) {
1350 int len;
1352 if (get_user(len, optlen))
1353 return -EFAULT;
1355 lock_sock(sk);
1356 err = nf_getsockopt(sk, PF_INET, optname, optval,
1357 &len);
1358 release_sock(sk);
1359 if (err >= 0)
1360 err = put_user(len, optlen);
1361 return err;
1363 #endif
1364 return err;
1366 EXPORT_SYMBOL(ip_getsockopt);
1368 #ifdef CONFIG_COMPAT
1369 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
1370 char __user *optval, int __user *optlen)
1372 int err;
1374 if (optname == MCAST_MSFILTER)
1375 return compat_mc_getsockopt(sk, level, optname, optval, optlen,
1376 ip_getsockopt);
1378 err = do_ip_getsockopt(sk, level, optname, optval, optlen,
1379 MSG_CMSG_COMPAT);
1381 #ifdef CONFIG_NETFILTER
1382 /* we need to exclude all possible ENOPROTOOPTs except default case */
1383 if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1384 !ip_mroute_opt(optname)) {
1385 int len;
1387 if (get_user(len, optlen))
1388 return -EFAULT;
1390 lock_sock(sk);
1391 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len);
1392 release_sock(sk);
1393 if (err >= 0)
1394 err = put_user(len, optlen);
1395 return err;
1397 #endif
1398 return err;
1400 EXPORT_SYMBOL(compat_ip_getsockopt);
1401 #endif