~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

TOMOYO Linux Cross Reference
Linux/net/ipv4/ip_sockglue.c

Version: ~ [ linux-5.4-rc3 ] ~ [ linux-5.3.6 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.79 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.149 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.196 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.196 ] ~ [ linux-4.3.6 ] ~ [ linux-4.2.8 ] ~ [ linux-4.1.52 ] ~ [ linux-4.0.9 ] ~ [ linux-3.19.8 ] ~ [ linux-3.18.140 ] ~ [ linux-3.17.8 ] ~ [ linux-3.16.75 ] ~ [ linux-3.15.10 ] ~ [ linux-3.14.79 ] ~ [ linux-3.13.11 ] ~ [ linux-3.12.74 ] ~ [ linux-3.11.10 ] ~ [ linux-3.10.108 ] ~ [ linux-3.9.11 ] ~ [ linux-3.8.13 ] ~ [ linux-3.7.10 ] ~ [ linux-3.6.11 ] ~ [ linux-3.5.7 ] ~ [ linux-3.4.113 ] ~ [ linux-3.3.8 ] ~ [ linux-3.2.102 ] ~ [ linux-3.1.10 ] ~ [ linux-3.0.101 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.5 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

  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 <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 #include <net/checksum.h>
 41 #if IS_ENABLED(CONFIG_IPV6)
 42 #include <net/transp_v6.h>
 43 #endif
 44 #include <net/ip_fib.h>
 45 
 46 #include <linux/errqueue.h>
 47 #include <linux/uaccess.h>
 48 
 49 /*
 50  *      SOL_IP control messages.
 51  */
 52 
 53 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
 54 {
 55         struct in_pktinfo info = *PKTINFO_SKB_CB(skb);
 56 
 57         info.ipi_addr.s_addr = ip_hdr(skb)->daddr;
 58 
 59         put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
 60 }
 61 
 62 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb)
 63 {
 64         int ttl = ip_hdr(skb)->ttl;
 65         put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl);
 66 }
 67 
 68 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb)
 69 {
 70         put_cmsg(msg, SOL_IP, IP_TOS, 1, &ip_hdr(skb)->tos);
 71 }
 72 
 73 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb)
 74 {
 75         if (IPCB(skb)->opt.optlen == 0)
 76                 return;
 77 
 78         put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen,
 79                  ip_hdr(skb) + 1);
 80 }
 81 
 82 
 83 static void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb)
 84 {
 85         unsigned char optbuf[sizeof(struct ip_options) + 40];
 86         struct ip_options *opt = (struct ip_options *)optbuf;
 87 
 88         if (IPCB(skb)->opt.optlen == 0)
 89                 return;
 90 
 91         if (ip_options_echo(opt, skb)) {
 92                 msg->msg_flags |= MSG_CTRUNC;
 93                 return;
 94         }
 95         ip_options_undo(opt);
 96 
 97         put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data);
 98 }
 99 
100 static void ip_cmsg_recv_fragsize(struct msghdr *msg, struct sk_buff *skb)
101 {
102         int val;
103 
104         if (IPCB(skb)->frag_max_size == 0)
105                 return;
106 
107         val = IPCB(skb)->frag_max_size;
108         put_cmsg(msg, SOL_IP, IP_RECVFRAGSIZE, sizeof(val), &val);
109 }
110 
111 static void ip_cmsg_recv_checksum(struct msghdr *msg, struct sk_buff *skb,
112                                   int tlen, int offset)
113 {
114         __wsum csum = skb->csum;
115 
116         if (skb->ip_summed != CHECKSUM_COMPLETE)
117                 return;
118 
119         if (offset != 0) {
120                 int tend_off = skb_transport_offset(skb) + tlen;
121                 csum = csum_sub(csum, skb_checksum(skb, tend_off, offset, 0));
122         }
123 
124         put_cmsg(msg, SOL_IP, IP_CHECKSUM, sizeof(__wsum), &csum);
125 }
126 
127 static void ip_cmsg_recv_security(struct msghdr *msg, struct sk_buff *skb)
128 {
129         char *secdata;
130         u32 seclen, secid;
131         int err;
132 
133         err = security_socket_getpeersec_dgram(NULL, skb, &secid);
134         if (err)
135                 return;
136 
137         err = security_secid_to_secctx(secid, &secdata, &seclen);
138         if (err)
139                 return;
140 
141         put_cmsg(msg, SOL_IP, SCM_SECURITY, seclen, secdata);
142         security_release_secctx(secdata, seclen);
143 }
144 
145 static void ip_cmsg_recv_dstaddr(struct msghdr *msg, struct sk_buff *skb)
146 {
147         struct sockaddr_in sin;
148         const struct iphdr *iph = ip_hdr(skb);
149         __be16 *ports = (__be16 *)skb_transport_header(skb);
150 
151         if (skb_transport_offset(skb) + 4 > (int)skb->len)
152                 return;
153 
154         /* All current transport protocols have the port numbers in the
155          * first four bytes of the transport header and this function is
156          * written with this assumption in mind.
157          */
158 
159         sin.sin_family = AF_INET;
160         sin.sin_addr.s_addr = iph->daddr;
161         sin.sin_port = ports[1];
162         memset(sin.sin_zero, 0, sizeof(sin.sin_zero));
163 
164         put_cmsg(msg, SOL_IP, IP_ORIGDSTADDR, sizeof(sin), &sin);
165 }
166 
167 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
168                          struct sk_buff *skb, int tlen, int offset)
169 {
170         struct inet_sock *inet = inet_sk(sk);
171         unsigned int flags = inet->cmsg_flags;
172 
173         /* Ordered by supposed usage frequency */
174         if (flags & IP_CMSG_PKTINFO) {
175                 ip_cmsg_recv_pktinfo(msg, skb);
176 
177                 flags &= ~IP_CMSG_PKTINFO;
178                 if (!flags)
179                         return;
180         }
181 
182         if (flags & IP_CMSG_TTL) {
183                 ip_cmsg_recv_ttl(msg, skb);
184 
185                 flags &= ~IP_CMSG_TTL;
186                 if (!flags)
187                         return;
188         }
189 
190         if (flags & IP_CMSG_TOS) {
191                 ip_cmsg_recv_tos(msg, skb);
192 
193                 flags &= ~IP_CMSG_TOS;
194                 if (!flags)
195                         return;
196         }
197 
198         if (flags & IP_CMSG_RECVOPTS) {
199                 ip_cmsg_recv_opts(msg, skb);
200 
201                 flags &= ~IP_CMSG_RECVOPTS;
202                 if (!flags)
203                         return;
204         }
205 
206         if (flags & IP_CMSG_RETOPTS) {
207                 ip_cmsg_recv_retopts(msg, skb);
208 
209                 flags &= ~IP_CMSG_RETOPTS;
210                 if (!flags)
211                         return;
212         }
213 
214         if (flags & IP_CMSG_PASSSEC) {
215                 ip_cmsg_recv_security(msg, skb);
216 
217                 flags &= ~IP_CMSG_PASSSEC;
218                 if (!flags)
219                         return;
220         }
221 
222         if (flags & IP_CMSG_ORIGDSTADDR) {
223                 ip_cmsg_recv_dstaddr(msg, skb);
224 
225                 flags &= ~IP_CMSG_ORIGDSTADDR;
226                 if (!flags)
227                         return;
228         }
229 
230         if (flags & IP_CMSG_CHECKSUM)
231                 ip_cmsg_recv_checksum(msg, skb, tlen, offset);
232 
233         if (flags & IP_CMSG_RECVFRAGSIZE)
234                 ip_cmsg_recv_fragsize(msg, skb);
235 }
236 EXPORT_SYMBOL(ip_cmsg_recv_offset);
237 
238 int ip_cmsg_send(struct sock *sk, struct msghdr *msg, struct ipcm_cookie *ipc,
239                  bool allow_ipv6)
240 {
241         int err, val;
242         struct cmsghdr *cmsg;
243         struct net *net = sock_net(sk);
244 
245         for_each_cmsghdr(cmsg, msg) {
246                 if (!CMSG_OK(msg, cmsg))
247                         return -EINVAL;
248 #if IS_ENABLED(CONFIG_IPV6)
249                 if (allow_ipv6 &&
250                     cmsg->cmsg_level == SOL_IPV6 &&
251                     cmsg->cmsg_type == IPV6_PKTINFO) {
252                         struct in6_pktinfo *src_info;
253 
254                         if (cmsg->cmsg_len < CMSG_LEN(sizeof(*src_info)))
255                                 return -EINVAL;
256                         src_info = (struct in6_pktinfo *)CMSG_DATA(cmsg);
257                         if (!ipv6_addr_v4mapped(&src_info->ipi6_addr))
258                                 return -EINVAL;
259                         ipc->oif = src_info->ipi6_ifindex;
260                         ipc->addr = src_info->ipi6_addr.s6_addr32[3];
261                         continue;
262                 }
263 #endif
264                 if (cmsg->cmsg_level == SOL_SOCKET) {
265                         err = __sock_cmsg_send(sk, msg, cmsg, &ipc->sockc);
266                         if (err)
267                                 return err;
268                         continue;
269                 }
270 
271                 if (cmsg->cmsg_level != SOL_IP)
272                         continue;
273                 switch (cmsg->cmsg_type) {
274                 case IP_RETOPTS:
275                         err = cmsg->cmsg_len - sizeof(struct cmsghdr);
276 
277                         /* Our caller is responsible for freeing ipc->opt */
278                         err = ip_options_get(net, &ipc->opt, CMSG_DATA(cmsg),
279                                              err < 40 ? err : 40);
280                         if (err)
281                                 return err;
282                         break;
283                 case IP_PKTINFO:
284                 {
285                         struct in_pktinfo *info;
286                         if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo)))
287                                 return -EINVAL;
288                         info = (struct in_pktinfo *)CMSG_DATA(cmsg);
289                         ipc->oif = info->ipi_ifindex;
290                         ipc->addr = info->ipi_spec_dst.s_addr;
291                         break;
292                 }
293                 case IP_TTL:
294                         if (cmsg->cmsg_len != CMSG_LEN(sizeof(int)))
295                                 return -EINVAL;
296                         val = *(int *)CMSG_DATA(cmsg);
297                         if (val < 1 || val > 255)
298                                 return -EINVAL;
299                         ipc->ttl = val;
300                         break;
301                 case IP_TOS:
302                         if (cmsg->cmsg_len == CMSG_LEN(sizeof(int)))
303                                 val = *(int *)CMSG_DATA(cmsg);
304                         else if (cmsg->cmsg_len == CMSG_LEN(sizeof(u8)))
305                                 val = *(u8 *)CMSG_DATA(cmsg);
306                         else
307                                 return -EINVAL;
308                         if (val < 0 || val > 255)
309                                 return -EINVAL;
310                         ipc->tos = val;
311                         ipc->priority = rt_tos2priority(ipc->tos);
312                         break;
313 
314                 default:
315                         return -EINVAL;
316                 }
317         }
318         return 0;
319 }
320 
321 
322 /* Special input handler for packets caught by router alert option.
323    They are selected only by protocol field, and then processed likely
324    local ones; but only if someone wants them! Otherwise, router
325    not running rsvpd will kill RSVP.
326 
327    It is user level problem, what it will make with them.
328    I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
329    but receiver should be enough clever f.e. to forward mtrace requests,
330    sent to multicast group to reach destination designated router.
331  */
332 struct ip_ra_chain __rcu *ip_ra_chain;
333 static DEFINE_SPINLOCK(ip_ra_lock);
334 
335 
336 static void ip_ra_destroy_rcu(struct rcu_head *head)
337 {
338         struct ip_ra_chain *ra = container_of(head, struct ip_ra_chain, rcu);
339 
340         sock_put(ra->saved_sk);
341         kfree(ra);
342 }
343 
344 int ip_ra_control(struct sock *sk, unsigned char on,
345                   void (*destructor)(struct sock *))
346 {
347         struct ip_ra_chain *ra, *new_ra;
348         struct ip_ra_chain __rcu **rap;
349 
350         if (sk->sk_type != SOCK_RAW || inet_sk(sk)->inet_num == IPPROTO_RAW)
351                 return -EINVAL;
352 
353         new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL;
354 
355         spin_lock_bh(&ip_ra_lock);
356         for (rap = &ip_ra_chain;
357              (ra = rcu_dereference_protected(*rap,
358                         lockdep_is_held(&ip_ra_lock))) != NULL;
359              rap = &ra->next) {
360                 if (ra->sk == sk) {
361                         if (on) {
362                                 spin_unlock_bh(&ip_ra_lock);
363                                 kfree(new_ra);
364                                 return -EADDRINUSE;
365                         }
366                         /* dont let ip_call_ra_chain() use sk again */
367                         ra->sk = NULL;
368                         RCU_INIT_POINTER(*rap, ra->next);
369                         spin_unlock_bh(&ip_ra_lock);
370 
371                         if (ra->destructor)
372                                 ra->destructor(sk);
373                         /*
374                          * Delay sock_put(sk) and kfree(ra) after one rcu grace
375                          * period. This guarantee ip_call_ra_chain() dont need
376                          * to mess with socket refcounts.
377                          */
378                         ra->saved_sk = sk;
379                         call_rcu(&ra->rcu, ip_ra_destroy_rcu);
380                         return 0;
381                 }
382         }
383         if (!new_ra) {
384                 spin_unlock_bh(&ip_ra_lock);
385                 return -ENOBUFS;
386         }
387         new_ra->sk = sk;
388         new_ra->destructor = destructor;
389 
390         RCU_INIT_POINTER(new_ra->next, ra);
391         rcu_assign_pointer(*rap, new_ra);
392         sock_hold(sk);
393         spin_unlock_bh(&ip_ra_lock);
394 
395         return 0;
396 }
397 
398 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
399                    __be16 port, u32 info, u8 *payload)
400 {
401         struct sock_exterr_skb *serr;
402 
403         skb = skb_clone(skb, GFP_ATOMIC);
404         if (!skb)
405                 return;
406 
407         serr = SKB_EXT_ERR(skb);
408         serr->ee.ee_errno = err;
409         serr->ee.ee_origin = SO_EE_ORIGIN_ICMP;
410         serr->ee.ee_type = icmp_hdr(skb)->type;
411         serr->ee.ee_code = icmp_hdr(skb)->code;
412         serr->ee.ee_pad = 0;
413         serr->ee.ee_info = info;
414         serr->ee.ee_data = 0;
415         serr->addr_offset = (u8 *)&(((struct iphdr *)(icmp_hdr(skb) + 1))->daddr) -
416                                    skb_network_header(skb);
417         serr->port = port;
418 
419         if (skb_pull(skb, payload - skb->data)) {
420                 skb_reset_transport_header(skb);
421                 if (sock_queue_err_skb(sk, skb) == 0)
422                         return;
423         }
424         kfree_skb(skb);
425 }
426 
427 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 port, u32 info)
428 {
429         struct inet_sock *inet = inet_sk(sk);
430         struct sock_exterr_skb *serr;
431         struct iphdr *iph;
432         struct sk_buff *skb;
433 
434         if (!inet->recverr)
435                 return;
436 
437         skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC);
438         if (!skb)
439                 return;
440 
441         skb_put(skb, sizeof(struct iphdr));
442         skb_reset_network_header(skb);
443         iph = ip_hdr(skb);
444         iph->daddr = daddr;
445 
446         serr = SKB_EXT_ERR(skb);
447         serr->ee.ee_errno = err;
448         serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
449         serr->ee.ee_type = 0;
450         serr->ee.ee_code = 0;
451         serr->ee.ee_pad = 0;
452         serr->ee.ee_info = info;
453         serr->ee.ee_data = 0;
454         serr->addr_offset = (u8 *)&iph->daddr - skb_network_header(skb);
455         serr->port = port;
456 
457         __skb_pull(skb, skb_tail_pointer(skb) - skb->data);
458         skb_reset_transport_header(skb);
459 
460         if (sock_queue_err_skb(sk, skb))
461                 kfree_skb(skb);
462 }
463 
464 /* For some errors we have valid addr_offset even with zero payload and
465  * zero port. Also, addr_offset should be supported if port is set.
466  */
467 static inline bool ipv4_datagram_support_addr(struct sock_exterr_skb *serr)
468 {
469         return serr->ee.ee_origin == SO_EE_ORIGIN_ICMP ||
470                serr->ee.ee_origin == SO_EE_ORIGIN_LOCAL || serr->port;
471 }
472 
473 /* IPv4 supports cmsg on all imcp errors and some timestamps
474  *
475  * Timestamp code paths do not initialize the fields expected by cmsg:
476  * the PKTINFO fields in skb->cb[]. Fill those in here.
477  */
478 static bool ipv4_datagram_support_cmsg(const struct sock *sk,
479                                        struct sk_buff *skb,
480                                        int ee_origin)
481 {
482         struct in_pktinfo *info;
483 
484         if (ee_origin == SO_EE_ORIGIN_ICMP)
485                 return true;
486 
487         if (ee_origin == SO_EE_ORIGIN_LOCAL)
488                 return false;
489 
490         /* Support IP_PKTINFO on tstamp packets if requested, to correlate
491          * timestamp with egress dev. Not possible for packets without iif
492          * or without payload (SOF_TIMESTAMPING_OPT_TSONLY).
493          */
494         info = PKTINFO_SKB_CB(skb);
495         if (!(sk->sk_tsflags & SOF_TIMESTAMPING_OPT_CMSG) ||
496             !info->ipi_ifindex)
497                 return false;
498 
499         info->ipi_spec_dst.s_addr = ip_hdr(skb)->saddr;
500         return true;
501 }
502 
503 /*
504  *      Handle MSG_ERRQUEUE
505  */
506 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len)
507 {
508         struct sock_exterr_skb *serr;
509         struct sk_buff *skb;
510         DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name);
511         struct {
512                 struct sock_extended_err ee;
513                 struct sockaddr_in       offender;
514         } errhdr;
515         int err;
516         int copied;
517 
518         WARN_ON_ONCE(sk->sk_family == AF_INET6);
519 
520         err = -EAGAIN;
521         skb = sock_dequeue_err_skb(sk);
522         if (!skb)
523                 goto out;
524 
525         copied = skb->len;
526         if (copied > len) {
527                 msg->msg_flags |= MSG_TRUNC;
528                 copied = len;
529         }
530         err = skb_copy_datagram_msg(skb, 0, msg, copied);
531         if (unlikely(err)) {
532                 kfree_skb(skb);
533                 return err;
534         }
535         sock_recv_timestamp(msg, sk, skb);
536 
537         serr = SKB_EXT_ERR(skb);
538 
539         if (sin && ipv4_datagram_support_addr(serr)) {
540                 sin->sin_family = AF_INET;
541                 sin->sin_addr.s_addr = *(__be32 *)(skb_network_header(skb) +
542                                                    serr->addr_offset);
543                 sin->sin_port = serr->port;
544                 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
545                 *addr_len = sizeof(*sin);
546         }
547 
548         memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err));
549         sin = &errhdr.offender;
550         memset(sin, 0, sizeof(*sin));
551 
552         if (ipv4_datagram_support_cmsg(sk, skb, serr->ee.ee_origin)) {
553                 sin->sin_family = AF_INET;
554                 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
555                 if (inet_sk(sk)->cmsg_flags)
556                         ip_cmsg_recv(msg, skb);
557         }
558 
559         put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr);
560 
561         /* Now we could try to dump offended packet options */
562 
563         msg->msg_flags |= MSG_ERRQUEUE;
564         err = copied;
565 
566         consume_skb(skb);
567 out:
568         return err;
569 }
570 
571 
572 /*
573  *      Socket option code for IP. This is the end of the line after any
574  *      TCP,UDP etc options on an IP socket.
575  */
576 static bool setsockopt_needs_rtnl(int optname)
577 {
578         switch (optname) {
579         case IP_ADD_MEMBERSHIP:
580         case IP_ADD_SOURCE_MEMBERSHIP:
581         case IP_BLOCK_SOURCE:
582         case IP_DROP_MEMBERSHIP:
583         case IP_DROP_SOURCE_MEMBERSHIP:
584         case IP_MSFILTER:
585         case IP_UNBLOCK_SOURCE:
586         case MCAST_BLOCK_SOURCE:
587         case MCAST_MSFILTER:
588         case MCAST_JOIN_GROUP:
589         case MCAST_JOIN_SOURCE_GROUP:
590         case MCAST_LEAVE_GROUP:
591         case MCAST_LEAVE_SOURCE_GROUP:
592         case MCAST_UNBLOCK_SOURCE:
593         case IP_ROUTER_ALERT:
594                 return true;
595         }
596         return false;
597 }
598 
599 static int do_ip_setsockopt(struct sock *sk, int level,
600                             int optname, char __user *optval, unsigned int optlen)
601 {
602         struct inet_sock *inet = inet_sk(sk);
603         struct net *net = sock_net(sk);
604         int val = 0, err;
605         bool needs_rtnl = setsockopt_needs_rtnl(optname);
606 
607         switch (optname) {
608         case IP_PKTINFO:
609         case IP_RECVTTL:
610         case IP_RECVOPTS:
611         case IP_RECVTOS:
612         case IP_RETOPTS:
613         case IP_TOS:
614         case IP_TTL:
615         case IP_HDRINCL:
616         case IP_MTU_DISCOVER:
617         case IP_RECVERR:
618         case IP_ROUTER_ALERT:
619         case IP_FREEBIND:
620         case IP_PASSSEC:
621         case IP_TRANSPARENT:
622         case IP_MINTTL:
623         case IP_NODEFRAG:
624         case IP_BIND_ADDRESS_NO_PORT:
625         case IP_UNICAST_IF:
626         case IP_MULTICAST_TTL:
627         case IP_MULTICAST_ALL:
628         case IP_MULTICAST_LOOP:
629         case IP_RECVORIGDSTADDR:
630         case IP_CHECKSUM:
631         case IP_RECVFRAGSIZE:
632                 if (optlen >= sizeof(int)) {
633                         if (get_user(val, (int __user *) optval))
634                                 return -EFAULT;
635                 } else if (optlen >= sizeof(char)) {
636                         unsigned char ucval;
637 
638                         if (get_user(ucval, (unsigned char __user *) optval))
639                                 return -EFAULT;
640                         val = (int) ucval;
641                 }
642         }
643 
644         /* If optlen==0, it is equivalent to val == 0 */
645 
646         if (ip_mroute_opt(optname))
647                 return ip_mroute_setsockopt(sk, optname, optval, optlen);
648 
649         err = 0;
650         if (needs_rtnl)
651                 rtnl_lock();
652         lock_sock(sk);
653 
654         switch (optname) {
655         case IP_OPTIONS:
656         {
657                 struct ip_options_rcu *old, *opt = NULL;
658 
659                 if (optlen > 40)
660                         goto e_inval;
661                 err = ip_options_get_from_user(sock_net(sk), &opt,
662                                                optval, optlen);
663                 if (err)
664                         break;
665                 old = rcu_dereference_protected(inet->inet_opt,
666                                                 lockdep_sock_is_held(sk));
667                 if (inet->is_icsk) {
668                         struct inet_connection_sock *icsk = inet_csk(sk);
669 #if IS_ENABLED(CONFIG_IPV6)
670                         if (sk->sk_family == PF_INET ||
671                             (!((1 << sk->sk_state) &
672                                (TCPF_LISTEN | TCPF_CLOSE)) &&
673                              inet->inet_daddr != LOOPBACK4_IPV6)) {
674 #endif
675                                 if (old)
676                                         icsk->icsk_ext_hdr_len -= old->opt.optlen;
677                                 if (opt)
678                                         icsk->icsk_ext_hdr_len += opt->opt.optlen;
679                                 icsk->icsk_sync_mss(sk, icsk->icsk_pmtu_cookie);
680 #if IS_ENABLED(CONFIG_IPV6)
681                         }
682 #endif
683                 }
684                 rcu_assign_pointer(inet->inet_opt, opt);
685                 if (old)
686                         kfree_rcu(old, rcu);
687                 break;
688         }
689         case IP_PKTINFO:
690                 if (val)
691                         inet->cmsg_flags |= IP_CMSG_PKTINFO;
692                 else
693                         inet->cmsg_flags &= ~IP_CMSG_PKTINFO;
694                 break;
695         case IP_RECVTTL:
696                 if (val)
697                         inet->cmsg_flags |=  IP_CMSG_TTL;
698                 else
699                         inet->cmsg_flags &= ~IP_CMSG_TTL;
700                 break;
701         case IP_RECVTOS:
702                 if (val)
703                         inet->cmsg_flags |=  IP_CMSG_TOS;
704                 else
705                         inet->cmsg_flags &= ~IP_CMSG_TOS;
706                 break;
707         case IP_RECVOPTS:
708                 if (val)
709                         inet->cmsg_flags |=  IP_CMSG_RECVOPTS;
710                 else
711                         inet->cmsg_flags &= ~IP_CMSG_RECVOPTS;
712                 break;
713         case IP_RETOPTS:
714                 if (val)
715                         inet->cmsg_flags |= IP_CMSG_RETOPTS;
716                 else
717                         inet->cmsg_flags &= ~IP_CMSG_RETOPTS;
718                 break;
719         case IP_PASSSEC:
720                 if (val)
721                         inet->cmsg_flags |= IP_CMSG_PASSSEC;
722                 else
723                         inet->cmsg_flags &= ~IP_CMSG_PASSSEC;
724                 break;
725         case IP_RECVORIGDSTADDR:
726                 if (val)
727                         inet->cmsg_flags |= IP_CMSG_ORIGDSTADDR;
728                 else
729                         inet->cmsg_flags &= ~IP_CMSG_ORIGDSTADDR;
730                 break;
731         case IP_CHECKSUM:
732                 if (val) {
733                         if (!(inet->cmsg_flags & IP_CMSG_CHECKSUM)) {
734                                 inet_inc_convert_csum(sk);
735                                 inet->cmsg_flags |= IP_CMSG_CHECKSUM;
736                         }
737                 } else {
738                         if (inet->cmsg_flags & IP_CMSG_CHECKSUM) {
739                                 inet_dec_convert_csum(sk);
740                                 inet->cmsg_flags &= ~IP_CMSG_CHECKSUM;
741                         }
742                 }
743                 break;
744         case IP_RECVFRAGSIZE:
745                 if (sk->sk_type != SOCK_RAW && sk->sk_type != SOCK_DGRAM)
746                         goto e_inval;
747                 if (val)
748                         inet->cmsg_flags |= IP_CMSG_RECVFRAGSIZE;
749                 else
750                         inet->cmsg_flags &= ~IP_CMSG_RECVFRAGSIZE;
751                 break;
752         case IP_TOS:    /* This sets both TOS and Precedence */
753                 if (sk->sk_type == SOCK_STREAM) {
754                         val &= ~INET_ECN_MASK;
755                         val |= inet->tos & INET_ECN_MASK;
756                 }
757                 if (inet->tos != val) {
758                         inet->tos = val;
759                         sk->sk_priority = rt_tos2priority(val);
760                         sk_dst_reset(sk);
761                 }
762                 break;
763         case IP_TTL:
764                 if (optlen < 1)
765                         goto e_inval;
766                 if (val != -1 && (val < 1 || val > 255))
767                         goto e_inval;
768                 inet->uc_ttl = val;
769                 break;
770         case IP_HDRINCL:
771                 if (sk->sk_type != SOCK_RAW) {
772                         err = -ENOPROTOOPT;
773                         break;
774                 }
775                 inet->hdrincl = val ? 1 : 0;
776                 break;
777         case IP_NODEFRAG:
778                 if (sk->sk_type != SOCK_RAW) {
779                         err = -ENOPROTOOPT;
780                         break;
781                 }
782                 inet->nodefrag = val ? 1 : 0;
783                 break;
784         case IP_BIND_ADDRESS_NO_PORT:
785                 inet->bind_address_no_port = val ? 1 : 0;
786                 break;
787         case IP_MTU_DISCOVER:
788                 if (val < IP_PMTUDISC_DONT || val > IP_PMTUDISC_OMIT)
789                         goto e_inval;
790                 inet->pmtudisc = val;
791                 break;
792         case IP_RECVERR:
793                 inet->recverr = !!val;
794                 if (!val)
795                         skb_queue_purge(&sk->sk_error_queue);
796                 break;
797         case IP_MULTICAST_TTL:
798                 if (sk->sk_type == SOCK_STREAM)
799                         goto e_inval;
800                 if (optlen < 1)
801                         goto e_inval;
802                 if (val == -1)
803                         val = 1;
804                 if (val < 0 || val > 255)
805                         goto e_inval;
806                 inet->mc_ttl = val;
807                 break;
808         case IP_MULTICAST_LOOP:
809                 if (optlen < 1)
810                         goto e_inval;
811                 inet->mc_loop = !!val;
812                 break;
813         case IP_UNICAST_IF:
814         {
815                 struct net_device *dev = NULL;
816                 int ifindex;
817 
818                 if (optlen != sizeof(int))
819                         goto e_inval;
820 
821                 ifindex = (__force int)ntohl((__force __be32)val);
822                 if (ifindex == 0) {
823                         inet->uc_index = 0;
824                         err = 0;
825                         break;
826                 }
827 
828                 dev = dev_get_by_index(sock_net(sk), ifindex);
829                 err = -EADDRNOTAVAIL;
830                 if (!dev)
831                         break;
832                 dev_put(dev);
833 
834                 err = -EINVAL;
835                 if (sk->sk_bound_dev_if)
836                         break;
837 
838                 inet->uc_index = ifindex;
839                 err = 0;
840                 break;
841         }
842         case IP_MULTICAST_IF:
843         {
844                 struct ip_mreqn mreq;
845                 struct net_device *dev = NULL;
846                 int midx;
847 
848                 if (sk->sk_type == SOCK_STREAM)
849                         goto e_inval;
850                 /*
851                  *      Check the arguments are allowable
852                  */
853 
854                 if (optlen < sizeof(struct in_addr))
855                         goto e_inval;
856 
857                 err = -EFAULT;
858                 if (optlen >= sizeof(struct ip_mreqn)) {
859                         if (copy_from_user(&mreq, optval, sizeof(mreq)))
860                                 break;
861                 } else {
862                         memset(&mreq, 0, sizeof(mreq));
863                         if (optlen >= sizeof(struct ip_mreq)) {
864                                 if (copy_from_user(&mreq, optval,
865                                                    sizeof(struct ip_mreq)))
866                                         break;
867                         } else if (optlen >= sizeof(struct in_addr)) {
868                                 if (copy_from_user(&mreq.imr_address, optval,
869                                                    sizeof(struct in_addr)))
870                                         break;
871                         }
872                 }
873 
874                 if (!mreq.imr_ifindex) {
875                         if (mreq.imr_address.s_addr == htonl(INADDR_ANY)) {
876                                 inet->mc_index = 0;
877                                 inet->mc_addr  = 0;
878                                 err = 0;
879                                 break;
880                         }
881                         dev = ip_dev_find(sock_net(sk), mreq.imr_address.s_addr);
882                         if (dev)
883                                 mreq.imr_ifindex = dev->ifindex;
884                 } else
885                         dev = dev_get_by_index(sock_net(sk), mreq.imr_ifindex);
886 
887 
888                 err = -EADDRNOTAVAIL;
889                 if (!dev)
890                         break;
891 
892                 midx = l3mdev_master_ifindex(dev);
893 
894                 dev_put(dev);
895 
896                 err = -EINVAL;
897                 if (sk->sk_bound_dev_if &&
898                     mreq.imr_ifindex != sk->sk_bound_dev_if &&
899                     (!midx || midx != sk->sk_bound_dev_if))
900                         break;
901 
902                 inet->mc_index = mreq.imr_ifindex;
903                 inet->mc_addr  = mreq.imr_address.s_addr;
904                 err = 0;
905                 break;
906         }
907 
908         case IP_ADD_MEMBERSHIP:
909         case IP_DROP_MEMBERSHIP:
910         {
911                 struct ip_mreqn mreq;
912 
913                 err = -EPROTO;
914                 if (inet_sk(sk)->is_icsk)
915                         break;
916 
917                 if (optlen < sizeof(struct ip_mreq))
918                         goto e_inval;
919                 err = -EFAULT;
920                 if (optlen >= sizeof(struct ip_mreqn)) {
921                         if (copy_from_user(&mreq, optval, sizeof(mreq)))
922                                 break;
923                 } else {
924                         memset(&mreq, 0, sizeof(mreq));
925                         if (copy_from_user(&mreq, optval, sizeof(struct ip_mreq)))
926                                 break;
927                 }
928 
929                 if (optname == IP_ADD_MEMBERSHIP)
930                         err = ip_mc_join_group(sk, &mreq);
931                 else
932                         err = ip_mc_leave_group(sk, &mreq);
933                 break;
934         }
935         case IP_MSFILTER:
936         {
937                 struct ip_msfilter *msf;
938 
939                 if (optlen < IP_MSFILTER_SIZE(0))
940                         goto e_inval;
941                 if (optlen > sysctl_optmem_max) {
942                         err = -ENOBUFS;
943                         break;
944                 }
945                 msf = kmalloc(optlen, GFP_KERNEL);
946                 if (!msf) {
947                         err = -ENOBUFS;
948                         break;
949                 }
950                 err = -EFAULT;
951                 if (copy_from_user(msf, optval, optlen)) {
952                         kfree(msf);
953                         break;
954                 }
955                 /* numsrc >= (1G-4) overflow in 32 bits */
956                 if (msf->imsf_numsrc >= 0x3ffffffcU ||
957                     msf->imsf_numsrc > net->ipv4.sysctl_igmp_max_msf) {
958                         kfree(msf);
959                         err = -ENOBUFS;
960                         break;
961                 }
962                 if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) {
963                         kfree(msf);
964                         err = -EINVAL;
965                         break;
966                 }
967                 err = ip_mc_msfilter(sk, msf, 0);
968                 kfree(msf);
969                 break;
970         }
971         case IP_BLOCK_SOURCE:
972         case IP_UNBLOCK_SOURCE:
973         case IP_ADD_SOURCE_MEMBERSHIP:
974         case IP_DROP_SOURCE_MEMBERSHIP:
975         {
976                 struct ip_mreq_source mreqs;
977                 int omode, add;
978 
979                 if (optlen != sizeof(struct ip_mreq_source))
980                         goto e_inval;
981                 if (copy_from_user(&mreqs, optval, sizeof(mreqs))) {
982                         err = -EFAULT;
983                         break;
984                 }
985                 if (optname == IP_BLOCK_SOURCE) {
986                         omode = MCAST_EXCLUDE;
987                         add = 1;
988                 } else if (optname == IP_UNBLOCK_SOURCE) {
989                         omode = MCAST_EXCLUDE;
990                         add = 0;
991                 } else if (optname == IP_ADD_SOURCE_MEMBERSHIP) {
992                         struct ip_mreqn mreq;
993 
994                         mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr;
995                         mreq.imr_address.s_addr = mreqs.imr_interface;
996                         mreq.imr_ifindex = 0;
997                         err = ip_mc_join_group(sk, &mreq);
998                         if (err && err != -EADDRINUSE)
999                                 break;
1000                         omode = MCAST_INCLUDE;
1001                         add = 1;
1002                 } else /* IP_DROP_SOURCE_MEMBERSHIP */ {
1003                         omode = MCAST_INCLUDE;
1004                         add = 0;
1005                 }
1006                 err = ip_mc_source(add, omode, sk, &mreqs, 0);
1007                 break;
1008         }
1009         case MCAST_JOIN_GROUP:
1010         case MCAST_LEAVE_GROUP:
1011         {
1012                 struct group_req greq;
1013                 struct sockaddr_in *psin;
1014                 struct ip_mreqn mreq;
1015 
1016                 if (optlen < sizeof(struct group_req))
1017                         goto e_inval;
1018                 err = -EFAULT;
1019                 if (copy_from_user(&greq, optval, sizeof(greq)))
1020                         break;
1021                 psin = (struct sockaddr_in *)&greq.gr_group;
1022                 if (psin->sin_family != AF_INET)
1023                         goto e_inval;
1024                 memset(&mreq, 0, sizeof(mreq));
1025                 mreq.imr_multiaddr = psin->sin_addr;
1026                 mreq.imr_ifindex = greq.gr_interface;
1027 
1028                 if (optname == MCAST_JOIN_GROUP)
1029                         err = ip_mc_join_group(sk, &mreq);
1030                 else
1031                         err = ip_mc_leave_group(sk, &mreq);
1032                 break;
1033         }
1034         case MCAST_JOIN_SOURCE_GROUP:
1035         case MCAST_LEAVE_SOURCE_GROUP:
1036         case MCAST_BLOCK_SOURCE:
1037         case MCAST_UNBLOCK_SOURCE:
1038         {
1039                 struct group_source_req greqs;
1040                 struct ip_mreq_source mreqs;
1041                 struct sockaddr_in *psin;
1042                 int omode, add;
1043 
1044                 if (optlen != sizeof(struct group_source_req))
1045                         goto e_inval;
1046                 if (copy_from_user(&greqs, optval, sizeof(greqs))) {
1047                         err = -EFAULT;
1048                         break;
1049                 }
1050                 if (greqs.gsr_group.ss_family != AF_INET ||
1051                     greqs.gsr_source.ss_family != AF_INET) {
1052                         err = -EADDRNOTAVAIL;
1053                         break;
1054                 }
1055                 psin = (struct sockaddr_in *)&greqs.gsr_group;
1056                 mreqs.imr_multiaddr = psin->sin_addr.s_addr;
1057                 psin = (struct sockaddr_in *)&greqs.gsr_source;
1058                 mreqs.imr_sourceaddr = psin->sin_addr.s_addr;
1059                 mreqs.imr_interface = 0; /* use index for mc_source */
1060 
1061                 if (optname == MCAST_BLOCK_SOURCE) {
1062                         omode = MCAST_EXCLUDE;
1063                         add = 1;
1064                 } else if (optname == MCAST_UNBLOCK_SOURCE) {
1065                         omode = MCAST_EXCLUDE;
1066                         add = 0;
1067                 } else if (optname == MCAST_JOIN_SOURCE_GROUP) {
1068                         struct ip_mreqn mreq;
1069 
1070                         psin = (struct sockaddr_in *)&greqs.gsr_group;
1071                         mreq.imr_multiaddr = psin->sin_addr;
1072                         mreq.imr_address.s_addr = 0;
1073                         mreq.imr_ifindex = greqs.gsr_interface;
1074                         err = ip_mc_join_group(sk, &mreq);
1075                         if (err && err != -EADDRINUSE)
1076                                 break;
1077                         greqs.gsr_interface = mreq.imr_ifindex;
1078                         omode = MCAST_INCLUDE;
1079                         add = 1;
1080                 } else /* MCAST_LEAVE_SOURCE_GROUP */ {
1081                         omode = MCAST_INCLUDE;
1082                         add = 0;
1083                 }
1084                 err = ip_mc_source(add, omode, sk, &mreqs,
1085                                    greqs.gsr_interface);
1086                 break;
1087         }
1088         case MCAST_MSFILTER:
1089         {
1090                 struct sockaddr_in *psin;
1091                 struct ip_msfilter *msf = NULL;
1092                 struct group_filter *gsf = NULL;
1093                 int msize, i, ifindex;
1094 
1095                 if (optlen < GROUP_FILTER_SIZE(0))
1096                         goto e_inval;
1097                 if (optlen > sysctl_optmem_max) {
1098                         err = -ENOBUFS;
1099                         break;
1100                 }
1101                 gsf = kmalloc(optlen, GFP_KERNEL);
1102                 if (!gsf) {
1103                         err = -ENOBUFS;
1104                         break;
1105                 }
1106                 err = -EFAULT;
1107                 if (copy_from_user(gsf, optval, optlen))
1108                         goto mc_msf_out;
1109 
1110                 /* numsrc >= (4G-140)/128 overflow in 32 bits */
1111                 if (gsf->gf_numsrc >= 0x1ffffff ||
1112                     gsf->gf_numsrc > net->ipv4.sysctl_igmp_max_msf) {
1113                         err = -ENOBUFS;
1114                         goto mc_msf_out;
1115                 }
1116                 if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) {
1117                         err = -EINVAL;
1118                         goto mc_msf_out;
1119                 }
1120                 msize = IP_MSFILTER_SIZE(gsf->gf_numsrc);
1121                 msf = kmalloc(msize, GFP_KERNEL);
1122                 if (!msf) {
1123                         err = -ENOBUFS;
1124                         goto mc_msf_out;
1125                 }
1126                 ifindex = gsf->gf_interface;
1127                 psin = (struct sockaddr_in *)&gsf->gf_group;
1128                 if (psin->sin_family != AF_INET) {
1129                         err = -EADDRNOTAVAIL;
1130                         goto mc_msf_out;
1131                 }
1132                 msf->imsf_multiaddr = psin->sin_addr.s_addr;
1133                 msf->imsf_interface = 0;
1134                 msf->imsf_fmode = gsf->gf_fmode;
1135                 msf->imsf_numsrc = gsf->gf_numsrc;
1136                 err = -EADDRNOTAVAIL;
1137                 for (i = 0; i < gsf->gf_numsrc; ++i) {
1138                         psin = (struct sockaddr_in *)&gsf->gf_slist[i];
1139 
1140                         if (psin->sin_family != AF_INET)
1141                                 goto mc_msf_out;
1142                         msf->imsf_slist[i] = psin->sin_addr.s_addr;
1143                 }
1144                 kfree(gsf);
1145                 gsf = NULL;
1146 
1147                 err = ip_mc_msfilter(sk, msf, ifindex);
1148 mc_msf_out:
1149                 kfree(msf);
1150                 kfree(gsf);
1151                 break;
1152         }
1153         case IP_MULTICAST_ALL:
1154                 if (optlen < 1)
1155                         goto e_inval;
1156                 if (val != 0 && val != 1)
1157                         goto e_inval;
1158                 inet->mc_all = val;
1159                 break;
1160         case IP_ROUTER_ALERT:
1161                 err = ip_ra_control(sk, val ? 1 : 0, NULL);
1162                 break;
1163 
1164         case IP_FREEBIND:
1165                 if (optlen < 1)
1166                         goto e_inval;
1167                 inet->freebind = !!val;
1168                 break;
1169 
1170         case IP_IPSEC_POLICY:
1171         case IP_XFRM_POLICY:
1172                 err = -EPERM;
1173                 if (!ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN))
1174                         break;
1175                 err = xfrm_user_policy(sk, optname, optval, optlen);
1176                 break;
1177 
1178         case IP_TRANSPARENT:
1179                 if (!!val && !ns_capable(sock_net(sk)->user_ns, CAP_NET_RAW) &&
1180                     !ns_capable(sock_net(sk)->user_ns, CAP_NET_ADMIN)) {
1181                         err = -EPERM;
1182                         break;
1183                 }
1184                 if (optlen < 1)
1185                         goto e_inval;
1186                 inet->transparent = !!val;
1187                 break;
1188 
1189         case IP_MINTTL:
1190                 if (optlen < 1)
1191                         goto e_inval;
1192                 if (val < 0 || val > 255)
1193                         goto e_inval;
1194                 inet->min_ttl = val;
1195                 break;
1196 
1197         default:
1198                 err = -ENOPROTOOPT;
1199                 break;
1200         }
1201         release_sock(sk);
1202         if (needs_rtnl)
1203                 rtnl_unlock();
1204         return err;
1205 
1206 e_inval:
1207         release_sock(sk);
1208         if (needs_rtnl)
1209                 rtnl_unlock();
1210         return -EINVAL;
1211 }
1212 
1213 /**
1214  * ipv4_pktinfo_prepare - transfer some info from rtable to skb
1215  * @sk: socket
1216  * @skb: buffer
1217  *
1218  * To support IP_CMSG_PKTINFO option, we store rt_iif and specific
1219  * destination in skb->cb[] before dst drop.
1220  * This way, receiver doesn't make cache line misses to read rtable.
1221  */
1222 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb)
1223 {
1224         struct in_pktinfo *pktinfo = PKTINFO_SKB_CB(skb);
1225         bool prepare = (inet_sk(sk)->cmsg_flags & IP_CMSG_PKTINFO) ||
1226                        ipv6_sk_rxinfo(sk);
1227 
1228         if (prepare && skb_rtable(skb)) {
1229                 /* skb->cb is overloaded: prior to this point it is IP{6}CB
1230                  * which has interface index (iif) as the first member of the
1231                  * underlying inet{6}_skb_parm struct. This code then overlays
1232                  * PKTINFO_SKB_CB and in_pktinfo also has iif as the first
1233                  * element so the iif is picked up from the prior IPCB. If iif
1234                  * is the loopback interface, then return the sending interface
1235                  * (e.g., process binds socket to eth0 for Tx which is
1236                  * redirected to loopback in the rtable/dst).
1237                  */
1238                 if (pktinfo->ipi_ifindex == LOOPBACK_IFINDEX)
1239                         pktinfo->ipi_ifindex = inet_iif(skb);
1240 
1241                 pktinfo->ipi_spec_dst.s_addr = fib_compute_spec_dst(skb);
1242         } else {
1243                 pktinfo->ipi_ifindex = 0;
1244                 pktinfo->ipi_spec_dst.s_addr = 0;
1245         }
1246         /* We need to keep the dst for __ip_options_echo()
1247          * We could restrict the test to opt.ts_needtime || opt.srr,
1248          * but the following is good enough as IP options are not often used.
1249          */
1250         if (unlikely(IPCB(skb)->opt.optlen))
1251                 skb_dst_force(skb);
1252         else
1253                 skb_dst_drop(skb);
1254 }
1255 
1256 int ip_setsockopt(struct sock *sk, int level,
1257                 int optname, char __user *optval, unsigned int optlen)
1258 {
1259         int err;
1260 
1261         if (level != SOL_IP)
1262                 return -ENOPROTOOPT;
1263 
1264         err = do_ip_setsockopt(sk, level, optname, optval, optlen);
1265 #ifdef CONFIG_NETFILTER
1266         /* we need to exclude all possible ENOPROTOOPTs except default case */
1267         if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
1268                         optname != IP_IPSEC_POLICY &&
1269                         optname != IP_XFRM_POLICY &&
1270                         !ip_mroute_opt(optname)) {
1271                 lock_sock(sk);
1272                 err = nf_setsockopt(sk, PF_INET, optname, optval, optlen);
1273                 release_sock(sk);
1274         }
1275 #endif
1276         return err;
1277 }
1278 EXPORT_SYMBOL(ip_setsockopt);
1279 
1280 #ifdef CONFIG_COMPAT
1281 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
1282                          char __user *optval, unsigned int optlen)
1283 {
1284         int err;
1285 
1286         if (level != SOL_IP)
1287                 return -ENOPROTOOPT;
1288 
1289         if (optname >= MCAST_JOIN_GROUP && optname <= MCAST_MSFILTER)
1290                 return compat_mc_setsockopt(sk, level, optname, optval, optlen,
1291                         ip_setsockopt);
1292 
1293         err = do_ip_setsockopt(sk, level, optname, optval, optlen);
1294 #ifdef CONFIG_NETFILTER
1295         /* we need to exclude all possible ENOPROTOOPTs except default case */
1296         if (err == -ENOPROTOOPT && optname != IP_HDRINCL &&
1297                         optname != IP_IPSEC_POLICY &&
1298                         optname != IP_XFRM_POLICY &&
1299                         !ip_mroute_opt(optname)) {
1300                 lock_sock(sk);
1301                 err = compat_nf_setsockopt(sk, PF_INET, optname,
1302                                            optval, optlen);
1303                 release_sock(sk);
1304         }
1305 #endif
1306         return err;
1307 }
1308 EXPORT_SYMBOL(compat_ip_setsockopt);
1309 #endif
1310 
1311 /*
1312  *      Get the options. Note for future reference. The GET of IP options gets
1313  *      the _received_ ones. The set sets the _sent_ ones.
1314  */
1315 
1316 static bool getsockopt_needs_rtnl(int optname)
1317 {
1318         switch (optname) {
1319         case IP_MSFILTER:
1320         case MCAST_MSFILTER:
1321                 return true;
1322         }
1323         return false;
1324 }
1325 
1326 static int do_ip_getsockopt(struct sock *sk, int level, int optname,
1327                             char __user *optval, int __user *optlen, unsigned int flags)
1328 {
1329         struct inet_sock *inet = inet_sk(sk);
1330         bool needs_rtnl = getsockopt_needs_rtnl(optname);
1331         int val, err = 0;
1332         int len;
1333 
1334         if (level != SOL_IP)
1335                 return -EOPNOTSUPP;
1336 
1337         if (ip_mroute_opt(optname))
1338                 return ip_mroute_getsockopt(sk, optname, optval, optlen);
1339 
1340         if (get_user(len, optlen))
1341                 return -EFAULT;
1342         if (len < 0)
1343                 return -EINVAL;
1344 
1345         if (needs_rtnl)
1346                 rtnl_lock();
1347         lock_sock(sk);
1348 
1349         switch (optname) {
1350         case IP_OPTIONS:
1351         {
1352                 unsigned char optbuf[sizeof(struct ip_options)+40];
1353                 struct ip_options *opt = (struct ip_options *)optbuf;
1354                 struct ip_options_rcu *inet_opt;
1355 
1356                 inet_opt = rcu_dereference_protected(inet->inet_opt,
1357                                                      lockdep_sock_is_held(sk));
1358                 opt->optlen = 0;
1359                 if (inet_opt)
1360                         memcpy(optbuf, &inet_opt->opt,
1361                                sizeof(struct ip_options) +
1362                                inet_opt->opt.optlen);
1363                 release_sock(sk);
1364 
1365                 if (opt->optlen == 0)
1366                         return put_user(0, optlen);
1367 
1368                 ip_options_undo(opt);
1369 
1370                 len = min_t(unsigned int, len, opt->optlen);
1371                 if (put_user(len, optlen))
1372                         return -EFAULT;
1373                 if (copy_to_user(optval, opt->__data, len))
1374                         return -EFAULT;
1375                 return 0;
1376         }
1377         case IP_PKTINFO:
1378                 val = (inet->cmsg_flags & IP_CMSG_PKTINFO) != 0;
1379                 break;
1380         case IP_RECVTTL:
1381                 val = (inet->cmsg_flags & IP_CMSG_TTL) != 0;
1382                 break;
1383         case IP_RECVTOS:
1384                 val = (inet->cmsg_flags & IP_CMSG_TOS) != 0;
1385                 break;
1386         case IP_RECVOPTS:
1387                 val = (inet->cmsg_flags & IP_CMSG_RECVOPTS) != 0;
1388                 break;
1389         case IP_RETOPTS:
1390                 val = (inet->cmsg_flags & IP_CMSG_RETOPTS) != 0;
1391                 break;
1392         case IP_PASSSEC:
1393                 val = (inet->cmsg_flags & IP_CMSG_PASSSEC) != 0;
1394                 break;
1395         case IP_RECVORIGDSTADDR:
1396                 val = (inet->cmsg_flags & IP_CMSG_ORIGDSTADDR) != 0;
1397                 break;
1398         case IP_CHECKSUM:
1399                 val = (inet->cmsg_flags & IP_CMSG_CHECKSUM) != 0;
1400                 break;
1401         case IP_RECVFRAGSIZE:
1402                 val = (inet->cmsg_flags & IP_CMSG_RECVFRAGSIZE) != 0;
1403                 break;
1404         case IP_TOS:
1405                 val = inet->tos;
1406                 break;
1407         case IP_TTL:
1408         {
1409                 struct net *net = sock_net(sk);
1410                 val = (inet->uc_ttl == -1 ?
1411                        net->ipv4.sysctl_ip_default_ttl :
1412                        inet->uc_ttl);
1413                 break;
1414         }
1415         case IP_HDRINCL:
1416                 val = inet->hdrincl;
1417                 break;
1418         case IP_NODEFRAG:
1419                 val = inet->nodefrag;
1420                 break;
1421         case IP_BIND_ADDRESS_NO_PORT:
1422                 val = inet->bind_address_no_port;
1423                 break;
1424         case IP_MTU_DISCOVER:
1425                 val = inet->pmtudisc;
1426                 break;
1427         case IP_MTU:
1428         {
1429                 struct dst_entry *dst;
1430                 val = 0;
1431                 dst = sk_dst_get(sk);
1432                 if (dst) {
1433                         val = dst_mtu(dst);
1434                         dst_release(dst);
1435                 }
1436                 if (!val) {
1437                         release_sock(sk);
1438                         return -ENOTCONN;
1439                 }
1440                 break;
1441         }
1442         case IP_RECVERR:
1443                 val = inet->recverr;
1444                 break;
1445         case IP_MULTICAST_TTL:
1446                 val = inet->mc_ttl;
1447                 break;
1448         case IP_MULTICAST_LOOP:
1449                 val = inet->mc_loop;
1450                 break;
1451         case IP_UNICAST_IF:
1452                 val = (__force int)htonl((__u32) inet->uc_index);
1453                 break;
1454         case IP_MULTICAST_IF:
1455         {
1456                 struct in_addr addr;
1457                 len = min_t(unsigned int, len, sizeof(struct in_addr));
1458                 addr.s_addr = inet->mc_addr;
1459                 release_sock(sk);
1460 
1461                 if (put_user(len, optlen))
1462                         return -EFAULT;
1463                 if (copy_to_user(optval, &addr, len))
1464                         return -EFAULT;
1465                 return 0;
1466         }
1467         case IP_MSFILTER:
1468         {
1469                 struct ip_msfilter msf;
1470 
1471                 if (len < IP_MSFILTER_SIZE(0)) {
1472                         err = -EINVAL;
1473                         goto out;
1474                 }
1475                 if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) {
1476                         err = -EFAULT;
1477                         goto out;
1478                 }
1479                 err = ip_mc_msfget(sk, &msf,
1480                                    (struct ip_msfilter __user *)optval, optlen);
1481                 goto out;
1482         }
1483         case MCAST_MSFILTER:
1484         {
1485                 struct group_filter gsf;
1486 
1487                 if (len < GROUP_FILTER_SIZE(0)) {
1488                         err = -EINVAL;
1489                         goto out;
1490                 }
1491                 if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) {
1492                         err = -EFAULT;
1493                         goto out;
1494                 }
1495                 err = ip_mc_gsfget(sk, &gsf,
1496                                    (struct group_filter __user *)optval,
1497                                    optlen);
1498                 goto out;
1499         }
1500         case IP_MULTICAST_ALL:
1501                 val = inet->mc_all;
1502                 break;
1503         case IP_PKTOPTIONS:
1504         {
1505                 struct msghdr msg;
1506 
1507                 release_sock(sk);
1508 
1509                 if (sk->sk_type != SOCK_STREAM)
1510                         return -ENOPROTOOPT;
1511 
1512                 msg.msg_control = (__force void *) optval;
1513                 msg.msg_controllen = len;
1514                 msg.msg_flags = flags;
1515 
1516                 if (inet->cmsg_flags & IP_CMSG_PKTINFO) {
1517                         struct in_pktinfo info;
1518 
1519                         info.ipi_addr.s_addr = inet->inet_rcv_saddr;
1520                         info.ipi_spec_dst.s_addr = inet->inet_rcv_saddr;
1521                         info.ipi_ifindex = inet->mc_index;
1522                         put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
1523                 }
1524                 if (inet->cmsg_flags & IP_CMSG_TTL) {
1525                         int hlim = inet->mc_ttl;
1526                         put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim);
1527                 }
1528                 if (inet->cmsg_flags & IP_CMSG_TOS) {
1529                         int tos = inet->rcv_tos;
1530                         put_cmsg(&msg, SOL_IP, IP_TOS, sizeof(tos), &tos);
1531                 }
1532                 len -= msg.msg_controllen;
1533                 return put_user(len, optlen);
1534         }
1535         case IP_FREEBIND:
1536                 val = inet->freebind;
1537                 break;
1538         case IP_TRANSPARENT:
1539                 val = inet->transparent;
1540                 break;
1541         case IP_MINTTL:
1542                 val = inet->min_ttl;
1543                 break;
1544         default:
1545                 release_sock(sk);
1546                 return -ENOPROTOOPT;
1547         }
1548         release_sock(sk);
1549 
1550         if (len < sizeof(int) && len > 0 && val >= 0 && val <= 255) {
1551                 unsigned char ucval = (unsigned char)val;
1552                 len = 1;
1553                 if (put_user(len, optlen))
1554                         return -EFAULT;
1555                 if (copy_to_user(optval, &ucval, 1))
1556                         return -EFAULT;
1557         } else {
1558                 len = min_t(unsigned int, sizeof(int), len);
1559                 if (put_user(len, optlen))
1560                         return -EFAULT;
1561                 if (copy_to_user(optval, &val, len))
1562                         return -EFAULT;
1563         }
1564         return 0;
1565 
1566 out:
1567         release_sock(sk);
1568         if (needs_rtnl)
1569                 rtnl_unlock();
1570         return err;
1571 }
1572 
1573 int ip_getsockopt(struct sock *sk, int level,
1574                   int optname, char __user *optval, int __user *optlen)
1575 {
1576         int err;
1577 
1578         err = do_ip_getsockopt(sk, level, optname, optval, optlen, 0);
1579 #ifdef CONFIG_NETFILTER
1580         /* we need to exclude all possible ENOPROTOOPTs except default case */
1581         if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1582                         !ip_mroute_opt(optname)) {
1583                 int len;
1584 
1585                 if (get_user(len, optlen))
1586                         return -EFAULT;
1587 
1588                 lock_sock(sk);
1589                 err = nf_getsockopt(sk, PF_INET, optname, optval,
1590                                 &len);
1591                 release_sock(sk);
1592                 if (err >= 0)
1593                         err = put_user(len, optlen);
1594                 return err;
1595         }
1596 #endif
1597         return err;
1598 }
1599 EXPORT_SYMBOL(ip_getsockopt);
1600 
1601 #ifdef CONFIG_COMPAT
1602 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
1603                          char __user *optval, int __user *optlen)
1604 {
1605         int err;
1606 
1607         if (optname == MCAST_MSFILTER)
1608                 return compat_mc_getsockopt(sk, level, optname, optval, optlen,
1609                         ip_getsockopt);
1610 
1611         err = do_ip_getsockopt(sk, level, optname, optval, optlen,
1612                 MSG_CMSG_COMPAT);
1613 
1614 #ifdef CONFIG_NETFILTER
1615         /* we need to exclude all possible ENOPROTOOPTs except default case */
1616         if (err == -ENOPROTOOPT && optname != IP_PKTOPTIONS &&
1617                         !ip_mroute_opt(optname)) {
1618                 int len;
1619 
1620                 if (get_user(len, optlen))
1621                         return -EFAULT;
1622 
1623                 lock_sock(sk);
1624                 err = compat_nf_getsockopt(sk, PF_INET, optname, optval, &len);
1625                 release_sock(sk);
1626                 if (err >= 0)
1627                         err = put_user(len, optlen);
1628                 return err;
1629         }
1630 #endif
1631         return err;
1632 }
1633 EXPORT_SYMBOL(compat_ip_getsockopt);
1634 #endif
1635 

~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

kernel.org | git.kernel.org | LWN.net | Project Home | Wiki (Japanese) | Wiki (English) | SVN repository | Mail admin

Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.

osdn.jp