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

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

Version: ~ [ linux-5.3-rc5 ] ~ [ linux-5.2.9 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.67 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.139 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.189 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.189 ] ~ [ 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-3.16.72 ] ~ [ 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  *              RAW - implementation of IP "raw" sockets.
  7  *
  8  * Authors:     Ross Biro
  9  *              Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
 10  *
 11  * Fixes:
 12  *              Alan Cox        :       verify_area() fixed up
 13  *              Alan Cox        :       ICMP error handling
 14  *              Alan Cox        :       EMSGSIZE if you send too big a packet
 15  *              Alan Cox        :       Now uses generic datagrams and shared
 16  *                                      skbuff library. No more peek crashes,
 17  *                                      no more backlogs
 18  *              Alan Cox        :       Checks sk->broadcast.
 19  *              Alan Cox        :       Uses skb_free_datagram/skb_copy_datagram
 20  *              Alan Cox        :       Raw passes ip options too
 21  *              Alan Cox        :       Setsocketopt added
 22  *              Alan Cox        :       Fixed error return for broadcasts
 23  *              Alan Cox        :       Removed wake_up calls
 24  *              Alan Cox        :       Use ttl/tos
 25  *              Alan Cox        :       Cleaned up old debugging
 26  *              Alan Cox        :       Use new kernel side addresses
 27  *      Arnt Gulbrandsen        :       Fixed MSG_DONTROUTE in raw sockets.
 28  *              Alan Cox        :       BSD style RAW socket demultiplexing.
 29  *              Alan Cox        :       Beginnings of mrouted support.
 30  *              Alan Cox        :       Added IP_HDRINCL option.
 31  *              Alan Cox        :       Skip broadcast check if BSDism set.
 32  *              David S. Miller :       New socket lookup architecture.
 33  *
 34  *              This program is free software; you can redistribute it and/or
 35  *              modify it under the terms of the GNU General Public License
 36  *              as published by the Free Software Foundation; either version
 37  *              2 of the License, or (at your option) any later version.
 38  */
 39 
 40 #include <linux/types.h>
 41 #include <linux/atomic.h>
 42 #include <asm/byteorder.h>
 43 #include <asm/current.h>
 44 #include <asm/uaccess.h>
 45 #include <asm/ioctls.h>
 46 #include <linux/stddef.h>
 47 #include <linux/slab.h>
 48 #include <linux/errno.h>
 49 #include <linux/aio.h>
 50 #include <linux/kernel.h>
 51 #include <linux/export.h>
 52 #include <linux/spinlock.h>
 53 #include <linux/sockios.h>
 54 #include <linux/socket.h>
 55 #include <linux/in.h>
 56 #include <linux/mroute.h>
 57 #include <linux/netdevice.h>
 58 #include <linux/in_route.h>
 59 #include <linux/route.h>
 60 #include <linux/skbuff.h>
 61 #include <net/net_namespace.h>
 62 #include <net/dst.h>
 63 #include <net/sock.h>
 64 #include <linux/ip.h>
 65 #include <linux/net.h>
 66 #include <net/ip.h>
 67 #include <net/icmp.h>
 68 #include <net/udp.h>
 69 #include <net/raw.h>
 70 #include <net/snmp.h>
 71 #include <net/tcp_states.h>
 72 #include <net/inet_common.h>
 73 #include <net/checksum.h>
 74 #include <net/xfrm.h>
 75 #include <linux/rtnetlink.h>
 76 #include <linux/proc_fs.h>
 77 #include <linux/seq_file.h>
 78 #include <linux/netfilter.h>
 79 #include <linux/netfilter_ipv4.h>
 80 #include <linux/compat.h>
 81 
 82 static struct raw_hashinfo raw_v4_hashinfo = {
 83         .lock = __RW_LOCK_UNLOCKED(raw_v4_hashinfo.lock),
 84 };
 85 
 86 void raw_hash_sk(struct sock *sk)
 87 {
 88         struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
 89         struct hlist_head *head;
 90 
 91         head = &h->ht[inet_sk(sk)->inet_num & (RAW_HTABLE_SIZE - 1)];
 92 
 93         write_lock_bh(&h->lock);
 94         sk_add_node(sk, head);
 95         sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
 96         write_unlock_bh(&h->lock);
 97 }
 98 EXPORT_SYMBOL_GPL(raw_hash_sk);
 99 
100 void raw_unhash_sk(struct sock *sk)
101 {
102         struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
103 
104         write_lock_bh(&h->lock);
105         if (sk_del_node_init(sk))
106                 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
107         write_unlock_bh(&h->lock);
108 }
109 EXPORT_SYMBOL_GPL(raw_unhash_sk);
110 
111 static struct sock *__raw_v4_lookup(struct net *net, struct sock *sk,
112                 unsigned short num, __be32 raddr, __be32 laddr, int dif)
113 {
114         struct hlist_node *node;
115 
116         sk_for_each_from(sk, node) {
117                 struct inet_sock *inet = inet_sk(sk);
118 
119                 if (net_eq(sock_net(sk), net) && inet->inet_num == num  &&
120                     !(inet->inet_daddr && inet->inet_daddr != raddr)    &&
121                     !(inet->inet_rcv_saddr && inet->inet_rcv_saddr != laddr) &&
122                     !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
123                         goto found; /* gotcha */
124         }
125         sk = NULL;
126 found:
127         return sk;
128 }
129 
130 /*
131  *      0 - deliver
132  *      1 - block
133  */
134 static int icmp_filter(const struct sock *sk, const struct sk_buff *skb)
135 {
136         struct icmphdr _hdr;
137         const struct icmphdr *hdr;
138 
139         hdr = skb_header_pointer(skb, skb_transport_offset(skb),
140                                  sizeof(_hdr), &_hdr);
141         if (!hdr)
142                 return 1;
143 
144         if (hdr->type < 32) {
145                 __u32 data = raw_sk(sk)->filter.data;
146 
147                 return ((1U << hdr->type) & data) != 0;
148         }
149 
150         /* Do not block unknown ICMP types */
151         return 0;
152 }
153 
154 /* IP input processing comes here for RAW socket delivery.
155  * Caller owns SKB, so we must make clones.
156  *
157  * RFC 1122: SHOULD pass TOS value up to the transport layer.
158  * -> It does. And not only TOS, but all IP header.
159  */
160 static int raw_v4_input(struct sk_buff *skb, const struct iphdr *iph, int hash)
161 {
162         struct sock *sk;
163         struct hlist_head *head;
164         int delivered = 0;
165         struct net *net;
166 
167         read_lock(&raw_v4_hashinfo.lock);
168         head = &raw_v4_hashinfo.ht[hash];
169         if (hlist_empty(head))
170                 goto out;
171 
172         net = dev_net(skb->dev);
173         sk = __raw_v4_lookup(net, __sk_head(head), iph->protocol,
174                              iph->saddr, iph->daddr,
175                              skb->dev->ifindex);
176 
177         while (sk) {
178                 delivered = 1;
179                 if (iph->protocol != IPPROTO_ICMP || !icmp_filter(sk, skb)) {
180                         struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
181 
182                         /* Not releasing hash table! */
183                         if (clone)
184                                 raw_rcv(sk, clone);
185                 }
186                 sk = __raw_v4_lookup(net, sk_next(sk), iph->protocol,
187                                      iph->saddr, iph->daddr,
188                                      skb->dev->ifindex);
189         }
190 out:
191         read_unlock(&raw_v4_hashinfo.lock);
192         return delivered;
193 }
194 
195 int raw_local_deliver(struct sk_buff *skb, int protocol)
196 {
197         int hash;
198         struct sock *raw_sk;
199 
200         hash = protocol & (RAW_HTABLE_SIZE - 1);
201         raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
202 
203         /* If there maybe a raw socket we must check - if not we
204          * don't care less
205          */
206         if (raw_sk && !raw_v4_input(skb, ip_hdr(skb), hash))
207                 raw_sk = NULL;
208 
209         return raw_sk != NULL;
210 
211 }
212 
213 static void raw_err(struct sock *sk, struct sk_buff *skb, u32 info)
214 {
215         struct inet_sock *inet = inet_sk(sk);
216         const int type = icmp_hdr(skb)->type;
217         const int code = icmp_hdr(skb)->code;
218         int err = 0;
219         int harderr = 0;
220 
221         /* Report error on raw socket, if:
222            1. User requested ip_recverr.
223            2. Socket is connected (otherwise the error indication
224               is useless without ip_recverr and error is hard.
225          */
226         if (!inet->recverr && sk->sk_state != TCP_ESTABLISHED)
227                 return;
228 
229         switch (type) {
230         default:
231         case ICMP_TIME_EXCEEDED:
232                 err = EHOSTUNREACH;
233                 break;
234         case ICMP_SOURCE_QUENCH:
235                 return;
236         case ICMP_PARAMETERPROB:
237                 err = EPROTO;
238                 harderr = 1;
239                 break;
240         case ICMP_DEST_UNREACH:
241                 err = EHOSTUNREACH;
242                 if (code > NR_ICMP_UNREACH)
243                         break;
244                 err = icmp_err_convert[code].errno;
245                 harderr = icmp_err_convert[code].fatal;
246                 if (code == ICMP_FRAG_NEEDED) {
247                         harderr = inet->pmtudisc != IP_PMTUDISC_DONT;
248                         err = EMSGSIZE;
249                 }
250         }
251 
252         if (inet->recverr) {
253                 const struct iphdr *iph = (const struct iphdr *)skb->data;
254                 u8 *payload = skb->data + (iph->ihl << 2);
255 
256                 if (inet->hdrincl)
257                         payload = skb->data;
258                 ip_icmp_error(sk, skb, err, 0, info, payload);
259         }
260 
261         if (inet->recverr || harderr) {
262                 sk->sk_err = err;
263                 sk->sk_error_report(sk);
264         }
265 }
266 
267 void raw_icmp_error(struct sk_buff *skb, int protocol, u32 info)
268 {
269         int hash;
270         struct sock *raw_sk;
271         const struct iphdr *iph;
272         struct net *net;
273 
274         hash = protocol & (RAW_HTABLE_SIZE - 1);
275 
276         read_lock(&raw_v4_hashinfo.lock);
277         raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
278         if (raw_sk != NULL) {
279                 iph = (const struct iphdr *)skb->data;
280                 net = dev_net(skb->dev);
281 
282                 while ((raw_sk = __raw_v4_lookup(net, raw_sk, protocol,
283                                                 iph->daddr, iph->saddr,
284                                                 skb->dev->ifindex)) != NULL) {
285                         raw_err(raw_sk, skb, info);
286                         raw_sk = sk_next(raw_sk);
287                         iph = (const struct iphdr *)skb->data;
288                 }
289         }
290         read_unlock(&raw_v4_hashinfo.lock);
291 }
292 
293 static int raw_rcv_skb(struct sock * sk, struct sk_buff * skb)
294 {
295         /* Charge it to the socket. */
296 
297         ipv4_pktinfo_prepare(skb);
298         if (sock_queue_rcv_skb(sk, skb) < 0) {
299                 kfree_skb(skb);
300                 return NET_RX_DROP;
301         }
302 
303         return NET_RX_SUCCESS;
304 }
305 
306 int raw_rcv(struct sock *sk, struct sk_buff *skb)
307 {
308         if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
309                 atomic_inc(&sk->sk_drops);
310                 kfree_skb(skb);
311                 return NET_RX_DROP;
312         }
313         nf_reset(skb);
314 
315         skb_push(skb, skb->data - skb_network_header(skb));
316 
317         raw_rcv_skb(sk, skb);
318         return 0;
319 }
320 
321 static int raw_send_hdrinc(struct sock *sk, struct flowi4 *fl4,
322                            void *from, size_t length,
323                            struct rtable **rtp,
324                            unsigned int flags)
325 {
326         struct inet_sock *inet = inet_sk(sk);
327         struct net *net = sock_net(sk);
328         struct iphdr *iph;
329         struct sk_buff *skb;
330         unsigned int iphlen;
331         int err;
332         struct rtable *rt = *rtp;
333         int hlen, tlen;
334 
335         if (length > rt->dst.dev->mtu) {
336                 ip_local_error(sk, EMSGSIZE, fl4->daddr, inet->inet_dport,
337                                rt->dst.dev->mtu);
338                 return -EMSGSIZE;
339         }
340         if (flags&MSG_PROBE)
341                 goto out;
342 
343         hlen = LL_RESERVED_SPACE(rt->dst.dev);
344         tlen = rt->dst.dev->needed_tailroom;
345         skb = sock_alloc_send_skb(sk,
346                                   length + hlen + tlen + 15,
347                                   flags & MSG_DONTWAIT, &err);
348         if (skb == NULL)
349                 goto error;
350         skb_reserve(skb, hlen);
351 
352         skb->priority = sk->sk_priority;
353         skb->mark = sk->sk_mark;
354         skb_dst_set(skb, &rt->dst);
355         *rtp = NULL;
356 
357         skb_reset_network_header(skb);
358         iph = ip_hdr(skb);
359         skb_put(skb, length);
360 
361         skb->ip_summed = CHECKSUM_NONE;
362 
363         skb->transport_header = skb->network_header;
364         err = -EFAULT;
365         if (memcpy_fromiovecend((void *)iph, from, 0, length))
366                 goto error_free;
367 
368         iphlen = iph->ihl * 4;
369 
370         /*
371          * We don't want to modify the ip header, but we do need to
372          * be sure that it won't cause problems later along the network
373          * stack.  Specifically we want to make sure that iph->ihl is a
374          * sane value.  If ihl points beyond the length of the buffer passed
375          * in, reject the frame as invalid
376          */
377         err = -EINVAL;
378         if (iphlen > length)
379                 goto error_free;
380 
381         if (iphlen >= sizeof(*iph)) {
382                 if (!iph->saddr)
383                         iph->saddr = fl4->saddr;
384                 iph->check   = 0;
385                 iph->tot_len = htons(length);
386                 if (!iph->id)
387                         ip_select_ident(skb, NULL);
388 
389                 iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
390         }
391         if (iph->protocol == IPPROTO_ICMP)
392                 icmp_out_count(net, ((struct icmphdr *)
393                         skb_transport_header(skb))->type);
394 
395         err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT, skb, NULL,
396                       rt->dst.dev, dst_output);
397         if (err > 0)
398                 err = net_xmit_errno(err);
399         if (err)
400                 goto error;
401 out:
402         return 0;
403 
404 error_free:
405         kfree_skb(skb);
406 error:
407         IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
408         if (err == -ENOBUFS && !inet->recverr)
409                 err = 0;
410         return err;
411 }
412 
413 static int raw_probe_proto_opt(struct flowi4 *fl4, struct msghdr *msg)
414 {
415         struct iovec *iov;
416         u8 __user *type = NULL;
417         u8 __user *code = NULL;
418         int probed = 0;
419         unsigned int i;
420 
421         if (!msg->msg_iov)
422                 return 0;
423 
424         for (i = 0; i < msg->msg_iovlen; i++) {
425                 iov = &msg->msg_iov[i];
426                 if (!iov)
427                         continue;
428 
429                 switch (fl4->flowi4_proto) {
430                 case IPPROTO_ICMP:
431                         /* check if one-byte field is readable or not. */
432                         if (iov->iov_base && iov->iov_len < 1)
433                                 break;
434 
435                         if (!type) {
436                                 type = iov->iov_base;
437                                 /* check if code field is readable or not. */
438                                 if (iov->iov_len > 1)
439                                         code = type + 1;
440                         } else if (!code)
441                                 code = iov->iov_base;
442 
443                         if (type && code) {
444                                 if (get_user(fl4->fl4_icmp_type, type) ||
445                                     get_user(fl4->fl4_icmp_code, code))
446                                         return -EFAULT;
447                                 probed = 1;
448                         }
449                         break;
450                 default:
451                         probed = 1;
452                         break;
453                 }
454                 if (probed)
455                         break;
456         }
457         return 0;
458 }
459 
460 static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
461                        size_t len)
462 {
463         struct inet_sock *inet = inet_sk(sk);
464         struct ipcm_cookie ipc;
465         struct rtable *rt = NULL;
466         struct flowi4 fl4;
467         int free = 0;
468         __be32 daddr;
469         __be32 saddr;
470         u8  tos;
471         int err;
472         struct ip_options_data opt_copy;
473 
474         err = -EMSGSIZE;
475         if (len > 0xFFFF)
476                 goto out;
477 
478         /*
479          *      Check the flags.
480          */
481 
482         err = -EOPNOTSUPP;
483         if (msg->msg_flags & MSG_OOB)   /* Mirror BSD error message */
484                 goto out;               /* compatibility */
485 
486         /*
487          *      Get and verify the address.
488          */
489 
490         if (msg->msg_namelen) {
491                 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
492                 err = -EINVAL;
493                 if (msg->msg_namelen < sizeof(*usin))
494                         goto out;
495                 if (usin->sin_family != AF_INET) {
496                         pr_info_once("%s: %s forgot to set AF_INET. Fix it!\n",
497                                      __func__, current->comm);
498                         err = -EAFNOSUPPORT;
499                         if (usin->sin_family)
500                                 goto out;
501                 }
502                 daddr = usin->sin_addr.s_addr;
503                 /* ANK: I did not forget to get protocol from port field.
504                  * I just do not know, who uses this weirdness.
505                  * IP_HDRINCL is much more convenient.
506                  */
507         } else {
508                 err = -EDESTADDRREQ;
509                 if (sk->sk_state != TCP_ESTABLISHED)
510                         goto out;
511                 daddr = inet->inet_daddr;
512         }
513 
514         ipc.addr = inet->inet_saddr;
515         ipc.opt = NULL;
516         ipc.tx_flags = 0;
517         ipc.oif = sk->sk_bound_dev_if;
518 
519         if (msg->msg_controllen) {
520                 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
521                 if (err)
522                         goto out;
523                 if (ipc.opt)
524                         free = 1;
525         }
526 
527         saddr = ipc.addr;
528         ipc.addr = daddr;
529 
530         if (!ipc.opt) {
531                 struct ip_options_rcu *inet_opt;
532 
533                 rcu_read_lock();
534                 inet_opt = rcu_dereference(inet->inet_opt);
535                 if (inet_opt) {
536                         memcpy(&opt_copy, inet_opt,
537                                sizeof(*inet_opt) + inet_opt->opt.optlen);
538                         ipc.opt = &opt_copy.opt;
539                 }
540                 rcu_read_unlock();
541         }
542 
543         if (ipc.opt) {
544                 err = -EINVAL;
545                 /* Linux does not mangle headers on raw sockets,
546                  * so that IP options + IP_HDRINCL is non-sense.
547                  */
548                 if (inet->hdrincl)
549                         goto done;
550                 if (ipc.opt->opt.srr) {
551                         if (!daddr)
552                                 goto done;
553                         daddr = ipc.opt->opt.faddr;
554                 }
555         }
556         tos = RT_CONN_FLAGS(sk);
557         if (msg->msg_flags & MSG_DONTROUTE)
558                 tos |= RTO_ONLINK;
559 
560         if (ipv4_is_multicast(daddr)) {
561                 if (!ipc.oif)
562                         ipc.oif = inet->mc_index;
563                 if (!saddr)
564                         saddr = inet->mc_addr;
565         } else if (!ipc.oif)
566                 ipc.oif = inet->uc_index;
567 
568         flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
569                            RT_SCOPE_UNIVERSE,
570                            inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol,
571                            inet_sk_flowi_flags(sk) | FLOWI_FLAG_CAN_SLEEP,
572                            daddr, saddr, 0, 0);
573 
574         if (!inet->hdrincl) {
575                 err = raw_probe_proto_opt(&fl4, msg);
576                 if (err)
577                         goto done;
578         }
579 
580         security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
581         rt = ip_route_output_flow(sock_net(sk), &fl4, sk);
582         if (IS_ERR(rt)) {
583                 err = PTR_ERR(rt);
584                 rt = NULL;
585                 goto done;
586         }
587 
588         err = -EACCES;
589         if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST))
590                 goto done;
591 
592         if (msg->msg_flags & MSG_CONFIRM)
593                 goto do_confirm;
594 back_from_confirm:
595 
596         if (inet->hdrincl)
597                 err = raw_send_hdrinc(sk, &fl4, msg->msg_iov, len,
598                                       &rt, msg->msg_flags);
599 
600          else {
601                 if (!ipc.addr)
602                         ipc.addr = fl4.daddr;
603                 lock_sock(sk);
604                 err = ip_append_data(sk, &fl4, ip_generic_getfrag,
605                                      msg->msg_iov, len, 0,
606                                      &ipc, &rt, msg->msg_flags);
607                 if (err)
608                         ip_flush_pending_frames(sk);
609                 else if (!(msg->msg_flags & MSG_MORE)) {
610                         err = ip_push_pending_frames(sk, &fl4);
611                         if (err == -ENOBUFS && !inet->recverr)
612                                 err = 0;
613                 }
614                 release_sock(sk);
615         }
616 done:
617         if (free)
618                 kfree(ipc.opt);
619         ip_rt_put(rt);
620 
621 out:
622         if (err < 0)
623                 return err;
624         return len;
625 
626 do_confirm:
627         dst_confirm(&rt->dst);
628         if (!(msg->msg_flags & MSG_PROBE) || len)
629                 goto back_from_confirm;
630         err = 0;
631         goto done;
632 }
633 
634 static void raw_close(struct sock *sk, long timeout)
635 {
636         /*
637          * Raw sockets may have direct kernel references. Kill them.
638          */
639         ip_ra_control(sk, 0, NULL);
640 
641         sk_common_release(sk);
642 }
643 
644 static void raw_destroy(struct sock *sk)
645 {
646         lock_sock(sk);
647         ip_flush_pending_frames(sk);
648         release_sock(sk);
649 }
650 
651 /* This gets rid of all the nasties in af_inet. -DaveM */
652 static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
653 {
654         struct inet_sock *inet = inet_sk(sk);
655         struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
656         int ret = -EINVAL;
657         int chk_addr_ret;
658 
659         if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
660                 goto out;
661         chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
662         ret = -EADDRNOTAVAIL;
663         if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
664             chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
665                 goto out;
666         inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
667         if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
668                 inet->inet_saddr = 0;  /* Use device */
669         sk_dst_reset(sk);
670         ret = 0;
671 out:    return ret;
672 }
673 
674 /*
675  *      This should be easy, if there is something there
676  *      we return it, otherwise we block.
677  */
678 
679 static int raw_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
680                        size_t len, int noblock, int flags, int *addr_len)
681 {
682         struct inet_sock *inet = inet_sk(sk);
683         size_t copied = 0;
684         int err = -EOPNOTSUPP;
685         struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
686         struct sk_buff *skb;
687 
688         if (flags & MSG_OOB)
689                 goto out;
690 
691         if (flags & MSG_ERRQUEUE) {
692                 err = ip_recv_error(sk, msg, len, addr_len);
693                 goto out;
694         }
695 
696         skb = skb_recv_datagram(sk, flags, noblock, &err);
697         if (!skb)
698                 goto out;
699         if (ccs_socket_post_recvmsg_permission(sk, skb, flags)) {
700                 err = -EAGAIN; /* Hope less harmful than -EPERM. */
701                 goto out;
702         }
703 
704         copied = skb->len;
705         if (len < copied) {
706                 msg->msg_flags |= MSG_TRUNC;
707                 copied = len;
708         }
709 
710         err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
711         if (err)
712                 goto done;
713 
714         sock_recv_ts_and_drops(msg, sk, skb);
715 
716         /* Copy the address. */
717         if (sin) {
718                 sin->sin_family = AF_INET;
719                 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
720                 sin->sin_port = 0;
721                 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
722                 *addr_len = sizeof(*sin);
723         }
724         if (inet->cmsg_flags)
725                 ip_cmsg_recv(msg, skb);
726         if (flags & MSG_TRUNC)
727                 copied = skb->len;
728 done:
729         skb_free_datagram(sk, skb);
730 out:
731         if (err)
732                 return err;
733         return copied;
734 }
735 
736 static int raw_init(struct sock *sk)
737 {
738         struct raw_sock *rp = raw_sk(sk);
739 
740         if (inet_sk(sk)->inet_num == IPPROTO_ICMP)
741                 memset(&rp->filter, 0, sizeof(rp->filter));
742         return 0;
743 }
744 
745 static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen)
746 {
747         if (optlen > sizeof(struct icmp_filter))
748                 optlen = sizeof(struct icmp_filter);
749         if (copy_from_user(&raw_sk(sk)->filter, optval, optlen))
750                 return -EFAULT;
751         return 0;
752 }
753 
754 static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen)
755 {
756         int len, ret = -EFAULT;
757 
758         if (get_user(len, optlen))
759                 goto out;
760         ret = -EINVAL;
761         if (len < 0)
762                 goto out;
763         if (len > sizeof(struct icmp_filter))
764                 len = sizeof(struct icmp_filter);
765         ret = -EFAULT;
766         if (put_user(len, optlen) ||
767             copy_to_user(optval, &raw_sk(sk)->filter, len))
768                 goto out;
769         ret = 0;
770 out:    return ret;
771 }
772 
773 static int do_raw_setsockopt(struct sock *sk, int level, int optname,
774                           char __user *optval, unsigned int optlen)
775 {
776         if (optname == ICMP_FILTER) {
777                 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
778                         return -EOPNOTSUPP;
779                 else
780                         return raw_seticmpfilter(sk, optval, optlen);
781         }
782         return -ENOPROTOOPT;
783 }
784 
785 static int raw_setsockopt(struct sock *sk, int level, int optname,
786                           char __user *optval, unsigned int optlen)
787 {
788         if (level != SOL_RAW)
789                 return ip_setsockopt(sk, level, optname, optval, optlen);
790         return do_raw_setsockopt(sk, level, optname, optval, optlen);
791 }
792 
793 #ifdef CONFIG_COMPAT
794 static int compat_raw_setsockopt(struct sock *sk, int level, int optname,
795                                  char __user *optval, unsigned int optlen)
796 {
797         if (level != SOL_RAW)
798                 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
799         return do_raw_setsockopt(sk, level, optname, optval, optlen);
800 }
801 #endif
802 
803 static int do_raw_getsockopt(struct sock *sk, int level, int optname,
804                           char __user *optval, int __user *optlen)
805 {
806         if (optname == ICMP_FILTER) {
807                 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
808                         return -EOPNOTSUPP;
809                 else
810                         return raw_geticmpfilter(sk, optval, optlen);
811         }
812         return -ENOPROTOOPT;
813 }
814 
815 static int raw_getsockopt(struct sock *sk, int level, int optname,
816                           char __user *optval, int __user *optlen)
817 {
818         if (level != SOL_RAW)
819                 return ip_getsockopt(sk, level, optname, optval, optlen);
820         return do_raw_getsockopt(sk, level, optname, optval, optlen);
821 }
822 
823 #ifdef CONFIG_COMPAT
824 static int compat_raw_getsockopt(struct sock *sk, int level, int optname,
825                                  char __user *optval, int __user *optlen)
826 {
827         if (level != SOL_RAW)
828                 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
829         return do_raw_getsockopt(sk, level, optname, optval, optlen);
830 }
831 #endif
832 
833 static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg)
834 {
835         switch (cmd) {
836         case SIOCOUTQ: {
837                 int amount = sk_wmem_alloc_get(sk);
838 
839                 return put_user(amount, (int __user *)arg);
840         }
841         case SIOCINQ: {
842                 struct sk_buff *skb;
843                 int amount = 0;
844 
845                 spin_lock_bh(&sk->sk_receive_queue.lock);
846                 skb = skb_peek(&sk->sk_receive_queue);
847                 if (skb != NULL)
848                         amount = skb->len;
849                 spin_unlock_bh(&sk->sk_receive_queue.lock);
850                 return put_user(amount, (int __user *)arg);
851         }
852 
853         default:
854 #ifdef CONFIG_IP_MROUTE
855                 return ipmr_ioctl(sk, cmd, (void __user *)arg);
856 #else
857                 return -ENOIOCTLCMD;
858 #endif
859         }
860 }
861 
862 #ifdef CONFIG_COMPAT
863 static int compat_raw_ioctl(struct sock *sk, unsigned int cmd, unsigned long arg)
864 {
865         switch (cmd) {
866         case SIOCOUTQ:
867         case SIOCINQ:
868                 return -ENOIOCTLCMD;
869         default:
870 #ifdef CONFIG_IP_MROUTE
871                 return ipmr_compat_ioctl(sk, cmd, compat_ptr(arg));
872 #else
873                 return -ENOIOCTLCMD;
874 #endif
875         }
876 }
877 #endif
878 
879 struct proto raw_prot = {
880         .name              = "RAW",
881         .owner             = THIS_MODULE,
882         .close             = raw_close,
883         .destroy           = raw_destroy,
884         .connect           = ip4_datagram_connect,
885         .disconnect        = udp_disconnect,
886         .ioctl             = raw_ioctl,
887         .init              = raw_init,
888         .setsockopt        = raw_setsockopt,
889         .getsockopt        = raw_getsockopt,
890         .sendmsg           = raw_sendmsg,
891         .recvmsg           = raw_recvmsg,
892         .bind              = raw_bind,
893         .backlog_rcv       = raw_rcv_skb,
894         .hash              = raw_hash_sk,
895         .unhash            = raw_unhash_sk,
896         .obj_size          = sizeof(struct raw_sock),
897         .h.raw_hash        = &raw_v4_hashinfo,
898 #ifdef CONFIG_COMPAT
899         .compat_setsockopt = compat_raw_setsockopt,
900         .compat_getsockopt = compat_raw_getsockopt,
901         .compat_ioctl      = compat_raw_ioctl,
902 #endif
903 };
904 
905 #ifdef CONFIG_PROC_FS
906 static struct sock *raw_get_first(struct seq_file *seq)
907 {
908         struct sock *sk;
909         struct raw_iter_state *state = raw_seq_private(seq);
910 
911         for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE;
912                         ++state->bucket) {
913                 struct hlist_node *node;
914 
915                 sk_for_each(sk, node, &state->h->ht[state->bucket])
916                         if (sock_net(sk) == seq_file_net(seq))
917                                 goto found;
918         }
919         sk = NULL;
920 found:
921         return sk;
922 }
923 
924 static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk)
925 {
926         struct raw_iter_state *state = raw_seq_private(seq);
927 
928         do {
929                 sk = sk_next(sk);
930 try_again:
931                 ;
932         } while (sk && sock_net(sk) != seq_file_net(seq));
933 
934         if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
935                 sk = sk_head(&state->h->ht[state->bucket]);
936                 goto try_again;
937         }
938         return sk;
939 }
940 
941 static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos)
942 {
943         struct sock *sk = raw_get_first(seq);
944 
945         if (sk)
946                 while (pos && (sk = raw_get_next(seq, sk)) != NULL)
947                         --pos;
948         return pos ? NULL : sk;
949 }
950 
951 void *raw_seq_start(struct seq_file *seq, loff_t *pos)
952 {
953         struct raw_iter_state *state = raw_seq_private(seq);
954 
955         read_lock(&state->h->lock);
956         return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
957 }
958 EXPORT_SYMBOL_GPL(raw_seq_start);
959 
960 void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos)
961 {
962         struct sock *sk;
963 
964         if (v == SEQ_START_TOKEN)
965                 sk = raw_get_first(seq);
966         else
967                 sk = raw_get_next(seq, v);
968         ++*pos;
969         return sk;
970 }
971 EXPORT_SYMBOL_GPL(raw_seq_next);
972 
973 void raw_seq_stop(struct seq_file *seq, void *v)
974 {
975         struct raw_iter_state *state = raw_seq_private(seq);
976 
977         read_unlock(&state->h->lock);
978 }
979 EXPORT_SYMBOL_GPL(raw_seq_stop);
980 
981 static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
982 {
983         struct inet_sock *inet = inet_sk(sp);
984         __be32 dest = inet->inet_daddr,
985                src = inet->inet_rcv_saddr;
986         __u16 destp = 0,
987               srcp  = inet->inet_num;
988 
989         seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
990                 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d\n",
991                 i, src, srcp, dest, destp, sp->sk_state,
992                 sk_wmem_alloc_get(sp),
993                 sk_rmem_alloc_get(sp),
994                 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
995                 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
996 }
997 
998 static int raw_seq_show(struct seq_file *seq, void *v)
999 {
1000         if (v == SEQ_START_TOKEN)
1001                 seq_printf(seq, "  sl  local_address rem_address   st tx_queue "
1002                                 "rx_queue tr tm->when retrnsmt   uid  timeout "
1003                                 "inode ref pointer drops\n");
1004         else
1005                 raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
1006         return 0;
1007 }
1008 
1009 static const struct seq_operations raw_seq_ops = {
1010         .start = raw_seq_start,
1011         .next  = raw_seq_next,
1012         .stop  = raw_seq_stop,
1013         .show  = raw_seq_show,
1014 };
1015 
1016 int raw_seq_open(struct inode *ino, struct file *file,
1017                  struct raw_hashinfo *h, const struct seq_operations *ops)
1018 {
1019         int err;
1020         struct raw_iter_state *i;
1021 
1022         err = seq_open_net(ino, file, ops, sizeof(struct raw_iter_state));
1023         if (err < 0)
1024                 return err;
1025 
1026         i = raw_seq_private((struct seq_file *)file->private_data);
1027         i->h = h;
1028         return 0;
1029 }
1030 EXPORT_SYMBOL_GPL(raw_seq_open);
1031 
1032 static int raw_v4_seq_open(struct inode *inode, struct file *file)
1033 {
1034         return raw_seq_open(inode, file, &raw_v4_hashinfo, &raw_seq_ops);
1035 }
1036 
1037 static const struct file_operations raw_seq_fops = {
1038         .owner   = THIS_MODULE,
1039         .open    = raw_v4_seq_open,
1040         .read    = seq_read,
1041         .llseek  = seq_lseek,
1042         .release = seq_release_net,
1043 };
1044 
1045 static __net_init int raw_init_net(struct net *net)
1046 {
1047         if (!proc_net_fops_create(net, "raw", S_IRUGO, &raw_seq_fops))
1048                 return -ENOMEM;
1049 
1050         return 0;
1051 }
1052 
1053 static __net_exit void raw_exit_net(struct net *net)
1054 {
1055         proc_net_remove(net, "raw");
1056 }
1057 
1058 static __net_initdata struct pernet_operations raw_net_ops = {
1059         .init = raw_init_net,
1060         .exit = raw_exit_net,
1061 };
1062 
1063 int __init raw_proc_init(void)
1064 {
1065         return register_pernet_subsys(&raw_net_ops);
1066 }
1067 
1068 void __init raw_proc_exit(void)
1069 {
1070         unregister_pernet_subsys(&raw_net_ops);
1071 }
1072 #endif /* CONFIG_PROC_FS */
1073 

~ [ 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