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

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

Version: ~ [ linux-5.9 ] ~ [ linux-5.8.14 ] ~ [ linux-5.7.19 ] ~ [ linux-5.6.19 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.70 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.150 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.200 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.238 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.238 ] ~ [ 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-3.16.85 ] ~ [ 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-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 <linux/igmp.h>
 62 #include <net/net_namespace.h>
 63 #include <net/dst.h>
 64 #include <net/sock.h>
 65 #include <linux/ip.h>
 66 #include <linux/net.h>
 67 #include <net/ip.h>
 68 #include <net/icmp.h>
 69 #include <net/udp.h>
 70 #include <net/raw.h>
 71 #include <net/snmp.h>
 72 #include <net/tcp_states.h>
 73 #include <net/inet_common.h>
 74 #include <net/checksum.h>
 75 #include <net/xfrm.h>
 76 #include <linux/rtnetlink.h>
 77 #include <linux/proc_fs.h>
 78 #include <linux/seq_file.h>
 79 #include <linux/netfilter.h>
 80 #include <linux/netfilter_ipv4.h>
 81 #include <linux/compat.h>
 82 #include <linux/uio.h>
 83 
 84 struct raw_frag_vec {
 85         struct msghdr *msg;
 86         union {
 87                 struct icmphdr icmph;
 88                 char c[1];
 89         } hdr;
 90         int hlen;
 91 };
 92 
 93 static struct raw_hashinfo raw_v4_hashinfo = {
 94         .lock = __RW_LOCK_UNLOCKED(raw_v4_hashinfo.lock),
 95 };
 96 
 97 void raw_hash_sk(struct sock *sk)
 98 {
 99         struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
100         struct hlist_head *head;
101 
102         head = &h->ht[inet_sk(sk)->inet_num & (RAW_HTABLE_SIZE - 1)];
103 
104         write_lock_bh(&h->lock);
105         sk_add_node(sk, head);
106         sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
107         write_unlock_bh(&h->lock);
108 }
109 EXPORT_SYMBOL_GPL(raw_hash_sk);
110 
111 void raw_unhash_sk(struct sock *sk)
112 {
113         struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
114 
115         write_lock_bh(&h->lock);
116         if (sk_del_node_init(sk))
117                 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
118         write_unlock_bh(&h->lock);
119 }
120 EXPORT_SYMBOL_GPL(raw_unhash_sk);
121 
122 static struct sock *__raw_v4_lookup(struct net *net, struct sock *sk,
123                 unsigned short num, __be32 raddr, __be32 laddr, int dif)
124 {
125         sk_for_each_from(sk) {
126                 struct inet_sock *inet = inet_sk(sk);
127 
128                 if (net_eq(sock_net(sk), net) && inet->inet_num == num  &&
129                     !(inet->inet_daddr && inet->inet_daddr != raddr)    &&
130                     !(inet->inet_rcv_saddr && inet->inet_rcv_saddr != laddr) &&
131                     !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
132                         goto found; /* gotcha */
133         }
134         sk = NULL;
135 found:
136         return sk;
137 }
138 
139 /*
140  *      0 - deliver
141  *      1 - block
142  */
143 static int icmp_filter(const struct sock *sk, const struct sk_buff *skb)
144 {
145         struct icmphdr _hdr;
146         const struct icmphdr *hdr;
147 
148         hdr = skb_header_pointer(skb, skb_transport_offset(skb),
149                                  sizeof(_hdr), &_hdr);
150         if (!hdr)
151                 return 1;
152 
153         if (hdr->type < 32) {
154                 __u32 data = raw_sk(sk)->filter.data;
155 
156                 return ((1U << hdr->type) & data) != 0;
157         }
158 
159         /* Do not block unknown ICMP types */
160         return 0;
161 }
162 
163 /* IP input processing comes here for RAW socket delivery.
164  * Caller owns SKB, so we must make clones.
165  *
166  * RFC 1122: SHOULD pass TOS value up to the transport layer.
167  * -> It does. And not only TOS, but all IP header.
168  */
169 static int raw_v4_input(struct sk_buff *skb, const struct iphdr *iph, int hash)
170 {
171         struct sock *sk;
172         struct hlist_head *head;
173         int delivered = 0;
174         struct net *net;
175 
176         read_lock(&raw_v4_hashinfo.lock);
177         head = &raw_v4_hashinfo.ht[hash];
178         if (hlist_empty(head))
179                 goto out;
180 
181         net = dev_net(skb->dev);
182         sk = __raw_v4_lookup(net, __sk_head(head), iph->protocol,
183                              iph->saddr, iph->daddr,
184                              skb->dev->ifindex);
185 
186         while (sk) {
187                 delivered = 1;
188                 if ((iph->protocol != IPPROTO_ICMP || !icmp_filter(sk, skb)) &&
189                     ip_mc_sf_allow(sk, iph->daddr, iph->saddr,
190                                    skb->dev->ifindex)) {
191                         struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
192 
193                         /* Not releasing hash table! */
194                         if (clone)
195                                 raw_rcv(sk, clone);
196                 }
197                 sk = __raw_v4_lookup(net, sk_next(sk), iph->protocol,
198                                      iph->saddr, iph->daddr,
199                                      skb->dev->ifindex);
200         }
201 out:
202         read_unlock(&raw_v4_hashinfo.lock);
203         return delivered;
204 }
205 
206 int raw_local_deliver(struct sk_buff *skb, int protocol)
207 {
208         int hash;
209         struct sock *raw_sk;
210 
211         hash = protocol & (RAW_HTABLE_SIZE - 1);
212         raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
213 
214         /* If there maybe a raw socket we must check - if not we
215          * don't care less
216          */
217         if (raw_sk && !raw_v4_input(skb, ip_hdr(skb), hash))
218                 raw_sk = NULL;
219 
220         return raw_sk != NULL;
221 
222 }
223 
224 static void raw_err(struct sock *sk, struct sk_buff *skb, u32 info)
225 {
226         struct inet_sock *inet = inet_sk(sk);
227         const int type = icmp_hdr(skb)->type;
228         const int code = icmp_hdr(skb)->code;
229         int err = 0;
230         int harderr = 0;
231 
232         if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED)
233                 ipv4_sk_update_pmtu(skb, sk, info);
234         else if (type == ICMP_REDIRECT) {
235                 ipv4_sk_redirect(skb, sk);
236                 return;
237         }
238 
239         /* Report error on raw socket, if:
240            1. User requested ip_recverr.
241            2. Socket is connected (otherwise the error indication
242               is useless without ip_recverr and error is hard.
243          */
244         if (!inet->recverr && sk->sk_state != TCP_ESTABLISHED)
245                 return;
246 
247         switch (type) {
248         default:
249         case ICMP_TIME_EXCEEDED:
250                 err = EHOSTUNREACH;
251                 break;
252         case ICMP_SOURCE_QUENCH:
253                 return;
254         case ICMP_PARAMETERPROB:
255                 err = EPROTO;
256                 harderr = 1;
257                 break;
258         case ICMP_DEST_UNREACH:
259                 err = EHOSTUNREACH;
260                 if (code > NR_ICMP_UNREACH)
261                         break;
262                 err = icmp_err_convert[code].errno;
263                 harderr = icmp_err_convert[code].fatal;
264                 if (code == ICMP_FRAG_NEEDED) {
265                         harderr = inet->pmtudisc != IP_PMTUDISC_DONT;
266                         err = EMSGSIZE;
267                 }
268         }
269 
270         if (inet->recverr) {
271                 const struct iphdr *iph = (const struct iphdr *)skb->data;
272                 u8 *payload = skb->data + (iph->ihl << 2);
273 
274                 if (inet->hdrincl)
275                         payload = skb->data;
276                 ip_icmp_error(sk, skb, err, 0, info, payload);
277         }
278 
279         if (inet->recverr || harderr) {
280                 sk->sk_err = err;
281                 sk->sk_error_report(sk);
282         }
283 }
284 
285 void raw_icmp_error(struct sk_buff *skb, int protocol, u32 info)
286 {
287         int hash;
288         struct sock *raw_sk;
289         const struct iphdr *iph;
290         struct net *net;
291 
292         hash = protocol & (RAW_HTABLE_SIZE - 1);
293 
294         read_lock(&raw_v4_hashinfo.lock);
295         raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
296         if (raw_sk != NULL) {
297                 iph = (const struct iphdr *)skb->data;
298                 net = dev_net(skb->dev);
299 
300                 while ((raw_sk = __raw_v4_lookup(net, raw_sk, protocol,
301                                                 iph->daddr, iph->saddr,
302                                                 skb->dev->ifindex)) != NULL) {
303                         raw_err(raw_sk, skb, info);
304                         raw_sk = sk_next(raw_sk);
305                         iph = (const struct iphdr *)skb->data;
306                 }
307         }
308         read_unlock(&raw_v4_hashinfo.lock);
309 }
310 
311 static int raw_rcv_skb(struct sock *sk, struct sk_buff *skb)
312 {
313         /* Charge it to the socket. */
314 
315         ipv4_pktinfo_prepare(sk, skb);
316         if (sock_queue_rcv_skb(sk, skb) < 0) {
317                 kfree_skb(skb);
318                 return NET_RX_DROP;
319         }
320 
321         return NET_RX_SUCCESS;
322 }
323 
324 int raw_rcv(struct sock *sk, struct sk_buff *skb)
325 {
326         if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
327                 atomic_inc(&sk->sk_drops);
328                 kfree_skb(skb);
329                 return NET_RX_DROP;
330         }
331         nf_reset(skb);
332 
333         skb_push(skb, skb->data - skb_network_header(skb));
334 
335         raw_rcv_skb(sk, skb);
336         return 0;
337 }
338 
339 static int raw_send_hdrinc(struct sock *sk, struct flowi4 *fl4,
340                            struct msghdr *msg, size_t length,
341                            struct rtable **rtp,
342                            unsigned int flags)
343 {
344         struct inet_sock *inet = inet_sk(sk);
345         struct net *net = sock_net(sk);
346         struct iphdr *iph;
347         struct sk_buff *skb;
348         unsigned int iphlen;
349         int err;
350         struct rtable *rt = *rtp;
351         int hlen, tlen;
352 
353         if (length > rt->dst.dev->mtu) {
354                 ip_local_error(sk, EMSGSIZE, fl4->daddr, inet->inet_dport,
355                                rt->dst.dev->mtu);
356                 return -EMSGSIZE;
357         }
358         if (flags&MSG_PROBE)
359                 goto out;
360 
361         hlen = LL_RESERVED_SPACE(rt->dst.dev);
362         tlen = rt->dst.dev->needed_tailroom;
363         skb = sock_alloc_send_skb(sk,
364                                   length + hlen + tlen + 15,
365                                   flags & MSG_DONTWAIT, &err);
366         if (skb == NULL)
367                 goto error;
368         skb_reserve(skb, hlen);
369 
370         skb->priority = sk->sk_priority;
371         skb->mark = sk->sk_mark;
372         skb_dst_set(skb, &rt->dst);
373         *rtp = NULL;
374 
375         skb_reset_network_header(skb);
376         iph = ip_hdr(skb);
377         skb_put(skb, length);
378 
379         skb->ip_summed = CHECKSUM_NONE;
380 
381         sock_tx_timestamp(sk, &skb_shinfo(skb)->tx_flags);
382 
383         skb->transport_header = skb->network_header;
384         err = -EFAULT;
385         if (memcpy_from_msg(iph, msg, length))
386                 goto error_free;
387 
388         iphlen = iph->ihl * 4;
389 
390         /*
391          * We don't want to modify the ip header, but we do need to
392          * be sure that it won't cause problems later along the network
393          * stack.  Specifically we want to make sure that iph->ihl is a
394          * sane value.  If ihl points beyond the length of the buffer passed
395          * in, reject the frame as invalid
396          */
397         err = -EINVAL;
398         if (iphlen > length)
399                 goto error_free;
400 
401         if (iphlen >= sizeof(*iph)) {
402                 if (!iph->saddr)
403                         iph->saddr = fl4->saddr;
404                 iph->check   = 0;
405                 iph->tot_len = htons(length);
406                 if (!iph->id)
407                         ip_select_ident(skb, NULL);
408 
409                 iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
410         }
411         if (iph->protocol == IPPROTO_ICMP)
412                 icmp_out_count(net, ((struct icmphdr *)
413                         skb_transport_header(skb))->type);
414 
415         err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT, skb, NULL,
416                       rt->dst.dev, dst_output);
417         if (err > 0)
418                 err = net_xmit_errno(err);
419         if (err)
420                 goto error;
421 out:
422         return 0;
423 
424 error_free:
425         kfree_skb(skb);
426 error:
427         IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
428         if (err == -ENOBUFS && !inet->recverr)
429                 err = 0;
430         return err;
431 }
432 
433 static int raw_probe_proto_opt(struct raw_frag_vec *rfv, struct flowi4 *fl4)
434 {
435         int err;
436 
437         if (fl4->flowi4_proto != IPPROTO_ICMP)
438                 return 0;
439 
440         /* We only need the first two bytes. */
441         rfv->hlen = 2;
442 
443         err = memcpy_from_msg(rfv->hdr.c, rfv->msg, rfv->hlen);
444         if (err)
445                 return err;
446 
447         fl4->fl4_icmp_type = rfv->hdr.icmph.type;
448         fl4->fl4_icmp_code = rfv->hdr.icmph.code;
449 
450         return 0;
451 }
452 
453 static int raw_getfrag(void *from, char *to, int offset, int len, int odd,
454                        struct sk_buff *skb)
455 {
456         struct raw_frag_vec *rfv = from;
457 
458         if (offset < rfv->hlen) {
459                 int copy = min(rfv->hlen - offset, len);
460 
461                 if (skb->ip_summed == CHECKSUM_PARTIAL)
462                         memcpy(to, rfv->hdr.c + offset, copy);
463                 else
464                         skb->csum = csum_block_add(
465                                 skb->csum,
466                                 csum_partial_copy_nocheck(rfv->hdr.c + offset,
467                                                           to, copy, 0),
468                                 odd);
469 
470                 odd = 0;
471                 offset += copy;
472                 to += copy;
473                 len -= copy;
474 
475                 if (!len)
476                         return 0;
477         }
478 
479         offset -= rfv->hlen;
480 
481         return ip_generic_getfrag(rfv->msg, to, offset, len, odd, skb);
482 }
483 
484 static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
485                        size_t len)
486 {
487         struct inet_sock *inet = inet_sk(sk);
488         struct ipcm_cookie ipc;
489         struct rtable *rt = NULL;
490         struct flowi4 fl4;
491         int free = 0;
492         __be32 daddr;
493         __be32 saddr;
494         u8  tos;
495         int err;
496         struct ip_options_data opt_copy;
497         struct raw_frag_vec rfv;
498 
499         err = -EMSGSIZE;
500         if (len > 0xFFFF)
501                 goto out;
502 
503         /*
504          *      Check the flags.
505          */
506 
507         err = -EOPNOTSUPP;
508         if (msg->msg_flags & MSG_OOB)   /* Mirror BSD error message */
509                 goto out;               /* compatibility */
510 
511         /*
512          *      Get and verify the address.
513          */
514 
515         if (msg->msg_namelen) {
516                 DECLARE_SOCKADDR(struct sockaddr_in *, usin, msg->msg_name);
517                 err = -EINVAL;
518                 if (msg->msg_namelen < sizeof(*usin))
519                         goto out;
520                 if (usin->sin_family != AF_INET) {
521                         pr_info_once("%s: %s forgot to set AF_INET. Fix it!\n",
522                                      __func__, current->comm);
523                         err = -EAFNOSUPPORT;
524                         if (usin->sin_family)
525                                 goto out;
526                 }
527                 daddr = usin->sin_addr.s_addr;
528                 /* ANK: I did not forget to get protocol from port field.
529                  * I just do not know, who uses this weirdness.
530                  * IP_HDRINCL is much more convenient.
531                  */
532         } else {
533                 err = -EDESTADDRREQ;
534                 if (sk->sk_state != TCP_ESTABLISHED)
535                         goto out;
536                 daddr = inet->inet_daddr;
537         }
538 
539         ipc.addr = inet->inet_saddr;
540         ipc.opt = NULL;
541         ipc.tx_flags = 0;
542         ipc.ttl = 0;
543         ipc.tos = -1;
544         ipc.oif = sk->sk_bound_dev_if;
545 
546         if (msg->msg_controllen) {
547                 err = ip_cmsg_send(sock_net(sk), msg, &ipc, false);
548                 if (err)
549                         goto out;
550                 if (ipc.opt)
551                         free = 1;
552         }
553 
554         saddr = ipc.addr;
555         ipc.addr = daddr;
556 
557         if (!ipc.opt) {
558                 struct ip_options_rcu *inet_opt;
559 
560                 rcu_read_lock();
561                 inet_opt = rcu_dereference(inet->inet_opt);
562                 if (inet_opt) {
563                         memcpy(&opt_copy, inet_opt,
564                                sizeof(*inet_opt) + inet_opt->opt.optlen);
565                         ipc.opt = &opt_copy.opt;
566                 }
567                 rcu_read_unlock();
568         }
569 
570         if (ipc.opt) {
571                 err = -EINVAL;
572                 /* Linux does not mangle headers on raw sockets,
573                  * so that IP options + IP_HDRINCL is non-sense.
574                  */
575                 if (inet->hdrincl)
576                         goto done;
577                 if (ipc.opt->opt.srr) {
578                         if (!daddr)
579                                 goto done;
580                         daddr = ipc.opt->opt.faddr;
581                 }
582         }
583         tos = get_rtconn_flags(&ipc, sk);
584         if (msg->msg_flags & MSG_DONTROUTE)
585                 tos |= RTO_ONLINK;
586 
587         if (ipv4_is_multicast(daddr)) {
588                 if (!ipc.oif)
589                         ipc.oif = inet->mc_index;
590                 if (!saddr)
591                         saddr = inet->mc_addr;
592         } else if (!ipc.oif)
593                 ipc.oif = inet->uc_index;
594 
595         flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
596                            RT_SCOPE_UNIVERSE,
597                            inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol,
598                            inet_sk_flowi_flags(sk) |
599                             (inet->hdrincl ? FLOWI_FLAG_KNOWN_NH : 0),
600                            daddr, saddr, 0, 0);
601 
602         if (!inet->hdrincl) {
603                 rfv.msg = msg;
604                 rfv.hlen = 0;
605 
606                 err = raw_probe_proto_opt(&rfv, &fl4);
607                 if (err)
608                         goto done;
609         }
610 
611         security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
612         rt = ip_route_output_flow(sock_net(sk), &fl4, sk);
613         if (IS_ERR(rt)) {
614                 err = PTR_ERR(rt);
615                 rt = NULL;
616                 goto done;
617         }
618 
619         err = -EACCES;
620         if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST))
621                 goto done;
622 
623         if (msg->msg_flags & MSG_CONFIRM)
624                 goto do_confirm;
625 back_from_confirm:
626 
627         if (inet->hdrincl)
628                 err = raw_send_hdrinc(sk, &fl4, msg, len,
629                                       &rt, msg->msg_flags);
630 
631          else {
632                 sock_tx_timestamp(sk, &ipc.tx_flags);
633 
634                 if (!ipc.addr)
635                         ipc.addr = fl4.daddr;
636                 lock_sock(sk);
637                 err = ip_append_data(sk, &fl4, raw_getfrag,
638                                      &rfv, len, 0,
639                                      &ipc, &rt, msg->msg_flags);
640                 if (err)
641                         ip_flush_pending_frames(sk);
642                 else if (!(msg->msg_flags & MSG_MORE)) {
643                         err = ip_push_pending_frames(sk, &fl4);
644                         if (err == -ENOBUFS && !inet->recverr)
645                                 err = 0;
646                 }
647                 release_sock(sk);
648         }
649 done:
650         if (free)
651                 kfree(ipc.opt);
652         ip_rt_put(rt);
653 
654 out:
655         if (err < 0)
656                 return err;
657         return len;
658 
659 do_confirm:
660         dst_confirm(&rt->dst);
661         if (!(msg->msg_flags & MSG_PROBE) || len)
662                 goto back_from_confirm;
663         err = 0;
664         goto done;
665 }
666 
667 static void raw_close(struct sock *sk, long timeout)
668 {
669         /*
670          * Raw sockets may have direct kernel references. Kill them.
671          */
672         ip_ra_control(sk, 0, NULL);
673 
674         sk_common_release(sk);
675 }
676 
677 static void raw_destroy(struct sock *sk)
678 {
679         lock_sock(sk);
680         ip_flush_pending_frames(sk);
681         release_sock(sk);
682 }
683 
684 /* This gets rid of all the nasties in af_inet. -DaveM */
685 static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
686 {
687         struct inet_sock *inet = inet_sk(sk);
688         struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
689         int ret = -EINVAL;
690         int chk_addr_ret;
691 
692         if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
693                 goto out;
694         chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
695         ret = -EADDRNOTAVAIL;
696         if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
697             chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
698                 goto out;
699         inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
700         if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
701                 inet->inet_saddr = 0;  /* Use device */
702         sk_dst_reset(sk);
703         ret = 0;
704 out:    return ret;
705 }
706 
707 /*
708  *      This should be easy, if there is something there
709  *      we return it, otherwise we block.
710  */
711 
712 static int raw_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
713                        size_t len, int noblock, int flags, int *addr_len)
714 {
715         struct inet_sock *inet = inet_sk(sk);
716         size_t copied = 0;
717         int err = -EOPNOTSUPP;
718         DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name);
719         struct sk_buff *skb;
720 
721         if (flags & MSG_OOB)
722                 goto out;
723 
724         if (flags & MSG_ERRQUEUE) {
725                 err = ip_recv_error(sk, msg, len, addr_len);
726                 goto out;
727         }
728 
729         skb = skb_recv_datagram(sk, flags, noblock, &err);
730         if (!skb)
731                 goto out;
732         if (ccs_socket_post_recvmsg_permission(sk, skb, flags)) {
733                 err = -EAGAIN; /* Hope less harmful than -EPERM. */
734                 goto out;
735         }
736 
737         copied = skb->len;
738         if (len < copied) {
739                 msg->msg_flags |= MSG_TRUNC;
740                 copied = len;
741         }
742 
743         err = skb_copy_datagram_msg(skb, 0, msg, copied);
744         if (err)
745                 goto done;
746 
747         sock_recv_ts_and_drops(msg, sk, skb);
748 
749         /* Copy the address. */
750         if (sin) {
751                 sin->sin_family = AF_INET;
752                 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
753                 sin->sin_port = 0;
754                 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
755                 *addr_len = sizeof(*sin);
756         }
757         if (inet->cmsg_flags)
758                 ip_cmsg_recv(msg, skb);
759         if (flags & MSG_TRUNC)
760                 copied = skb->len;
761 done:
762         skb_free_datagram(sk, skb);
763 out:
764         if (err)
765                 return err;
766         return copied;
767 }
768 
769 static int raw_init(struct sock *sk)
770 {
771         struct raw_sock *rp = raw_sk(sk);
772 
773         if (inet_sk(sk)->inet_num == IPPROTO_ICMP)
774                 memset(&rp->filter, 0, sizeof(rp->filter));
775         return 0;
776 }
777 
778 static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen)
779 {
780         if (optlen > sizeof(struct icmp_filter))
781                 optlen = sizeof(struct icmp_filter);
782         if (copy_from_user(&raw_sk(sk)->filter, optval, optlen))
783                 return -EFAULT;
784         return 0;
785 }
786 
787 static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen)
788 {
789         int len, ret = -EFAULT;
790 
791         if (get_user(len, optlen))
792                 goto out;
793         ret = -EINVAL;
794         if (len < 0)
795                 goto out;
796         if (len > sizeof(struct icmp_filter))
797                 len = sizeof(struct icmp_filter);
798         ret = -EFAULT;
799         if (put_user(len, optlen) ||
800             copy_to_user(optval, &raw_sk(sk)->filter, len))
801                 goto out;
802         ret = 0;
803 out:    return ret;
804 }
805 
806 static int do_raw_setsockopt(struct sock *sk, int level, int optname,
807                           char __user *optval, unsigned int optlen)
808 {
809         if (optname == ICMP_FILTER) {
810                 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
811                         return -EOPNOTSUPP;
812                 else
813                         return raw_seticmpfilter(sk, optval, optlen);
814         }
815         return -ENOPROTOOPT;
816 }
817 
818 static int raw_setsockopt(struct sock *sk, int level, int optname,
819                           char __user *optval, unsigned int optlen)
820 {
821         if (level != SOL_RAW)
822                 return ip_setsockopt(sk, level, optname, optval, optlen);
823         return do_raw_setsockopt(sk, level, optname, optval, optlen);
824 }
825 
826 #ifdef CONFIG_COMPAT
827 static int compat_raw_setsockopt(struct sock *sk, int level, int optname,
828                                  char __user *optval, unsigned int optlen)
829 {
830         if (level != SOL_RAW)
831                 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
832         return do_raw_setsockopt(sk, level, optname, optval, optlen);
833 }
834 #endif
835 
836 static int do_raw_getsockopt(struct sock *sk, int level, int optname,
837                           char __user *optval, int __user *optlen)
838 {
839         if (optname == ICMP_FILTER) {
840                 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
841                         return -EOPNOTSUPP;
842                 else
843                         return raw_geticmpfilter(sk, optval, optlen);
844         }
845         return -ENOPROTOOPT;
846 }
847 
848 static int raw_getsockopt(struct sock *sk, int level, int optname,
849                           char __user *optval, int __user *optlen)
850 {
851         if (level != SOL_RAW)
852                 return ip_getsockopt(sk, level, optname, optval, optlen);
853         return do_raw_getsockopt(sk, level, optname, optval, optlen);
854 }
855 
856 #ifdef CONFIG_COMPAT
857 static int compat_raw_getsockopt(struct sock *sk, int level, int optname,
858                                  char __user *optval, int __user *optlen)
859 {
860         if (level != SOL_RAW)
861                 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
862         return do_raw_getsockopt(sk, level, optname, optval, optlen);
863 }
864 #endif
865 
866 static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg)
867 {
868         switch (cmd) {
869         case SIOCOUTQ: {
870                 int amount = sk_wmem_alloc_get(sk);
871 
872                 return put_user(amount, (int __user *)arg);
873         }
874         case SIOCINQ: {
875                 struct sk_buff *skb;
876                 int amount = 0;
877 
878                 spin_lock_bh(&sk->sk_receive_queue.lock);
879                 skb = skb_peek(&sk->sk_receive_queue);
880                 if (skb != NULL)
881                         amount = skb->len;
882                 spin_unlock_bh(&sk->sk_receive_queue.lock);
883                 return put_user(amount, (int __user *)arg);
884         }
885 
886         default:
887 #ifdef CONFIG_IP_MROUTE
888                 return ipmr_ioctl(sk, cmd, (void __user *)arg);
889 #else
890                 return -ENOIOCTLCMD;
891 #endif
892         }
893 }
894 
895 #ifdef CONFIG_COMPAT
896 static int compat_raw_ioctl(struct sock *sk, unsigned int cmd, unsigned long arg)
897 {
898         switch (cmd) {
899         case SIOCOUTQ:
900         case SIOCINQ:
901                 return -ENOIOCTLCMD;
902         default:
903 #ifdef CONFIG_IP_MROUTE
904                 return ipmr_compat_ioctl(sk, cmd, compat_ptr(arg));
905 #else
906                 return -ENOIOCTLCMD;
907 #endif
908         }
909 }
910 #endif
911 
912 struct proto raw_prot = {
913         .name              = "RAW",
914         .owner             = THIS_MODULE,
915         .close             = raw_close,
916         .destroy           = raw_destroy,
917         .connect           = ip4_datagram_connect,
918         .disconnect        = udp_disconnect,
919         .ioctl             = raw_ioctl,
920         .init              = raw_init,
921         .setsockopt        = raw_setsockopt,
922         .getsockopt        = raw_getsockopt,
923         .sendmsg           = raw_sendmsg,
924         .recvmsg           = raw_recvmsg,
925         .bind              = raw_bind,
926         .backlog_rcv       = raw_rcv_skb,
927         .release_cb        = ip4_datagram_release_cb,
928         .hash              = raw_hash_sk,
929         .unhash            = raw_unhash_sk,
930         .obj_size          = sizeof(struct raw_sock),
931         .h.raw_hash        = &raw_v4_hashinfo,
932 #ifdef CONFIG_COMPAT
933         .compat_setsockopt = compat_raw_setsockopt,
934         .compat_getsockopt = compat_raw_getsockopt,
935         .compat_ioctl      = compat_raw_ioctl,
936 #endif
937 };
938 
939 #ifdef CONFIG_PROC_FS
940 static struct sock *raw_get_first(struct seq_file *seq)
941 {
942         struct sock *sk;
943         struct raw_iter_state *state = raw_seq_private(seq);
944 
945         for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE;
946                         ++state->bucket) {
947                 sk_for_each(sk, &state->h->ht[state->bucket])
948                         if (sock_net(sk) == seq_file_net(seq))
949                                 goto found;
950         }
951         sk = NULL;
952 found:
953         return sk;
954 }
955 
956 static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk)
957 {
958         struct raw_iter_state *state = raw_seq_private(seq);
959 
960         do {
961                 sk = sk_next(sk);
962 try_again:
963                 ;
964         } while (sk && sock_net(sk) != seq_file_net(seq));
965 
966         if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
967                 sk = sk_head(&state->h->ht[state->bucket]);
968                 goto try_again;
969         }
970         return sk;
971 }
972 
973 static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos)
974 {
975         struct sock *sk = raw_get_first(seq);
976 
977         if (sk)
978                 while (pos && (sk = raw_get_next(seq, sk)) != NULL)
979                         --pos;
980         return pos ? NULL : sk;
981 }
982 
983 void *raw_seq_start(struct seq_file *seq, loff_t *pos)
984 {
985         struct raw_iter_state *state = raw_seq_private(seq);
986 
987         read_lock(&state->h->lock);
988         return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
989 }
990 EXPORT_SYMBOL_GPL(raw_seq_start);
991 
992 void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos)
993 {
994         struct sock *sk;
995 
996         if (v == SEQ_START_TOKEN)
997                 sk = raw_get_first(seq);
998         else
999                 sk = raw_get_next(seq, v);
1000         ++*pos;
1001         return sk;
1002 }
1003 EXPORT_SYMBOL_GPL(raw_seq_next);
1004 
1005 void raw_seq_stop(struct seq_file *seq, void *v)
1006 {
1007         struct raw_iter_state *state = raw_seq_private(seq);
1008 
1009         read_unlock(&state->h->lock);
1010 }
1011 EXPORT_SYMBOL_GPL(raw_seq_stop);
1012 
1013 static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
1014 {
1015         struct inet_sock *inet = inet_sk(sp);
1016         __be32 dest = inet->inet_daddr,
1017                src = inet->inet_rcv_saddr;
1018         __u16 destp = 0,
1019               srcp  = inet->inet_num;
1020 
1021         seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
1022                 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d\n",
1023                 i, src, srcp, dest, destp, sp->sk_state,
1024                 sk_wmem_alloc_get(sp),
1025                 sk_rmem_alloc_get(sp),
1026                 0, 0L, 0,
1027                 from_kuid_munged(seq_user_ns(seq), sock_i_uid(sp)),
1028                 0, sock_i_ino(sp),
1029                 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
1030 }
1031 
1032 static int raw_seq_show(struct seq_file *seq, void *v)
1033 {
1034         if (v == SEQ_START_TOKEN)
1035                 seq_printf(seq, "  sl  local_address rem_address   st tx_queue "
1036                                 "rx_queue tr tm->when retrnsmt   uid  timeout "
1037                                 "inode ref pointer drops\n");
1038         else
1039                 raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
1040         return 0;
1041 }
1042 
1043 static const struct seq_operations raw_seq_ops = {
1044         .start = raw_seq_start,
1045         .next  = raw_seq_next,
1046         .stop  = raw_seq_stop,
1047         .show  = raw_seq_show,
1048 };
1049 
1050 int raw_seq_open(struct inode *ino, struct file *file,
1051                  struct raw_hashinfo *h, const struct seq_operations *ops)
1052 {
1053         int err;
1054         struct raw_iter_state *i;
1055 
1056         err = seq_open_net(ino, file, ops, sizeof(struct raw_iter_state));
1057         if (err < 0)
1058                 return err;
1059 
1060         i = raw_seq_private((struct seq_file *)file->private_data);
1061         i->h = h;
1062         return 0;
1063 }
1064 EXPORT_SYMBOL_GPL(raw_seq_open);
1065 
1066 static int raw_v4_seq_open(struct inode *inode, struct file *file)
1067 {
1068         return raw_seq_open(inode, file, &raw_v4_hashinfo, &raw_seq_ops);
1069 }
1070 
1071 static const struct file_operations raw_seq_fops = {
1072         .owner   = THIS_MODULE,
1073         .open    = raw_v4_seq_open,
1074         .read    = seq_read,
1075         .llseek  = seq_lseek,
1076         .release = seq_release_net,
1077 };
1078 
1079 static __net_init int raw_init_net(struct net *net)
1080 {
1081         if (!proc_create("raw", S_IRUGO, net->proc_net, &raw_seq_fops))
1082                 return -ENOMEM;
1083 
1084         return 0;
1085 }
1086 
1087 static __net_exit void raw_exit_net(struct net *net)
1088 {
1089         remove_proc_entry("raw", net->proc_net);
1090 }
1091 
1092 static __net_initdata struct pernet_operations raw_net_ops = {
1093         .init = raw_init_net,
1094         .exit = raw_exit_net,
1095 };
1096 
1097 int __init raw_proc_init(void)
1098 {
1099         return register_pernet_subsys(&raw_net_ops);
1100 }
1101 
1102 void __init raw_proc_exit(void)
1103 {
1104         unregister_pernet_subsys(&raw_net_ops);
1105 }
1106 #endif /* CONFIG_PROC_FS */
1107 

~ [ 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