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  * Version:	$Id: ip_sockglue.c,v 1.61 2001/10/20 00:00:11 davem Exp $
9  *
10  * Authors:	see ip.c
11  *
12  * Fixes:
13  *		Many		:	Split from ip.c , see ip.c for history.
14  *		Martin Mares	:	TOS setting fixed.
15  *		Alan Cox	:	Fixed a couple of oopses in Martin's
16  *					TOS tweaks.
17  *		Mike McLagan	:	Routing by source
18  */
19 
20 #include <linux/config.h>
21 #include <linux/types.h>
22 #include <linux/mm.h>
23 #include <linux/sched.h>
24 #include <linux/skbuff.h>
25 #include <linux/ip.h>
26 #include <linux/icmp.h>
27 #include <linux/netdevice.h>
28 #include <net/sock.h>
29 #include <net/ip.h>
30 #include <net/icmp.h>
31 #include <net/tcp.h>
32 #include <linux/tcp.h>
33 #include <linux/udp.h>
34 #include <linux/igmp.h>
35 #include <linux/netfilter.h>
36 #include <linux/route.h>
37 #include <linux/mroute.h>
38 #include <net/route.h>
39 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
40 #include <net/transp_v6.h>
41 #endif
42 
43 #include <linux/errqueue.h>
44 #include <asm/uaccess.h>
45 
46 #define IP_CMSG_PKTINFO		1
47 #define IP_CMSG_TTL		2
48 #define IP_CMSG_TOS		4
49 #define IP_CMSG_RECVOPTS	8
50 #define IP_CMSG_RETOPTS		16
51 
52 /*
53  *	SOL_IP control messages.
54  */
55 
ip_cmsg_recv_pktinfo(struct msghdr * msg,struct sk_buff * skb)56 static void ip_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
57 {
58 	struct in_pktinfo info;
59 	struct rtable *rt = (struct rtable *)skb->dst;
60 
61 	info.ipi_addr.s_addr = skb->nh.iph->daddr;
62 	if (rt) {
63 		info.ipi_ifindex = rt->rt_iif;
64 		info.ipi_spec_dst.s_addr = rt->rt_spec_dst;
65 	} else {
66 		info.ipi_ifindex = 0;
67 		info.ipi_spec_dst.s_addr = 0;
68 	}
69 
70 	put_cmsg(msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
71 }
72 
ip_cmsg_recv_ttl(struct msghdr * msg,struct sk_buff * skb)73 static void ip_cmsg_recv_ttl(struct msghdr *msg, struct sk_buff *skb)
74 {
75 	int ttl = skb->nh.iph->ttl;
76 	put_cmsg(msg, SOL_IP, IP_TTL, sizeof(int), &ttl);
77 }
78 
ip_cmsg_recv_tos(struct msghdr * msg,struct sk_buff * skb)79 static void ip_cmsg_recv_tos(struct msghdr *msg, struct sk_buff *skb)
80 {
81 	put_cmsg(msg, SOL_IP, IP_TOS, 1, &skb->nh.iph->tos);
82 }
83 
ip_cmsg_recv_opts(struct msghdr * msg,struct sk_buff * skb)84 static void ip_cmsg_recv_opts(struct msghdr *msg, struct sk_buff *skb)
85 {
86 	if (IPCB(skb)->opt.optlen == 0)
87 		return;
88 
89 	put_cmsg(msg, SOL_IP, IP_RECVOPTS, IPCB(skb)->opt.optlen, skb->nh.iph+1);
90 }
91 
92 
ip_cmsg_recv_retopts(struct msghdr * msg,struct sk_buff * skb)93 void ip_cmsg_recv_retopts(struct msghdr *msg, struct sk_buff *skb)
94 {
95 	unsigned char optbuf[sizeof(struct ip_options) + 40];
96 	struct ip_options * opt = (struct ip_options*)optbuf;
97 
98 	if (IPCB(skb)->opt.optlen == 0)
99 		return;
100 
101 	if (ip_options_echo(opt, skb)) {
102 		msg->msg_flags |= MSG_CTRUNC;
103 		return;
104 	}
105 	ip_options_undo(opt);
106 
107 	put_cmsg(msg, SOL_IP, IP_RETOPTS, opt->optlen, opt->__data);
108 }
109 
110 
ip_cmsg_recv(struct msghdr * msg,struct sk_buff * skb)111 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
112 {
113 	unsigned flags = skb->sk->protinfo.af_inet.cmsg_flags;
114 
115 	/* Ordered by supposed usage frequency */
116 	if (flags & 1)
117 		ip_cmsg_recv_pktinfo(msg, skb);
118 	if ((flags>>=1) == 0)
119 		return;
120 
121 	if (flags & 1)
122 		ip_cmsg_recv_ttl(msg, skb);
123 	if ((flags>>=1) == 0)
124 		return;
125 
126 	if (flags & 1)
127 		ip_cmsg_recv_tos(msg, skb);
128 	if ((flags>>=1) == 0)
129 		return;
130 
131 	if (flags & 1)
132 		ip_cmsg_recv_opts(msg, skb);
133 	if ((flags>>=1) == 0)
134 		return;
135 
136 	if (flags & 1)
137 		ip_cmsg_recv_retopts(msg, skb);
138 }
139 
ip_cmsg_send(struct msghdr * msg,struct ipcm_cookie * ipc)140 int ip_cmsg_send(struct msghdr *msg, struct ipcm_cookie *ipc)
141 {
142 	int err;
143 	struct cmsghdr *cmsg;
144 
145 	for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
146 		if (!CMSG_OK(msg, cmsg))
147 			return -EINVAL;
148 		if (cmsg->cmsg_level != SOL_IP)
149 			continue;
150 		switch (cmsg->cmsg_type) {
151 		case IP_RETOPTS:
152 			err = cmsg->cmsg_len - CMSG_ALIGN(sizeof(struct cmsghdr));
153 			err = ip_options_get(&ipc->opt, CMSG_DATA(cmsg), err < 40 ? err : 40, 0);
154 			if (err)
155 				return err;
156 			break;
157 		case IP_PKTINFO:
158 		{
159 			struct in_pktinfo *info;
160 			if (cmsg->cmsg_len != CMSG_LEN(sizeof(struct in_pktinfo)))
161 				return -EINVAL;
162 			info = (struct in_pktinfo *)CMSG_DATA(cmsg);
163 			ipc->oif = info->ipi_ifindex;
164 			ipc->addr = info->ipi_spec_dst.s_addr;
165 			break;
166 		}
167 		default:
168 			return -EINVAL;
169 		}
170 	}
171 	return 0;
172 }
173 
174 
175 /* Special input handler for packets catched by router alert option.
176    They are selected only by protocol field, and then processed likely
177    local ones; but only if someone wants them! Otherwise, router
178    not running rsvpd will kill RSVP.
179 
180    It is user level problem, what it will make with them.
181    I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
182    but receiver should be enough clever f.e. to forward mtrace requests,
183    sent to multicast group to reach destination designated router.
184  */
185 struct ip_ra_chain *ip_ra_chain;
186 rwlock_t ip_ra_lock = RW_LOCK_UNLOCKED;
187 
ip_ra_control(struct sock * sk,unsigned char on,void (* destructor)(struct sock *))188 int ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *))
189 {
190 	struct ip_ra_chain *ra, *new_ra, **rap;
191 
192 	if (sk->type != SOCK_RAW || sk->num == IPPROTO_RAW)
193 		return -EINVAL;
194 
195 	new_ra = on ? kmalloc(sizeof(*new_ra), GFP_KERNEL) : NULL;
196 
197 	write_lock_bh(&ip_ra_lock);
198 	for (rap = &ip_ra_chain; (ra=*rap) != NULL; rap = &ra->next) {
199 		if (ra->sk == sk) {
200 			if (on) {
201 				write_unlock_bh(&ip_ra_lock);
202 				if (new_ra)
203 					kfree(new_ra);
204 				return -EADDRINUSE;
205 			}
206 			*rap = ra->next;
207 			write_unlock_bh(&ip_ra_lock);
208 
209 			if (ra->destructor)
210 				ra->destructor(sk);
211 			sock_put(sk);
212 			kfree(ra);
213 			return 0;
214 		}
215 	}
216 	if (new_ra == NULL) {
217 		write_unlock_bh(&ip_ra_lock);
218 		return -ENOBUFS;
219 	}
220 	new_ra->sk = sk;
221 	new_ra->destructor = destructor;
222 
223 	new_ra->next = ra;
224 	*rap = new_ra;
225 	sock_hold(sk);
226 	write_unlock_bh(&ip_ra_lock);
227 
228 	return 0;
229 }
230 
ip_icmp_error(struct sock * sk,struct sk_buff * skb,int err,u16 port,u32 info,u8 * payload)231 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
232 		   u16 port, u32 info, u8 *payload)
233 {
234 	struct sock_exterr_skb *serr;
235 
236 	if (!sk->protinfo.af_inet.recverr)
237 		return;
238 
239 	skb = skb_clone(skb, GFP_ATOMIC);
240 	if (!skb)
241 		return;
242 
243 	serr = SKB_EXT_ERR(skb);
244 	serr->ee.ee_errno = err;
245 	serr->ee.ee_origin = SO_EE_ORIGIN_ICMP;
246 	serr->ee.ee_type = skb->h.icmph->type;
247 	serr->ee.ee_code = skb->h.icmph->code;
248 	serr->ee.ee_pad = 0;
249 	serr->ee.ee_info = info;
250 	serr->ee.ee_data = 0;
251 	serr->addr_offset = (u8*)&(((struct iphdr*)(skb->h.icmph+1))->daddr) - skb->nh.raw;
252 	serr->port = port;
253 
254 	skb->h.raw = payload;
255 	if (!skb_pull(skb, payload - skb->data) ||
256 	    sock_queue_err_skb(sk, skb))
257 		kfree_skb(skb);
258 }
259 
ip_local_error(struct sock * sk,int err,u32 daddr,u16 port,u32 info)260 void ip_local_error(struct sock *sk, int err, u32 daddr, u16 port, u32 info)
261 {
262 	struct sock_exterr_skb *serr;
263 	struct iphdr *iph;
264 	struct sk_buff *skb;
265 
266 	if (!sk->protinfo.af_inet.recverr)
267 		return;
268 
269 	skb = alloc_skb(sizeof(struct iphdr), GFP_ATOMIC);
270 	if (!skb)
271 		return;
272 
273 	iph = (struct iphdr*)skb_put(skb, sizeof(struct iphdr));
274 	skb->nh.iph = iph;
275 	iph->daddr = daddr;
276 
277 	serr = SKB_EXT_ERR(skb);
278 	serr->ee.ee_errno = err;
279 	serr->ee.ee_origin = SO_EE_ORIGIN_LOCAL;
280 	serr->ee.ee_type = 0;
281 	serr->ee.ee_code = 0;
282 	serr->ee.ee_pad = 0;
283 	serr->ee.ee_info = info;
284 	serr->ee.ee_data = 0;
285 	serr->addr_offset = (u8*)&iph->daddr - skb->nh.raw;
286 	serr->port = port;
287 
288 	skb->h.raw = skb->tail;
289 	__skb_pull(skb, skb->tail - skb->data);
290 
291 	if (sock_queue_err_skb(sk, skb))
292 		kfree_skb(skb);
293 }
294 
295 /*
296  *	Handle MSG_ERRQUEUE
297  */
ip_recv_error(struct sock * sk,struct msghdr * msg,int len)298 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len)
299 {
300 	struct sock_exterr_skb *serr;
301 	struct sk_buff *skb, *skb2;
302 	struct sockaddr_in *sin;
303 	struct {
304 		struct sock_extended_err ee;
305 		struct sockaddr_in	 offender;
306 	} errhdr;
307 	int err;
308 	int copied;
309 
310 	err = -EAGAIN;
311 	skb = skb_dequeue(&sk->error_queue);
312 	if (skb == NULL)
313 		goto out;
314 
315 	copied = skb->len;
316 	if (copied > len) {
317 		msg->msg_flags |= MSG_TRUNC;
318 		copied = len;
319 	}
320 	err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
321 	if (err)
322 		goto out_free_skb;
323 
324 	sock_recv_timestamp(msg, sk, skb);
325 
326 	serr = SKB_EXT_ERR(skb);
327 
328 	sin = (struct sockaddr_in *)msg->msg_name;
329 	if (sin) {
330 		sin->sin_family = AF_INET;
331 		sin->sin_addr.s_addr = *(u32*)(skb->nh.raw + serr->addr_offset);
332 		sin->sin_port = serr->port;
333 		memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
334 	}
335 
336 	memcpy(&errhdr.ee, &serr->ee, sizeof(struct sock_extended_err));
337 	sin = &errhdr.offender;
338 	sin->sin_family = AF_UNSPEC;
339 	if (serr->ee.ee_origin == SO_EE_ORIGIN_ICMP) {
340 		sin->sin_family = AF_INET;
341 		sin->sin_addr.s_addr = skb->nh.iph->saddr;
342 		sin->sin_port = 0;
343 		memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
344 		if (sk->protinfo.af_inet.cmsg_flags)
345 			ip_cmsg_recv(msg, skb);
346 	}
347 
348 	put_cmsg(msg, SOL_IP, IP_RECVERR, sizeof(errhdr), &errhdr);
349 
350 	/* Now we could try to dump offended packet options */
351 
352 	msg->msg_flags |= MSG_ERRQUEUE;
353 	err = copied;
354 
355 	/* Reset and regenerate socket error */
356 	spin_lock_irq(&sk->error_queue.lock);
357 	sk->err = 0;
358 	if ((skb2 = skb_peek(&sk->error_queue)) != NULL) {
359 		sk->err = SKB_EXT_ERR(skb2)->ee.ee_errno;
360 		spin_unlock_irq(&sk->error_queue.lock);
361 		sk->error_report(sk);
362 	} else {
363 		spin_unlock_irq(&sk->error_queue.lock);
364 	}
365 
366 out_free_skb:
367 	kfree_skb(skb);
368 out:
369 	return err;
370 }
371 
372 
373 /*
374  *	Socket option code for IP. This is the end of the line after any TCP,UDP etc options on
375  *	an IP socket.
376  */
377 
ip_setsockopt(struct sock * sk,int level,int optname,char * optval,int optlen)378 int ip_setsockopt(struct sock *sk, int level, int optname, char *optval, int optlen)
379 {
380 	int val=0,err;
381 
382 	if (level != SOL_IP)
383 		return -ENOPROTOOPT;
384 
385 	if (((1<<optname) & ((1<<IP_PKTINFO) | (1<<IP_RECVTTL) |
386 			    (1<<IP_RECVOPTS) | (1<<IP_RECVTOS) |
387 			    (1<<IP_RETOPTS) | (1<<IP_TOS) |
388 			    (1<<IP_TTL) | (1<<IP_HDRINCL) |
389 			    (1<<IP_MTU_DISCOVER) | (1<<IP_RECVERR) |
390 			    (1<<IP_ROUTER_ALERT) | (1<<IP_FREEBIND))) ||
391 				optname == IP_MULTICAST_TTL ||
392 				optname == IP_MULTICAST_LOOP) {
393 		if (optlen >= sizeof(int)) {
394 			if (get_user(val, (int *) optval))
395 				return -EFAULT;
396 		} else if (optlen >= sizeof(char)) {
397 			unsigned char ucval;
398 
399 			if (get_user(ucval, (unsigned char *) optval))
400 				return -EFAULT;
401 			val = (int) ucval;
402 		}
403 	}
404 
405 	/* If optlen==0, it is equivalent to val == 0 */
406 
407 #ifdef CONFIG_IP_MROUTE
408 	if (optname >= MRT_BASE && optname <= (MRT_BASE + 10))
409 		return ip_mroute_setsockopt(sk,optname,optval,optlen);
410 #endif
411 
412 	err = 0;
413 	lock_sock(sk);
414 
415 	switch (optname) {
416 		case IP_OPTIONS:
417 		{
418 			struct ip_options * opt = NULL;
419 			if (optlen > 40 || optlen < 0)
420 				goto e_inval;
421 			err = ip_options_get(&opt, optval, optlen, 1);
422 			if (err)
423 				break;
424 			if (sk->type == SOCK_STREAM) {
425 				struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
426 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
427 				if (sk->family == PF_INET ||
428 				    (!((1<<sk->state)&(TCPF_LISTEN|TCPF_CLOSE))
429 				     && sk->daddr != LOOPBACK4_IPV6)) {
430 #endif
431 					if (opt)
432 						tp->ext_header_len = opt->optlen;
433 					tcp_sync_mss(sk, tp->pmtu_cookie);
434 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
435 				}
436 #endif
437 			}
438 			opt = xchg(&sk->protinfo.af_inet.opt, opt);
439 			if (opt)
440 				kfree(opt);
441 			break;
442 		}
443 		case IP_PKTINFO:
444 			if (val)
445 				sk->protinfo.af_inet.cmsg_flags |= IP_CMSG_PKTINFO;
446 			else
447 				sk->protinfo.af_inet.cmsg_flags &= ~IP_CMSG_PKTINFO;
448 			break;
449 		case IP_RECVTTL:
450 			if (val)
451 				sk->protinfo.af_inet.cmsg_flags |=  IP_CMSG_TTL;
452 			else
453 				sk->protinfo.af_inet.cmsg_flags &= ~IP_CMSG_TTL;
454 			break;
455 		case IP_RECVTOS:
456 			if (val)
457 				sk->protinfo.af_inet.cmsg_flags |=  IP_CMSG_TOS;
458 			else
459 				sk->protinfo.af_inet.cmsg_flags &= ~IP_CMSG_TOS;
460 			break;
461 		case IP_RECVOPTS:
462 			if (val)
463 				sk->protinfo.af_inet.cmsg_flags |=  IP_CMSG_RECVOPTS;
464 			else
465 				sk->protinfo.af_inet.cmsg_flags &= ~IP_CMSG_RECVOPTS;
466 			break;
467 		case IP_RETOPTS:
468 			if (val)
469 				sk->protinfo.af_inet.cmsg_flags |= IP_CMSG_RETOPTS;
470 			else
471 				sk->protinfo.af_inet.cmsg_flags &= ~IP_CMSG_RETOPTS;
472 			break;
473 		case IP_TOS:	/* This sets both TOS and Precedence */
474 			if (sk->type == SOCK_STREAM) {
475 				val &= ~3;
476 				val |= sk->protinfo.af_inet.tos & 3;
477 			}
478 			if (IPTOS_PREC(val) >= IPTOS_PREC_CRITIC_ECP &&
479 			    !capable(CAP_NET_ADMIN)) {
480 				err = -EPERM;
481 				break;
482 			}
483 			if (sk->protinfo.af_inet.tos != val) {
484 				sk->protinfo.af_inet.tos=val;
485 				sk->priority = rt_tos2priority(val);
486 				sk_dst_reset(sk);
487 			}
488 			break;
489 		case IP_TTL:
490 			if (optlen<1)
491 				goto e_inval;
492 			if(val==-1)
493 				val = sysctl_ip_default_ttl;
494 			if(val<1||val>255)
495 				goto e_inval;
496 			sk->protinfo.af_inet.ttl=val;
497 			break;
498 		case IP_HDRINCL:
499 			if(sk->type!=SOCK_RAW) {
500 				err = -ENOPROTOOPT;
501 				break;
502 			}
503 			sk->protinfo.af_inet.hdrincl=val?1:0;
504 			break;
505 		case IP_MTU_DISCOVER:
506 			if (val<0 || val>2)
507 				goto e_inval;
508 			sk->protinfo.af_inet.pmtudisc = val;
509 			break;
510 		case IP_RECVERR:
511 			sk->protinfo.af_inet.recverr = !!val;
512 			if (!val)
513 				skb_queue_purge(&sk->error_queue);
514 			break;
515 		case IP_MULTICAST_TTL:
516 			if (sk->type == SOCK_STREAM)
517 				goto e_inval;
518 			if (optlen<1)
519 				goto e_inval;
520 			if (val==-1)
521 				val = 1;
522 			if (val < 0 || val > 255)
523 				goto e_inval;
524 			sk->protinfo.af_inet.mc_ttl=val;
525 	                break;
526 		case IP_MULTICAST_LOOP:
527 			if (optlen<1)
528 				goto e_inval;
529 			sk->protinfo.af_inet.mc_loop = val ? 1 : 0;
530 	                break;
531 		case IP_MULTICAST_IF:
532 		{
533 			struct ip_mreqn mreq;
534 			struct net_device *dev = NULL;
535 
536 			if (sk->type == SOCK_STREAM)
537 				goto e_inval;
538 			/*
539 			 *	Check the arguments are allowable
540 			 */
541 
542 			err = -EFAULT;
543 			if (optlen >= sizeof(struct ip_mreqn)) {
544 				if (copy_from_user(&mreq,optval,sizeof(mreq)))
545 					break;
546 			} else {
547 				memset(&mreq, 0, sizeof(mreq));
548 				if (optlen >= sizeof(struct in_addr) &&
549 				    copy_from_user(&mreq.imr_address,optval,sizeof(struct in_addr)))
550 					break;
551 			}
552 
553 			if (!mreq.imr_ifindex) {
554 				if (mreq.imr_address.s_addr == INADDR_ANY) {
555 					sk->protinfo.af_inet.mc_index = 0;
556 					sk->protinfo.af_inet.mc_addr  = 0;
557 					err = 0;
558 					break;
559 				}
560 				dev = ip_dev_find(mreq.imr_address.s_addr);
561 				if (dev) {
562 					mreq.imr_ifindex = dev->ifindex;
563 					dev_put(dev);
564 				}
565 			} else
566 				dev = __dev_get_by_index(mreq.imr_ifindex);
567 
568 
569 			err = -EADDRNOTAVAIL;
570 			if (!dev)
571 				break;
572 
573 			err = -EINVAL;
574 			if (sk->bound_dev_if && mreq.imr_ifindex != sk->bound_dev_if)
575 				break;
576 
577 			sk->protinfo.af_inet.mc_index = mreq.imr_ifindex;
578 			sk->protinfo.af_inet.mc_addr  = mreq.imr_address.s_addr;
579 			err = 0;
580 			break;
581 		}
582 
583 		case IP_ADD_MEMBERSHIP:
584 		case IP_DROP_MEMBERSHIP:
585 		{
586 			struct ip_mreqn mreq;
587 
588 			if (optlen < sizeof(struct ip_mreq))
589 				goto e_inval;
590 			err = -EFAULT;
591 			if (optlen >= sizeof(struct ip_mreqn)) {
592 				if(copy_from_user(&mreq,optval,sizeof(mreq)))
593 					break;
594 			} else {
595 				memset(&mreq, 0, sizeof(mreq));
596 				if (copy_from_user(&mreq,optval,sizeof(struct ip_mreq)))
597 					break;
598 			}
599 
600 			if (optname == IP_ADD_MEMBERSHIP)
601 				err = ip_mc_join_group(sk, &mreq);
602 			else
603 				err = ip_mc_leave_group(sk, &mreq);
604 			break;
605 		}
606 		case IP_MSFILTER:
607 		{
608 			extern int sysctl_optmem_max;
609 			extern int sysctl_igmp_max_msf;
610 			struct ip_msfilter *msf;
611 
612 			if (optlen < IP_MSFILTER_SIZE(0))
613 				goto e_inval;
614 			if (optlen > sysctl_optmem_max) {
615 				err = -ENOBUFS;
616 				break;
617 			}
618 			msf = (struct ip_msfilter *)kmalloc(optlen, GFP_KERNEL);
619 			if (msf == 0) {
620 				err = -ENOBUFS;
621 				break;
622 			}
623 			err = -EFAULT;
624 			if (copy_from_user(msf, optval, optlen)) {
625 				kfree(msf);
626 				break;
627 			}
628 			/* numsrc >= (1G-4) overflow in 32 bits */
629 			if (msf->imsf_numsrc >= 0x3ffffffcU ||
630 			    msf->imsf_numsrc > sysctl_igmp_max_msf) {
631 				kfree(msf);
632 				err = -ENOBUFS;
633 				break;
634 			}
635 			if (IP_MSFILTER_SIZE(msf->imsf_numsrc) > optlen) {
636 				kfree(msf);
637 				err = -EINVAL;
638 				break;
639 			}
640 			err = ip_mc_msfilter(sk, msf, 0);
641 			kfree(msf);
642 			break;
643 		}
644 		case IP_BLOCK_SOURCE:
645 		case IP_UNBLOCK_SOURCE:
646 		case IP_ADD_SOURCE_MEMBERSHIP:
647 		case IP_DROP_SOURCE_MEMBERSHIP:
648 		{
649 			struct ip_mreq_source mreqs;
650 			int omode, add;
651 
652 			if (optlen != sizeof(struct ip_mreq_source))
653 				goto e_inval;
654 			if (copy_from_user(&mreqs, optval, sizeof(mreqs))) {
655 				err = -EFAULT;
656 				break;
657 			}
658 			if (optname == IP_BLOCK_SOURCE) {
659 				omode = MCAST_EXCLUDE;
660 				add = 1;
661 			} else if (optname == IP_UNBLOCK_SOURCE) {
662 				omode = MCAST_EXCLUDE;
663 				add = 0;
664 			} else if (optname == IP_ADD_SOURCE_MEMBERSHIP) {
665 				struct ip_mreqn mreq;
666 
667 				mreq.imr_multiaddr.s_addr = mreqs.imr_multiaddr;
668 				mreq.imr_address.s_addr = mreqs.imr_interface;
669 				mreq.imr_ifindex = 0;
670 				err = ip_mc_join_group(sk, &mreq);
671 				if (err)
672 					break;
673 				omode = MCAST_INCLUDE;
674 				add = 1;
675 			} else /*IP_DROP_SOURCE_MEMBERSHIP */ {
676 				omode = MCAST_INCLUDE;
677 				add = 0;
678 			}
679 			err = ip_mc_source(add, omode, sk, &mreqs, 0);
680 			break;
681 		}
682 		case MCAST_JOIN_GROUP:
683 		case MCAST_LEAVE_GROUP:
684 		{
685 			struct group_req greq;
686 			struct sockaddr_in *psin;
687 			struct ip_mreqn mreq;
688 
689 			if (optlen < sizeof(struct group_req))
690 				goto e_inval;
691 			err = -EFAULT;
692 			if(copy_from_user(&greq, optval, sizeof(greq)))
693 				break;
694 			psin = (struct sockaddr_in *)&greq.gr_group;
695 			if (psin->sin_family != AF_INET)
696 				goto e_inval;
697 			memset(&mreq, 0, sizeof(mreq));
698 			mreq.imr_multiaddr = psin->sin_addr;
699 			mreq.imr_ifindex = greq.gr_interface;
700 
701 			if (optname == MCAST_JOIN_GROUP)
702 				err = ip_mc_join_group(sk, &mreq);
703 			else
704 				err = ip_mc_leave_group(sk, &mreq);
705 			break;
706 		}
707 		case MCAST_JOIN_SOURCE_GROUP:
708 		case MCAST_LEAVE_SOURCE_GROUP:
709 		case MCAST_BLOCK_SOURCE:
710 		case MCAST_UNBLOCK_SOURCE:
711 		{
712 			struct group_source_req greqs;
713 			struct ip_mreq_source mreqs;
714 			struct sockaddr_in *psin;
715 			int omode, add;
716 
717 			if (optlen != sizeof(struct group_source_req))
718 				goto e_inval;
719 			if (copy_from_user(&greqs, optval, sizeof(greqs))) {
720 				err = -EFAULT;
721 				break;
722 			}
723 			if (greqs.gsr_group.ss_family != AF_INET ||
724 			    greqs.gsr_source.ss_family != AF_INET) {
725 				err = -EADDRNOTAVAIL;
726 				break;
727 			}
728 			psin = (struct sockaddr_in *)&greqs.gsr_group;
729 			mreqs.imr_multiaddr = psin->sin_addr.s_addr;
730 			psin = (struct sockaddr_in *)&greqs.gsr_source;
731 			mreqs.imr_sourceaddr = psin->sin_addr.s_addr;
732 			mreqs.imr_interface = 0; /* use index for mc_source */
733 
734 			if (optname == MCAST_BLOCK_SOURCE) {
735 				omode = MCAST_EXCLUDE;
736 				add = 1;
737 			} else if (optname == MCAST_UNBLOCK_SOURCE) {
738 				omode = MCAST_EXCLUDE;
739 				add = 0;
740 			} else if (optname == MCAST_JOIN_SOURCE_GROUP) {
741 				struct ip_mreqn mreq;
742 
743 				psin = (struct sockaddr_in *)&greqs.gsr_group;
744 				mreq.imr_multiaddr = psin->sin_addr;
745 				mreq.imr_address.s_addr = 0;
746 				mreq.imr_ifindex = greqs.gsr_interface;
747 				err = ip_mc_join_group(sk, &mreq);
748 				if (err)
749 					break;
750 				greqs.gsr_interface = mreq.imr_ifindex;
751 				omode = MCAST_INCLUDE;
752 				add = 1;
753 			} else /* MCAST_LEAVE_SOURCE_GROUP */ {
754 				omode = MCAST_INCLUDE;
755 				add = 0;
756 			}
757 			err = ip_mc_source(add, omode, sk, &mreqs,
758 				greqs.gsr_interface);
759 			break;
760 		}
761 		case MCAST_MSFILTER:
762 		{
763 			extern int sysctl_optmem_max;
764 			extern int sysctl_igmp_max_msf;
765 			struct sockaddr_in *psin;
766 			struct ip_msfilter *msf = 0;
767 			struct group_filter *gsf = 0;
768 			int msize, i, ifindex;
769 
770 			if (optlen < GROUP_FILTER_SIZE(0))
771 				goto e_inval;
772 			if (optlen > sysctl_optmem_max) {
773 				err = -ENOBUFS;
774 				break;
775 			}
776 			gsf = (struct group_filter *)kmalloc(optlen,GFP_KERNEL);
777 			if (gsf == 0) {
778 				err = -ENOBUFS;
779 				break;
780 			}
781 			err = -EFAULT;
782 			if (copy_from_user(gsf, optval, optlen)) {
783 				goto mc_msf_out;
784 			}
785 			/* numsrc >= (4G-140)/128 overflow in 32 bits */
786 			if (gsf->gf_numsrc >= 0x1ffffff ||
787 			    gsf->gf_numsrc > sysctl_igmp_max_msf) {
788 				err = -ENOBUFS;
789 				goto mc_msf_out;
790 			}
791 			if (GROUP_FILTER_SIZE(gsf->gf_numsrc) > optlen) {
792 				err = -EINVAL;
793 				goto mc_msf_out;
794 			}
795 			msize = IP_MSFILTER_SIZE(gsf->gf_numsrc);
796 			msf = (struct ip_msfilter *)kmalloc(msize,GFP_KERNEL);
797 			if (msf == 0) {
798 				err = -ENOBUFS;
799 				goto mc_msf_out;
800 			}
801 			ifindex = gsf->gf_interface;
802 			psin = (struct sockaddr_in *)&gsf->gf_group;
803 			if (psin->sin_family != AF_INET) {
804 				err = -EADDRNOTAVAIL;
805 				goto mc_msf_out;
806 			}
807 			msf->imsf_multiaddr = psin->sin_addr.s_addr;
808 			msf->imsf_interface = 0;
809 			msf->imsf_fmode = gsf->gf_fmode;
810 			msf->imsf_numsrc = gsf->gf_numsrc;
811 			err = -EADDRNOTAVAIL;
812 			for (i=0; i<gsf->gf_numsrc; ++i) {
813 				psin = (struct sockaddr_in *)&gsf->gf_slist[i];
814 
815 				if (psin->sin_family != AF_INET)
816 					goto mc_msf_out;
817 				msf->imsf_slist[i] = psin->sin_addr.s_addr;
818 			}
819 			kfree(gsf);
820 			gsf = 0;
821 
822 			err = ip_mc_msfilter(sk, msf, ifindex);
823 mc_msf_out:
824 			if (msf)
825 				kfree(msf);
826 			if (gsf)
827 				kfree(gsf);
828 			break;
829 		}
830 		case IP_ROUTER_ALERT:
831 			err = ip_ra_control(sk, val ? 1 : 0, NULL);
832 			break;
833 
834 		case IP_FREEBIND:
835 			if (optlen<1)
836 				goto e_inval;
837 			sk->protinfo.af_inet.freebind = !!val;
838 	                break;
839 
840 		default:
841 #ifdef CONFIG_NETFILTER
842 			err = nf_setsockopt(sk, PF_INET, optname, optval,
843 					    optlen);
844 #else
845 			err = -ENOPROTOOPT;
846 #endif
847 			break;
848 	}
849 	release_sock(sk);
850 	return err;
851 
852 e_inval:
853 	release_sock(sk);
854 	return -EINVAL;
855 }
856 
857 /*
858  *	Get the options. Note for future reference. The GET of IP options gets the
859  *	_received_ ones. The set sets the _sent_ ones.
860  */
861 
ip_getsockopt(struct sock * sk,int level,int optname,char * optval,int * optlen)862 int ip_getsockopt(struct sock *sk, int level, int optname, char *optval, int *optlen)
863 {
864 	int val;
865 	int len;
866 
867 	if(level!=SOL_IP)
868 		return -EOPNOTSUPP;
869 
870 #ifdef CONFIG_IP_MROUTE
871 	if(optname>=MRT_BASE && optname <=MRT_BASE+10)
872 	{
873 		return ip_mroute_getsockopt(sk,optname,optval,optlen);
874 	}
875 #endif
876 
877 	if(get_user(len,optlen))
878 		return -EFAULT;
879 	if(len < 0)
880 		return -EINVAL;
881 
882 	lock_sock(sk);
883 
884 	switch(optname)	{
885 		case IP_OPTIONS:
886 			{
887 				unsigned char optbuf[sizeof(struct ip_options)+40];
888 				struct ip_options * opt = (struct ip_options*)optbuf;
889 				opt->optlen = 0;
890 				if (sk->protinfo.af_inet.opt)
891 					memcpy(optbuf, sk->protinfo.af_inet.opt,
892 					       sizeof(struct ip_options)+
893 					       sk->protinfo.af_inet.opt->optlen);
894 				release_sock(sk);
895 
896 				if (opt->optlen == 0)
897 					return put_user(0, optlen);
898 
899 				ip_options_undo(opt);
900 
901 				len = min_t(unsigned int, len, opt->optlen);
902 				if(put_user(len, optlen))
903 					return -EFAULT;
904 				if(copy_to_user(optval, opt->__data, len))
905 					return -EFAULT;
906 				return 0;
907 			}
908 		case IP_PKTINFO:
909 			val = (sk->protinfo.af_inet.cmsg_flags & IP_CMSG_PKTINFO) != 0;
910 			break;
911 		case IP_RECVTTL:
912 			val = (sk->protinfo.af_inet.cmsg_flags & IP_CMSG_TTL) != 0;
913 			break;
914 		case IP_RECVTOS:
915 			val = (sk->protinfo.af_inet.cmsg_flags & IP_CMSG_TOS) != 0;
916 			break;
917 		case IP_RECVOPTS:
918 			val = (sk->protinfo.af_inet.cmsg_flags & IP_CMSG_RECVOPTS) != 0;
919 			break;
920 		case IP_RETOPTS:
921 			val = (sk->protinfo.af_inet.cmsg_flags & IP_CMSG_RETOPTS) != 0;
922 			break;
923 		case IP_TOS:
924 			val=sk->protinfo.af_inet.tos;
925 			break;
926 		case IP_TTL:
927 			val=sk->protinfo.af_inet.ttl;
928 			break;
929 		case IP_HDRINCL:
930 			val=sk->protinfo.af_inet.hdrincl;
931 			break;
932 		case IP_MTU_DISCOVER:
933 			val=sk->protinfo.af_inet.pmtudisc;
934 			break;
935 		case IP_MTU:
936 		{
937 			struct dst_entry *dst;
938 			val = 0;
939 			dst = sk_dst_get(sk);
940 			if (dst) {
941 				val = dst->pmtu;
942 				dst_release(dst);
943 			}
944 			if (!val) {
945 				release_sock(sk);
946 				return -ENOTCONN;
947 			}
948 			break;
949 		}
950 		case IP_RECVERR:
951 			val=sk->protinfo.af_inet.recverr;
952 			break;
953 		case IP_MULTICAST_TTL:
954 			val=sk->protinfo.af_inet.mc_ttl;
955 			break;
956 		case IP_MULTICAST_LOOP:
957 			val=sk->protinfo.af_inet.mc_loop;
958 			break;
959 		case IP_MULTICAST_IF:
960 		{
961 			struct in_addr addr;
962 			len = min_t(unsigned int, len, sizeof(struct in_addr));
963 			addr.s_addr = sk->protinfo.af_inet.mc_addr;
964 			release_sock(sk);
965 
966   			if(put_user(len, optlen))
967   				return -EFAULT;
968 			if(copy_to_user((void *)optval, &addr, len))
969 				return -EFAULT;
970 			return 0;
971 		}
972 		case IP_MSFILTER:
973 		{
974 			struct ip_msfilter msf;
975 			int err;
976 
977 			if (len < IP_MSFILTER_SIZE(0)) {
978 				release_sock(sk);
979 				return -EINVAL;
980 			}
981 			if (copy_from_user(&msf, optval, IP_MSFILTER_SIZE(0))) {
982 				release_sock(sk);
983 				return -EFAULT;
984 			}
985 			err = ip_mc_msfget(sk, &msf,
986 				(struct ip_msfilter *)optval, optlen);
987 			release_sock(sk);
988 			return err;
989 		}
990 		case MCAST_MSFILTER:
991 		{
992 			struct group_filter gsf;
993 			int err;
994 
995 			if (len < GROUP_FILTER_SIZE(0)) {
996 				release_sock(sk);
997 				return -EINVAL;
998 			}
999 			if (copy_from_user(&gsf, optval, GROUP_FILTER_SIZE(0))) {
1000 				release_sock(sk);
1001 				return -EFAULT;
1002 			}
1003 			err = ip_mc_gsfget(sk, &gsf,
1004 				(struct group_filter *)optval, optlen);
1005 			release_sock(sk);
1006 			return err;
1007 		}
1008 		case IP_PKTOPTIONS:
1009 		{
1010 			struct msghdr msg;
1011 
1012 			release_sock(sk);
1013 
1014 			if (sk->type != SOCK_STREAM)
1015 				return -ENOPROTOOPT;
1016 
1017 			msg.msg_control = optval;
1018 			msg.msg_controllen = len;
1019 			msg.msg_flags = 0;
1020 
1021 			if (sk->protinfo.af_inet.cmsg_flags&IP_CMSG_PKTINFO) {
1022 				struct in_pktinfo info;
1023 
1024 				info.ipi_addr.s_addr = sk->rcv_saddr;
1025 				info.ipi_spec_dst.s_addr = sk->rcv_saddr;
1026 				info.ipi_ifindex = sk->protinfo.af_inet.mc_index;
1027 				put_cmsg(&msg, SOL_IP, IP_PKTINFO, sizeof(info), &info);
1028 			}
1029 			if (sk->protinfo.af_inet.cmsg_flags&IP_CMSG_TTL) {
1030 				int hlim = sk->protinfo.af_inet.mc_ttl;
1031 				put_cmsg(&msg, SOL_IP, IP_TTL, sizeof(hlim), &hlim);
1032 			}
1033 			len -= msg.msg_controllen;
1034 			return put_user(len, optlen);
1035 		}
1036 		case IP_FREEBIND:
1037 			val = sk->protinfo.af_inet.freebind;
1038 			break;
1039 		default:
1040 #ifdef CONFIG_NETFILTER
1041 			val = nf_getsockopt(sk, PF_INET, optname, optval,
1042 					    &len);
1043 			release_sock(sk);
1044 			if (val >= 0)
1045 				val = put_user(len, optlen);
1046 			return val;
1047 #else
1048 			release_sock(sk);
1049 			return -ENOPROTOOPT;
1050 #endif
1051 	}
1052 	release_sock(sk);
1053 
1054 	if (len < sizeof(int) && len > 0 && val>=0 && val<255) {
1055 		unsigned char ucval = (unsigned char)val;
1056 		len = 1;
1057 		if(put_user(len, optlen))
1058 			return -EFAULT;
1059 		if(copy_to_user(optval,&ucval,1))
1060 			return -EFAULT;
1061 	} else {
1062 		len = min_t(unsigned int, sizeof(int), len);
1063 		if(put_user(len, optlen))
1064 			return -EFAULT;
1065 		if(copy_to_user(optval,&val,len))
1066 			return -EFAULT;
1067 	}
1068 	return 0;
1069 }
1070