[linux.git] / net / ipv4 / ip_options.c

/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		The options processing module for ip.c
 *
 * Authors:	A.N.Kuznetsov
 *
 */

#include <linux/capability.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/types.h>
#include <asm/uaccess.h>
#include <linux/skbuff.h>
#include <linux/ip.h>
#include <linux/icmp.h>
#include <linux/netdevice.h>
#include <linux/rtnetlink.h>
#include <net/sock.h>
#include <net/ip.h>
#include <net/icmp.h>
#include <net/route.h>
#include <net/cipso_ipv4.h>

/*
 * Write options to IP header, record destination address to
 * source route option, address of outgoing interface
 * (we should already know it, so that this  function is allowed be
 * called only after routing decision) and timestamp,
 * if we originate this datagram.
 *
 * daddr is real destination address, next hop is recorded in IP header.
 * saddr is address of outgoing interface.
 */

void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
			    __be32 daddr, struct rtable *rt, int is_frag)
{
	unsigned char *iph = skb_network_header(skb);

	memcpy(&(IPCB(skb)->opt), opt, sizeof(struct ip_options));
	memcpy(iph+sizeof(struct iphdr), opt->__data, opt->optlen);
	opt = &(IPCB(skb)->opt);

	if (opt->srr)
		memcpy(iph+opt->srr+iph[opt->srr+1]-4, &daddr, 4);

	if (!is_frag) {
		if (opt->rr_needaddr)
			ip_rt_get_source(iph+opt->rr+iph[opt->rr+2]-5, rt);
		if (opt->ts_needaddr)
			ip_rt_get_source(iph+opt->ts+iph[opt->ts+2]-9, rt);
		if (opt->ts_needtime) {
			struct timespec tv;
			__be32 midtime;
			getnstimeofday(&tv);
			midtime = htonl((tv.tv_sec % 86400) * MSEC_PER_SEC + tv.tv_nsec / NSEC_PER_MSEC);
			memcpy(iph+opt->ts+iph[opt->ts+2]-5, &midtime, 4);
		}
		return;
	}
	if (opt->rr) {
		memset(iph+opt->rr, IPOPT_NOP, iph[opt->rr+1]);
		opt->rr = 0;
		opt->rr_needaddr = 0;
	}
	if (opt->ts) {
		memset(iph+opt->ts, IPOPT_NOP, iph[opt->ts+1]);
		opt->ts = 0;
		opt->ts_needaddr = opt->ts_needtime = 0;
	}
}

/*
 * Provided (sopt, skb) points to received options,
 * build in dopt compiled option set appropriate for answering.
 * i.e. invert SRR option, copy anothers,
 * and grab room in RR/TS options.
 *
 * NOTE: dopt cannot point to skb.
 */

int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb)
{
	const struct ip_options *sopt;
	unsigned char *sptr, *dptr;
	int soffset, doffset;
	int	optlen;
	__be32	daddr;

	memset(dopt, 0, sizeof(struct ip_options));

	sopt = &(IPCB(skb)->opt);

	if (sopt->optlen == 0)
		return 0;

	sptr = skb_network_header(skb);
	dptr = dopt->__data;

	daddr = skb_rtable(skb)->rt_spec_dst;

	if (sopt->rr) {
		optlen  = sptr[sopt->rr+1];
		soffset = sptr[sopt->rr+2];
		dopt->rr = dopt->optlen + sizeof(struct iphdr);
		memcpy(dptr, sptr+sopt->rr, optlen);
		if (sopt->rr_needaddr && soffset <= optlen) {
			if (soffset + 3 > optlen)
				return -EINVAL;
			dptr[2] = soffset + 4;
			dopt->rr_needaddr = 1;
		}
		dptr += optlen;
		dopt->optlen += optlen;
	}
	if (sopt->ts) {
		optlen = sptr[sopt->ts+1];
		soffset = sptr[sopt->ts+2];
		dopt->ts = dopt->optlen + sizeof(struct iphdr);
		memcpy(dptr, sptr+sopt->ts, optlen);
		if (soffset <= optlen) {
			if (sopt->ts_needaddr) {
				if (soffset + 3 > optlen)
					return -EINVAL;
				dopt->ts_needaddr = 1;
				soffset += 4;
			}
			if (sopt->ts_needtime) {
				if (soffset + 3 > optlen)
					return -EINVAL;
				if ((dptr[3]&0xF) != IPOPT_TS_PRESPEC) {
					dopt->ts_needtime = 1;
					soffset += 4;
				} else {
					dopt->ts_needtime = 0;

					if (soffset + 7 <= optlen) {
						__be32 addr;

						memcpy(&addr, dptr+soffset-1, 4);
						if (inet_addr_type(dev_net(skb_dst(skb)->dev), addr) != RTN_UNICAST) {
							dopt->ts_needtime = 1;
							soffset += 8;
						}
					}
				}
			}
			dptr[2] = soffset;
		}
		dptr += optlen;
		dopt->optlen += optlen;
	}
	if (sopt->srr) {
		unsigned char *start = sptr+sopt->srr;
		__be32 faddr;

		optlen  = start[1];
		soffset = start[2];
		doffset = 0;
		if (soffset > optlen)
			soffset = optlen + 1;
		soffset -= 4;
		if (soffset > 3) {
			memcpy(&faddr, &start[soffset-1], 4);
			for (soffset-=4, doffset=4; soffset > 3; soffset-=4, doffset+=4)
				memcpy(&dptr[doffset-1], &start[soffset-1], 4);
			/*
			 * RFC1812 requires to fix illegal source routes.
			 */
			if (memcmp(&ip_hdr(skb)->saddr,
				   &start[soffset + 3], 4) == 0)
				doffset -= 4;
		}
		if (doffset > 3) {
			memcpy(&start[doffset-1], &daddr, 4);
			dopt->faddr = faddr;
			dptr[0] = start[0];
			dptr[1] = doffset+3;
			dptr[2] = 4;
			dptr += doffset+3;
			dopt->srr = dopt->optlen + sizeof(struct iphdr);
			dopt->optlen += doffset+3;
			dopt->is_strictroute = sopt->is_strictroute;
		}
	}
	if (sopt->cipso) {
		optlen  = sptr[sopt->cipso+1];
		dopt->cipso = dopt->optlen+sizeof(struct iphdr);
		memcpy(dptr, sptr+sopt->cipso, optlen);
		dptr += optlen;
		dopt->optlen += optlen;
	}
	while (dopt->optlen & 3) {
		*dptr++ = IPOPT_END;
		dopt->optlen++;
	}
	return 0;
}

/*
 *	Options "fragmenting", just fill options not
 *	allowed in fragments with NOOPs.
 *	Simple and stupid 8), but the most efficient way.
 */

void ip_options_fragment(struct sk_buff * skb)
{
	unsigned char *optptr = skb_network_header(skb) + sizeof(struct iphdr);
	struct ip_options * opt = &(IPCB(skb)->opt);
	int  l = opt->optlen;
	int  optlen;

	while (l > 0) {
		switch (*optptr) {
		case IPOPT_END:
			return;
		case IPOPT_NOOP:
			l--;
			optptr++;
			continue;
		}
		optlen = optptr[1];
		if (optlen<2 || optlen>l)
		  return;
		if (!IPOPT_COPIED(*optptr))
			memset(optptr, IPOPT_NOOP, optlen);
		l -= optlen;
		optptr += optlen;
	}
	opt->ts = 0;
	opt->rr = 0;
	opt->rr_needaddr = 0;
	opt->ts_needaddr = 0;
	opt->ts_needtime = 0;
}

/*
 * Verify options and fill pointers in struct options.
 * Caller should clear *opt, and set opt->data.
 * If opt == NULL, then skb->data should point to IP header.
 */

int ip_options_compile(struct net *net,
		       struct ip_options * opt, struct sk_buff * skb)
{
	int l;
	unsigned char * iph;
	unsigned char * optptr;
	int optlen;
	unsigned char * pp_ptr = NULL;
	struct rtable *rt = NULL;

	if (skb != NULL) {
		rt = skb_rtable(skb);
		optptr = (unsigned char *)&(ip_hdr(skb)[1]);
	} else
		optptr = opt->__data;
	iph = optptr - sizeof(struct iphdr);

	for (l = opt->optlen; l > 0; ) {
		switch (*optptr) {
		      case IPOPT_END:
			for (optptr++, l--; l>0; optptr++, l--) {
				if (*optptr != IPOPT_END) {
					*optptr = IPOPT_END;
					opt->is_changed = 1;
				}
			}
			goto eol;
		      case IPOPT_NOOP:
			l--;
			optptr++;
			continue;
		}
		optlen = optptr[1];
		if (optlen<2 || optlen>l) {
			pp_ptr = optptr;
			goto error;
		}
		switch (*optptr) {
		      case IPOPT_SSRR:
		      case IPOPT_LSRR:
			if (optlen < 3) {
				pp_ptr = optptr + 1;
				goto error;
			}
			if (optptr[2] < 4) {
				pp_ptr = optptr + 2;
				goto error;
			}
			/* NB: cf RFC-1812 5.2.4.1 */
			if (opt->srr) {
				pp_ptr = optptr;
				goto error;
			}
			if (!skb) {
				if (optptr[2] != 4 || optlen < 7 || ((optlen-3) & 3)) {
					pp_ptr = optptr + 1;
					goto error;
				}
				memcpy(&opt->faddr, &optptr[3], 4);
				if (optlen > 7)
					memmove(&optptr[3], &optptr[7], optlen-7);
			}
			opt->is_strictroute = (optptr[0] == IPOPT_SSRR);
			opt->srr = optptr - iph;
			break;
		      case IPOPT_RR:
			if (opt->rr) {
				pp_ptr = optptr;
				goto error;
			}
			if (optlen < 3) {
				pp_ptr = optptr + 1;
				goto error;
			}
			if (optptr[2] < 4) {
				pp_ptr = optptr + 2;
				goto error;
			}
			if (optptr[2] <= optlen) {
				if (optptr[2]+3 > optlen) {
					pp_ptr = optptr + 2;
					goto error;
				}
				if (rt) {
					memcpy(&optptr[optptr[2]-1], &rt->rt_spec_dst, 4);
					opt->is_changed = 1;
				}
				optptr[2] += 4;
				opt->rr_needaddr = 1;
			}
			opt->rr = optptr - iph;
			break;
		      case IPOPT_TIMESTAMP:
			if (opt->ts) {
				pp_ptr = optptr;
				goto error;
			}
			if (optlen < 4) {
				pp_ptr = optptr + 1;
				goto error;
			}
			if (optptr[2] < 5) {
				pp_ptr = optptr + 2;
				goto error;
			}
			if (optptr[2] <= optlen) {
				__be32 *timeptr = NULL;
				if (optptr[2]+3 > optptr[1]) {
					pp_ptr = optptr + 2;
					goto error;
				}
				switch (optptr[3]&0xF) {
				      case IPOPT_TS_TSONLY:
					opt->ts = optptr - iph;
					if (skb)
						timeptr = (__be32*)&optptr[optptr[2]-1];
					opt->ts_needtime = 1;
					optptr[2] += 4;
					break;
				      case IPOPT_TS_TSANDADDR:
					if (optptr[2]+7 > optptr[1]) {
						pp_ptr = optptr + 2;
						goto error;
					}
					opt->ts = optptr - iph;
					if (rt)  {
						memcpy(&optptr[optptr[2]-1], &rt->rt_spec_dst, 4);
						timeptr = (__be32*)&optptr[optptr[2]+3];
					}
					opt->ts_needaddr = 1;
					opt->ts_needtime = 1;
					optptr[2] += 8;
					break;
				      case IPOPT_TS_PRESPEC:
					if (optptr[2]+7 > optptr[1]) {
						pp_ptr = optptr + 2;
						goto error;
					}
					opt->ts = optptr - iph;
					{
						__be32 addr;
						memcpy(&addr, &optptr[optptr[2]-1], 4);
						if (inet_addr_type(net, addr) == RTN_UNICAST)
							break;
						if (skb)
							timeptr = (__be32*)&optptr[optptr[2]+3];
					}
					opt->ts_needtime = 1;
					optptr[2] += 8;
					break;
				      default:
					if (!skb && !capable(CAP_NET_RAW)) {
						pp_ptr = optptr + 3;
						goto error;
					}
					break;
				}
				if (timeptr) {
					struct timespec tv;
					__be32  midtime;
					getnstimeofday(&tv);
					midtime = htonl((tv.tv_sec % 86400) * MSEC_PER_SEC + tv.tv_nsec / NSEC_PER_MSEC);
					memcpy(timeptr, &midtime, sizeof(__be32));
					opt->is_changed = 1;
				}
			} else {
				unsigned overflow = optptr[3]>>4;
				if (overflow == 15) {
					pp_ptr = optptr + 3;
					goto error;
				}
				opt->ts = optptr - iph;
				if (skb) {
					optptr[3] = (optptr[3]&0xF)|((overflow+1)<<4);
					opt->is_changed = 1;
				}
			}
			break;
		      case IPOPT_RA:
			if (optlen < 4) {
				pp_ptr = optptr + 1;
				goto error;
			}
			if (optptr[2] == 0 && optptr[3] == 0)
				opt->router_alert = optptr - iph;
			break;
		      case IPOPT_CIPSO:
			if ((!skb && !capable(CAP_NET_RAW)) || opt->cipso) {
				pp_ptr = optptr;
				goto error;
			}
			opt->cipso = optptr - iph;
			if (cipso_v4_validate(skb, &optptr)) {
				pp_ptr = optptr;
				goto error;
			}
			break;
		      case IPOPT_SEC:
		      case IPOPT_SID:
		      default:
			if (!skb && !capable(CAP_NET_RAW)) {
				pp_ptr = optptr;
				goto error;
			}
			break;
		}
		l -= optlen;
		optptr += optlen;
	}

eol:
	if (!pp_ptr)
		return 0;

error:
	if (skb) {
		icmp_send(skb, ICMP_PARAMETERPROB, 0, htonl((pp_ptr-iph)<<24));
	}
	return -EINVAL;
}
EXPORT_SYMBOL(ip_options_compile);

/*
 *	Undo all the changes done by ip_options_compile().
 */

void ip_options_undo(struct ip_options * opt)
{
	if (opt->srr) {
		unsigned  char * optptr = opt->__data+opt->srr-sizeof(struct  iphdr);
		memmove(optptr+7, optptr+3, optptr[1]-7);
		memcpy(optptr+3, &opt->faddr, 4);
	}
	if (opt->rr_needaddr) {
		unsigned  char * optptr = opt->__data+opt->rr-sizeof(struct  iphdr);
		optptr[2] -= 4;
		memset(&optptr[optptr[2]-1], 0, 4);
	}
	if (opt->ts) {
		unsigned  char * optptr = opt->__data+opt->ts-sizeof(struct  iphdr);
		if (opt->ts_needtime) {
			optptr[2] -= 4;
			memset(&optptr[optptr[2]-1], 0, 4);
			if ((optptr[3]&0xF) == IPOPT_TS_PRESPEC)
				optptr[2] -= 4;
		}
		if (opt->ts_needaddr) {
			optptr[2] -= 4;
			memset(&optptr[optptr[2]-1], 0, 4);
		}
	}
}

static struct ip_options_rcu *ip_options_get_alloc(const int optlen)
{
	return kzalloc(sizeof(struct ip_options_rcu) + ((optlen + 3) & ~3),
		       GFP_KERNEL);
}

static int ip_options_get_finish(struct net *net, struct ip_options_rcu **optp,
				 struct ip_options_rcu *opt, int optlen)
{
	while (optlen & 3)
		opt->opt.__data[optlen++] = IPOPT_END;
	opt->opt.optlen = optlen;
	if (optlen && ip_options_compile(net, &opt->opt, NULL)) {
		kfree(opt);
		return -EINVAL;
	}
	kfree(*optp);
	*optp = opt;
	return 0;
}

int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
			     unsigned char __user *data, int optlen)
{
	struct ip_options_rcu *opt = ip_options_get_alloc(optlen);

	if (!opt)
		return -ENOMEM;
	if (optlen && copy_from_user(opt->opt.__data, data, optlen)) {
		kfree(opt);
		return -EFAULT;
	}
	return ip_options_get_finish(net, optp, opt, optlen);
}

int ip_options_get(struct net *net, struct ip_options_rcu **optp,
		   unsigned char *data, int optlen)
{
	struct ip_options_rcu *opt = ip_options_get_alloc(optlen);

	if (!opt)
		return -ENOMEM;
	if (optlen)
		memcpy(opt->opt.__data, data, optlen);
	return ip_options_get_finish(net, optp, opt, optlen);
}

void ip_forward_options(struct sk_buff *skb)
{
	struct   ip_options * opt	= &(IPCB(skb)->opt);
	unsigned char * optptr;
	struct rtable *rt = skb_rtable(skb);
	unsigned char *raw = skb_network_header(skb);

	if (opt->rr_needaddr) {
		optptr = (unsigned char *)raw + opt->rr;
		ip_rt_get_source(&optptr[optptr[2]-5], rt);
		opt->is_changed = 1;
	}
	if (opt->srr_is_hit) {
		int srrptr, srrspace;

		optptr = raw + opt->srr;

		for ( srrptr=optptr[2], srrspace = optptr[1];
		     srrptr <= srrspace;
		     srrptr += 4
		     ) {
			if (srrptr + 3 > srrspace)
				break;
			if (memcmp(&rt->rt_dst, &optptr[srrptr-1], 4) == 0)
				break;
		}
		if (srrptr + 3 <= srrspace) {
			opt->is_changed = 1;
			ip_rt_get_source(&optptr[srrptr-1], rt);
			ip_hdr(skb)->daddr = rt->rt_dst;
			optptr[2] = srrptr+4;
		} else if (net_ratelimit())
			printk(KERN_CRIT "ip_forward(): Argh! Destination lost!\n");
		if (opt->ts_needaddr) {
			optptr = raw + opt->ts;
			ip_rt_get_source(&optptr[optptr[2]-9], rt);
			opt->is_changed = 1;
		}
	}
	if (opt->is_changed) {
		opt->is_changed = 0;
		ip_send_check(ip_hdr(skb));
	}
}

int ip_options_rcv_srr(struct sk_buff *skb)
{
	struct ip_options *opt = &(IPCB(skb)->opt);
	int srrspace, srrptr;
	__be32 nexthop;
	struct iphdr *iph = ip_hdr(skb);
	unsigned char *optptr = skb_network_header(skb) + opt->srr;
	struct rtable *rt = skb_rtable(skb);
	struct rtable *rt2;
	unsigned long orefdst;
	int err;

	if (!opt->srr || !rt)
		return 0;

	if (skb->pkt_type != PACKET_HOST)
		return -EINVAL;
	if (rt->rt_type == RTN_UNICAST) {
		if (!opt->is_strictroute)
			return 0;
		icmp_send(skb, ICMP_PARAMETERPROB, 0, htonl(16<<24));
		return -EINVAL;
	}
	if (rt->rt_type != RTN_LOCAL)
		return -EINVAL;

	for (srrptr=optptr[2], srrspace = optptr[1]; srrptr <= srrspace; srrptr += 4) {
		if (srrptr + 3 > srrspace) {
			icmp_send(skb, ICMP_PARAMETERPROB, 0, htonl((opt->srr+2)<<24));
			return -EINVAL;
		}
		memcpy(&nexthop, &optptr[srrptr-1], 4);

		orefdst = skb->_skb_refdst;
		skb_dst_set(skb, NULL);
		err = ip_route_input(skb, nexthop, iph->saddr, iph->tos, skb->dev);
		rt2 = skb_rtable(skb);
		if (err || (rt2->rt_type != RTN_UNICAST && rt2->rt_type != RTN_LOCAL)) {
			skb_dst_drop(skb);
			skb->_skb_refdst = orefdst;
			return -EINVAL;
		}
		refdst_drop(orefdst);
		if (rt2->rt_type != RTN_LOCAL)
			break;
		/* Superfast 8) loopback forward */
		memcpy(&iph->daddr, &optptr[srrptr-1], 4);
		opt->is_changed = 1;
	}
	if (srrptr <= srrspace) {
		opt->srr_is_hit = 1;
		opt->is_changed = 1;
	}
	return 0;
}
EXPORT_SYMBOL(ip_options_rcv_srr);
Commit	Line	Data
1da177e4 LT	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 options processing module for ip.c
	7	*
1da177e4	8	* Authors: A.N.Kuznetsov
e905a9ed	9	*
1da177e4 LT	10	*/
1da177e4 LT	11
4fc268d2	12	#include <linux/capability.h>
1da177e4	13	#include <linux/module.h>
5a0e3ad6	14	#include <linux/slab.h>
1da177e4 LT	15	#include <linux/types.h>
	16	#include <asm/uaccess.h>
	17	#include <linux/skbuff.h>
	18	#include <linux/ip.h>
	19	#include <linux/icmp.h>
	20	#include <linux/netdevice.h>
	21	#include <linux/rtnetlink.h>
	22	#include <net/sock.h>
	23	#include <net/ip.h>
	24	#include <net/icmp.h>
14c85021	25	#include <net/route.h>
11a03f78	26	#include <net/cipso_ipv4.h>
1da177e4	27
e905a9ed	28	/*
1da177e4 LT	29	* Write options to IP header, record destination address to
	30	* source route option, address of outgoing interface
	31	* (we should already know it, so that this function is allowed be
	32	* called only after routing decision) and timestamp,
	33	* if we originate this datagram.
	34	*
	35	* daddr is real destination address, next hop is recorded in IP header.
	36	* saddr is address of outgoing interface.
	37	*/
	38
f6d8bd05	39	void ip_options_build(struct sk_buff skb, struct ip_options opt,
8712f774	40	__be32 daddr, struct rtable *rt, int is_frag)
1da177e4	41	{
d56f90a7	42	unsigned char *iph = skb_network_header(skb);
1da177e4 LT	43
	44	memcpy(&(IPCB(skb)->opt), opt, sizeof(struct ip_options));
	45	memcpy(iph+sizeof(struct iphdr), opt->__data, opt->optlen);
	46	opt = &(IPCB(skb)->opt);
1da177e4 LT	47
	48	if (opt->srr)
	49	memcpy(iph+opt->srr+iph[opt->srr+1]-4, &daddr, 4);
	50
	51	if (!is_frag) {
	52	if (opt->rr_needaddr)
	53	ip_rt_get_source(iph+opt->rr+iph[opt->rr+2]-5, rt);
	54	if (opt->ts_needaddr)
	55	ip_rt_get_source(iph+opt->ts+iph[opt->ts+2]-9, rt);
	56	if (opt->ts_needtime) {
f25c3d61	57	struct timespec tv;
e25d2ca6	58	__be32 midtime;
f25c3d61 YH	59	getnstimeofday(&tv);
f25c3d61 YH	60	midtime = htonl((tv.tv_sec % 86400) * MSEC_PER_SEC + tv.tv_nsec / NSEC_PER_MSEC);
1da177e4 LT	61	memcpy(iph+opt->ts+iph[opt->ts+2]-5, &midtime, 4);
	62	}
	63	return;
	64	}
	65	if (opt->rr) {
	66	memset(iph+opt->rr, IPOPT_NOP, iph[opt->rr+1]);
	67	opt->rr = 0;
	68	opt->rr_needaddr = 0;
	69	}
	70	if (opt->ts) {
	71	memset(iph+opt->ts, IPOPT_NOP, iph[opt->ts+1]);
	72	opt->ts = 0;
	73	opt->ts_needaddr = opt->ts_needtime = 0;
	74	}
	75	}
	76
e905a9ed	77	/*
1da177e4 LT	78	* Provided (sopt, skb) points to received options,
	79	* build in dopt compiled option set appropriate for answering.
	80	* i.e. invert SRR option, copy anothers,
	81	* and grab room in RR/TS options.
	82	*
	83	* NOTE: dopt cannot point to skb.
	84	*/
	85
f6d8bd05	86	int ip_options_echo(struct ip_options dopt, struct sk_buff skb)
1da177e4	87	{
f6d8bd05	88	const struct ip_options *sopt;
1da177e4 LT	89	unsigned char sptr, dptr;
	90	int soffset, doffset;
	91	int optlen;
e25d2ca6	92	__be32 daddr;
1da177e4 LT	93
	94	memset(dopt, 0, sizeof(struct ip_options));
	95
1da177e4 LT	96	sopt = &(IPCB(skb)->opt);
1da177e4 LT	97
f6d8bd05	98	if (sopt->optlen == 0)
1da177e4	99	return 0;
1da177e4	100
d56f90a7	101	sptr = skb_network_header(skb);
1da177e4 LT	102	dptr = dopt->__data;
1da177e4 LT	103
511c3f92	104	daddr = skb_rtable(skb)->rt_spec_dst;
1da177e4 LT	105
	106	if (sopt->rr) {
	107	optlen = sptr[sopt->rr+1];
	108	soffset = sptr[sopt->rr+2];
	109	dopt->rr = dopt->optlen + sizeof(struct iphdr);
	110	memcpy(dptr, sptr+sopt->rr, optlen);
	111	if (sopt->rr_needaddr && soffset <= optlen) {
	112	if (soffset + 3 > optlen)
	113	return -EINVAL;
	114	dptr[2] = soffset + 4;
	115	dopt->rr_needaddr = 1;
	116	}
	117	dptr += optlen;
	118	dopt->optlen += optlen;
	119	}
	120	if (sopt->ts) {
	121	optlen = sptr[sopt->ts+1];
	122	soffset = sptr[sopt->ts+2];
	123	dopt->ts = dopt->optlen + sizeof(struct iphdr);
	124	memcpy(dptr, sptr+sopt->ts, optlen);
	125	if (soffset <= optlen) {
	126	if (sopt->ts_needaddr) {
	127	if (soffset + 3 > optlen)
	128	return -EINVAL;
	129	dopt->ts_needaddr = 1;
	130	soffset += 4;
	131	}
	132	if (sopt->ts_needtime) {
	133	if (soffset + 3 > optlen)
	134	return -EINVAL;
	135	if ((dptr[3]&0xF) != IPOPT_TS_PRESPEC) {
	136	dopt->ts_needtime = 1;
	137	soffset += 4;
	138	} else {
	139	dopt->ts_needtime = 0;
	140
8628bd8a	141	if (soffset + 7 <= optlen) {
fd683222	142	__be32 addr;
1da177e4	143
8628bd8a JL	144	memcpy(&addr, dptr+soffset-1, 4);
8628bd8a JL	145	if (inet_addr_type(dev_net(skb_dst(skb)->dev), addr) != RTN_UNICAST) {
1da177e4 LT	146	dopt->ts_needtime = 1;
	147	soffset += 8;
	148	}
	149	}
	150	}
	151	}
	152	dptr[2] = soffset;
	153	}
	154	dptr += optlen;
	155	dopt->optlen += optlen;
	156	}
	157	if (sopt->srr) {
f6d8bd05	158	unsigned char *start = sptr+sopt->srr;
3ca3c68e	159	__be32 faddr;
1da177e4 LT	160
	161	optlen = start[1];
	162	soffset = start[2];
	163	doffset = 0;
	164	if (soffset > optlen)
	165	soffset = optlen + 1;
	166	soffset -= 4;
	167	if (soffset > 3) {
	168	memcpy(&faddr, &start[soffset-1], 4);
	169	for (soffset-=4, doffset=4; soffset > 3; soffset-=4, doffset+=4)
	170	memcpy(&dptr[doffset-1], &start[soffset-1], 4);
	171	/*
	172	* RFC1812 requires to fix illegal source routes.
	173	*/
eddc9ec5 ACM	174	if (memcmp(&ip_hdr(skb)->saddr,
eddc9ec5 ACM	175	&start[soffset + 3], 4) == 0)
1da177e4 LT	176	doffset -= 4;
	177	}
	178	if (doffset > 3) {
	179	memcpy(&start[doffset-1], &daddr, 4);
	180	dopt->faddr = faddr;
	181	dptr[0] = start[0];
	182	dptr[1] = doffset+3;
	183	dptr[2] = 4;
	184	dptr += doffset+3;
	185	dopt->srr = dopt->optlen + sizeof(struct iphdr);
	186	dopt->optlen += doffset+3;
	187	dopt->is_strictroute = sopt->is_strictroute;
	188	}
	189	}
11a03f78 PM	190	if (sopt->cipso) {
	191	optlen = sptr[sopt->cipso+1];
	192	dopt->cipso = dopt->optlen+sizeof(struct iphdr);
	193	memcpy(dptr, sptr+sopt->cipso, optlen);
	194	dptr += optlen;
	195	dopt->optlen += optlen;
	196	}
1da177e4 LT	197	while (dopt->optlen & 3) {
	198	*dptr++ = IPOPT_END;
	199	dopt->optlen++;
	200	}
	201	return 0;
	202	}
	203
	204	/*
	205	* Options "fragmenting", just fill options not
	206	* allowed in fragments with NOOPs.
	207	* Simple and stupid 8), but the most efficient way.
	208	*/
	209
e905a9ed	210	void ip_options_fragment(struct sk_buff * skb)
1da177e4	211	{
d56f90a7	212	unsigned char *optptr = skb_network_header(skb) + sizeof(struct iphdr);
1da177e4 LT	213	struct ip_options * opt = &(IPCB(skb)->opt);
	214	int l = opt->optlen;
	215	int optlen;
	216
	217	while (l > 0) {
	218	switch (*optptr) {
	219	case IPOPT_END:
	220	return;
	221	case IPOPT_NOOP:
	222	l--;
	223	optptr++;
	224	continue;
	225	}
	226	optlen = optptr[1];
	227	if (optlen<2 \|\| optlen>l)
	228	return;
	229	if (!IPOPT_COPIED(*optptr))
	230	memset(optptr, IPOPT_NOOP, optlen);
	231	l -= optlen;
	232	optptr += optlen;
	233	}
	234	opt->ts = 0;
	235	opt->rr = 0;
	236	opt->rr_needaddr = 0;
	237	opt->ts_needaddr = 0;
	238	opt->ts_needtime = 0;
1da177e4 LT	239	}
	240
	241	/*
	242	* Verify options and fill pointers in struct options.
	243	* Caller should clear *opt, and set opt->data.
	244	* If opt == NULL, then skb->data should point to IP header.
	245	*/
	246
0e6bd4a1 DL	247	int ip_options_compile(struct net *net,
0e6bd4a1 DL	248	struct ip_options * opt, struct sk_buff * skb)
1da177e4 LT	249	{
	250	int l;
	251	unsigned char * iph;
	252	unsigned char * optptr;
	253	int optlen;
	254	unsigned char * pp_ptr = NULL;
22aba383	255	struct rtable *rt = NULL;
1da177e4	256
22aba383	257	if (skb != NULL) {
511c3f92	258	rt = skb_rtable(skb);
22aba383 DL	259	optptr = (unsigned char *)&(ip_hdr(skb)[1]);
22aba383 DL	260	} else
10fe7d85	261	optptr = opt->__data;
22aba383	262	iph = optptr - sizeof(struct iphdr);
1da177e4 LT	263
	264	for (l = opt->optlen; l > 0; ) {
	265	switch (*optptr) {
	266	case IPOPT_END:
	267	for (optptr++, l--; l>0; optptr++, l--) {
	268	if (*optptr != IPOPT_END) {
	269	*optptr = IPOPT_END;
	270	opt->is_changed = 1;
	271	}
	272	}
	273	goto eol;
	274	case IPOPT_NOOP:
	275	l--;
	276	optptr++;
	277	continue;
	278	}
	279	optlen = optptr[1];
	280	if (optlen<2 \|\| optlen>l) {
	281	pp_ptr = optptr;
	282	goto error;
	283	}
	284	switch (*optptr) {
	285	case IPOPT_SSRR:
	286	case IPOPT_LSRR:
	287	if (optlen < 3) {
	288	pp_ptr = optptr + 1;
	289	goto error;
	290	}
	291	if (optptr[2] < 4) {
	292	pp_ptr = optptr + 2;
	293	goto error;
	294	}
	295	/* NB: cf RFC-1812 5.2.4.1 */
	296	if (opt->srr) {
	297	pp_ptr = optptr;
	298	goto error;
	299	}
	300	if (!skb) {
	301	if (optptr[2] != 4 \|\| optlen < 7 \|\| ((optlen-3) & 3)) {
	302	pp_ptr = optptr + 1;
	303	goto error;
	304	}
	305	memcpy(&opt->faddr, &optptr[3], 4);
	306	if (optlen > 7)
	307	memmove(&optptr[3], &optptr[7], optlen-7);
	308	}
	309	opt->is_strictroute = (optptr[0] == IPOPT_SSRR);
	310	opt->srr = optptr - iph;
	311	break;
	312	case IPOPT_RR:
	313	if (opt->rr) {
	314	pp_ptr = optptr;
	315	goto error;
	316	}
	317	if (optlen < 3) {
	318	pp_ptr = optptr + 1;
	319	goto error;
	320	}
	321	if (optptr[2] < 4) {
	322	pp_ptr = optptr + 2;
	323	goto error;
	324	}
	325	if (optptr[2] <= optlen) {
	326	if (optptr[2]+3 > optlen) {
327	pp_ptr = optptr + 2;
328	goto error;
329	}
c65353da	330	if (rt) {
1da177e4 LT	331	memcpy(&optptr[optptr[2]-1], &rt->rt_spec_dst, 4);
	332	opt->is_changed = 1;
	333	}
	334	optptr[2] += 4;
	335	opt->rr_needaddr = 1;
	336	}
	337	opt->rr = optptr - iph;
	338	break;
	339	case IPOPT_TIMESTAMP:
	340	if (opt->ts) {
	341	pp_ptr = optptr;
	342	goto error;
	343	}
	344	if (optlen < 4) {
	345	pp_ptr = optptr + 1;
	346	goto error;
	347	}
	348	if (optptr[2] < 5) {
	349	pp_ptr = optptr + 2;
	350	goto error;
	351	}
	352	if (optptr[2] <= optlen) {
e25d2ca6	353	__be32 *timeptr = NULL;
1da177e4 LT	354	if (optptr[2]+3 > optptr[1]) {
	355	pp_ptr = optptr + 2;
	356	goto error;
	357	}
	358	switch (optptr[3]&0xF) {
	359	case IPOPT_TS_TSONLY:
	360	opt->ts = optptr - iph;
e905a9ed	361	if (skb)
e25d2ca6	362	timeptr = (__be32*)&optptr[optptr[2]-1];
1da177e4 LT	363	opt->ts_needtime = 1;
	364	optptr[2] += 4;
	365	break;
	366	case IPOPT_TS_TSANDADDR:
	367	if (optptr[2]+7 > optptr[1]) {
	368	pp_ptr = optptr + 2;
	369	goto error;
	370	}
	371	opt->ts = optptr - iph;
c65353da	372	if (rt) {
1da177e4	373	memcpy(&optptr[optptr[2]-1], &rt->rt_spec_dst, 4);
e25d2ca6	374	timeptr = (__be32*)&optptr[optptr[2]+3];
1da177e4 LT	375	}
	376	opt->ts_needaddr = 1;
	377	opt->ts_needtime = 1;
	378	optptr[2] += 8;
	379	break;
	380	case IPOPT_TS_PRESPEC:
	381	if (optptr[2]+7 > optptr[1]) {
	382	pp_ptr = optptr + 2;
	383	goto error;
	384	}
	385	opt->ts = optptr - iph;
	386	{
fd683222	387	__be32 addr;
1da177e4	388	memcpy(&addr, &optptr[optptr[2]-1], 4);
0e6bd4a1	389	if (inet_addr_type(net, addr) == RTN_UNICAST)
1da177e4 LT	390	break;
1da177e4 LT	391	if (skb)
e25d2ca6	392	timeptr = (__be32*)&optptr[optptr[2]+3];
1da177e4 LT	393	}
	394	opt->ts_needtime = 1;
	395	optptr[2] += 8;
	396	break;
	397	default:
	398	if (!skb && !capable(CAP_NET_RAW)) {
	399	pp_ptr = optptr + 3;
	400	goto error;
	401	}
	402	break;
	403	}
	404	if (timeptr) {
f25c3d61	405	struct timespec tv;
e25d2ca6	406	__be32 midtime;
f25c3d61 YH	407	getnstimeofday(&tv);
f25c3d61 YH	408	midtime = htonl((tv.tv_sec % 86400) * MSEC_PER_SEC + tv.tv_nsec / NSEC_PER_MSEC);
e25d2ca6	409	memcpy(timeptr, &midtime, sizeof(__be32));
1da177e4 LT	410	opt->is_changed = 1;
	411	}
	412	} else {
	413	unsigned overflow = optptr[3]>>4;
	414	if (overflow == 15) {
	415	pp_ptr = optptr + 3;
	416	goto error;
	417	}
	418	opt->ts = optptr - iph;
	419	if (skb) {
	420	optptr[3] = (optptr[3]&0xF)\|((overflow+1)<<4);
	421	opt->is_changed = 1;
	422	}
	423	}
	424	break;
	425	case IPOPT_RA:
	426	if (optlen < 4) {
	427	pp_ptr = optptr + 1;
	428	goto error;
	429	}
	430	if (optptr[2] == 0 && optptr[3] == 0)
	431	opt->router_alert = optptr - iph;
	432	break;
11a03f78	433	case IPOPT_CIPSO:
f8687afe	434	if ((!skb && !capable(CAP_NET_RAW)) \|\| opt->cipso) {
11a03f78 PM	435	pp_ptr = optptr;
	436	goto error;
	437	}
	438	opt->cipso = optptr - iph;
15c45f7b	439	if (cipso_v4_validate(skb, &optptr)) {
11a03f78 PM	440	pp_ptr = optptr;
	441	goto error;
	442	}
	443	break;
1da177e4 LT	444	case IPOPT_SEC:
	445	case IPOPT_SID:
	446	default:
	447	if (!skb && !capable(CAP_NET_RAW)) {
	448	pp_ptr = optptr;
	449	goto error;
	450	}
	451	break;
	452	}
	453	l -= optlen;
	454	optptr += optlen;
	455	}
	456
	457	eol:
	458	if (!pp_ptr)
	459	return 0;
	460
	461	error:
	462	if (skb) {
	463	icmp_send(skb, ICMP_PARAMETERPROB, 0, htonl((pp_ptr-iph)<<24));
	464	}
	465	return -EINVAL;
	466	}
462fb2af	467	EXPORT_SYMBOL(ip_options_compile);
1da177e4 LT	468
	469	/*
	470	* Undo all the changes done by ip_options_compile().
	471	*/
	472
	473	void ip_options_undo(struct ip_options * opt)
	474	{
	475	if (opt->srr) {
	476	unsigned char * optptr = opt->__data+opt->srr-sizeof(struct iphdr);
	477	memmove(optptr+7, optptr+3, optptr[1]-7);
	478	memcpy(optptr+3, &opt->faddr, 4);
	479	}
	480	if (opt->rr_needaddr) {
	481	unsigned char * optptr = opt->__data+opt->rr-sizeof(struct iphdr);
	482	optptr[2] -= 4;
	483	memset(&optptr[optptr[2]-1], 0, 4);
	484	}
	485	if (opt->ts) {
	486	unsigned char * optptr = opt->__data+opt->ts-sizeof(struct iphdr);
	487	if (opt->ts_needtime) {
	488	optptr[2] -= 4;
	489	memset(&optptr[optptr[2]-1], 0, 4);
	490	if ((optptr[3]&0xF) == IPOPT_TS_PRESPEC)
	491	optptr[2] -= 4;
	492	}
	493	if (opt->ts_needaddr) {
	494	optptr[2] -= 4;
	495	memset(&optptr[optptr[2]-1], 0, 4);
	496	}
	497	}
	498	}
	499
f6d8bd05	500	static struct ip_options_rcu *ip_options_get_alloc(const int optlen)
1da177e4	501	{
f6d8bd05	502	return kzalloc(sizeof(struct ip_options_rcu) + ((optlen + 3) & ~3),
37640703	503	GFP_KERNEL);
4c6ea29d	504	}
1da177e4	505
f6d8bd05 ED	506	static int ip_options_get_finish(struct net net, struct ip_options_rcu *optp,
f6d8bd05 ED	507	struct ip_options_rcu *opt, int optlen)
4c6ea29d	508	{
1da177e4	509	while (optlen & 3)
f6d8bd05 ED	510	opt->opt.__data[optlen++] = IPOPT_END;
	511	opt->opt.optlen = optlen;
	512	if (optlen && ip_options_compile(net, &opt->opt, NULL)) {
1da177e4 LT	513	kfree(opt);
	514	return -EINVAL;
	515	}
a51482bd	516	kfree(*optp);
1da177e4 LT	517	*optp = opt;
	518	return 0;
	519	}
	520
f6d8bd05	521	int ip_options_get_from_user(struct net net, struct ip_options_rcu *optp,
f2c4802b	522	unsigned char __user *data, int optlen)
4c6ea29d	523	{
f6d8bd05	524	struct ip_options_rcu *opt = ip_options_get_alloc(optlen);
4c6ea29d ACM	525
	526	if (!opt)
	527	return -ENOMEM;
f6d8bd05	528	if (optlen && copy_from_user(opt->opt.__data, data, optlen)) {
4c6ea29d ACM	529	kfree(opt);
	530	return -EFAULT;
	531	}
f2c4802b	532	return ip_options_get_finish(net, optp, opt, optlen);
4c6ea29d ACM	533	}
4c6ea29d ACM	534
f6d8bd05	535	int ip_options_get(struct net net, struct ip_options_rcu *optp,
f2c4802b	536	unsigned char *data, int optlen)
4c6ea29d	537	{
f6d8bd05	538	struct ip_options_rcu *opt = ip_options_get_alloc(optlen);
4c6ea29d ACM	539
	540	if (!opt)
	541	return -ENOMEM;
	542	if (optlen)
f6d8bd05	543	memcpy(opt->opt.__data, data, optlen);
f2c4802b	544	return ip_options_get_finish(net, optp, opt, optlen);
4c6ea29d ACM	545	}
4c6ea29d ACM	546
1da177e4 LT	547	void ip_forward_options(struct sk_buff *skb)
	548	{
	549	struct ip_options * opt = &(IPCB(skb)->opt);
	550	unsigned char * optptr;
511c3f92	551	struct rtable *rt = skb_rtable(skb);
d56f90a7	552	unsigned char *raw = skb_network_header(skb);
1da177e4 LT	553
	554	if (opt->rr_needaddr) {
	555	optptr = (unsigned char *)raw + opt->rr;
	556	ip_rt_get_source(&optptr[optptr[2]-5], rt);
	557	opt->is_changed = 1;
	558	}
	559	if (opt->srr_is_hit) {
	560	int srrptr, srrspace;
	561
	562	optptr = raw + opt->srr;
	563
	564	for ( srrptr=optptr[2], srrspace = optptr[1];
	565	srrptr <= srrspace;
	566	srrptr += 4
	567	) {
	568	if (srrptr + 3 > srrspace)
	569	break;
	570	if (memcmp(&rt->rt_dst, &optptr[srrptr-1], 4) == 0)
	571	break;
	572	}
	573	if (srrptr + 3 <= srrspace) {
	574	opt->is_changed = 1;
	575	ip_rt_get_source(&optptr[srrptr-1], rt);
eddc9ec5	576	ip_hdr(skb)->daddr = rt->rt_dst;
1da177e4 LT	577	optptr[2] = srrptr+4;
	578	} else if (net_ratelimit())
	579	printk(KERN_CRIT "ip_forward(): Argh! Destination lost!\n");
	580	if (opt->ts_needaddr) {
	581	optptr = raw + opt->ts;
	582	ip_rt_get_source(&optptr[optptr[2]-9], rt);
	583	opt->is_changed = 1;
	584	}
	585	}
	586	if (opt->is_changed) {
	587	opt->is_changed = 0;
eddc9ec5	588	ip_send_check(ip_hdr(skb));
1da177e4 LT	589	}
	590	}
	591
	592	int ip_options_rcv_srr(struct sk_buff *skb)
	593	{
	594	struct ip_options *opt = &(IPCB(skb)->opt);
	595	int srrspace, srrptr;
9e12bb22	596	__be32 nexthop;
eddc9ec5	597	struct iphdr *iph = ip_hdr(skb);
d56f90a7	598	unsigned char *optptr = skb_network_header(skb) + opt->srr;
511c3f92	599	struct rtable *rt = skb_rtable(skb);
1da177e4	600	struct rtable *rt2;
7fee226a	601	unsigned long orefdst;
1da177e4 LT	602	int err;
1da177e4 LT	603
c65353da	604	if (!opt->srr \|\| !rt)
1da177e4 LT	605	return 0;
	606
	607	if (skb->pkt_type != PACKET_HOST)
	608	return -EINVAL;
	609	if (rt->rt_type == RTN_UNICAST) {
	610	if (!opt->is_strictroute)
	611	return 0;
	612	icmp_send(skb, ICMP_PARAMETERPROB, 0, htonl(16<<24));
	613	return -EINVAL;
	614	}
	615	if (rt->rt_type != RTN_LOCAL)
	616	return -EINVAL;
	617
	618	for (srrptr=optptr[2], srrspace = optptr[1]; srrptr <= srrspace; srrptr += 4) {
	619	if (srrptr + 3 > srrspace) {
	620	icmp_send(skb, ICMP_PARAMETERPROB, 0, htonl((opt->srr+2)<<24));
	621	return -EINVAL;
	622	}
	623	memcpy(&nexthop, &optptr[srrptr-1], 4);
	624
7fee226a	625	orefdst = skb->_skb_refdst;
adf30907	626	skb_dst_set(skb, NULL);
1da177e4	627	err = ip_route_input(skb, nexthop, iph->saddr, iph->tos, skb->dev);
511c3f92	628	rt2 = skb_rtable(skb);
1da177e4	629	if (err \|\| (rt2->rt_type != RTN_UNICAST && rt2->rt_type != RTN_LOCAL)) {
7fee226a ED	630	skb_dst_drop(skb);
7fee226a ED	631	skb->_skb_refdst = orefdst;
1da177e4 LT	632	return -EINVAL;
1da177e4 LT	633	}
7fee226a	634	refdst_drop(orefdst);
1da177e4 LT	635	if (rt2->rt_type != RTN_LOCAL)
	636	break;
	637	/* Superfast 8) loopback forward */
	638	memcpy(&iph->daddr, &optptr[srrptr-1], 4);
	639	opt->is_changed = 1;
	640	}
	641	if (srrptr <= srrspace) {
	642	opt->srr_is_hit = 1;
	643	opt->is_changed = 1;
	644	}
	645	return 0;
	646	}
462fb2af	647	EXPORT_SYMBOL(ip_options_rcv_srr);