[deliverable/linux.git] / net / ipv4 / ip_input.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 Internet Protocol (IP) module.
 *
 * Authors:	Ross Biro
 *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
 *		Donald Becker, <becker@super.org>
 *		Alan Cox, <alan@lxorguk.ukuu.org.uk>
 *		Richard Underwood
 *		Stefan Becker, <stefanb@yello.ping.de>
 *		Jorge Cwik, <jorge@laser.satlink.net>
 *		Arnt Gulbrandsen, <agulbra@nvg.unit.no>
 *
 *
 * Fixes:
 *		Alan Cox	:	Commented a couple of minor bits of surplus code
 *		Alan Cox	:	Undefining IP_FORWARD doesn't include the code
 *					(just stops a compiler warning).
 *		Alan Cox	:	Frames with >=MAX_ROUTE record routes, strict routes or loose routes
 *					are junked rather than corrupting things.
 *		Alan Cox	:	Frames to bad broadcast subnets are dumped
 *					We used to process them non broadcast and
 *					boy could that cause havoc.
 *		Alan Cox	:	ip_forward sets the free flag on the
 *					new frame it queues. Still crap because
 *					it copies the frame but at least it
 *					doesn't eat memory too.
 *		Alan Cox	:	Generic queue code and memory fixes.
 *		Fred Van Kempen :	IP fragment support (borrowed from NET2E)
 *		Gerhard Koerting:	Forward fragmented frames correctly.
 *		Gerhard Koerting: 	Fixes to my fix of the above 8-).
 *		Gerhard Koerting:	IP interface addressing fix.
 *		Linus Torvalds	:	More robustness checks
 *		Alan Cox	:	Even more checks: Still not as robust as it ought to be
 *		Alan Cox	:	Save IP header pointer for later
 *		Alan Cox	:	ip option setting
 *		Alan Cox	:	Use ip_tos/ip_ttl settings
 *		Alan Cox	:	Fragmentation bogosity removed
 *					(Thanks to Mark.Bush@prg.ox.ac.uk)
 *		Dmitry Gorodchanin :	Send of a raw packet crash fix.
 *		Alan Cox	:	Silly ip bug when an overlength
 *					fragment turns up. Now frees the
 *					queue.
 *		Linus Torvalds/ :	Memory leakage on fragmentation
 *		Alan Cox	:	handling.
 *		Gerhard Koerting:	Forwarding uses IP priority hints
 *		Teemu Rantanen	:	Fragment problems.
 *		Alan Cox	:	General cleanup, comments and reformat
 *		Alan Cox	:	SNMP statistics
 *		Alan Cox	:	BSD address rule semantics. Also see
 *					UDP as there is a nasty checksum issue
 *					if you do things the wrong way.
 *		Alan Cox	:	Always defrag, moved IP_FORWARD to the config.in file
 *		Alan Cox	: 	IP options adjust sk->priority.
 *		Pedro Roque	:	Fix mtu/length error in ip_forward.
 *		Alan Cox	:	Avoid ip_chk_addr when possible.
 *	Richard Underwood	:	IP multicasting.
 *		Alan Cox	:	Cleaned up multicast handlers.
 *		Alan Cox	:	RAW sockets demultiplex in the BSD style.
 *		Gunther Mayer	:	Fix the SNMP reporting typo
 *		Alan Cox	:	Always in group 224.0.0.1
 *	Pauline Middelink	:	Fast ip_checksum update when forwarding
 *					Masquerading support.
 *		Alan Cox	:	Multicast loopback error for 224.0.0.1
 *		Alan Cox	:	IP_MULTICAST_LOOP option.
 *		Alan Cox	:	Use notifiers.
 *		Bjorn Ekwall	:	Removed ip_csum (from slhc.c too)
 *		Bjorn Ekwall	:	Moved ip_fast_csum to ip.h (inline!)
 *		Stefan Becker   :       Send out ICMP HOST REDIRECT
 *	Arnt Gulbrandsen	:	ip_build_xmit
 *		Alan Cox	:	Per socket routing cache
 *		Alan Cox	:	Fixed routing cache, added header cache.
 *		Alan Cox	:	Loopback didn't work right in original ip_build_xmit - fixed it.
 *		Alan Cox	:	Only send ICMP_REDIRECT if src/dest are the same net.
 *		Alan Cox	:	Incoming IP option handling.
 *		Alan Cox	:	Set saddr on raw output frames as per BSD.
 *		Alan Cox	:	Stopped broadcast source route explosions.
 *		Alan Cox	:	Can disable source routing
 *		Takeshi Sone    :	Masquerading didn't work.
 *	Dave Bonn,Alan Cox	:	Faster IP forwarding whenever possible.
 *		Alan Cox	:	Memory leaks, tramples, misc debugging.
 *		Alan Cox	:	Fixed multicast (by popular demand 8))
 *		Alan Cox	:	Fixed forwarding (by even more popular demand 8))
 *		Alan Cox	:	Fixed SNMP statistics [I think]
 *	Gerhard Koerting	:	IP fragmentation forwarding fix
 *		Alan Cox	:	Device lock against page fault.
 *		Alan Cox	:	IP_HDRINCL facility.
 *	Werner Almesberger	:	Zero fragment bug
 *		Alan Cox	:	RAW IP frame length bug
 *		Alan Cox	:	Outgoing firewall on build_xmit
 *		A.N.Kuznetsov	:	IP_OPTIONS support throughout the kernel
 *		Alan Cox	:	Multicast routing hooks
 *		Jos Vos		:	Do accounting *before* call_in_firewall
 *	Willy Konynenberg	:	Transparent proxying support
 *
 *
 *
 * To Fix:
 *		IP fragmentation wants rewriting cleanly. The RFC815 algorithm is much more efficient
 *		and could be made very efficient with the addition of some virtual memory hacks to permit
 *		the allocation of a buffer that can then be 'grown' by twiddling page tables.
 *		Output fragmentation wants updating along with the buffer management to use a single
 *		interleaved copy algorithm so that fragmenting has a one copy overhead. Actual packet
 *		output should probably do its own fragmentation at the UDP/RAW layer. TCP shouldn't cause
 *		fragmentation anyway.
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 */

#define pr_fmt(fmt) "IPv4: " fmt

#include <linux/module.h>
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/slab.h>

#include <linux/net.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/in.h>
#include <linux/inet.h>
#include <linux/inetdevice.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>

#include <net/snmp.h>
#include <net/ip.h>
#include <net/protocol.h>
#include <net/route.h>
#include <linux/skbuff.h>
#include <net/sock.h>
#include <net/arp.h>
#include <net/icmp.h>
#include <net/raw.h>
#include <net/checksum.h>
#include <linux/netfilter_ipv4.h>
#include <net/xfrm.h>
#include <linux/mroute.h>
#include <linux/netlink.h>

/*
 *	Process Router Attention IP option (RFC 2113)
 */
bool ip_call_ra_chain(struct sk_buff *skb)
{
	struct ip_ra_chain *ra;
	u8 protocol = ip_hdr(skb)->protocol;
	struct sock *last = NULL;
	struct net_device *dev = skb->dev;

	for (ra = rcu_dereference(ip_ra_chain); ra; ra = rcu_dereference(ra->next)) {
		struct sock *sk = ra->sk;

		/* If socket is bound to an interface, only report
		 * the packet if it came  from that interface.
		 */
		if (sk && inet_sk(sk)->inet_num == protocol &&
		    (!sk->sk_bound_dev_if ||
		     sk->sk_bound_dev_if == dev->ifindex) &&
		    net_eq(sock_net(sk), dev_net(dev))) {
			if (ip_is_fragment(ip_hdr(skb))) {
				if (ip_defrag(skb, IP_DEFRAG_CALL_RA_CHAIN))
					return true;
			}
			if (last) {
				struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
				if (skb2)
					raw_rcv(last, skb2);
			}
			last = sk;
		}
	}

	if (last) {
		raw_rcv(last, skb);
		return true;
	}
	return false;
}

static int ip_local_deliver_finish(struct sk_buff *skb)
{
	struct net *net = dev_net(skb->dev);

	__skb_pull(skb, ip_hdrlen(skb));

	/* Point into the IP datagram, just past the header. */
	skb_reset_transport_header(skb);

	rcu_read_lock();
	{
		int protocol = ip_hdr(skb)->protocol;
		const struct net_protocol *ipprot;
		int raw;

	resubmit:
		raw = raw_local_deliver(skb, protocol);

		ipprot = rcu_dereference(inet_protos[protocol]);
		if (ipprot != NULL) {
			int ret;

			if (!net_eq(net, &init_net) && !ipprot->netns_ok) {
				net_info_ratelimited("%s: proto %d isn't netns-ready\n",
						     __func__, protocol);
				kfree_skb(skb);
				goto out;
			}

			if (!ipprot->no_policy) {
				if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
					kfree_skb(skb);
					goto out;
				}
				nf_reset(skb);
			}
			ret = ipprot->handler(skb);
			if (ret < 0) {
				protocol = -ret;
				goto resubmit;
			}
			IP_INC_STATS_BH(net, IPSTATS_MIB_INDELIVERS);
		} else {
			if (!raw) {
				if (xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
					IP_INC_STATS_BH(net, IPSTATS_MIB_INUNKNOWNPROTOS);
					icmp_send(skb, ICMP_DEST_UNREACH,
						  ICMP_PROT_UNREACH, 0);
				}
			} else
				IP_INC_STATS_BH(net, IPSTATS_MIB_INDELIVERS);
			kfree_skb(skb);
		}
	}
 out:
	rcu_read_unlock();

	return 0;
}

/*
 * 	Deliver IP Packets to the higher protocol layers.
 */
int ip_local_deliver(struct sk_buff *skb)
{
	/*
	 *	Reassemble IP fragments.
	 */

	if (ip_is_fragment(ip_hdr(skb))) {
		if (ip_defrag(skb, IP_DEFRAG_LOCAL_DELIVER))
			return 0;
	}

	return NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_IN, skb, skb->dev, NULL,
		       ip_local_deliver_finish);
}

static inline bool ip_rcv_options(struct sk_buff *skb)
{
	struct ip_options *opt;
	const struct iphdr *iph;
	struct net_device *dev = skb->dev;

	/* It looks as overkill, because not all
	   IP options require packet mangling.
	   But it is the easiest for now, especially taking
	   into account that combination of IP options
	   and running sniffer is extremely rare condition.
					      --ANK (980813)
	*/
	if (skb_cow(skb, skb_headroom(skb))) {
		IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS);
		goto drop;
	}

	iph = ip_hdr(skb);
	opt = &(IPCB(skb)->opt);
	opt->optlen = iph->ihl*4 - sizeof(struct iphdr);

	if (ip_options_compile(dev_net(dev), opt, skb)) {
		IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INHDRERRORS);
		goto drop;
	}

	if (unlikely(opt->srr)) {
		struct in_device *in_dev = __in_dev_get_rcu(dev);

		if (in_dev) {
			if (!IN_DEV_SOURCE_ROUTE(in_dev)) {
				if (IN_DEV_LOG_MARTIANS(in_dev))
					net_info_ratelimited("source route option %pI4 -> %pI4\n",
							     &iph->saddr,
							     &iph->daddr);
				goto drop;
			}
		}

		if (ip_options_rcv_srr(skb))
			goto drop;
	}

	return false;
drop:
	return true;
}

int sysctl_ip_early_demux __read_mostly = 1;

static int ip_rcv_finish(struct sk_buff *skb)
{
	const struct iphdr *iph = ip_hdr(skb);
	struct rtable *rt;

	if (sysctl_ip_early_demux && !skb_dst(skb)) {
		const struct net_protocol *ipprot;
		int protocol = iph->protocol;

		rcu_read_lock();
		ipprot = rcu_dereference(inet_protos[protocol]);
		if (ipprot && ipprot->early_demux) {
			ipprot->early_demux(skb);
			/* must reload iph, skb->head might have changed */
			iph = ip_hdr(skb);
		}
		rcu_read_unlock();
	}

	/*
	 *	Initialise the virtual path cache for the packet. It describes
	 *	how the packet travels inside Linux networking.
	 */
	if (!skb_dst(skb)) {
		int err = ip_route_input(skb, iph->daddr, iph->saddr,
					 iph->tos, skb->dev);
		if (unlikely(err)) {
			if (err == -EXDEV)
				NET_INC_STATS_BH(dev_net(skb->dev),
						 LINUX_MIB_IPRPFILTER);
			goto drop;
		}
	}

#ifdef CONFIG_IP_ROUTE_CLASSID
	if (unlikely(skb_dst(skb)->tclassid)) {
		struct ip_rt_acct *st = this_cpu_ptr(ip_rt_acct);
		u32 idx = skb_dst(skb)->tclassid;
		st[idx&0xFF].o_packets++;
		st[idx&0xFF].o_bytes += skb->len;
		st[(idx>>16)&0xFF].i_packets++;
		st[(idx>>16)&0xFF].i_bytes += skb->len;
	}
#endif

	if (iph->ihl > 5 && ip_rcv_options(skb))
		goto drop;

	rt = skb_rtable(skb);
	if (rt->rt_type == RTN_MULTICAST) {
		IP_UPD_PO_STATS_BH(dev_net(rt->dst.dev), IPSTATS_MIB_INMCAST,
				skb->len);
	} else if (rt->rt_type == RTN_BROADCAST)
		IP_UPD_PO_STATS_BH(dev_net(rt->dst.dev), IPSTATS_MIB_INBCAST,
				skb->len);

	return dst_input(skb);

drop:
	kfree_skb(skb);
	return NET_RX_DROP;
}

/*
 * 	Main IP Receive routine.
 */
int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, struct net_device *orig_dev)
{
	const struct iphdr *iph;
	u32 len;

	/* When the interface is in promisc. mode, drop all the crap
	 * that it receives, do not try to analyse it.
	 */
	if (skb->pkt_type == PACKET_OTHERHOST)
		goto drop;


	IP_UPD_PO_STATS_BH(dev_net(dev), IPSTATS_MIB_IN, skb->len);

	if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL) {
		IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS);
		goto out;
	}

	if (!pskb_may_pull(skb, sizeof(struct iphdr)))
		goto inhdr_error;

	iph = ip_hdr(skb);

	/*
	 *	RFC1122: 3.2.1.2 MUST silently discard any IP frame that fails the checksum.
	 *
	 *	Is the datagram acceptable?
	 *
	 *	1.	Length at least the size of an ip header
	 *	2.	Version of 4
	 *	3.	Checksums correctly. [Speed optimisation for later, skip loopback checksums]
	 *	4.	Doesn't have a bogus length
	 */

	if (iph->ihl < 5 || iph->version != 4)
		goto inhdr_error;

	if (!pskb_may_pull(skb, iph->ihl*4))
		goto inhdr_error;

	iph = ip_hdr(skb);

	if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl)))
		goto inhdr_error;

	len = ntohs(iph->tot_len);
	if (skb->len < len) {
		IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INTRUNCATEDPKTS);
		goto drop;
	} else if (len < (iph->ihl*4))
		goto inhdr_error;

	/* Our transport medium may have padded the buffer out. Now we know it
	 * is IP we can trim to the true length of the frame.
	 * Note this now means skb->len holds ntohs(iph->tot_len).
	 */
	if (pskb_trim_rcsum(skb, len)) {
		IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS);
		goto drop;
	}

	/* Remove any debris in the socket control block */
	memset(IPCB(skb), 0, sizeof(struct inet_skb_parm));

	/* Must drop socket now because of tproxy. */
	skb_orphan(skb);

	return NF_HOOK(NFPROTO_IPV4, NF_INET_PRE_ROUTING, skb, dev, NULL,
		       ip_rcv_finish);

inhdr_error:
	IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INHDRERRORS);
drop:
	kfree_skb(skb);
out:
	return NET_RX_DROP;
}
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 Internet Protocol (IP) module.
	7	*
02c30a84	8	* Authors: Ross Biro
1da177e4 LT	9	* Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
1da177e4 LT	10	* Donald Becker, <becker@super.org>
113aa838	11	* Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4 LT	12	* Richard Underwood
	13	* Stefan Becker, <stefanb@yello.ping.de>
	14	* Jorge Cwik, <jorge@laser.satlink.net>
	15	* Arnt Gulbrandsen, <agulbra@nvg.unit.no>
e905a9ed	16	*
1da177e4 LT	17	*
	18	* Fixes:
	19	* Alan Cox : Commented a couple of minor bits of surplus code
	20	* Alan Cox : Undefining IP_FORWARD doesn't include the code
	21	* (just stops a compiler warning).
	22	* Alan Cox : Frames with >=MAX_ROUTE record routes, strict routes or loose routes
	23	* are junked rather than corrupting things.
	24	* Alan Cox : Frames to bad broadcast subnets are dumped
	25	* We used to process them non broadcast and
	26	* boy could that cause havoc.
	27	* Alan Cox : ip_forward sets the free flag on the
	28	* new frame it queues. Still crap because
	29	* it copies the frame but at least it
	30	* doesn't eat memory too.
	31	* Alan Cox : Generic queue code and memory fixes.
	32	* Fred Van Kempen : IP fragment support (borrowed from NET2E)
	33	* Gerhard Koerting: Forward fragmented frames correctly.
	34	* Gerhard Koerting: Fixes to my fix of the above 8-).
	35	* Gerhard Koerting: IP interface addressing fix.
	36	* Linus Torvalds : More robustness checks
	37	* Alan Cox : Even more checks: Still not as robust as it ought to be
	38	* Alan Cox : Save IP header pointer for later
	39	* Alan Cox : ip option setting
	40	* Alan Cox : Use ip_tos/ip_ttl settings
	41	* Alan Cox : Fragmentation bogosity removed
	42	* (Thanks to Mark.Bush@prg.ox.ac.uk)
	43	* Dmitry Gorodchanin : Send of a raw packet crash fix.
	44	* Alan Cox : Silly ip bug when an overlength
	45	* fragment turns up. Now frees the
	46	* queue.
	47	* Linus Torvalds/ : Memory leakage on fragmentation
	48	* Alan Cox : handling.
	49	* Gerhard Koerting: Forwarding uses IP priority hints
	50	* Teemu Rantanen : Fragment problems.
	51	* Alan Cox : General cleanup, comments and reformat
	52	* Alan Cox : SNMP statistics
	53	* Alan Cox : BSD address rule semantics. Also see
	54	* UDP as there is a nasty checksum issue
	55	* if you do things the wrong way.
	56	* Alan Cox : Always defrag, moved IP_FORWARD to the config.in file
	57	* Alan Cox : IP options adjust sk->priority.
	58	* Pedro Roque : Fix mtu/length error in ip_forward.
	59	* Alan Cox : Avoid ip_chk_addr when possible.
	60	* Richard Underwood : IP multicasting.
	61	* Alan Cox : Cleaned up multicast handlers.
	62	* Alan Cox : RAW sockets demultiplex in the BSD style.
	63	* Gunther Mayer : Fix the SNMP reporting typo
	64	* Alan Cox : Always in group 224.0.0.1
	65	* Pauline Middelink : Fast ip_checksum update when forwarding
	66	* Masquerading support.
	67	* Alan Cox : Multicast loopback error for 224.0.0.1
	68	* Alan Cox : IP_MULTICAST_LOOP option.
	69	* Alan Cox : Use notifiers.
	70	* Bjorn Ekwall : Removed ip_csum (from slhc.c too)
	71	* Bjorn Ekwall : Moved ip_fast_csum to ip.h (inline!)
	72	* Stefan Becker : Send out ICMP HOST REDIRECT
	73	* Arnt Gulbrandsen : ip_build_xmit
	74	* Alan Cox : Per socket routing cache
	75	* Alan Cox : Fixed routing cache, added header cache.
	76	* Alan Cox : Loopback didn't work right in original ip_build_xmit - fixed it.
	77	* Alan Cox : Only send ICMP_REDIRECT if src/dest are the same net.
	78	* Alan Cox : Incoming IP option handling.
	79	* Alan Cox : Set saddr on raw output frames as per BSD.
	80	* Alan Cox : Stopped broadcast source route explosions.
81	* Alan Cox : Can disable source routing
82	* Takeshi Sone : Masquerading didn't work.
83	* Dave Bonn,Alan Cox : Faster IP forwarding whenever possible.
84	* Alan Cox : Memory leaks, tramples, misc debugging.
85	* Alan Cox : Fixed multicast (by popular demand 8))
86	* Alan Cox : Fixed forwarding (by even more popular demand 8))
87	* Alan Cox : Fixed SNMP statistics [I think]
88	* Gerhard Koerting : IP fragmentation forwarding fix
89	* Alan Cox : Device lock against page fault.
90	* Alan Cox : IP_HDRINCL facility.
91	* Werner Almesberger : Zero fragment bug
92	* Alan Cox : RAW IP frame length bug
93	* Alan Cox : Outgoing firewall on build_xmit
94	* A.N.Kuznetsov : IP_OPTIONS support throughout the kernel
95	* Alan Cox : Multicast routing hooks
96	* Jos Vos : Do accounting before call_in_firewall
97	* Willy Konynenberg : Transparent proxying support
98	*
e905a9ed	99	*
1da177e4 LT	100	*
	101	* To Fix:
	102	* IP fragmentation wants rewriting cleanly. The RFC815 algorithm is much more efficient
	103	* and could be made very efficient with the addition of some virtual memory hacks to permit
	104	* the allocation of a buffer that can then be 'grown' by twiddling page tables.
e905a9ed	105	* Output fragmentation wants updating along with the buffer management to use a single
1da177e4 LT	106	* interleaved copy algorithm so that fragmenting has a one copy overhead. Actual packet
	107	* output should probably do its own fragmentation at the UDP/RAW layer. TCP shouldn't cause
	108	* fragmentation anyway.
	109	*
	110	* This program is free software; you can redistribute it and/or
	111	* modify it under the terms of the GNU General Public License
	112	* as published by the Free Software Foundation; either version
	113	* 2 of the License, or (at your option) any later version.
	114	*/
	115
afd46503 JP	116	#define pr_fmt(fmt) "IPv4: " fmt
afd46503 JP	117
1da177e4 LT	118	#include <linux/module.h>
	119	#include <linux/types.h>
	120	#include <linux/kernel.h>
	121	#include <linux/string.h>
	122	#include <linux/errno.h>
5a0e3ad6	123	#include <linux/slab.h>
1da177e4 LT	124
	125	#include <linux/net.h>
	126	#include <linux/socket.h>
	127	#include <linux/sockios.h>
	128	#include <linux/in.h>
	129	#include <linux/inet.h>
14c85021	130	#include <linux/inetdevice.h>
1da177e4 LT	131	#include <linux/netdevice.h>
	132	#include <linux/etherdevice.h>
	133
	134	#include <net/snmp.h>
	135	#include <net/ip.h>
	136	#include <net/protocol.h>
	137	#include <net/route.h>
	138	#include <linux/skbuff.h>
	139	#include <net/sock.h>
	140	#include <net/arp.h>
	141	#include <net/icmp.h>
	142	#include <net/raw.h>
	143	#include <net/checksum.h>
	144	#include <linux/netfilter_ipv4.h>
	145	#include <net/xfrm.h>
	146	#include <linux/mroute.h>
	147	#include <linux/netlink.h>
	148
1da177e4	149	/*
66018506	150	* Process Router Attention IP option (RFC 2113)
e905a9ed	151	*/
ba57b4db	152	bool ip_call_ra_chain(struct sk_buff *skb)
1da177e4 LT	153	{
1da177e4 LT	154	struct ip_ra_chain *ra;
eddc9ec5	155	u8 protocol = ip_hdr(skb)->protocol;
1da177e4	156	struct sock *last = NULL;
cb84663e	157	struct net_device *dev = skb->dev;
1da177e4	158
66018506	159	for (ra = rcu_dereference(ip_ra_chain); ra; ra = rcu_dereference(ra->next)) {
1da177e4 LT	160	struct sock *sk = ra->sk;
	161
	162	/* If socket is bound to an interface, only report
	163	* the packet if it came from that interface.
	164	*/
c720c7e8	165	if (sk && inet_sk(sk)->inet_num == protocol &&
1da177e4	166	(!sk->sk_bound_dev_if \|\|
cb84663e	167	sk->sk_bound_dev_if == dev->ifindex) &&
09ad9bc7	168	net_eq(sock_net(sk), dev_net(dev))) {
56f8a75c	169	if (ip_is_fragment(ip_hdr(skb))) {
66018506	170	if (ip_defrag(skb, IP_DEFRAG_CALL_RA_CHAIN))
ba57b4db	171	return true;
1da177e4 LT	172	}
	173	if (last) {
	174	struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
	175	if (skb2)
	176	raw_rcv(last, skb2);
	177	}
	178	last = sk;
	179	}
	180	}
	181
	182	if (last) {
	183	raw_rcv(last, skb);
ba57b4db	184	return true;
1da177e4	185	}
ba57b4db	186	return false;
1da177e4 LT	187	}
1da177e4 LT	188
861d0486	189	static int ip_local_deliver_finish(struct sk_buff *skb)
1da177e4	190	{
c346dca1	191	struct net *net = dev_net(skb->dev);
f145049a	192
c9bdd4b5	193	__skb_pull(skb, ip_hdrlen(skb));
1da177e4	194
e905a9ed	195	/* Point into the IP datagram, just past the header. */
badff6d0	196	skb_reset_transport_header(skb);
1da177e4 LT	197
	198	rcu_read_lock();
	199	{
eddc9ec5	200	int protocol = ip_hdr(skb)->protocol;
32613090	201	const struct net_protocol *ipprot;
f9242b6b	202	int raw;
1da177e4 LT	203
1da177e4 LT	204	resubmit:
7bc54c90	205	raw = raw_local_deliver(skb, protocol);
1da177e4	206
f9242b6b	207	ipprot = rcu_dereference(inet_protos[protocol]);
9c0188ac	208	if (ipprot != NULL) {
1da177e4 LT	209	int ret;
1da177e4 LT	210
9c0188ac	211	if (!net_eq(net, &init_net) && !ipprot->netns_ok) {
e87cc472 JP	212	net_info_ratelimited("%s: proto %d isn't netns-ready\n",
e87cc472 JP	213	__func__, protocol);
9c0188ac AD	214	kfree_skb(skb);
	215	goto out;
	216	}
	217
b59c2701 PM	218	if (!ipprot->no_policy) {
	219	if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
	220	kfree_skb(skb);
	221	goto out;
	222	}
	223	nf_reset(skb);
1da177e4 LT	224	}
	225	ret = ipprot->handler(skb);
	226	if (ret < 0) {
	227	protocol = -ret;
	228	goto resubmit;
	229	}
7c73a6fa	230	IP_INC_STATS_BH(net, IPSTATS_MIB_INDELIVERS);
1da177e4	231	} else {
7bc54c90	232	if (!raw) {
1da177e4	233	if (xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
7c73a6fa	234	IP_INC_STATS_BH(net, IPSTATS_MIB_INUNKNOWNPROTOS);
1da177e4 LT	235	icmp_send(skb, ICMP_DEST_UNREACH,
	236	ICMP_PROT_UNREACH, 0);
	237	}
	238	} else
7c73a6fa	239	IP_INC_STATS_BH(net, IPSTATS_MIB_INDELIVERS);
1da177e4 LT	240	kfree_skb(skb);
	241	}
	242	}
	243	out:
	244	rcu_read_unlock();
	245
	246	return 0;
	247	}
	248
	249	/*
	250	* Deliver IP Packets to the higher protocol layers.
e905a9ed	251	*/
1da177e4 LT	252	int ip_local_deliver(struct sk_buff *skb)
	253	{
	254	/*
	255	* Reassemble IP fragments.
	256	*/
	257
56f8a75c	258	if (ip_is_fragment(ip_hdr(skb))) {
776c729e	259	if (ip_defrag(skb, IP_DEFRAG_LOCAL_DELIVER))
1da177e4 LT	260	return 0;
	261	}
	262
9bbc768a	263	return NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_IN, skb, skb->dev, NULL,
1da177e4 LT	264	ip_local_deliver_finish);
	265	}
	266
6a91395f	267	static inline bool ip_rcv_options(struct sk_buff *skb)
d245407e TG	268	{
d245407e TG	269	struct ip_options *opt;
b71d1d42	270	const struct iphdr *iph;
d245407e TG	271	struct net_device *dev = skb->dev;
	272
	273	/* It looks as overkill, because not all
	274	IP options require packet mangling.
	275	But it is the easiest for now, especially taking
	276	into account that combination of IP options
	277	and running sniffer is extremely rare condition.
	278	--ANK (980813)
	279	*/
	280	if (skb_cow(skb, skb_headroom(skb))) {
7c73a6fa	281	IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS);
d245407e TG	282	goto drop;
	283	}
	284
eddc9ec5	285	iph = ip_hdr(skb);
22aba383 DL	286	opt = &(IPCB(skb)->opt);
22aba383 DL	287	opt->optlen = iph->ihl*4 - sizeof(struct iphdr);
d245407e	288
c346dca1	289	if (ip_options_compile(dev_net(dev), opt, skb)) {
7c73a6fa	290	IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INHDRERRORS);
d245407e TG	291	goto drop;
	292	}
	293
d245407e	294	if (unlikely(opt->srr)) {
6e8b11b4 ED	295	struct in_device *in_dev = __in_dev_get_rcu(dev);
6e8b11b4 ED	296
d245407e TG	297	if (in_dev) {
d245407e TG	298	if (!IN_DEV_SOURCE_ROUTE(in_dev)) {
e87cc472 JP	299	if (IN_DEV_LOG_MARTIANS(in_dev))
	300	net_info_ratelimited("source route option %pI4 -> %pI4\n",
	301	&iph->saddr,
	302	&iph->daddr);
d245407e TG	303	goto drop;
d245407e TG	304	}
d245407e TG	305	}
	306
	307	if (ip_options_rcv_srr(skb))
	308	goto drop;
	309	}
	310
6a91395f	311	return false;
d245407e	312	drop:
6a91395f	313	return true;
d245407e TG	314	}
d245407e TG	315
6648bd7e AD	316	int sysctl_ip_early_demux __read_mostly = 1;
6648bd7e AD	317
861d0486	318	static int ip_rcv_finish(struct sk_buff *skb)
1da177e4	319	{
eddc9ec5	320	const struct iphdr *iph = ip_hdr(skb);
5506b54b	321	struct rtable *rt;
1da177e4	322
160eb5a6 DM	323	if (sysctl_ip_early_demux && !skb_dst(skb)) {
	324	const struct net_protocol *ipprot;
	325	int protocol = iph->protocol;
	326
	327	rcu_read_lock();
	328	ipprot = rcu_dereference(inet_protos[protocol]);
9cb429d6	329	if (ipprot && ipprot->early_demux) {
160eb5a6	330	ipprot->early_demux(skb);
9cb429d6 ED	331	/* must reload iph, skb->head might have changed */
	332	iph = ip_hdr(skb);
	333	}
160eb5a6 DM	334	rcu_read_unlock();
	335	}
	336
1da177e4 LT	337	/*
	338	* Initialise the virtual path cache for the packet. It describes
	339	* how the packet travels inside Linux networking.
e905a9ed	340	*/
160eb5a6	341	if (!skb_dst(skb)) {
38a424e4 DM	342	int err = ip_route_input(skb, iph->daddr, iph->saddr,
38a424e4 DM	343	iph->tos, skb->dev);
160eb5a6 DM	344	if (unlikely(err)) {
	345	if (err == -EXDEV)
	346	NET_INC_STATS_BH(dev_net(skb->dev),
	347	LINUX_MIB_IPRPFILTER);
	348	goto drop;
2c2910a4	349	}
1da177e4 LT	350	}
1da177e4 LT	351
c7066f70	352	#ifdef CONFIG_IP_ROUTE_CLASSID
adf30907	353	if (unlikely(skb_dst(skb)->tclassid)) {
7a9b2d59	354	struct ip_rt_acct *st = this_cpu_ptr(ip_rt_acct);
adf30907	355	u32 idx = skb_dst(skb)->tclassid;
1da177e4	356	st[idx&0xFF].o_packets++;
fd3f8c4c	357	st[idx&0xFF].o_bytes += skb->len;
1da177e4	358	st[(idx>>16)&0xFF].i_packets++;
fd3f8c4c	359	st[(idx>>16)&0xFF].i_bytes += skb->len;
1da177e4 LT	360	}
	361	#endif
	362
d245407e TG	363	if (iph->ihl > 5 && ip_rcv_options(skb))
d245407e TG	364	goto drop;
1da177e4	365
511c3f92	366	rt = skb_rtable(skb);
edf391ff	367	if (rt->rt_type == RTN_MULTICAST) {
d8d1f30b	368	IP_UPD_PO_STATS_BH(dev_net(rt->dst.dev), IPSTATS_MIB_INMCAST,
edf391ff NH	369	skb->len);
edf391ff NH	370	} else if (rt->rt_type == RTN_BROADCAST)
d8d1f30b	371	IP_UPD_PO_STATS_BH(dev_net(rt->dst.dev), IPSTATS_MIB_INBCAST,
edf391ff	372	skb->len);
5506b54b	373
1da177e4 LT	374	return dst_input(skb);
1da177e4 LT	375
1da177e4	376	drop:
e905a9ed YH	377	kfree_skb(skb);
e905a9ed YH	378	return NET_RX_DROP;
1da177e4 LT	379	}
	380
	381	/*
	382	* Main IP Receive routine.
e905a9ed	383	*/
f2ccd8fa	384	int ip_rcv(struct sk_buff skb, struct net_device dev, struct packet_type pt, struct net_device orig_dev)
1da177e4	385	{
b71d1d42	386	const struct iphdr *iph;
58615242	387	u32 len;
1da177e4 LT	388
	389	/* When the interface is in promisc. mode, drop all the crap
	390	* that it receives, do not try to analyse it.
	391	*/
	392	if (skb->pkt_type == PACKET_OTHERHOST)
	393	goto drop;
	394
edf391ff NH	395
edf391ff NH	396	IP_UPD_PO_STATS_BH(dev_net(dev), IPSTATS_MIB_IN, skb->len);
1da177e4 LT	397
1da177e4 LT	398	if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL) {
7c73a6fa	399	IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS);
1da177e4 LT	400	goto out;
	401	}
	402
	403	if (!pskb_may_pull(skb, sizeof(struct iphdr)))
	404	goto inhdr_error;
	405
eddc9ec5	406	iph = ip_hdr(skb);
1da177e4 LT	407
1da177e4 LT	408	/*
c67fa027	409	* RFC1122: 3.2.1.2 MUST silently discard any IP frame that fails the checksum.
1da177e4 LT	410	*
	411	* Is the datagram acceptable?
	412	*
	413	* 1. Length at least the size of an ip header
	414	* 2. Version of 4
	415	* 3. Checksums correctly. [Speed optimisation for later, skip loopback checksums]
	416	* 4. Doesn't have a bogus length
	417	*/
	418
	419	if (iph->ihl < 5 \|\| iph->version != 4)
58615242	420	goto inhdr_error;
1da177e4 LT	421
	422	if (!pskb_may_pull(skb, iph->ihl*4))
	423	goto inhdr_error;
	424
eddc9ec5	425	iph = ip_hdr(skb);
1da177e4	426
e9c60422	427	if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl)))
58615242	428	goto inhdr_error;
1da177e4	429
58615242	430	len = ntohs(iph->tot_len);
704aed53	431	if (skb->len < len) {
7c73a6fa	432	IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INTRUNCATEDPKTS);
704aed53 MC	433	goto drop;
704aed53 MC	434	} else if (len < (iph->ihl*4))
58615242	435	goto inhdr_error;
1da177e4	436
58615242 TG	437	/* Our transport medium may have padded the buffer out. Now we know it
	438	* is IP we can trim to the true length of the frame.
	439	* Note this now means skb->len holds ntohs(iph->tot_len).
	440	*/
	441	if (pskb_trim_rcsum(skb, len)) {
7c73a6fa	442	IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS);
58615242	443	goto drop;
1da177e4 LT	444	}
1da177e4 LT	445
53602f92	446	/* Remove any debris in the socket control block */
d569f1d7	447	memset(IPCB(skb), 0, sizeof(struct inet_skb_parm));
53602f92	448
71f9dacd HX	449	/* Must drop socket now because of tproxy. */
	450	skb_orphan(skb);
	451
9bbc768a	452	return NF_HOOK(NFPROTO_IPV4, NF_INET_PRE_ROUTING, skb, dev, NULL,
1da177e4 LT	453	ip_rcv_finish);
	454
	455	inhdr_error:
7c73a6fa	456	IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INHDRERRORS);
1da177e4	457	drop:
e905a9ed	458	kfree_skb(skb);
1da177e4	459	out:
e905a9ed	460	return NET_RX_DROP;
1da177e4	461	}