[deliverable/linux.git] / net / netfilter / ipvs / ip_vs_proto_udp.c

/*
 * ip_vs_proto_udp.c:	UDP load balancing support for IPVS
 *
 * Authors:     Wensong Zhang <wensong@linuxvirtualserver.org>
 *              Julian Anastasov <ja@ssi.bg>
 *
 *              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.
 *
 * Changes:     Hans Schillstrom <hans.schillstrom@ericsson.com>
 *              Network name space (netns) aware.
 *
 */

#define KMSG_COMPONENT "IPVS"
#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt

#include <linux/in.h>
#include <linux/ip.h>
#include <linux/kernel.h>
#include <linux/netfilter.h>
#include <linux/netfilter_ipv4.h>
#include <linux/udp.h>

#include <net/ip_vs.h>
#include <net/ip.h>
#include <net/ip6_checksum.h>

static int
udp_conn_schedule(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
		  int *verdict, struct ip_vs_conn **cpp,
		  struct ip_vs_iphdr *iph)
{
	struct net *net;
	struct ip_vs_service *svc;
	struct udphdr _udph, *uh;

	/* IPv6 fragments, only first fragment will hit this */
	uh = skb_header_pointer(skb, iph->len, sizeof(_udph), &_udph);
	if (uh == NULL) {
		*verdict = NF_DROP;
		return 0;
	}
	net = skb_net(skb);
	rcu_read_lock();
	svc = ip_vs_service_find(net, af, skb->mark, iph->protocol,
				 &iph->daddr, uh->dest);
	if (svc) {
		int ignored;

		if (ip_vs_todrop(net_ipvs(net))) {
			/*
			 * It seems that we are very loaded.
			 * We have to drop this packet :(
			 */
			rcu_read_unlock();
			*verdict = NF_DROP;
			return 0;
		}

		/*
		 * Let the virtual server select a real server for the
		 * incoming connection, and create a connection entry.
		 */
		*cpp = ip_vs_schedule(svc, skb, pd, &ignored, iph);
		if (!*cpp && ignored <= 0) {
			if (!ignored)
				*verdict = ip_vs_leave(svc, skb, pd, iph);
			else
				*verdict = NF_DROP;
			rcu_read_unlock();
			return 0;
		}
	}
	rcu_read_unlock();
	/* NF_ACCEPT */
	return 1;
}


static inline void
udp_fast_csum_update(int af, struct udphdr *uhdr,
		     const union nf_inet_addr *oldip,
		     const union nf_inet_addr *newip,
		     __be16 oldport, __be16 newport)
{
#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6)
		uhdr->check =
			csum_fold(ip_vs_check_diff16(oldip->ip6, newip->ip6,
					 ip_vs_check_diff2(oldport, newport,
						~csum_unfold(uhdr->check))));
	else
#endif
		uhdr->check =
			csum_fold(ip_vs_check_diff4(oldip->ip, newip->ip,
					 ip_vs_check_diff2(oldport, newport,
						~csum_unfold(uhdr->check))));
	if (!uhdr->check)
		uhdr->check = CSUM_MANGLED_0;
}

static inline void
udp_partial_csum_update(int af, struct udphdr *uhdr,
		     const union nf_inet_addr *oldip,
		     const union nf_inet_addr *newip,
		     __be16 oldlen, __be16 newlen)
{
#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6)
		uhdr->check =
			~csum_fold(ip_vs_check_diff16(oldip->ip6, newip->ip6,
					 ip_vs_check_diff2(oldlen, newlen,
						csum_unfold(uhdr->check))));
	else
#endif
	uhdr->check =
		~csum_fold(ip_vs_check_diff4(oldip->ip, newip->ip,
				ip_vs_check_diff2(oldlen, newlen,
						csum_unfold(uhdr->check))));
}


static int
udp_snat_handler(struct sk_buff *skb, struct ip_vs_protocol *pp,
		 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph)
{
	struct udphdr *udph;
	unsigned int udphoff = iph->len;
	int oldlen;
	int payload_csum = 0;

#ifdef CONFIG_IP_VS_IPV6
	if (cp->af == AF_INET6 && iph->fragoffs)
		return 1;
#endif
	oldlen = skb->len - udphoff;

	/* csum_check requires unshared skb */
	if (!skb_make_writable(skb, udphoff+sizeof(*udph)))
		return 0;

	if (unlikely(cp->app != NULL)) {
		int ret;

		/* Some checks before mangling */
		if (pp->csum_check && !pp->csum_check(cp->af, skb, pp))
			return 0;

		/*
		 *	Call application helper if needed
		 */
		if (!(ret = ip_vs_app_pkt_out(cp, skb)))
			return 0;
		/* ret=2: csum update is needed after payload mangling */
		if (ret == 1)
			oldlen = skb->len - udphoff;
		else
			payload_csum = 1;
	}

	udph = (void *)skb_network_header(skb) + udphoff;
	udph->source = cp->vport;

	/*
	 *	Adjust UDP checksums
	 */
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
		udp_partial_csum_update(cp->af, udph, &cp->daddr, &cp->vaddr,
					htons(oldlen),
					htons(skb->len - udphoff));
	} else if (!payload_csum && (udph->check != 0)) {
		/* Only port and addr are changed, do fast csum update */
		udp_fast_csum_update(cp->af, udph, &cp->daddr, &cp->vaddr,
				     cp->dport, cp->vport);
		if (skb->ip_summed == CHECKSUM_COMPLETE)
			skb->ip_summed = (cp->app && pp->csum_check) ?
					 CHECKSUM_UNNECESSARY : CHECKSUM_NONE;
	} else {
		/* full checksum calculation */
		udph->check = 0;
		skb->csum = skb_checksum(skb, udphoff, skb->len - udphoff, 0);
#ifdef CONFIG_IP_VS_IPV6
		if (cp->af == AF_INET6)
			udph->check = csum_ipv6_magic(&cp->vaddr.in6,
						      &cp->caddr.in6,
						      skb->len - udphoff,
						      cp->protocol, skb->csum);
		else
#endif
			udph->check = csum_tcpudp_magic(cp->vaddr.ip,
							cp->caddr.ip,
							skb->len - udphoff,
							cp->protocol,
							skb->csum);
		if (udph->check == 0)
			udph->check = CSUM_MANGLED_0;
		skb->ip_summed = CHECKSUM_UNNECESSARY;
		IP_VS_DBG(11, "O-pkt: %s O-csum=%d (+%zd)\n",
			  pp->name, udph->check,
			  (char*)&(udph->check) - (char*)udph);
	}
	return 1;
}


static int
udp_dnat_handler(struct sk_buff *skb, struct ip_vs_protocol *pp,
		 struct ip_vs_conn *cp, struct ip_vs_iphdr *iph)
{
	struct udphdr *udph;
	unsigned int udphoff = iph->len;
	int oldlen;
	int payload_csum = 0;

#ifdef CONFIG_IP_VS_IPV6
	if (cp->af == AF_INET6 && iph->fragoffs)
		return 1;
#endif
	oldlen = skb->len - udphoff;

	/* csum_check requires unshared skb */
	if (!skb_make_writable(skb, udphoff+sizeof(*udph)))
		return 0;

	if (unlikely(cp->app != NULL)) {
		int ret;

		/* Some checks before mangling */
		if (pp->csum_check && !pp->csum_check(cp->af, skb, pp))
			return 0;

		/*
		 *	Attempt ip_vs_app call.
		 *	It will fix ip_vs_conn
		 */
		if (!(ret = ip_vs_app_pkt_in(cp, skb)))
			return 0;
		/* ret=2: csum update is needed after payload mangling */
		if (ret == 1)
			oldlen = skb->len - udphoff;
		else
			payload_csum = 1;
	}

	udph = (void *)skb_network_header(skb) + udphoff;
	udph->dest = cp->dport;

	/*
	 *	Adjust UDP checksums
	 */
	if (skb->ip_summed == CHECKSUM_PARTIAL) {
		udp_partial_csum_update(cp->af, udph, &cp->vaddr, &cp->daddr,
					htons(oldlen),
					htons(skb->len - udphoff));
	} else if (!payload_csum && (udph->check != 0)) {
		/* Only port and addr are changed, do fast csum update */
		udp_fast_csum_update(cp->af, udph, &cp->vaddr, &cp->daddr,
				     cp->vport, cp->dport);
		if (skb->ip_summed == CHECKSUM_COMPLETE)
			skb->ip_summed = (cp->app && pp->csum_check) ?
					 CHECKSUM_UNNECESSARY : CHECKSUM_NONE;
	} else {
		/* full checksum calculation */
		udph->check = 0;
		skb->csum = skb_checksum(skb, udphoff, skb->len - udphoff, 0);
#ifdef CONFIG_IP_VS_IPV6
		if (cp->af == AF_INET6)
			udph->check = csum_ipv6_magic(&cp->caddr.in6,
						      &cp->daddr.in6,
						      skb->len - udphoff,
						      cp->protocol, skb->csum);
		else
#endif
			udph->check = csum_tcpudp_magic(cp->caddr.ip,
							cp->daddr.ip,
							skb->len - udphoff,
							cp->protocol,
							skb->csum);
		if (udph->check == 0)
			udph->check = CSUM_MANGLED_0;
		skb->ip_summed = CHECKSUM_UNNECESSARY;
	}
	return 1;
}


static int
udp_csum_check(int af, struct sk_buff *skb, struct ip_vs_protocol *pp)
{
	struct udphdr _udph, *uh;
	unsigned int udphoff;

#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6)
		udphoff = sizeof(struct ipv6hdr);
	else
#endif
		udphoff = ip_hdrlen(skb);

	uh = skb_header_pointer(skb, udphoff, sizeof(_udph), &_udph);
	if (uh == NULL)
		return 0;

	if (uh->check != 0) {
		switch (skb->ip_summed) {
		case CHECKSUM_NONE:
			skb->csum = skb_checksum(skb, udphoff,
						 skb->len - udphoff, 0);
		case CHECKSUM_COMPLETE:
#ifdef CONFIG_IP_VS_IPV6
			if (af == AF_INET6) {
				if (csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
						    &ipv6_hdr(skb)->daddr,
						    skb->len - udphoff,
						    ipv6_hdr(skb)->nexthdr,
						    skb->csum)) {
					IP_VS_DBG_RL_PKT(0, af, pp, skb, 0,
							 "Failed checksum for");
					return 0;
				}
			} else
#endif
				if (csum_tcpudp_magic(ip_hdr(skb)->saddr,
						      ip_hdr(skb)->daddr,
						      skb->len - udphoff,
						      ip_hdr(skb)->protocol,
						      skb->csum)) {
					IP_VS_DBG_RL_PKT(0, af, pp, skb, 0,
							 "Failed checksum for");
					return 0;
				}
			break;
		default:
			/* No need to checksum. */
			break;
		}
	}
	return 1;
}

static inline __u16 udp_app_hashkey(__be16 port)
{
	return (((__force u16)port >> UDP_APP_TAB_BITS) ^ (__force u16)port)
		& UDP_APP_TAB_MASK;
}


static int udp_register_app(struct net *net, struct ip_vs_app *inc)
{
	struct ip_vs_app *i;
	__u16 hash;
	__be16 port = inc->port;
	int ret = 0;
	struct netns_ipvs *ipvs = net_ipvs(net);
	struct ip_vs_proto_data *pd = ip_vs_proto_data_get(net, IPPROTO_UDP);

	hash = udp_app_hashkey(port);

	list_for_each_entry(i, &ipvs->udp_apps[hash], p_list) {
		if (i->port == port) {
			ret = -EEXIST;
			goto out;
		}
	}
	list_add_rcu(&inc->p_list, &ipvs->udp_apps[hash]);
	atomic_inc(&pd->appcnt);

  out:
	return ret;
}


static void
udp_unregister_app(struct net *net, struct ip_vs_app *inc)
{
	struct ip_vs_proto_data *pd = ip_vs_proto_data_get(net, IPPROTO_UDP);

	atomic_dec(&pd->appcnt);
	list_del_rcu(&inc->p_list);
}


static int udp_app_conn_bind(struct ip_vs_conn *cp)
{
	struct netns_ipvs *ipvs = net_ipvs(ip_vs_conn_net(cp));
	int hash;
	struct ip_vs_app *inc;
	int result = 0;

	/* Default binding: bind app only for NAT */
	if (IP_VS_FWD_METHOD(cp) != IP_VS_CONN_F_MASQ)
		return 0;

	/* Lookup application incarnations and bind the right one */
	hash = udp_app_hashkey(cp->vport);

	rcu_read_lock();
	list_for_each_entry_rcu(inc, &ipvs->udp_apps[hash], p_list) {
		if (inc->port == cp->vport) {
			if (unlikely(!ip_vs_app_inc_get(inc)))
				break;
			rcu_read_unlock();

			IP_VS_DBG_BUF(9, "%s(): Binding conn %s:%u->"
				      "%s:%u to app %s on port %u\n",
				      __func__,
				      IP_VS_DBG_ADDR(cp->af, &cp->caddr),
				      ntohs(cp->cport),
				      IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
				      ntohs(cp->vport),
				      inc->name, ntohs(inc->port));

			cp->app = inc;
			if (inc->init_conn)
				result = inc->init_conn(inc, cp);
			goto out;
		}
	}
	rcu_read_unlock();

  out:
	return result;
}


static const int udp_timeouts[IP_VS_UDP_S_LAST+1] = {
	[IP_VS_UDP_S_NORMAL]		=	5*60*HZ,
	[IP_VS_UDP_S_LAST]		=	2*HZ,
};

static const char *const udp_state_name_table[IP_VS_UDP_S_LAST+1] = {
	[IP_VS_UDP_S_NORMAL]		=	"UDP",
	[IP_VS_UDP_S_LAST]		=	"BUG!",
};

static const char * udp_state_name(int state)
{
	if (state >= IP_VS_UDP_S_LAST)
		return "ERR!";
	return udp_state_name_table[state] ? udp_state_name_table[state] : "?";
}

static void
udp_state_transition(struct ip_vs_conn *cp, int direction,
		     const struct sk_buff *skb,
		     struct ip_vs_proto_data *pd)
{
	if (unlikely(!pd)) {
		pr_err("UDP no ns data\n");
		return;
	}

	cp->timeout = pd->timeout_table[IP_VS_UDP_S_NORMAL];
}

static int __udp_init(struct net *net, struct ip_vs_proto_data *pd)
{
	struct netns_ipvs *ipvs = net_ipvs(net);

	ip_vs_init_hash_table(ipvs->udp_apps, UDP_APP_TAB_SIZE);
	pd->timeout_table = ip_vs_create_timeout_table((int *)udp_timeouts,
							sizeof(udp_timeouts));
	if (!pd->timeout_table)
		return -ENOMEM;
	return 0;
}

static void __udp_exit(struct net *net, struct ip_vs_proto_data *pd)
{
	kfree(pd->timeout_table);
}


struct ip_vs_protocol ip_vs_protocol_udp = {
	.name =			"UDP",
	.protocol =		IPPROTO_UDP,
	.num_states =		IP_VS_UDP_S_LAST,
	.dont_defrag =		0,
	.init =			NULL,
	.exit =			NULL,
	.init_netns =		__udp_init,
	.exit_netns =		__udp_exit,
	.conn_schedule =	udp_conn_schedule,
	.conn_in_get =		ip_vs_conn_in_get_proto,
	.conn_out_get =		ip_vs_conn_out_get_proto,
	.snat_handler =		udp_snat_handler,
	.dnat_handler =		udp_dnat_handler,
	.csum_check =		udp_csum_check,
	.state_transition =	udp_state_transition,
	.state_name =		udp_state_name,
	.register_app =		udp_register_app,
	.unregister_app =	udp_unregister_app,
	.app_conn_bind =	udp_app_conn_bind,
	.debug_packet =		ip_vs_tcpudp_debug_packet,
	.timeout_change =	NULL,
};
Commit	Line	Data
1da177e4 LT	1	/*
	2	* ip_vs_proto_udp.c: UDP load balancing support for IPVS
	3	*
1da177e4 LT	4	* Authors: Wensong Zhang <wensong@linuxvirtualserver.org>
	5	* Julian Anastasov <ja@ssi.bg>
	6	*
	7	* This program is free software; you can redistribute it and/or
	8	* modify it under the terms of the GNU General Public License
	9	* as published by the Free Software Foundation; either version
	10	* 2 of the License, or (at your option) any later version.
	11	*
78b16bde HS	12	* Changes: Hans Schillstrom <hans.schillstrom@ericsson.com>
78b16bde HS	13	* Network name space (netns) aware.
1da177e4 LT	14	*
	15	*/
	16
9aada7ac HE	17	#define KMSG_COMPONENT "IPVS"
	18	#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
	19
14c85021 ACM	20	#include <linux/in.h>
14c85021 ACM	21	#include <linux/ip.h>
1da177e4	22	#include <linux/kernel.h>
af1e1cf0	23	#include <linux/netfilter.h>
1da177e4	24	#include <linux/netfilter_ipv4.h>
14c85021	25	#include <linux/udp.h>
1da177e4 LT	26
1da177e4 LT	27	#include <net/ip_vs.h>
c9bdd4b5	28	#include <net/ip.h>
63f2c046	29	#include <net/ip6_checksum.h>
1da177e4	30
1da177e4	31	static int
9330419d	32	udp_conn_schedule(int af, struct sk_buff skb, struct ip_vs_proto_data pd,
d4383f04 JDB	33	int verdict, struct ip_vs_conn *cpp,
d4383f04 JDB	34	struct ip_vs_iphdr *iph)
1da177e4	35	{
fc723250	36	struct net *net;
1da177e4 LT	37	struct ip_vs_service *svc;
	38	struct udphdr _udph, *uh;
	39
d4383f04 JDB	40	/* IPv6 fragments, only first fragment will hit this */
d4383f04 JDB	41	uh = skb_header_pointer(skb, iph->len, sizeof(_udph), &_udph);
1da177e4 LT	42	if (uh == NULL) {
	43	*verdict = NF_DROP;
	44	return 0;
	45	}
fc723250	46	net = skb_net(skb);
ceec4c38 JA	47	rcu_read_lock();
	48	svc = ip_vs_service_find(net, af, skb->mark, iph->protocol,
	49	&iph->daddr, uh->dest);
3c2e0505	50	if (svc) {
190ecd27 JA	51	int ignored;
190ecd27 JA	52
a0840e2e	53	if (ip_vs_todrop(net_ipvs(net))) {
1da177e4 LT	54	/*
	55	* It seems that we are very loaded.
	56	* We have to drop this packet :(
	57	*/
ceec4c38	58	rcu_read_unlock();
1da177e4 LT	59	*verdict = NF_DROP;
	60	return 0;
	61	}
	62
	63	/*
	64	* Let the virtual server select a real server for the
	65	* incoming connection, and create a connection entry.
	66	*/
d4383f04	67	*cpp = ip_vs_schedule(svc, skb, pd, &ignored, iph);
a5959d53 HS	68	if (!*cpp && ignored <= 0) {
a5959d53 HS	69	if (!ignored)
d4383f04	70	*verdict = ip_vs_leave(svc, skb, pd, iph);
ceec4c38	71	else
a5959d53	72	*verdict = NF_DROP;
ceec4c38	73	rcu_read_unlock();
1da177e4 LT	74	return 0;
1da177e4 LT	75	}
1da177e4	76	}
ceec4c38	77	rcu_read_unlock();
a5959d53	78	/* NF_ACCEPT */
1da177e4 LT	79	return 1;
	80	}
	81
	82
	83	static inline void
0bbdd42b JV	84	udp_fast_csum_update(int af, struct udphdr *uhdr,
	85	const union nf_inet_addr *oldip,
	86	const union nf_inet_addr *newip,
014d730d	87	__be16 oldport, __be16 newport)
1da177e4	88	{
0bbdd42b JV	89	#ifdef CONFIG_IP_VS_IPV6
	90	if (af == AF_INET6)
	91	uhdr->check =
	92	csum_fold(ip_vs_check_diff16(oldip->ip6, newip->ip6,
	93	ip_vs_check_diff2(oldport, newport,
	94	~csum_unfold(uhdr->check))));
	95	else
	96	#endif
	97	uhdr->check =
	98	csum_fold(ip_vs_check_diff4(oldip->ip, newip->ip,
	99	ip_vs_check_diff2(oldport, newport,
	100	~csum_unfold(uhdr->check))));
1da177e4	101	if (!uhdr->check)
f6ab0288	102	uhdr->check = CSUM_MANGLED_0;
1da177e4 LT	103	}
1da177e4 LT	104
503e81f6 SH	105	static inline void
	106	udp_partial_csum_update(int af, struct udphdr *uhdr,
	107	const union nf_inet_addr *oldip,
	108	const union nf_inet_addr *newip,
	109	__be16 oldlen, __be16 newlen)
	110	{
	111	#ifdef CONFIG_IP_VS_IPV6
	112	if (af == AF_INET6)
	113	uhdr->check =
5bc9068e	114	~csum_fold(ip_vs_check_diff16(oldip->ip6, newip->ip6,
503e81f6	115	ip_vs_check_diff2(oldlen, newlen,
5bc9068e	116	csum_unfold(uhdr->check))));
503e81f6 SH	117	else
	118	#endif
	119	uhdr->check =
5bc9068e	120	~csum_fold(ip_vs_check_diff4(oldip->ip, newip->ip,
503e81f6	121	ip_vs_check_diff2(oldlen, newlen,
5bc9068e	122	csum_unfold(uhdr->check))));
503e81f6 SH	123	}
	124
	125
1da177e4	126	static int
d4383f04 JDB	127	udp_snat_handler(struct sk_buff skb, struct ip_vs_protocol pp,
d4383f04 JDB	128	struct ip_vs_conn cp, struct ip_vs_iphdr iph)
1da177e4 LT	129	{
1da177e4 LT	130	struct udphdr *udph;
d4383f04	131	unsigned int udphoff = iph->len;
503e81f6	132	int oldlen;
8b27b10f	133	int payload_csum = 0;
0bbdd42b JV	134
0bbdd42b JV	135	#ifdef CONFIG_IP_VS_IPV6
d4383f04	136	if (cp->af == AF_INET6 && iph->fragoffs)
63dca2c0	137	return 1;
0bbdd42b	138	#endif
503e81f6	139	oldlen = skb->len - udphoff;
1da177e4 LT	140
1da177e4 LT	141	/* csum_check requires unshared skb */
3db05fea	142	if (!skb_make_writable(skb, udphoff+sizeof(*udph)))
1da177e4 LT	143	return 0;
	144
	145	if (unlikely(cp->app != NULL)) {
8b27b10f JA	146	int ret;
8b27b10f JA	147
1da177e4	148	/* Some checks before mangling */
0bbdd42b	149	if (pp->csum_check && !pp->csum_check(cp->af, skb, pp))
1da177e4 LT	150	return 0;
	151
	152	/*
	153	* Call application helper if needed
	154	*/
8b27b10f	155	if (!(ret = ip_vs_app_pkt_out(cp, skb)))
1da177e4	156	return 0;
8b27b10f JA	157	/* ret=2: csum update is needed after payload mangling */
	158	if (ret == 1)
	159	oldlen = skb->len - udphoff;
	160	else
	161	payload_csum = 1;
1da177e4 LT	162	}
1da177e4 LT	163
0bbdd42b	164	udph = (void *)skb_network_header(skb) + udphoff;
1da177e4 LT	165	udph->source = cp->vport;
	166
	167	/*
	168	* Adjust UDP checksums
	169	*/
503e81f6 SH	170	if (skb->ip_summed == CHECKSUM_PARTIAL) {
503e81f6 SH	171	udp_partial_csum_update(cp->af, udph, &cp->daddr, &cp->vaddr,
ca62059b HH	172	htons(oldlen),
ca62059b HH	173	htons(skb->len - udphoff));
8b27b10f	174	} else if (!payload_csum && (udph->check != 0)) {
1da177e4	175	/* Only port and addr are changed, do fast csum update */
0bbdd42b	176	udp_fast_csum_update(cp->af, udph, &cp->daddr, &cp->vaddr,
1da177e4	177	cp->dport, cp->vport);
3db05fea	178	if (skb->ip_summed == CHECKSUM_COMPLETE)
8b27b10f JA	179	skb->ip_summed = (cp->app && pp->csum_check) ?
8b27b10f JA	180	CHECKSUM_UNNECESSARY : CHECKSUM_NONE;
1da177e4 LT	181	} else {
	182	/* full checksum calculation */
	183	udph->check = 0;
3db05fea	184	skb->csum = skb_checksum(skb, udphoff, skb->len - udphoff, 0);
0bbdd42b JV	185	#ifdef CONFIG_IP_VS_IPV6
	186	if (cp->af == AF_INET6)
	187	udph->check = csum_ipv6_magic(&cp->vaddr.in6,
	188	&cp->caddr.in6,
	189	skb->len - udphoff,
	190	cp->protocol, skb->csum);
	191	else
	192	#endif
	193	udph->check = csum_tcpudp_magic(cp->vaddr.ip,
	194	cp->caddr.ip,
	195	skb->len - udphoff,
	196	cp->protocol,
	197	skb->csum);
1da177e4	198	if (udph->check == 0)
f6ab0288	199	udph->check = CSUM_MANGLED_0;
8b27b10f	200	skb->ip_summed = CHECKSUM_UNNECESSARY;
1da177e4 LT	201	IP_VS_DBG(11, "O-pkt: %s O-csum=%d (+%zd)\n",
	202	pp->name, udph->check,
	203	(char)&(udph->check) - (char)udph);
	204	}
	205	return 1;
	206	}
	207
	208
	209	static int
d4383f04 JDB	210	udp_dnat_handler(struct sk_buff skb, struct ip_vs_protocol pp,
d4383f04 JDB	211	struct ip_vs_conn cp, struct ip_vs_iphdr iph)
1da177e4 LT	212	{
1da177e4 LT	213	struct udphdr *udph;
d4383f04	214	unsigned int udphoff = iph->len;
503e81f6	215	int oldlen;
8b27b10f	216	int payload_csum = 0;
0bbdd42b JV	217
0bbdd42b JV	218	#ifdef CONFIG_IP_VS_IPV6
d4383f04	219	if (cp->af == AF_INET6 && iph->fragoffs)
63dca2c0	220	return 1;
0bbdd42b	221	#endif
503e81f6	222	oldlen = skb->len - udphoff;
1da177e4 LT	223
1da177e4 LT	224	/* csum_check requires unshared skb */
3db05fea	225	if (!skb_make_writable(skb, udphoff+sizeof(*udph)))
1da177e4 LT	226	return 0;
	227
	228	if (unlikely(cp->app != NULL)) {
8b27b10f JA	229	int ret;
8b27b10f JA	230
1da177e4	231	/* Some checks before mangling */
0bbdd42b	232	if (pp->csum_check && !pp->csum_check(cp->af, skb, pp))
1da177e4 LT	233	return 0;
	234
	235	/*
	236	* Attempt ip_vs_app call.
	237	* It will fix ip_vs_conn
	238	*/
8b27b10f	239	if (!(ret = ip_vs_app_pkt_in(cp, skb)))
1da177e4	240	return 0;
8b27b10f JA	241	/* ret=2: csum update is needed after payload mangling */
	242	if (ret == 1)
	243	oldlen = skb->len - udphoff;
	244	else
	245	payload_csum = 1;
1da177e4 LT	246	}
1da177e4 LT	247
0bbdd42b	248	udph = (void *)skb_network_header(skb) + udphoff;
1da177e4 LT	249	udph->dest = cp->dport;
	250
	251	/*
	252	* Adjust UDP checksums
	253	*/
503e81f6	254	if (skb->ip_summed == CHECKSUM_PARTIAL) {
5bc9068e	255	udp_partial_csum_update(cp->af, udph, &cp->vaddr, &cp->daddr,
ca62059b HH	256	htons(oldlen),
ca62059b HH	257	htons(skb->len - udphoff));
8b27b10f	258	} else if (!payload_csum && (udph->check != 0)) {
1da177e4	259	/* Only port and addr are changed, do fast csum update */
0bbdd42b	260	udp_fast_csum_update(cp->af, udph, &cp->vaddr, &cp->daddr,
1da177e4	261	cp->vport, cp->dport);
3db05fea	262	if (skb->ip_summed == CHECKSUM_COMPLETE)
8b27b10f JA	263	skb->ip_summed = (cp->app && pp->csum_check) ?
8b27b10f JA	264	CHECKSUM_UNNECESSARY : CHECKSUM_NONE;
1da177e4 LT	265	} else {
	266	/* full checksum calculation */
	267	udph->check = 0;
3db05fea	268	skb->csum = skb_checksum(skb, udphoff, skb->len - udphoff, 0);
0bbdd42b JV	269	#ifdef CONFIG_IP_VS_IPV6
	270	if (cp->af == AF_INET6)
	271	udph->check = csum_ipv6_magic(&cp->caddr.in6,
	272	&cp->daddr.in6,
	273	skb->len - udphoff,
	274	cp->protocol, skb->csum);
	275	else
	276	#endif
	277	udph->check = csum_tcpudp_magic(cp->caddr.ip,
	278	cp->daddr.ip,
	279	skb->len - udphoff,
	280	cp->protocol,
	281	skb->csum);
1da177e4	282	if (udph->check == 0)
f6ab0288	283	udph->check = CSUM_MANGLED_0;
3db05fea	284	skb->ip_summed = CHECKSUM_UNNECESSARY;
1da177e4 LT	285	}
	286	return 1;
	287	}
	288
	289
	290	static int
51ef348b	291	udp_csum_check(int af, struct sk_buff skb, struct ip_vs_protocol pp)
1da177e4 LT	292	{
1da177e4 LT	293	struct udphdr _udph, *uh;
51ef348b JV	294	unsigned int udphoff;
	295
	296	#ifdef CONFIG_IP_VS_IPV6
	297	if (af == AF_INET6)
	298	udphoff = sizeof(struct ipv6hdr);
	299	else
	300	#endif
	301	udphoff = ip_hdrlen(skb);
1da177e4 LT	302
	303	uh = skb_header_pointer(skb, udphoff, sizeof(_udph), &_udph);
	304	if (uh == NULL)
	305	return 0;
	306
	307	if (uh->check != 0) {
	308	switch (skb->ip_summed) {
	309	case CHECKSUM_NONE:
	310	skb->csum = skb_checksum(skb, udphoff,
	311	skb->len - udphoff, 0);
84fa7933	312	case CHECKSUM_COMPLETE:
51ef348b JV	313	#ifdef CONFIG_IP_VS_IPV6
	314	if (af == AF_INET6) {
	315	if (csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
	316	&ipv6_hdr(skb)->daddr,
	317	skb->len - udphoff,
	318	ipv6_hdr(skb)->nexthdr,
	319	skb->csum)) {
0d79641a	320	IP_VS_DBG_RL_PKT(0, af, pp, skb, 0,
51ef348b JV	321	"Failed checksum for");
	322	return 0;
	323	}
	324	} else
	325	#endif
	326	if (csum_tcpudp_magic(ip_hdr(skb)->saddr,
	327	ip_hdr(skb)->daddr,
	328	skb->len - udphoff,
	329	ip_hdr(skb)->protocol,
	330	skb->csum)) {
0d79641a	331	IP_VS_DBG_RL_PKT(0, af, pp, skb, 0,
51ef348b JV	332	"Failed checksum for");
	333	return 0;
	334	}
1da177e4 LT	335	break;
1da177e4 LT	336	default:
84fa7933	337	/* No need to checksum. */
1da177e4 LT	338	break;
	339	}
	340	}
	341	return 1;
	342	}
	343
75e7ce66	344	static inline __u16 udp_app_hashkey(__be16 port)
1da177e4	345	{
75e7ce66 AV	346	return (((__force u16)port >> UDP_APP_TAB_BITS) ^ (__force u16)port)
75e7ce66 AV	347	& UDP_APP_TAB_MASK;
1da177e4 LT	348	}
	349
	350
ab8a5e84	351	static int udp_register_app(struct net net, struct ip_vs_app inc)
1da177e4 LT	352	{
1da177e4 LT	353	struct ip_vs_app *i;
75e7ce66 AV	354	__u16 hash;
75e7ce66 AV	355	__be16 port = inc->port;
1da177e4	356	int ret = 0;
ab8a5e84 HS	357	struct netns_ipvs *ipvs = net_ipvs(net);
ab8a5e84 HS	358	struct ip_vs_proto_data *pd = ip_vs_proto_data_get(net, IPPROTO_UDP);
1da177e4 LT	359
	360	hash = udp_app_hashkey(port);
	361
78b16bde	362	list_for_each_entry(i, &ipvs->udp_apps[hash], p_list) {
1da177e4 LT	363	if (i->port == port) {
	364	ret = -EEXIST;
	365	goto out;
	366	}
	367	}
363c97d7	368	list_add_rcu(&inc->p_list, &ipvs->udp_apps[hash]);
9bbac6a9	369	atomic_inc(&pd->appcnt);
1da177e4 LT	370
1da177e4 LT	371	out:
1da177e4 LT	372	return ret;
	373	}
	374
	375
	376	static void
ab8a5e84	377	udp_unregister_app(struct net net, struct ip_vs_app inc)
1da177e4	378	{
ab8a5e84	379	struct ip_vs_proto_data *pd = ip_vs_proto_data_get(net, IPPROTO_UDP);
78b16bde	380
9bbac6a9	381	atomic_dec(&pd->appcnt);
363c97d7	382	list_del_rcu(&inc->p_list);
1da177e4 LT	383	}
	384
	385
	386	static int udp_app_conn_bind(struct ip_vs_conn *cp)
	387	{
6e67e586	388	struct netns_ipvs *ipvs = net_ipvs(ip_vs_conn_net(cp));
1da177e4 LT	389	int hash;
	390	struct ip_vs_app *inc;
	391	int result = 0;
	392
	393	/* Default binding: bind app only for NAT */
	394	if (IP_VS_FWD_METHOD(cp) != IP_VS_CONN_F_MASQ)
	395	return 0;
	396
	397	/* Lookup application incarnations and bind the right one */
	398	hash = udp_app_hashkey(cp->vport);
	399
363c97d7 JA	400	rcu_read_lock();
363c97d7 JA	401	list_for_each_entry_rcu(inc, &ipvs->udp_apps[hash], p_list) {
1da177e4 LT	402	if (inc->port == cp->vport) {
	403	if (unlikely(!ip_vs_app_inc_get(inc)))
	404	break;
363c97d7	405	rcu_read_unlock();
1da177e4	406
1e3e238e	407	IP_VS_DBG_BUF(9, "%s(): Binding conn %s:%u->"
cfc78c5a JV	408	"%s:%u to app %s on port %u\n",
	409	__func__,
	410	IP_VS_DBG_ADDR(cp->af, &cp->caddr),
	411	ntohs(cp->cport),
	412	IP_VS_DBG_ADDR(cp->af, &cp->vaddr),
	413	ntohs(cp->vport),
	414	inc->name, ntohs(inc->port));
	415
1da177e4 LT	416	cp->app = inc;
	417	if (inc->init_conn)
	418	result = inc->init_conn(inc, cp);
	419	goto out;
	420	}
	421	}
363c97d7	422	rcu_read_unlock();
1da177e4 LT	423
	424	out:
	425	return result;
	426	}
	427
	428
78b16bde	429	static const int udp_timeouts[IP_VS_UDP_S_LAST+1] = {
1da177e4 LT	430	[IP_VS_UDP_S_NORMAL] = 560HZ,
	431	[IP_VS_UDP_S_LAST] = 2*HZ,
	432	};
	433
36cbd3dc	434	static const char *const udp_state_name_table[IP_VS_UDP_S_LAST+1] = {
1da177e4 LT	435	[IP_VS_UDP_S_NORMAL] = "UDP",
	436	[IP_VS_UDP_S_LAST] = "BUG!",
	437	};
	438
1da177e4 LT	439	static const char * udp_state_name(int state)
	440	{
	441	if (state >= IP_VS_UDP_S_LAST)
	442	return "ERR!";
	443	return udp_state_name_table[state] ? udp_state_name_table[state] : "?";
	444	}
	445
4a516f11	446	static void
1da177e4 LT	447	udp_state_transition(struct ip_vs_conn *cp, int direction,
1da177e4 LT	448	const struct sk_buff *skb,
9330419d	449	struct ip_vs_proto_data *pd)
1da177e4	450	{
78b16bde HS	451	if (unlikely(!pd)) {
78b16bde HS	452	pr_err("UDP no ns data\n");
4a516f11	453	return;
78b16bde HS	454	}
	455
	456	cp->timeout = pd->timeout_table[IP_VS_UDP_S_NORMAL];
1da177e4 LT	457	}
1da177e4 LT	458
582b8e3e	459	static int __udp_init(struct net net, struct ip_vs_proto_data pd)
1da177e4	460	{
78b16bde HS	461	struct netns_ipvs *ipvs = net_ipvs(net);
	462
	463	ip_vs_init_hash_table(ipvs->udp_apps, UDP_APP_TAB_SIZE);
78b16bde HS	464	pd->timeout_table = ip_vs_create_timeout_table((int *)udp_timeouts,
78b16bde HS	465	sizeof(udp_timeouts));
582b8e3e HS	466	if (!pd->timeout_table)
	467	return -ENOMEM;
	468	return 0;
1da177e4 LT	469	}
1da177e4 LT	470
78b16bde	471	static void __udp_exit(struct net net, struct ip_vs_proto_data pd)
1da177e4	472	{
78b16bde	473	kfree(pd->timeout_table);
1da177e4 LT	474	}
	475
	476
	477	struct ip_vs_protocol ip_vs_protocol_udp = {
	478	.name = "UDP",
	479	.protocol = IPPROTO_UDP,
2ad17def	480	.num_states = IP_VS_UDP_S_LAST,
1da177e4	481	.dont_defrag = 0,
78b16bde HS	482	.init = NULL,
	483	.exit = NULL,
	484	.init_netns = __udp_init,
	485	.exit_netns = __udp_exit,
1da177e4	486	.conn_schedule = udp_conn_schedule,
5c0d2374 SH	487	.conn_in_get = ip_vs_conn_in_get_proto,
5c0d2374 SH	488	.conn_out_get = ip_vs_conn_out_get_proto,
1da177e4 LT	489	.snat_handler = udp_snat_handler,
	490	.dnat_handler = udp_dnat_handler,
	491	.csum_check = udp_csum_check,
	492	.state_transition = udp_state_transition,
	493	.state_name = udp_state_name,
	494	.register_app = udp_register_app,
	495	.unregister_app = udp_unregister_app,
	496	.app_conn_bind = udp_app_conn_bind,
	497	.debug_packet = ip_vs_tcpudp_debug_packet,
	498	.timeout_change = NULL,
1da177e4	499	};