xref: /linux/net/netfilter/ipvs/ip_vs_core.c (revision e290145564886d6a3038810c621f738c1fe9fa51)

// SPDX-License-Identifier: GPL-2.0-or-later
/*
 * IPVS         An implementation of the IP virtual server support for the
 *              LINUX operating system.  IPVS is now implemented as a module
 *              over the Netfilter framework. IPVS can be used to build a
 *              high-performance and highly available server based on a
 *              cluster of servers.
 *
 * Authors:     Wensong Zhang <wensong@linuxvirtualserver.org>
 *              Peter Kese <peter.kese@ijs.si>
 *              Julian Anastasov <ja@ssi.bg>
 *
 * The IPVS code for kernel 2.2 was done by Wensong Zhang and Peter Kese,
 * with changes/fixes from Julian Anastasov, Lars Marowsky-Bree, Horms
 * and others.
 *
 * Changes:
 *	Paul `Rusty' Russell		properly handle non-linear skbs
 *	Harald Welte			don't use nfcache
 */

#define pr_fmt(fmt) "IPVS: " fmt

#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/ip.h>
#include <linux/tcp.h>
#include <linux/sctp.h>
#include <linux/icmp.h>
#include <linux/slab.h>

#include <net/ip.h>
#include <net/tcp.h>
#include <net/udp.h>
#include <net/icmp.h>                   /* for icmp_send */
#include <net/gue.h>
#include <net/gre.h>
#include <net/route.h>
#include <net/ip6_checksum.h>
#include <net/netns/generic.h>		/* net_generic() */

#include <linux/netfilter.h>
#include <linux/netfilter_ipv4.h>

#ifdef CONFIG_IP_VS_IPV6
#include <net/ipv6.h>
#include <linux/netfilter_ipv6.h>
#include <net/ip6_route.h>
#endif

#include <net/ip_vs.h>
#include <linux/indirect_call_wrapper.h>


EXPORT_SYMBOL(register_ip_vs_scheduler);
EXPORT_SYMBOL(unregister_ip_vs_scheduler);
EXPORT_SYMBOL(ip_vs_proto_name);
EXPORT_SYMBOL(ip_vs_conn_new);
EXPORT_SYMBOL(ip_vs_conn_in_get);
EXPORT_SYMBOL(ip_vs_conn_out_get);
#ifdef CONFIG_IP_VS_PROTO_TCP
EXPORT_SYMBOL(ip_vs_tcp_conn_listen);
#endif
EXPORT_SYMBOL(ip_vs_conn_put);
#ifdef CONFIG_IP_VS_DEBUG
EXPORT_SYMBOL(ip_vs_get_debug_level);
#endif
EXPORT_SYMBOL(ip_vs_new_conn_out);

#if defined(CONFIG_IP_VS_PROTO_TCP) && defined(CONFIG_IP_VS_PROTO_UDP)
#define SNAT_CALL(f, ...) \
	INDIRECT_CALL_2(f, tcp_snat_handler, udp_snat_handler, __VA_ARGS__)
#elif defined(CONFIG_IP_VS_PROTO_TCP)
#define SNAT_CALL(f, ...) INDIRECT_CALL_1(f, tcp_snat_handler, __VA_ARGS__)
#elif defined(CONFIG_IP_VS_PROTO_UDP)
#define SNAT_CALL(f, ...) INDIRECT_CALL_1(f, udp_snat_handler, __VA_ARGS__)
#else
#define SNAT_CALL(f, ...) f(__VA_ARGS__)
#endif

static unsigned int ip_vs_net_id __read_mostly;
/* netns cnt used for uniqueness */
static atomic_t ipvs_netns_cnt = ATOMIC_INIT(0);

/* ID used in ICMP lookups */
#define icmp_id(icmph)          (((icmph)->un).echo.id)
#define icmpv6_id(icmph)        (icmph->icmp6_dataun.u_echo.identifier)

const char *ip_vs_proto_name(unsigned int proto)
{
	static char buf[20];

	switch (proto) {
	case IPPROTO_IP:
		return "IP";
	case IPPROTO_UDP:
		return "UDP";
	case IPPROTO_TCP:
		return "TCP";
	case IPPROTO_SCTP:
		return "SCTP";
	case IPPROTO_ICMP:
		return "ICMP";
#ifdef CONFIG_IP_VS_IPV6
	case IPPROTO_ICMPV6:
		return "ICMPv6";
#endif
	default:
		sprintf(buf, "IP_%u", proto);
		return buf;
	}
}

void ip_vs_init_hash_table(struct list_head *table, int rows)
{
	while (--rows >= 0)
		INIT_LIST_HEAD(&table[rows]);
}

/* IPVS Resizable Hash Tables:
 * - list_bl buckets with bit lock
 *
 * Goals:
 * - RCU lookup for entry can run in parallel with add/del/move operations
 * - hash keys can be on non-contiguous memory
 * - support entries with duplicate keys
 * - unlink entries without lookup, use the saved table and bucket id
 * - resizing can trigger on load change or depending on key refresh period
 * - customizable load factor to balance between speed and memory usage
 * - add/del/move operations should be allowed for any context
 *
 * Resizing:
 * - new table is attached to the current table and all entries are moved
 * with new hash key. Finally, the new table is installed as current one and
 * the old table is released after RCU grace period.
 * - RCU read-side critical sections will walk two tables while resizing is
 * in progress
 * - new entries are added to the new table
 * - entries will be deleted from the old or from the new table, the table_id
 * can be saved into entry as part of the hash key to know where the entry is
 * hashed
 * - move operations may delay readers or to cause retry for the modified
 * bucket. As result, searched entry will be found but walkers that operate
 * on multiple entries may see same entry twice if bucket walking is retried.
 * - for fast path the number of entries (load) can be compared to u_thresh
 * and l_thresh to decide when to trigger table growing/shrinking. They
 * are calculated based on load factor (shift count), negative value allows
 * load to be below 100% to reduce collisions by maintaining larger table
 * while positive value tolerates collisions by using smaller table and load
 * above 100%: u_thresh(load) = size * (2 ^ lfactor)
 *
 * Locking:
 * - lock: protect seqc if other context except resizer can move entries
 * - seqc: seqcount_t, delay/retry readers while entries are moved to
 * new table on resizing
 * - bit lock: serialize bucket modifications
 * - writers may use other locking mechanisms to serialize operations for
 * resizing, moving and installing new tables
 */

void ip_vs_rht_free(struct ip_vs_rht *t)
{
	kvfree(t->buckets);
	kvfree(t->seqc);
	kvfree(t->lock);
	kfree(t);
}

void ip_vs_rht_rcu_free(struct rcu_head *head)
{
	struct ip_vs_rht *t;

	t = container_of(head, struct ip_vs_rht, rcu_head);
	ip_vs_rht_free(t);
}

struct ip_vs_rht *ip_vs_rht_alloc(int buckets, int scounts, int locks)
{
	struct ip_vs_rht *t = kzalloc_obj(*t);
	int i;

	if (!t)
		return NULL;
	if (scounts) {
		int ml = roundup_pow_of_two(nr_cpu_ids);

		scounts = min(scounts, buckets);
		scounts = min(scounts, ml);
		t->seqc = kvmalloc_objs(*t->seqc, scounts);
		if (!t->seqc)
			goto err;
		for (i = 0; i < scounts; i++)
			seqcount_init(&t->seqc[i]);

		if (locks) {
			locks = min(locks, scounts);
			t->lock = kvmalloc_objs(*t->lock, locks);
			if (!t->lock)
				goto err;
			for (i = 0; i < locks; i++)
				spin_lock_init(&t->lock[i].l);
		}
	}

	t->buckets = kvmalloc_objs(*t->buckets, buckets);
	if (!t->buckets)
		goto err;
	for (i = 0; i < buckets; i++)
		INIT_HLIST_BL_HEAD(&t->buckets[i]);
	t->mask = buckets - 1;
	t->size = buckets;
	t->seqc_mask = scounts - 1;
	t->lock_mask = locks - 1;
	t->u_thresh = buckets;
	t->l_thresh = buckets >> 4;
	t->bits = order_base_2(buckets);
	/* new_tbl points to self if no new table is filled */
	RCU_INIT_POINTER(t->new_tbl, t);
	get_random_bytes(&t->hash_key, sizeof(t->hash_key));
	return t;

err:
	ip_vs_rht_free(t);
	return NULL;
}

/* Get the desired table size for n entries based on current table size and
 * by using the formula size = n / (2^lfactor)
 * lfactor: shift value for the load factor:
 * - >0: u_thresh=size << lfactor, for load factor above 100%
 * - <0: u_thresh=size >> -lfactor, for load factor below 100%
 * - 0: for load factor of 100%
 */
int ip_vs_rht_desired_size(struct netns_ipvs *ipvs, struct ip_vs_rht *t, int n,
			   int lfactor, int min_bits, int max_bits)
{
	if (!t)
		return 1 << min_bits;
	n = n > 0 ? roundup_pow_of_two(n) : 1;
	if (lfactor < 0) {
		int factor = min(-lfactor, max_bits);

		n = min(n, 1 << (max_bits - factor));
		n <<= factor;
	} else {
		n = min(n >> lfactor, 1 << max_bits);
	}
	if (lfactor != t->lfactor)
		return clamp(n, 1 << min_bits, 1 << max_bits);
	if (n > t->size)
		return n;
	if (n > t->size >> 4)
		return t->size;
	/* Shrink but keep it n * 2 to prevent frequent resizing */
	return clamp(n << 1, 1 << min_bits, 1 << max_bits);
}

/* Set thresholds based on table size and load factor:
 * u_thresh = size * (2^lfactor)
 * l_thresh = u_thresh / 16
 * u_thresh/l_thresh can be used to check if load triggers a table grow/shrink
 */
void ip_vs_rht_set_thresholds(struct ip_vs_rht *t, int size, int lfactor,
			      int min_bits, int max_bits)
{
	if (size >= 1 << max_bits)
		t->u_thresh = INT_MAX;	/* stop growing */
	else if (lfactor <= 0)
		t->u_thresh = size >> min(-lfactor, max_bits);
	else
		t->u_thresh = min(size, 1 << (30 - lfactor)) << lfactor;

	/* l_thresh: shrink when load is 16 times lower, can be 0 */
	if (size >= 1 << max_bits)
		t->l_thresh = (1 << max_bits) >> 4;
	else if (size > 1 << min_bits)
		t->l_thresh = t->u_thresh >> 4;
	else
		t->l_thresh = 0;	/* stop shrinking */
}

/* Return hash value for local info (fast, insecure) */
u32 ip_vs_rht_hash_linfo(struct ip_vs_rht *t, int af,
			 const union nf_inet_addr *addr, u32 v1, u32 v2)
{
	u32 v3;

#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6)
		v3 = ipv6_addr_hash(&addr->in6);
	else
#endif
		v3 = addr->all[0];

	return jhash_3words(v1, v2, v3, (u32)t->hash_key.key[0]);
}

static inline void
ip_vs_in_stats(struct ip_vs_conn *cp, struct sk_buff *skb)
{
	struct ip_vs_dest *dest = cp->dest;
	struct netns_ipvs *ipvs = cp->ipvs;

	if (dest && (dest->cflags & IP_VS_DEST_CF_AVAILABLE)) {
		struct ip_vs_cpu_stats *s;
		struct ip_vs_service *svc;

		local_bh_disable();

		s = this_cpu_ptr(dest->stats.cpustats);
		u64_stats_update_begin(&s->syncp);
		u64_stats_inc(&s->cnt.inpkts);
		u64_stats_add(&s->cnt.inbytes, skb->len);
		u64_stats_update_end(&s->syncp);

		svc = rcu_dereference(dest->svc);
		s = this_cpu_ptr(svc->stats.cpustats);
		u64_stats_update_begin(&s->syncp);
		u64_stats_inc(&s->cnt.inpkts);
		u64_stats_add(&s->cnt.inbytes, skb->len);
		u64_stats_update_end(&s->syncp);

		s = this_cpu_ptr(ipvs->tot_stats->s.cpustats);
		u64_stats_update_begin(&s->syncp);
		u64_stats_inc(&s->cnt.inpkts);
		u64_stats_add(&s->cnt.inbytes, skb->len);
		u64_stats_update_end(&s->syncp);

		local_bh_enable();
	}
}


static inline void
ip_vs_out_stats(struct ip_vs_conn *cp, struct sk_buff *skb)
{
	struct ip_vs_dest *dest = cp->dest;
	struct netns_ipvs *ipvs = cp->ipvs;

	if (dest && (dest->cflags & IP_VS_DEST_CF_AVAILABLE)) {
		struct ip_vs_cpu_stats *s;
		struct ip_vs_service *svc;

		local_bh_disable();

		s = this_cpu_ptr(dest->stats.cpustats);
		u64_stats_update_begin(&s->syncp);
		u64_stats_inc(&s->cnt.outpkts);
		u64_stats_add(&s->cnt.outbytes, skb->len);
		u64_stats_update_end(&s->syncp);

		svc = rcu_dereference(dest->svc);
		s = this_cpu_ptr(svc->stats.cpustats);
		u64_stats_update_begin(&s->syncp);
		u64_stats_inc(&s->cnt.outpkts);
		u64_stats_add(&s->cnt.outbytes, skb->len);
		u64_stats_update_end(&s->syncp);

		s = this_cpu_ptr(ipvs->tot_stats->s.cpustats);
		u64_stats_update_begin(&s->syncp);
		u64_stats_inc(&s->cnt.outpkts);
		u64_stats_add(&s->cnt.outbytes, skb->len);
		u64_stats_update_end(&s->syncp);

		local_bh_enable();
	}
}


static inline void
ip_vs_conn_stats(struct ip_vs_conn *cp, struct ip_vs_service *svc)
{
	struct netns_ipvs *ipvs = svc->ipvs;
	struct ip_vs_cpu_stats *s;

	local_bh_disable();

	s = this_cpu_ptr(cp->dest->stats.cpustats);
	u64_stats_update_begin(&s->syncp);
	u64_stats_inc(&s->cnt.conns);
	u64_stats_update_end(&s->syncp);

	s = this_cpu_ptr(svc->stats.cpustats);
	u64_stats_update_begin(&s->syncp);
	u64_stats_inc(&s->cnt.conns);
	u64_stats_update_end(&s->syncp);

	s = this_cpu_ptr(ipvs->tot_stats->s.cpustats);
	u64_stats_update_begin(&s->syncp);
	u64_stats_inc(&s->cnt.conns);
	u64_stats_update_end(&s->syncp);

	local_bh_enable();
}


static inline void
ip_vs_set_state(struct ip_vs_conn *cp, int direction,
		const struct sk_buff *skb,
		struct ip_vs_proto_data *pd, unsigned int iph_len)
{
	if (likely(pd->pp->state_transition))
		pd->pp->state_transition(cp, direction, skb, pd, iph_len);
}

static inline int
ip_vs_conn_fill_param_persist(const struct ip_vs_service *svc,
			      struct sk_buff *skb, int protocol,
			      const union nf_inet_addr *caddr, __be16 cport,
			      const union nf_inet_addr *vaddr, __be16 vport,
			      struct ip_vs_conn_param *p)
{
	ip_vs_conn_fill_param(svc->ipvs, svc->af, protocol, caddr, cport, vaddr,
			      vport, p);
	p->pe = rcu_dereference(svc->pe);
	if (p->pe && p->pe->fill_param)
		return p->pe->fill_param(p, skb);

	return 0;
}

/*
 *  IPVS persistent scheduling function
 *  It creates a connection entry according to its template if exists,
 *  or selects a server and creates a connection entry plus a template.
 *  Locking: we are svc user (svc->refcnt), so we hold all dests too
 *  Protocols supported: TCP, UDP
 */
static struct ip_vs_conn *
ip_vs_sched_persist(struct ip_vs_service *svc,
		    struct sk_buff *skb, __be16 src_port, __be16 dst_port,
		    int *ignored, struct ip_vs_iphdr *iph)
{
	struct ip_vs_conn *cp = NULL;
	struct ip_vs_dest *dest;
	struct ip_vs_conn *ct;
	__be16 dport = 0;		/* destination port to forward */
	unsigned int flags;
	struct ip_vs_conn_param param;
	const union nf_inet_addr fwmark = { .ip = htonl(svc->fwmark) };
	union nf_inet_addr snet;	/* source network of the client,
					   after masking */
	const union nf_inet_addr *src_addr, *dst_addr;

	if (likely(!ip_vs_iph_inverse(iph))) {
		src_addr = &iph->saddr;
		dst_addr = &iph->daddr;
	} else {
		src_addr = &iph->daddr;
		dst_addr = &iph->saddr;
	}


	/* Mask saddr with the netmask to adjust template granularity */
#ifdef CONFIG_IP_VS_IPV6
	if (svc->af == AF_INET6)
		ipv6_addr_prefix(&snet.in6, &src_addr->in6,
				 (__force __u32) svc->netmask);
	else
#endif
		snet.ip = src_addr->ip & svc->netmask;

	IP_VS_DBG_BUF(6, "p-schedule: src %s:%u dest %s:%u "
		      "mnet %s\n",
		      IP_VS_DBG_ADDR(svc->af, src_addr), ntohs(src_port),
		      IP_VS_DBG_ADDR(svc->af, dst_addr), ntohs(dst_port),
		      IP_VS_DBG_ADDR(svc->af, &snet));

	/*
	 * As far as we know, FTP is a very complicated network protocol, and
	 * it uses control connection and data connections. For active FTP,
	 * FTP server initialize data connection to the client, its source port
	 * is often 20. For passive FTP, FTP server tells the clients the port
	 * that it passively listens to,  and the client issues the data
	 * connection. In the tunneling or direct routing mode, the load
	 * balancer is on the client-to-server half of connection, the port
	 * number is unknown to the load balancer. So, a conn template like
	 * <caddr, 0, vaddr, 0, daddr, 0> is created for persistent FTP
	 * service, and a template like <caddr, 0, vaddr, vport, daddr, dport>
	 * is created for other persistent services.
	 */
	{
		int protocol = iph->protocol;
		const union nf_inet_addr *vaddr = dst_addr;
		__be16 vport = 0;

		if (dst_port == svc->port) {
			/* non-FTP template:
			 * <protocol, caddr, 0, vaddr, vport, daddr, dport>
			 * FTP template:
			 * <protocol, caddr, 0, vaddr, 0, daddr, 0>
			 */
			if (svc->port != FTPPORT)
				vport = dst_port;
		} else {
			/* Note: persistent fwmark-based services and
			 * persistent port zero service are handled here.
			 * fwmark template:
			 * <IPPROTO_IP,caddr,0,fwmark,0,daddr,0>
			 * port zero template:
			 * <protocol,caddr,0,vaddr,0,daddr,0>
			 */
			if (svc->fwmark) {
				protocol = IPPROTO_IP;
				vaddr = &fwmark;
			}
		}
		/* return *ignored = -1 so NF_DROP can be used */
		if (ip_vs_conn_fill_param_persist(svc, skb, protocol, &snet, 0,
						  vaddr, vport, &param) < 0) {
			*ignored = -1;
			return NULL;
		}
	}

	/* Check if a template already exists */
	ct = ip_vs_ct_in_get(&param);
	if (!ct || !ip_vs_check_template(ct, NULL)) {
		struct ip_vs_scheduler *sched;

		/*
		 * No template found or the dest of the connection
		 * template is not available.
		 * return *ignored=0 i.e. ICMP and NF_DROP
		 */
		sched = rcu_dereference(svc->scheduler);
		if (sched) {
			/* read svc->sched_data after svc->scheduler */
			smp_rmb();
			dest = sched->schedule(svc, skb, iph);
		} else {
			dest = NULL;
		}
		if (!dest) {
			IP_VS_DBG(1, "p-schedule: no dest found.\n");
			kfree(param.pe_data);
			*ignored = 0;
			return NULL;
		}

		if (dst_port == svc->port && svc->port != FTPPORT)
			dport = dest->port;

		/* Create a template
		 * This adds param.pe_data to the template,
		 * and thus param.pe_data will be destroyed
		 * when the template expires */
		ct = ip_vs_conn_new(&param, dest->af, &dest->addr, dport,
				    IP_VS_CONN_F_TEMPLATE, dest, skb->mark);
		if (ct == NULL) {
			kfree(param.pe_data);
			*ignored = -1;
			return NULL;
		}

		ct->timeout = svc->timeout;
	} else {
		/* set destination with the found template */
		dest = ct->dest;
		kfree(param.pe_data);
	}

	dport = dst_port;
	if (dport == svc->port && dest->port)
		dport = dest->port;

	flags = (svc->flags & IP_VS_SVC_F_ONEPACKET
		 && iph->protocol == IPPROTO_UDP) ?
		IP_VS_CONN_F_ONE_PACKET : 0;

	/*
	 *    Create a new connection according to the template
	 */
	ip_vs_conn_fill_param(svc->ipvs, svc->af, iph->protocol, src_addr,
			      src_port, dst_addr, dst_port, &param);

	cp = ip_vs_conn_new(&param, dest->af, &dest->addr, dport, flags, dest,
			    skb->mark);
	if (cp == NULL) {
		ip_vs_conn_put(ct);
		*ignored = -1;
		return NULL;
	}

	/*
	 *    Add its control
	 */
	ip_vs_control_add(cp, ct);
	ip_vs_conn_put(ct);

	ip_vs_conn_stats(cp, svc);
	return cp;
}


/*
 *  IPVS main scheduling function
 *  It selects a server according to the virtual service, and
 *  creates a connection entry.
 *  Protocols supported: TCP, UDP
 *
 *  Usage of *ignored
 *
 * 1 :   protocol tried to schedule (eg. on SYN), found svc but the
 *       svc/scheduler decides that this packet should be accepted with
 *       NF_ACCEPT because it must not be scheduled.
 *
 * 0 :   scheduler can not find destination, so try bypass or
 *       return ICMP and then NF_DROP (ip_vs_leave).
 *
 * -1 :  scheduler tried to schedule but fatal error occurred, eg.
 *       ip_vs_conn_new failure (ENOMEM) or ip_vs_sip_fill_param
 *       failure such as missing Call-ID, ENOMEM on skb_linearize
 *       or pe_data. In this case we should return NF_DROP without
 *       any attempts to send ICMP with ip_vs_leave.
 */
struct ip_vs_conn *
ip_vs_schedule(struct ip_vs_service *svc, struct sk_buff *skb,
	       struct ip_vs_proto_data *pd, int *ignored,
	       struct ip_vs_iphdr *iph)
{
	struct ip_vs_protocol *pp = pd->pp;
	struct ip_vs_conn *cp = NULL;
	struct ip_vs_scheduler *sched;
	struct ip_vs_dest *dest;
	__be16 _ports[2], *pptr, cport, vport;
	const void *caddr, *vaddr;
	unsigned int flags;

	*ignored = 1;
	/*
	 * IPv6 frags, only the first hit here.
	 */
	pptr = frag_safe_skb_hp(skb, iph->len, sizeof(_ports), _ports);
	if (pptr == NULL)
		return NULL;

	if (likely(!ip_vs_iph_inverse(iph))) {
		cport = pptr[0];
		caddr = &iph->saddr;
		vport = pptr[1];
		vaddr = &iph->daddr;
	} else {
		cport = pptr[1];
		caddr = &iph->daddr;
		vport = pptr[0];
		vaddr = &iph->saddr;
	}

	/*
	 * FTPDATA needs this check when using local real server.
	 * Never schedule Active FTPDATA connections from real server.
	 * For LVS-NAT they must be already created. For other methods
	 * with persistence the connection is created on SYN+ACK.
	 */
	if (cport == FTPDATA) {
		IP_VS_DBG_PKT(12, svc->af, pp, skb, iph->off,
			      "Not scheduling FTPDATA");
		return NULL;
	}

	/*
	 *    Do not schedule replies from local real server.
	 */
	if ((!skb->dev || skb->dev->flags & IFF_LOOPBACK)) {
		iph->hdr_flags ^= IP_VS_HDR_INVERSE;
		cp = INDIRECT_CALL_1(pp->conn_in_get,
				     ip_vs_conn_in_get_proto, svc->ipvs,
				     svc->af, skb, iph);
		iph->hdr_flags ^= IP_VS_HDR_INVERSE;

		if (cp) {
			IP_VS_DBG_PKT(12, svc->af, pp, skb, iph->off,
				      "Not scheduling reply for existing"
				      " connection");
			__ip_vs_conn_put(cp);
			return NULL;
		}
	}

	/*
	 *    Persistent service
	 */
	if (svc->flags & IP_VS_SVC_F_PERSISTENT)
		return ip_vs_sched_persist(svc, skb, cport, vport, ignored,
					   iph);

	*ignored = 0;

	/*
	 *    Non-persistent service
	 */
	if (!svc->fwmark && vport != svc->port) {
		if (!svc->port)
			pr_err("Schedule: port zero only supported "
			       "in persistent services, "
			       "check your ipvs configuration\n");
		return NULL;
	}

	sched = rcu_dereference(svc->scheduler);
	if (sched) {
		/* read svc->sched_data after svc->scheduler */
		smp_rmb();
		dest = sched->schedule(svc, skb, iph);
	} else {
		dest = NULL;
	}
	if (dest == NULL) {
		IP_VS_DBG(1, "Schedule: no dest found.\n");
		return NULL;
	}

	flags = (svc->flags & IP_VS_SVC_F_ONEPACKET
		 && iph->protocol == IPPROTO_UDP) ?
		IP_VS_CONN_F_ONE_PACKET : 0;

	/*
	 *    Create a connection entry.
	 */
	{
		struct ip_vs_conn_param p;

		ip_vs_conn_fill_param(svc->ipvs, svc->af, iph->protocol,
				      caddr, cport, vaddr, vport, &p);
		cp = ip_vs_conn_new(&p, dest->af, &dest->addr,
				    dest->port ? dest->port : vport,
				    flags, dest, skb->mark);
		if (!cp) {
			*ignored = -1;
			return NULL;
		}
	}

	IP_VS_DBG_BUF(6, "Schedule fwd:%c c:%s:%u v:%s:%u "
		      "d:%s:%u conn->flags:%X conn->refcnt:%d\n",
		      ip_vs_fwd_tag(cp),
		      IP_VS_DBG_ADDR(cp->af, &cp->caddr), ntohs(cp->cport),
		      IP_VS_DBG_ADDR(cp->af, &cp->vaddr), ntohs(cp->vport),
		      IP_VS_DBG_ADDR(cp->daf, &cp->daddr), ntohs(cp->dport),
		      cp->flags, refcount_read(&cp->refcnt));

	ip_vs_conn_stats(cp, svc);
	return cp;
}

static inline int ip_vs_addr_is_unicast(struct net *net, int af,
					union nf_inet_addr *addr)
{
#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6)
		return ipv6_addr_type(&addr->in6) & IPV6_ADDR_UNICAST;
#endif
	return (inet_addr_type(net, addr->ip) == RTN_UNICAST);
}

/*
 *  Pass or drop the packet.
 *  Called by ip_vs_in, when the virtual service is available but
 *  no destination is available for a new connection.
 */
int ip_vs_leave(struct ip_vs_service *svc, struct sk_buff *skb,
		struct ip_vs_proto_data *pd, struct ip_vs_iphdr *iph)
{
	__be16 _ports[2], *pptr, dport;
	struct netns_ipvs *ipvs = svc->ipvs;
	struct net *net = ipvs->net;

	pptr = frag_safe_skb_hp(skb, iph->len, sizeof(_ports), _ports);
	if (!pptr)
		return NF_DROP;
	dport = likely(!ip_vs_iph_inverse(iph)) ? pptr[1] : pptr[0];

	/* if it is fwmark-based service, the cache_bypass sysctl is up
	   and the destination is a non-local unicast, then create
	   a cache_bypass connection entry */
	if (sysctl_cache_bypass(ipvs) && svc->fwmark &&
	    !(iph->hdr_flags & (IP_VS_HDR_INVERSE | IP_VS_HDR_ICMP)) &&
	    ip_vs_addr_is_unicast(net, svc->af, &iph->daddr)) {
		int ret;
		struct ip_vs_conn *cp;
		unsigned int flags = (svc->flags & IP_VS_SVC_F_ONEPACKET &&
				      iph->protocol == IPPROTO_UDP) ?
				      IP_VS_CONN_F_ONE_PACKET : 0;
		union nf_inet_addr daddr = { .all = { 0, 0, 0, 0 } };

		/* create a new connection entry */
		IP_VS_DBG(6, "%s(): create a cache_bypass entry\n", __func__);
		{
			struct ip_vs_conn_param p;
			ip_vs_conn_fill_param(svc->ipvs, svc->af, iph->protocol,
					      &iph->saddr, pptr[0],
					      &iph->daddr, pptr[1], &p);
			cp = ip_vs_conn_new(&p, svc->af, &daddr, 0,
					    IP_VS_CONN_F_BYPASS | flags,
					    NULL, skb->mark);
			if (!cp)
				return NF_DROP;
		}

		/* statistics */
		ip_vs_in_stats(cp, skb);

		/* set state */
		ip_vs_set_state(cp, IP_VS_DIR_INPUT, skb, pd, iph->len);

		/* transmit the first SYN packet */
		ret = cp->packet_xmit(skb, cp, pd->pp, iph);
		/* do not touch skb anymore */

		if ((cp->flags & IP_VS_CONN_F_ONE_PACKET) && cp->control)
			atomic_inc(&cp->control->in_pkts);
		else
			atomic_inc(&cp->in_pkts);
		ip_vs_conn_put(cp);
		return ret;
	}

	/*
	 * When the virtual ftp service is presented, packets destined
	 * for other services on the VIP may get here (except services
	 * listed in the ipvs table), pass the packets, because it is
	 * not ipvs job to decide to drop the packets.
	 */
	if (svc->port == FTPPORT && dport != FTPPORT)
		return NF_ACCEPT;

	if (unlikely(ip_vs_iph_icmp(iph)))
		return NF_DROP;

	/*
	 * Notify the client that the destination is unreachable, and
	 * release the socket buffer.
	 * Since it is in IP layer, the TCP socket is not actually
	 * created, the TCP RST packet cannot be sent, instead that
	 * ICMP_PORT_UNREACH is sent here no matter it is TCP/UDP. --WZ
	 */
#ifdef CONFIG_IP_VS_IPV6
	if (svc->af == AF_INET6) {
		if (!skb->dev)
			skb->dev = net->loopback_dev;
		icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_PORT_UNREACH, 0);
	} else
#endif
		icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PORT_UNREACH, 0);

	return NF_DROP;
}

#ifdef CONFIG_SYSCTL

static int sysctl_snat_reroute(struct netns_ipvs *ipvs)
{
	return ipvs->sysctl_snat_reroute;
}

static int sysctl_nat_icmp_send(struct netns_ipvs *ipvs)
{
	return ipvs->sysctl_nat_icmp_send;
}

#else

static int sysctl_snat_reroute(struct netns_ipvs *ipvs) { return 0; }
static int sysctl_nat_icmp_send(struct netns_ipvs *ipvs) { return 0; }

#endif

static __sum16 ip_vs_checksum_complete(struct sk_buff *skb, int offset)
{
	return csum_fold(skb_checksum(skb, offset, skb->len - offset, 0));
}

static inline enum ip_defrag_users ip_vs_defrag_user(unsigned int hooknum)
{
	if (NF_INET_LOCAL_IN == hooknum)
		return IP_DEFRAG_VS_IN;
	if (NF_INET_FORWARD == hooknum)
		return IP_DEFRAG_VS_FWD;
	return IP_DEFRAG_VS_OUT;
}

static inline int ip_vs_gather_frags(struct netns_ipvs *ipvs,
				     struct sk_buff *skb, u_int32_t user)
{
	int err;

	local_bh_disable();
	err = ip_defrag(ipvs->net, skb, user);
	local_bh_enable();
	if (!err)
		ip_send_check(ip_hdr(skb));

	return err;
}

static int ip_vs_route_me_harder(struct netns_ipvs *ipvs, int af,
				 struct sk_buff *skb, unsigned int hooknum)
{
	if (!sysctl_snat_reroute(ipvs))
		return 0;
	/* Reroute replies only to remote clients (FORWARD and LOCAL_OUT) */
	if (NF_INET_LOCAL_IN == hooknum)
		return 0;
#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6) {
		struct dst_entry *dst = skb_dst(skb);

		if (dst->dev && !(dst->dev->flags & IFF_LOOPBACK) &&
		    ip6_route_me_harder(ipvs->net, skb->sk, skb) != 0)
			return 1;
	} else
#endif
		if (!(skb_rtable(skb)->rt_flags & RTCF_LOCAL) &&
		    ip_route_me_harder(ipvs->net, skb->sk, skb, RTN_LOCAL) != 0)
			return 1;

	return 0;
}

/*
 * Packet has been made sufficiently writable in caller
 * - inout: 1=in->out, 0=out->in
 */
bool ip_vs_nat_icmp(struct sk_buff *skb, struct ip_vs_protocol *pp,
		    struct ip_vs_conn *cp, int inout, unsigned int toff,
		    bool has_ports, struct ip_vs_iphdr *ciph)
{
	struct iphdr *iph	 = ip_hdr(skb);
	struct icmphdr *icmph	 = (struct icmphdr *)(skb->data + toff);
	struct iphdr *cih	 = (struct iphdr *)(icmph + 1);

	/* Before now we may used ihl from skb frag, revalidate it after
	 * copying it into skb head to prevent out-of-bounds access
	 */
	if (cih->ihl * 4 != ciph->len - ciph->off)
		return false;
	if (inout) {
		iph->saddr = cp->vaddr.ip;
		ip_send_check(iph);
		cih->daddr = cp->vaddr.ip;
		ip_send_check(cih);
	} else {
		iph->daddr = cp->daddr.ip;
		ip_send_check(iph);
		cih->saddr = cp->daddr.ip;
		ip_send_check(cih);
	}

	/* the TCP/UDP/SCTP port */
	if (has_ports) {
		__be16 *ports = (void *)(skb->data + ciph->len);

		if (inout)
			ports[1] = cp->vport;
		else
			ports[0] = cp->dport;
	}

	/* And finally the ICMP checksum */
	icmph->checksum = 0;
	icmph->checksum = ip_vs_checksum_complete(skb, toff);
	skb->ip_summed = CHECKSUM_UNNECESSARY;

	if (inout)
		IP_VS_DBG_PKT(11, AF_INET, pp, skb, ciph->off,
			      "Forwarding altered outgoing ICMP");
	else
		IP_VS_DBG_PKT(11, AF_INET, pp, skb, ciph->off,
			      "Forwarding altered incoming ICMP");
	return true;
}

#ifdef CONFIG_IP_VS_IPV6
void ip_vs_nat_icmp_v6(struct sk_buff *skb, struct ip_vs_protocol *pp,
		       struct ip_vs_conn *cp, int inout, unsigned int toff,
		       bool has_ports, struct ip_vs_iphdr *ciph)
{
	struct ipv6hdr *iph	 = ipv6_hdr(skb);
	struct icmp6hdr *icmph;
	struct ipv6hdr *cih;

	icmph = (struct icmp6hdr *)(skb->data + toff);
	cih = (struct ipv6hdr *)(skb->data + ciph->off);

	if (inout) {
		iph->saddr = cp->vaddr.in6;
		cih->daddr = cp->vaddr.in6;
	} else {
		iph->daddr = cp->daddr.in6;
		cih->saddr = cp->daddr.in6;
	}

	/* the TCP/UDP/SCTP port */
	if (has_ports) {
		__be16 *ports = (void *)(skb->data + ciph->len);

		IP_VS_DBG(11, "%s() changed port %d to %d\n", __func__,
			      ntohs(inout ? ports[1] : ports[0]),
			      ntohs(inout ? cp->vport : cp->dport));
		if (inout)
			ports[1] = cp->vport;
		else
			ports[0] = cp->dport;
	}

	/* And finally the ICMP checksum */
	icmph->icmp6_cksum = ~csum_ipv6_magic(&iph->saddr, &iph->daddr,
					      skb->len - toff,
					      IPPROTO_ICMPV6, 0);
	skb->csum_start = skb_headroom(skb) + toff;
	skb->csum_offset = offsetof(struct icmp6hdr, icmp6_cksum);
	skb->ip_summed = CHECKSUM_PARTIAL;

	if (inout)
		IP_VS_DBG_PKT(11, AF_INET6, pp, skb, ciph->off,
			      "Forwarding altered outgoing ICMPv6");
	else
		IP_VS_DBG_PKT(11, AF_INET6, pp, skb, ciph->off,
			      "Forwarding altered incoming ICMPv6");
}
#endif

/* Handle relevant response ICMP messages - forward to the right
 * destination host.
 */
static int handle_response_icmp(int af, struct sk_buff *skb,
				union nf_inet_addr *snet,
				struct ip_vs_conn *cp,
				struct ip_vs_protocol *pp,
				struct ip_vs_iphdr *ciph,
				unsigned int toff, unsigned int hooknum)
{
	int iproto = af == AF_INET6 ? IPPROTO_ICMPV6 : IPPROTO_ICMP;
	unsigned int verdict = NF_DROP;
	unsigned int ctoff = ciph->len;
	bool has_ports = false;

	if (IP_VS_FWD_METHOD(cp) != IP_VS_CONN_F_MASQ)
		goto after_nat;

	/* Ensure the checksum is correct */
	if (!ip_vs_checksum_common_check(skb, toff, iproto, af)) {
		/* Failed checksum! */
		IP_VS_DBG_BUF(1, "Forward ICMP: failed checksum from %s!\n",
			      IP_VS_DBG_ADDR(af, snet));
		goto out;
	}

	if (ciph->protocol == IPPROTO_TCP || ciph->protocol == IPPROTO_UDP ||
	    ciph->protocol == IPPROTO_SCTP) {
		ctoff += 2 * sizeof(__u16);
		has_ports = true;
	}
	if (skb_ensure_writable(skb, ctoff))
		goto out;

#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6)
		ip_vs_nat_icmp_v6(skb, pp, cp, 1, toff, has_ports, ciph);
	else
#endif
		if (!ip_vs_nat_icmp(skb, pp, cp, 1, toff, has_ports, ciph))
			goto out;

	if (ip_vs_route_me_harder(cp->ipvs, af, skb, hooknum))
		goto out;

after_nat:
	/* do the statistics and put it back */
	ip_vs_out_stats(cp, skb);

	skb->ipvs_property = 1;
	if (!(cp->flags & IP_VS_CONN_F_NFCT))
		ip_vs_notrack(skb);
	else
		ip_vs_update_conntrack(skb, cp, 0);
	verdict = NF_ACCEPT;

out:
	__ip_vs_conn_put(cp);

	return verdict;
}

/*
 *	Handle ICMP messages in the inside-to-outside direction (outgoing).
 *	Find any that might be relevant, check against existing connections.
 *	Currently handles error types - unreachable, quench, ttl exceeded.
 */
static int ip_vs_out_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb,
			  int *related, unsigned int hooknum,
			  struct ip_vs_iphdr *ipvsh)
{
	struct icmphdr	_icmph, *ic;
	struct iphdr	_ciph, *cih;	/* The ip header contained within the ICMP */
	struct ip_vs_iphdr ciph;
	struct ip_vs_conn *cp;
	struct ip_vs_protocol *pp;
	unsigned int offset;
	union nf_inet_addr snet;

	*related = 1;

	/* reassemble IP fragments */
	if (ip_is_fragment(ip_hdr(skb))) {
		if (ip_vs_gather_frags(ipvs, skb, ip_vs_defrag_user(hooknum)))
			return NF_STOLEN;
		if (!ip_vs_fill_iph_skb(AF_INET, skb, false, ipvsh))
			return NF_ACCEPT;
	}

	offset = ipvsh->len;
	ic = skb_header_pointer(skb, offset, sizeof(_icmph), &_icmph);
	if (ic == NULL)
		return NF_DROP;

	IP_VS_DBG(12, "Outgoing ICMP (%d,%d) %pI4->%pI4\n",
		  ic->type, ntohs(icmp_id(ic)),
		  &ipvsh->saddr.ip, &ipvsh->daddr.ip);

	/*
	 * Work through seeing if this is for us.
	 * These checks are supposed to be in an order that means easy
	 * things are checked first to speed up processing.... however
	 * this means that some packets will manage to get a long way
	 * down this stack and then be rejected, but that's life.
	 */
	if ((ic->type != ICMP_DEST_UNREACH) &&
	    (ic->type != ICMP_SOURCE_QUENCH) &&
	    (ic->type != ICMP_TIME_EXCEEDED)) {
		*related = 0;
		return NF_ACCEPT;
	}

	/* Now find the contained IP header */
	offset += sizeof(_icmph);
	if (!ip_vs_fill_iph_skb_icmp(AF_INET, skb, offset, true, &ciph))
		return NF_ACCEPT; /* The packet looks wrong, ignore */

	cih = skb_header_pointer(skb, offset, sizeof(_ciph), &_ciph);
	if (!(cih && cih->version == 4 &&
	      ciph.len - ciph.off >= sizeof(struct iphdr)))
		return NF_ACCEPT; /* The packet looks wrong, ignore */

	pp = ip_vs_proto_get(ciph.protocol);
	if (!pp)
		return NF_ACCEPT;

	/* Is the embedded protocol header present? */
	if (unlikely(cih->frag_off & htons(IP_OFFSET) && !pp->dont_defrag))
		return NF_ACCEPT;

	IP_VS_DBG_PKT(11, AF_INET, pp, skb, offset,
		      "Checking outgoing ICMP for");

	/* The embedded headers contain source and dest in reverse order */
	cp = INDIRECT_CALL_1(pp->conn_out_get, ip_vs_conn_out_get_proto,
			     ipvs, AF_INET, skb, &ciph);
	if (!cp)
		return NF_ACCEPT;

	snet.ip = ipvsh->saddr.ip;
	return handle_response_icmp(AF_INET, skb, &snet, cp, pp, &ciph,
				    ipvsh->len, hooknum);
}

#ifdef CONFIG_IP_VS_IPV6
static int ip_vs_out_icmp_v6(struct netns_ipvs *ipvs, struct sk_buff *skb,
			     int *related,  unsigned int hooknum,
			     struct ip_vs_iphdr *ipvsh)
{
	struct icmp6hdr	_icmph, *ic;
	struct ip_vs_iphdr ciph = {.flags = 0, .fragoffs = 0};/*Contained IP */
	struct ip_vs_conn *cp;
	struct ip_vs_protocol *pp;
	union nf_inet_addr snet;

	*related = 1;
	ic = frag_safe_skb_hp(skb, ipvsh->len, sizeof(_icmph), &_icmph);
	if (ic == NULL)
		return NF_DROP;

	/*
	 * Work through seeing if this is for us.
	 * These checks are supposed to be in an order that means easy
	 * things are checked first to speed up processing.... however
	 * this means that some packets will manage to get a long way
	 * down this stack and then be rejected, but that's life.
	 */
	if (ic->icmp6_type & ICMPV6_INFOMSG_MASK) {
		*related = 0;
		return NF_ACCEPT;
	}
	/* Fragment header that is before ICMP header tells us that:
	 * it's not an error message since they can't be fragmented.
	 */
	if (ipvsh->flags & IP6_FH_F_FRAG)
		return NF_DROP;

	IP_VS_DBG(8, "Outgoing ICMPv6 (%d,%d) %pI6c->%pI6c\n",
		  ic->icmp6_type, ntohs(icmpv6_id(ic)),
		  &ipvsh->saddr, &ipvsh->daddr);

	if (!ip_vs_fill_iph_skb_icmp(AF_INET6, skb, ipvsh->len + sizeof(_icmph),
				     true, &ciph))
		return NF_ACCEPT; /* The packet looks wrong, ignore */

	pp = ip_vs_proto_get(ciph.protocol);
	if (!pp)
		return NF_ACCEPT;

	/* Is the embedded protocol header present? */
	if (unlikely(ciph.fragoffs && !pp->dont_defrag))
		return NF_ACCEPT;

	/* The embedded headers contain source and dest in reverse order */
	cp = INDIRECT_CALL_1(pp->conn_out_get, ip_vs_conn_out_get_proto,
			     ipvs, AF_INET6, skb, &ciph);
	if (!cp)
		return NF_ACCEPT;

	snet.in6 = ciph.saddr.in6;
	return handle_response_icmp(AF_INET6, skb, &snet, cp, pp, &ciph,
				    ipvsh->len, hooknum);
}
#endif

/*
 * Check if sctp chunc is ABORT chunk
 */
static inline int is_sctp_abort(const struct sk_buff *skb, int nh_len)
{
	struct sctp_chunkhdr *sch, schunk;
	sch = skb_header_pointer(skb, nh_len + sizeof(struct sctphdr),
				 sizeof(schunk), &schunk);
	if (sch == NULL)
		return 0;
	if (sch->type == SCTP_CID_ABORT)
		return 1;
	return 0;
}

static inline int is_tcp_reset(const struct sk_buff *skb, int nh_len)
{
	struct tcphdr _tcph, *th;

	th = skb_header_pointer(skb, nh_len, sizeof(_tcph), &_tcph);
	if (th == NULL)
		return 0;
	return th->rst;
}

static inline bool is_new_conn(const struct sk_buff *skb,
			       struct ip_vs_iphdr *iph)
{
	switch (iph->protocol) {
	case IPPROTO_TCP: {
		struct tcphdr _tcph, *th;

		th = skb_header_pointer(skb, iph->len, sizeof(_tcph), &_tcph);
		if (th == NULL)
			return false;
		return th->syn;
	}
	case IPPROTO_SCTP: {
		struct sctp_chunkhdr *sch, schunk;

		sch = skb_header_pointer(skb, iph->len + sizeof(struct sctphdr),
					 sizeof(schunk), &schunk);
		if (sch == NULL)
			return false;
		return sch->type == SCTP_CID_INIT;
	}
	default:
		return false;
	}
}

static inline bool is_new_conn_expected(const struct ip_vs_conn *cp,
					int conn_reuse_mode)
{
	/* Controlled (FTP DATA or persistence)? */
	if (cp->control)
		return false;

	switch (cp->protocol) {
	case IPPROTO_TCP:
		return (cp->state == IP_VS_TCP_S_TIME_WAIT) ||
		       (cp->state == IP_VS_TCP_S_CLOSE) ||
			((conn_reuse_mode & 2) &&
			 (cp->state == IP_VS_TCP_S_FIN_WAIT) &&
			 (cp->flags & IP_VS_CONN_F_NOOUTPUT));
	case IPPROTO_SCTP:
		return cp->state == IP_VS_SCTP_S_CLOSED;
	default:
		return false;
	}
}

/* Generic function to create new connections for outgoing RS packets
 *
 * Pre-requisites for successful connection creation:
 * 1) Virtual Service is NOT fwmark based:
 *    In fwmark-VS actual vaddr and vport are unknown to IPVS
 * 2) Real Server and Virtual Service were NOT configured without port:
 *    This is to allow match of different VS to the same RS ip-addr
 */
struct ip_vs_conn *ip_vs_new_conn_out(struct ip_vs_service *svc,
				      struct ip_vs_dest *dest,
				      struct sk_buff *skb,
				      const struct ip_vs_iphdr *iph,
				      __be16 dport,
				      __be16 cport)
{
	struct ip_vs_conn_param param;
	struct ip_vs_conn *ct = NULL, *cp = NULL;
	const union nf_inet_addr *vaddr, *daddr, *caddr;
	union nf_inet_addr snet;
	__be16 vport;
	unsigned int flags;

	vaddr = &svc->addr;
	vport = svc->port;
	daddr = &iph->saddr;
	caddr = &iph->daddr;

	/* check pre-requisites are satisfied */
	if (svc->fwmark)
		return NULL;
	if (!vport || !dport)
		return NULL;

	/* for persistent service first create connection template */
	if (svc->flags & IP_VS_SVC_F_PERSISTENT) {
		/* apply netmask the same way ingress-side does */
#ifdef CONFIG_IP_VS_IPV6
		if (svc->af == AF_INET6)
			ipv6_addr_prefix(&snet.in6, &caddr->in6,
					 (__force __u32)svc->netmask);
		else
#endif
			snet.ip = caddr->ip & svc->netmask;
		/* fill params and create template if not existent */
		if (ip_vs_conn_fill_param_persist(svc, skb, iph->protocol,
						  &snet, 0, vaddr,
						  vport, &param) < 0)
			return NULL;
		ct = ip_vs_ct_in_get(&param);
		/* check if template exists and points to the same dest */
		if (!ct || !ip_vs_check_template(ct, dest)) {
			ct = ip_vs_conn_new(&param, dest->af, daddr, dport,
					    IP_VS_CONN_F_TEMPLATE, dest, 0);
			if (!ct) {
				kfree(param.pe_data);
				return NULL;
			}
			ct->timeout = svc->timeout;
		} else {
			kfree(param.pe_data);
		}
	}

	/* connection flags */
	flags = ((svc->flags & IP_VS_SVC_F_ONEPACKET) &&
		 iph->protocol == IPPROTO_UDP) ? IP_VS_CONN_F_ONE_PACKET : 0;
	/* create connection */
	ip_vs_conn_fill_param(svc->ipvs, svc->af, iph->protocol,
			      caddr, cport, vaddr, vport, &param);
	cp = ip_vs_conn_new(&param, dest->af, daddr, dport, flags, dest, 0);
	if (!cp) {
		if (ct)
			ip_vs_conn_put(ct);
		return NULL;
	}
	if (ct) {
		ip_vs_control_add(cp, ct);
		ip_vs_conn_put(ct);
	}
	ip_vs_conn_stats(cp, svc);

	/* return connection (will be used to handle outgoing packet) */
	IP_VS_DBG_BUF(6, "New connection RS-initiated:%c c:%s:%u v:%s:%u "
		      "d:%s:%u conn->flags:%X conn->refcnt:%d\n",
		      ip_vs_fwd_tag(cp),
		      IP_VS_DBG_ADDR(cp->af, &cp->caddr), ntohs(cp->cport),
		      IP_VS_DBG_ADDR(cp->af, &cp->vaddr), ntohs(cp->vport),
		      IP_VS_DBG_ADDR(cp->af, &cp->daddr), ntohs(cp->dport),
		      cp->flags, refcount_read(&cp->refcnt));
	return cp;
}

/* Handle outgoing packets which are considered requests initiated by
 * real servers, so that subsequent responses from external client can be
 * routed to the right real server.
 * Used also for outgoing responses in OPS mode.
 *
 * Connection management is handled by persistent-engine specific callback.
 */
static struct ip_vs_conn *__ip_vs_rs_conn_out(unsigned int hooknum,
					      struct netns_ipvs *ipvs,
					      int af, struct sk_buff *skb,
					      const struct ip_vs_iphdr *iph)
{
	struct ip_vs_dest *dest;
	struct ip_vs_conn *cp = NULL;
	__be16 _ports[2], *pptr;

	if (hooknum == NF_INET_LOCAL_IN)
		return NULL;

	pptr = frag_safe_skb_hp(skb, iph->len,
				sizeof(_ports), _ports);
	if (!pptr)
		return NULL;

	dest = ip_vs_find_real_service(ipvs, af, iph->protocol,
				       &iph->saddr, pptr[0]);
	if (dest) {
		struct ip_vs_service *svc;
		struct ip_vs_pe *pe;

		svc = rcu_dereference(dest->svc);
		if (svc) {
			pe = rcu_dereference(svc->pe);
			if (pe && pe->conn_out)
				cp = pe->conn_out(svc, dest, skb, iph,
						  pptr[0], pptr[1]);
		}
	}

	return cp;
}

/* Handle response packets: rewrite addresses and send away...
 */
static unsigned int
handle_response(int af, struct sk_buff *skb, struct ip_vs_proto_data *pd,
		struct ip_vs_conn *cp, struct ip_vs_iphdr *iph,
		unsigned int hooknum)
{
	struct ip_vs_protocol *pp = pd->pp;

	if (IP_VS_FWD_METHOD(cp) != IP_VS_CONN_F_MASQ)
		goto after_nat;

	IP_VS_DBG_PKT(11, af, pp, skb, iph->off, "Outgoing packet");

	if (skb_ensure_writable(skb, iph->len))
		goto drop;

	/* mangle the packet */
	if (pp->snat_handler &&
	    !SNAT_CALL(pp->snat_handler, skb, pp, cp, iph))
		goto drop;

#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6)
		ipv6_hdr(skb)->saddr = cp->vaddr.in6;
	else
#endif
	{
		ip_hdr(skb)->saddr = cp->vaddr.ip;
		ip_send_check(ip_hdr(skb));
	}

	/*
	 * nf_iterate does not expect change in the skb->dst->dev.
	 * It looks like it is not fatal to enable this code for hooks
	 * where our handlers are at the end of the chain list and
	 * when all next handlers use skb->dst->dev and not outdev.
	 * It will definitely route properly the inout NAT traffic
	 * when multiple paths are used.
	 */

	/* For policy routing, packets originating from this
	 * machine itself may be routed differently to packets
	 * passing through.  We want this packet to be routed as
	 * if it came from this machine itself.  So re-compute
	 * the routing information.
	 */
	if (ip_vs_route_me_harder(cp->ipvs, af, skb, hooknum))
		goto drop;

	IP_VS_DBG_PKT(10, af, pp, skb, iph->off, "After SNAT");

after_nat:
	ip_vs_out_stats(cp, skb);
	ip_vs_set_state(cp, IP_VS_DIR_OUTPUT, skb, pd, iph->len);
	skb->ipvs_property = 1;
	if (!(cp->flags & IP_VS_CONN_F_NFCT))
		ip_vs_notrack(skb);
	else
		ip_vs_update_conntrack(skb, cp, 0);
	ip_vs_conn_put(cp);

	return NF_ACCEPT;

drop:
	ip_vs_conn_put(cp);
	kfree_skb(skb);
	return NF_STOLEN;
}

/*
 *	Check if outgoing packet belongs to the established ip_vs_conn.
 */
static unsigned int
ip_vs_out_hook(void *priv, struct sk_buff *skb, const struct nf_hook_state *state)
{
	struct netns_ipvs *ipvs = net_ipvs(state->net);
	unsigned int hooknum = state->hook;
	struct ip_vs_iphdr iph;
	struct ip_vs_protocol *pp;
	struct ip_vs_proto_data *pd;
	struct ip_vs_conn *cp;
	int af = state->pf;
	struct sock *sk;

	/* Already marked as IPVS request or reply? */
	if (skb->ipvs_property)
		return NF_ACCEPT;

	sk = skb_to_full_sk(skb);
	/* Bad... Do not break raw sockets */
	if (unlikely(sk && hooknum == NF_INET_LOCAL_OUT &&
		     af == AF_INET)) {

		if (sk->sk_family == PF_INET && inet_test_bit(NODEFRAG, sk))
			return NF_ACCEPT;
	}

	if (unlikely(!skb_dst(skb)))
		return NF_ACCEPT;

	ip_vs_fill_iph_skb(af, skb, false, &iph);
#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6) {
		if (unlikely(iph.protocol == IPPROTO_ICMPV6)) {
			int related;
			int verdict = ip_vs_out_icmp_v6(ipvs, skb, &related,
							hooknum, &iph);

			if (related)
				return verdict;
		}
	} else
#endif
		if (unlikely(iph.protocol == IPPROTO_ICMP)) {
			int related;
			int verdict = ip_vs_out_icmp(ipvs, skb, &related,
						     hooknum, &iph);

			if (related)
				return verdict;
		}

	pd = ip_vs_proto_data_get(ipvs, iph.protocol);
	if (unlikely(!pd))
		return NF_ACCEPT;
	pp = pd->pp;

	/* reassemble IP fragments */
#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET)
#endif
		if (unlikely(ip_is_fragment(ip_hdr(skb)) && !pp->dont_defrag)) {
			if (ip_vs_gather_frags(ipvs, skb,
					       ip_vs_defrag_user(hooknum)))
				return NF_STOLEN;

			ip_vs_fill_iph_skb(AF_INET, skb, false, &iph);
		}

	/*
	 * Check if the packet belongs to an existing entry
	 */
	cp = INDIRECT_CALL_1(pp->conn_out_get, ip_vs_conn_out_get_proto,
			     ipvs, af, skb, &iph);

	if (likely(cp))
		return handle_response(af, skb, pd, cp, &iph, hooknum);

	/* Check for real-server-started requests */
	if (atomic_read(&ipvs->conn_out_counter[ip_vs_af_index(af)])) {
		/* Currently only for UDP:
		 * connection oriented protocols typically use
		 * ephemeral ports for outgoing connections, so
		 * related incoming responses would not match any VS
		 */
		if (pp->protocol == IPPROTO_UDP) {
			cp = __ip_vs_rs_conn_out(hooknum, ipvs, af, skb, &iph);
			if (likely(cp))
				return handle_response(af, skb, pd, cp, &iph,
						       hooknum);
		}
	}

	if (sysctl_nat_icmp_send(ipvs) &&
	    (pp->protocol == IPPROTO_TCP ||
	     pp->protocol == IPPROTO_UDP ||
	     pp->protocol == IPPROTO_SCTP)) {
		__be16 _ports[2], *pptr;

		pptr = frag_safe_skb_hp(skb, iph.len,
					 sizeof(_ports), _ports);
		if (pptr == NULL)
			return NF_ACCEPT;	/* Not for me */
		if (ip_vs_has_real_service(ipvs, af, iph.protocol, &iph.saddr,
					   pptr[0])) {
			/*
			 * Notify the real server: there is no
			 * existing entry if it is not RST
			 * packet or not TCP packet.
			 */
			if ((iph.protocol != IPPROTO_TCP &&
			     iph.protocol != IPPROTO_SCTP)
			     || ((iph.protocol == IPPROTO_TCP
				  && !is_tcp_reset(skb, iph.len))
				 || (iph.protocol == IPPROTO_SCTP
					&& !is_sctp_abort(skb,
						iph.len)))) {
#ifdef CONFIG_IP_VS_IPV6
				if (af == AF_INET6) {
					if (!skb->dev)
						skb->dev = ipvs->net->loopback_dev;
					icmpv6_send(skb,
						    ICMPV6_DEST_UNREACH,
						    ICMPV6_PORT_UNREACH,
						    0);
				} else
#endif
					icmp_send(skb,
						  ICMP_DEST_UNREACH,
						  ICMP_PORT_UNREACH, 0);
				return NF_DROP;
			}
		}
	}

	IP_VS_DBG_PKT(12, af, pp, skb, iph.off,
		      "ip_vs_out: packet continues traversal as normal");
	return NF_ACCEPT;
}

static unsigned int
ip_vs_try_to_schedule(struct netns_ipvs *ipvs, 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 ip_vs_protocol *pp = pd->pp;

	if (!iph->fragoffs) {
		/* No (second) fragments need to enter here, as nf_defrag_ipv6
		 * replayed fragment zero will already have created the cp
		 */

		/* Schedule and create new connection entry into cpp */
		if (!pp->conn_schedule(ipvs, af, skb, pd, verdict, cpp, iph))
			return 0;
	}

	if (unlikely(!*cpp)) {
		/* sorry, all this trouble for a no-hit :) */
		IP_VS_DBG_PKT(12, af, pp, skb, iph->off,
			      "ip_vs_in: packet continues traversal as normal");

		/* Fragment couldn't be mapped to a conn entry */
		if (iph->fragoffs)
			IP_VS_DBG_PKT(7, af, pp, skb, iph->off,
				      "unhandled fragment");

		*verdict = NF_ACCEPT;
		return 0;
	}

	return 1;
}

/* Check the UDP tunnel and return its header length */
static int ipvs_udp_decap(struct netns_ipvs *ipvs, struct sk_buff *skb,
			  unsigned int offset, __u16 af,
			  const union nf_inet_addr *daddr, __u8 *proto)
{
	struct udphdr _udph, *udph;
	struct ip_vs_dest *dest;

	udph = skb_header_pointer(skb, offset, sizeof(_udph), &_udph);
	if (!udph)
		goto unk;
	offset += sizeof(struct udphdr);
	dest = ip_vs_find_tunnel(ipvs, af, daddr, udph->dest);
	if (!dest)
		goto unk;
	if (dest->tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GUE) {
		struct guehdr _gueh, *gueh;

		gueh = skb_header_pointer(skb, offset, sizeof(_gueh), &_gueh);
		if (!gueh)
			goto unk;
		if (gueh->control != 0 || gueh->version != 0)
			goto unk;
		/* Later we can support also IPPROTO_IPV6 */
		if (gueh->proto_ctype != IPPROTO_IPIP)
			goto unk;
		*proto = gueh->proto_ctype;
		return sizeof(struct udphdr) + sizeof(struct guehdr) +
		       (gueh->hlen << 2);
	}

unk:
	return 0;
}

/* Check the GRE tunnel and return its header length */
static int ipvs_gre_decap(struct netns_ipvs *ipvs, struct sk_buff *skb,
			  unsigned int offset, __u16 af,
			  const union nf_inet_addr *daddr, __u8 *proto)
{
	struct gre_base_hdr _greh, *greh;
	struct ip_vs_dest *dest;

	greh = skb_header_pointer(skb, offset, sizeof(_greh), &_greh);
	if (!greh)
		goto unk;
	dest = ip_vs_find_tunnel(ipvs, af, daddr, 0);
	if (!dest)
		goto unk;
	if (dest->tun_type == IP_VS_CONN_F_TUNNEL_TYPE_GRE) {
		IP_TUNNEL_DECLARE_FLAGS(flags);
		__be16 type;

		/* Only support version 0 and C (csum) */
		if ((greh->flags & ~GRE_CSUM) != 0)
			goto unk;
		type = greh->protocol;
		/* Later we can support also IPPROTO_IPV6 */
		if (type != htons(ETH_P_IP))
			goto unk;
		*proto = IPPROTO_IPIP;

		gre_flags_to_tnl_flags(flags, greh->flags);

		return gre_calc_hlen(flags);
	}

unk:
	return 0;
}

/*
 *	Handle ICMP messages in the outside-to-inside direction (incoming).
 *	Find any that might be relevant, check against existing connections,
 *	forward to the right destination host if relevant.
 *	Currently handles error types - unreachable, quench, ttl exceeded.
 */
static int
ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related,
	      unsigned int hooknum, struct ip_vs_iphdr *iph)
{
	struct icmphdr	_icmph, *ic;
	struct iphdr	_ciph, *cih;	/* The ip header contained within the ICMP */
	struct ip_vs_iphdr ciph;
	struct ip_vs_conn *cp;
	struct ip_vs_protocol *pp;
	struct ip_vs_proto_data *pd;
	unsigned int offset, offset2, ihl, verdict;
	bool tunnel, new_cp = false;
	union nf_inet_addr *raddr;
	char *outer_proto __maybe_unused = "IPIP";
	unsigned int hlen_ipip;
	int ulen = 0;

	*related = 1;

	/* reassemble IP fragments */
	if (ip_is_fragment(ip_hdr(skb))) {
		if (ip_vs_gather_frags(ipvs, skb, ip_vs_defrag_user(hooknum)))
			return NF_STOLEN;
		if (!ip_vs_fill_iph_skb(AF_INET, skb, false, iph))
			return NF_ACCEPT;
	}

	ihl = iph->len;
	offset = iph->len;
	ic = skb_header_pointer(skb, offset, sizeof(_icmph), &_icmph);
	if (ic == NULL)
		return NF_DROP;

	IP_VS_DBG(12, "Incoming ICMP (%d,%d) %pI4->%pI4\n",
		  ic->type, ntohs(icmp_id(ic)),
		  &iph->saddr.ip, &iph->daddr.ip);

	/*
	 * Work through seeing if this is for us.
	 * These checks are supposed to be in an order that means easy
	 * things are checked first to speed up processing.... however
	 * this means that some packets will manage to get a long way
	 * down this stack and then be rejected, but that's life.
	 */
	if ((ic->type != ICMP_DEST_UNREACH) &&
	    (ic->type != ICMP_SOURCE_QUENCH) &&
	    (ic->type != ICMP_TIME_EXCEEDED)) {
		*related = 0;
		return NF_ACCEPT;
	}

	/* Now find the contained IP header */
	offset += sizeof(_icmph);
	cih = skb_header_pointer(skb, offset, sizeof(_ciph), &_ciph);
	if (!cih)
		return NF_ACCEPT; /* The packet looks wrong, ignore */
	hlen_ipip = cih->ihl * 4;
	if (!(cih->version == 4 && hlen_ipip >= sizeof(struct iphdr)))
		return NF_ACCEPT; /* The packet looks wrong, ignore */
	raddr = (union nf_inet_addr *)&cih->daddr;

	/* Special case for errors for IPIP/UDP/GRE tunnel packets */
	tunnel = false;
	if (cih->protocol == IPPROTO_IPIP) {
		struct ip_vs_dest *dest;

		if (unlikely(cih->frag_off & htons(IP_OFFSET)))
			return NF_ACCEPT;
		/* Error for our IPIP must arrive at LOCAL_IN */
		if (!(skb_rtable(skb)->rt_flags & RTCF_LOCAL))
			return NF_ACCEPT;
		dest = ip_vs_find_tunnel(ipvs, AF_INET, raddr, 0);
		/* Only for known tunnel */
		if (!dest || dest->tun_type != IP_VS_CONN_F_TUNNEL_TYPE_IPIP)
			return NF_ACCEPT;
		offset += hlen_ipip;
		tunnel = true;
	} else if ((cih->protocol == IPPROTO_UDP ||	/* Can be UDP encap */
		    cih->protocol == IPPROTO_GRE) &&	/* Can be GRE encap */
		   /* Error for our tunnel must arrive at LOCAL_IN */
		   (skb_rtable(skb)->rt_flags & RTCF_LOCAL)) {
		__u8 iproto;

		/* Non-first fragment has no UDP/GRE header */
		if (unlikely(cih->frag_off & htons(IP_OFFSET)))
			return NF_ACCEPT;
		offset2 = offset + hlen_ipip;
		if (cih->protocol == IPPROTO_UDP) {
			ulen = ipvs_udp_decap(ipvs, skb, offset2, AF_INET,
					      raddr, &iproto);
			outer_proto = "UDP";
		} else {
			ulen = ipvs_gre_decap(ipvs, skb, offset2, AF_INET,
					      raddr, &iproto);
			outer_proto = "GRE";
		}
		if (ulen > 0) {
			/* Skip IP and UDP/GRE tunnel headers */
			offset = offset2 + ulen;
			/* Now we should be at the original IP header */
			if (iproto == IPPROTO_IPIP)
				tunnel = true;
			else
				return NF_ACCEPT;
		}
	}

	if (!ip_vs_fill_iph_skb_icmp(AF_INET, skb, offset, !tunnel, &ciph))
		return NF_ACCEPT;
	pd = ip_vs_proto_data_get(ipvs, ciph.protocol);
	if (!pd)
		return NF_ACCEPT;
	pp = pd->pp;

	cih = skb_header_pointer(skb, offset, sizeof(_ciph), &_ciph);
	if (!(cih && cih->version == 4 &&
	      ciph.len - ciph.off >= sizeof(struct iphdr)))
		return NF_ACCEPT; /* The packet looks wrong, ignore */

	/* Is the embedded protocol header present? */
	if (unlikely(cih->frag_off & htons(IP_OFFSET) && !pp->dont_defrag))
		return NF_ACCEPT;

	IP_VS_DBG_PKT(11, AF_INET, pp, skb, offset,
		      "Checking incoming ICMP for");

	/* The embedded headers contain source and dest in reverse order.
	 * For IPIP/UDP/GRE tunnel this is error for request, not for reply.
	 */
	cp = INDIRECT_CALL_1(pp->conn_in_get, ip_vs_conn_in_get_proto,
			     ipvs, AF_INET, skb, &ciph);

	if (!cp) {
		int v;

		if (tunnel || !sysctl_schedule_icmp(ipvs))
			return NF_ACCEPT;

		if (!ip_vs_try_to_schedule(ipvs, AF_INET, skb, pd, &v, &cp, &ciph))
			return v;
		new_cp = true;
	}

	verdict = NF_DROP;

	/* Ensure the checksum is correct */
	if ((IP_VS_FWD_METHOD(cp) == IP_VS_CONN_F_MASQ || tunnel) &&
	    !ip_vs_checksum_common_check(skb, ihl, IPPROTO_ICMP, AF_INET)) {
		/* Failed checksum! */
		IP_VS_DBG(1, "Incoming ICMP: failed checksum from %pI4!\n",
			  &iph->saddr.ip);
		goto out;
	}

	if (tunnel) {
		unsigned int hlen_orig = ciph.len - ciph.off;
		__be32 info = ic->un.gateway;
		__u8 type = ic->type;
		__u8 code = ic->code;

		offset2 = offset;
		/* Update the MTU */
		if (ic->type == ICMP_DEST_UNREACH &&
		    ic->code == ICMP_FRAG_NEEDED) {
			struct ip_vs_dest *dest = cp->dest;
			u32 mtu = ntohs(ic->un.frag.mtu);
			__be16 frag_off = cih->frag_off;

			/* Strip outer IP and ICMP, go to IPIP/UDP/GRE header */
			if (pskb_pull(skb, ihl + sizeof(_icmph)) == NULL)
				goto ignore_tunnel;
			offset2 -= ihl + sizeof(_icmph);
			skb_reset_network_header(skb);
			/* Ensure the IP header is present in headroom */
			if (!pskb_may_pull(skb, hlen_ipip))
				goto ignore_tunnel;
			IP_VS_DBG(12, "ICMP for %s %pI4->%pI4: mtu=%u\n",
				  outer_proto, &ip_hdr(skb)->saddr,
				  &ip_hdr(skb)->daddr, mtu);
			ipv4_update_pmtu(skb, ipvs->net, mtu, 0, 0);
			/* Client uses PMTUD? */
			if (!(frag_off & htons(IP_DF)))
				goto ignore_tunnel;
			/* Prefer the resulting PMTU */
			if (dest) {
				struct ip_vs_dest_dst *dest_dst;

				dest_dst = rcu_dereference(dest->dest_dst);
				if (dest_dst)
					mtu = dst_mtu(dest_dst->dst_cache);
			}
			if (mtu > 68 + hlen_ipip + ulen)
				mtu -= hlen_ipip + ulen;
			info = htonl(mtu);
		}
		/* Strip outer IP, ICMP and IPIP/UDP/GRE, go to IP header of
		 * original request.
		 */
		if (pskb_pull(skb, offset2) == NULL)
			goto ignore_tunnel;
		skb_reset_network_header(skb);
		memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
		/* Ensure the IP header is present in headroom */
		if (!pskb_may_pull(skb, hlen_orig))
			goto ignore_tunnel;
		skb_set_transport_header(skb, hlen_orig);
		/* Before now we may used ihl from skb frag, revalidate it after
		 * copying it into skb head to prevent out-of-bounds access
		 */
		if (ip_hdr(skb)->ihl * 4 != hlen_orig)
			goto ignore_tunnel;
		IP_VS_DBG(12, "Sending ICMP for %pI4->%pI4: t=%u, c=%u, i=%u\n",
			&ip_hdr(skb)->saddr, &ip_hdr(skb)->daddr,
			type, code, ntohl(info));
		icmp_send(skb, type, code, info);
		/* ICMP can be shorter but anyways, account it */
		ip_vs_out_stats(cp, skb);

ignore_tunnel:
		consume_skb(skb);
		verdict = NF_STOLEN;
		goto out;
	}

	/* do the statistics and put it back */
	ip_vs_in_stats(cp, skb);
	verdict = ip_vs_icmp_xmit(skb, cp, pp, iph->len, hooknum, &ciph);

out:
	if (likely(!new_cp))
		__ip_vs_conn_put(cp);
	else
		ip_vs_conn_put(cp);

	return verdict;
}

#ifdef CONFIG_IP_VS_IPV6
static int ip_vs_in_icmp_v6(struct netns_ipvs *ipvs, struct sk_buff *skb,
			    int *related, unsigned int hooknum,
			    struct ip_vs_iphdr *iph)
{
	struct icmp6hdr	_icmph, *ic;
	struct ip_vs_iphdr ciph = {.flags = 0, .fragoffs = 0};/*Contained IP */
	struct ip_vs_conn *cp;
	struct ip_vs_protocol *pp;
	struct ip_vs_proto_data *pd;
	unsigned int offset, verdict;
	bool new_cp = false;

	*related = 1;

	ic = frag_safe_skb_hp(skb, iph->len, sizeof(_icmph), &_icmph);
	if (ic == NULL)
		return NF_DROP;

	/*
	 * Work through seeing if this is for us.
	 * These checks are supposed to be in an order that means easy
	 * things are checked first to speed up processing.... however
	 * this means that some packets will manage to get a long way
	 * down this stack and then be rejected, but that's life.
	 */
	if (ic->icmp6_type & ICMPV6_INFOMSG_MASK) {
		*related = 0;
		return NF_ACCEPT;
	}
	/* Fragment header that is before ICMP header tells us that:
	 * it's not an error message since they can't be fragmented.
	 */
	if (iph->flags & IP6_FH_F_FRAG)
		return NF_DROP;

	IP_VS_DBG(8, "Incoming ICMPv6 (%d,%d) %pI6c->%pI6c\n",
		  ic->icmp6_type, ntohs(icmpv6_id(ic)),
		  &iph->saddr, &iph->daddr);

	offset = iph->len + sizeof(_icmph);
	if (!ip_vs_fill_iph_skb_icmp(AF_INET6, skb, offset, true, &ciph))
		return NF_ACCEPT;

	pd = ip_vs_proto_data_get(ipvs, ciph.protocol);
	if (!pd)
		return NF_ACCEPT;
	pp = pd->pp;

	/* Is the embedded protocol header present? */
	if (ciph.fragoffs && !pp->dont_defrag)
		return NF_ACCEPT;

	IP_VS_DBG_PKT(11, AF_INET6, pp, skb, offset,
		      "Checking incoming ICMPv6 for");

	/* The embedded headers contain source and dest in reverse order
	 * if not from localhost
	 */
	cp = INDIRECT_CALL_1(pp->conn_in_get, ip_vs_conn_in_get_proto,
			     ipvs, AF_INET6, skb, &ciph);

	if (!cp) {
		int v;

		if (!sysctl_schedule_icmp(ipvs))
			return NF_ACCEPT;

		if (!ip_vs_try_to_schedule(ipvs, AF_INET6, skb, pd, &v, &cp, &ciph))
			return v;

		new_cp = true;
	}

	verdict = NF_DROP;

	/* Ensure the checksum is correct */
	if (IP_VS_FWD_METHOD(cp) == IP_VS_CONN_F_MASQ &&
	    !ip_vs_checksum_common_check(skb, iph->len, IPPROTO_ICMPV6,
					 AF_INET6)) {
		/* Failed checksum! */
		IP_VS_DBG(1, "Incoming ICMPv6: failed checksum from %pI6c!\n",
			  &iph->saddr);
		goto out;
	}

	/* do the statistics and put it back */
	ip_vs_in_stats(cp, skb);

	verdict = ip_vs_icmp_xmit_v6(skb, cp, pp, iph->len, hooknum, &ciph);

out:
	if (likely(!new_cp))
		__ip_vs_conn_put(cp);
	else
		ip_vs_conn_put(cp);

	return verdict;
}
#endif


/*
 *	Check if it's for virtual services, look it up,
 *	and send it on its way...
 */
static unsigned int
ip_vs_in_hook(void *priv, struct sk_buff *skb, const struct nf_hook_state *state)
{
	struct netns_ipvs *ipvs = net_ipvs(state->net);
	unsigned int hooknum = state->hook;
	struct ip_vs_iphdr iph;
	struct ip_vs_protocol *pp;
	struct ip_vs_proto_data *pd;
	struct ip_vs_conn *cp;
	int ret, pkts;
	struct sock *sk;
	int af = state->pf;

	/* Already marked as IPVS request or reply? */
	if (skb->ipvs_property)
		return NF_ACCEPT;

	/*
	 *	Big tappo:
	 *	- remote client: only PACKET_HOST
	 *	- route: used for struct net when skb->dev is unset
	 */
	if (unlikely((skb->pkt_type != PACKET_HOST &&
		      hooknum != NF_INET_LOCAL_OUT) ||
		     !skb_dst(skb))) {
		ip_vs_fill_iph_skb(af, skb, false, &iph);
		IP_VS_DBG_BUF(12, "packet type=%d proto=%d daddr=%s"
			      " ignored in hook %u\n",
			      skb->pkt_type, iph.protocol,
			      IP_VS_DBG_ADDR(af, &iph.daddr), hooknum);
		return NF_ACCEPT;
	}
	/* ipvs enabled in this netns ? */
	if (unlikely(sysctl_backup_only(ipvs)))
		return NF_ACCEPT;

	ip_vs_fill_iph_skb(af, skb, false, &iph);

	/* Bad... Do not break raw sockets */
	sk = skb_to_full_sk(skb);
	if (unlikely(sk && hooknum == NF_INET_LOCAL_OUT &&
		     af == AF_INET)) {

		if (sk->sk_family == PF_INET && inet_test_bit(NODEFRAG, sk))
			return NF_ACCEPT;
	}

#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6) {
		if (unlikely(iph.protocol == IPPROTO_ICMPV6)) {
			int related;
			int verdict = ip_vs_in_icmp_v6(ipvs, skb, &related,
						       hooknum, &iph);

			if (related)
				return verdict;
		}
	} else
#endif
		if (unlikely(iph.protocol == IPPROTO_ICMP)) {
			int related;
			int verdict = ip_vs_in_icmp(ipvs, skb, &related,
						    hooknum, &iph);

			if (related)
				return verdict;
		}

	/* Protocol supported? */
	pd = ip_vs_proto_data_get(ipvs, iph.protocol);
	if (unlikely(!pd)) {
		/* The only way we'll see this packet again is if it's
		 * encapsulated, so mark it with ipvs_property=1 so we
		 * skip it if we're ignoring tunneled packets
		 */
		if (sysctl_ignore_tunneled(ipvs))
			skb->ipvs_property = 1;

		return NF_ACCEPT;
	}
	pp = pd->pp;
	/*
	 * Check if the packet belongs to an existing connection entry
	 */
	cp = INDIRECT_CALL_1(pp->conn_in_get, ip_vs_conn_in_get_proto,
			     ipvs, af, skb, &iph);

	if (!iph.fragoffs && is_new_conn(skb, &iph) && cp) {
		int conn_reuse_mode = sysctl_conn_reuse_mode(ipvs);
		bool old_ct = false, resched = false;

		if (unlikely(sysctl_expire_nodest_conn(ipvs)) && cp->dest &&
		    unlikely(!atomic_read(&cp->dest->weight))) {
			resched = true;
			old_ct = ip_vs_conn_uses_old_conntrack(cp, skb);
		} else if (conn_reuse_mode &&
			   is_new_conn_expected(cp, conn_reuse_mode)) {
			old_ct = ip_vs_conn_uses_old_conntrack(cp, skb);
			if (!atomic_read(&cp->n_control)) {
				resched = true;
			} else {
				/* Do not reschedule controlling connection
				 * that uses conntrack while it is still
				 * referenced by controlled connection(s).
				 */
				resched = !old_ct;
			}
		}

		if (resched) {
			if (!old_ct) {
				spin_lock_bh(&cp->lock);
				cp->flags &= ~IP_VS_CONN_F_NFCT;
				spin_unlock_bh(&cp->lock);
			}
			if (!atomic_read(&cp->n_control))
				ip_vs_conn_expire_now(cp);
			__ip_vs_conn_put(cp);
			if (old_ct)
				return NF_DROP;
			cp = NULL;
		}
	}

	/* Check the server status */
	if (cp && cp->dest && !(cp->dest->cflags & IP_VS_DEST_CF_AVAILABLE)) {
		/* the destination server is not available */
		if (sysctl_expire_nodest_conn(ipvs)) {
			bool old_ct = ip_vs_conn_uses_old_conntrack(cp, skb);

			if (!old_ct) {
				spin_lock_bh(&cp->lock);
				cp->flags &= ~IP_VS_CONN_F_NFCT;
				spin_unlock_bh(&cp->lock);
			}

			ip_vs_conn_expire_now(cp);
			__ip_vs_conn_put(cp);
			if (old_ct)
				return NF_DROP;
			cp = NULL;
		} else {
			__ip_vs_conn_put(cp);
			return NF_DROP;
		}
	}

	if (unlikely(!cp)) {
		int v;

		if (!ip_vs_try_to_schedule(ipvs, af, skb, pd, &v, &cp, &iph))
			return v;
	}

	IP_VS_DBG_PKT(11, af, pp, skb, iph.off, "Incoming packet");

	ip_vs_in_stats(cp, skb);
	ip_vs_set_state(cp, IP_VS_DIR_INPUT, skb, pd, iph.len);
	if (cp->packet_xmit)
		ret = cp->packet_xmit(skb, cp, pp, &iph);
		/* do not touch skb anymore */
	else {
		IP_VS_DBG_RL("warning: packet_xmit is null");
		ret = NF_ACCEPT;
	}

	/* Increase its packet counter and check if it is needed
	 * to be synchronized
	 *
	 * Sync connection if it is about to close to
	 * encorage the standby servers to update the connections timeout
	 *
	 * For ONE_PKT let ip_vs_sync_conn() do the filter work.
	 */

	if (cp->flags & IP_VS_CONN_F_ONE_PACKET)
		pkts = sysctl_sync_threshold(ipvs);
	else
		pkts = atomic_inc_return(&cp->in_pkts);

	if (ipvs->sync_state & IP_VS_STATE_MASTER)
		ip_vs_sync_conn(ipvs, cp, pkts);
	else if ((cp->flags & IP_VS_CONN_F_ONE_PACKET) && cp->control)
		/* increment is done inside ip_vs_sync_conn too */
		atomic_inc(&cp->control->in_pkts);

	ip_vs_conn_put(cp);
	return ret;
}

/*
 *	It is hooked at the NF_INET_FORWARD chain, in order to catch ICMP
 *      related packets destined for 0.0.0.0/0.
 *      When fwmark-based virtual service is used, such as transparent
 *      cache cluster, TCP packets can be marked and routed to ip_vs_in,
 *      but ICMP destined for 0.0.0.0/0 cannot not be easily marked and
 *      sent to ip_vs_in_icmp. So, catch them at the NF_INET_FORWARD chain
 *      and send them to ip_vs_in_icmp.
 */
static unsigned int
ip_vs_forward_icmp(void *priv, struct sk_buff *skb,
		   const struct nf_hook_state *state)
{
	struct netns_ipvs *ipvs = net_ipvs(state->net);
	struct ip_vs_iphdr iphdr;
	int r;

	/* ipvs enabled in this netns ? */
	if (unlikely(sysctl_backup_only(ipvs)))
		return NF_ACCEPT;

	if (state->pf == NFPROTO_IPV4) {
		if (ip_hdr(skb)->protocol != IPPROTO_ICMP)
			return NF_ACCEPT;
		ip_vs_fill_iph_skb(AF_INET, skb, false, &iphdr);
#ifdef CONFIG_IP_VS_IPV6
	} else {
		ip_vs_fill_iph_skb(AF_INET6, skb, false, &iphdr);

		if (iphdr.protocol != IPPROTO_ICMPV6)
			return NF_ACCEPT;

		return ip_vs_in_icmp_v6(ipvs, skb, &r, state->hook, &iphdr);
#endif
	}

	return ip_vs_in_icmp(ipvs, skb, &r, state->hook, &iphdr);
}

static const struct nf_hook_ops ip_vs_ops4[] = {
	/* After packet filtering, change source only for VS/NAT */
	{
		.hook		= ip_vs_out_hook,
		.pf		= NFPROTO_IPV4,
		.hooknum	= NF_INET_LOCAL_IN,
		.priority	= NF_IP_PRI_NAT_SRC - 2,
	},
	/* After packet filtering, forward packet through VS/DR, VS/TUN,
	 * or VS/NAT(change destination), so that filtering rules can be
	 * applied to IPVS. */
	{
		.hook		= ip_vs_in_hook,
		.pf		= NFPROTO_IPV4,
		.hooknum	= NF_INET_LOCAL_IN,
		.priority	= NF_IP_PRI_NAT_SRC - 1,
	},
	/* Before ip_vs_in, change source only for VS/NAT */
	{
		.hook		= ip_vs_out_hook,
		.pf		= NFPROTO_IPV4,
		.hooknum	= NF_INET_LOCAL_OUT,
		.priority	= NF_IP_PRI_NAT_DST + 1,
	},
	/* After mangle, schedule and forward local requests */
	{
		.hook		= ip_vs_in_hook,
		.pf		= NFPROTO_IPV4,
		.hooknum	= NF_INET_LOCAL_OUT,
		.priority	= NF_IP_PRI_NAT_DST + 2,
	},
	/* After packet filtering (but before ip_vs_out_icmp), catch icmp
	 * destined for 0.0.0.0/0, which is for incoming IPVS connections */
	{
		.hook		= ip_vs_forward_icmp,
		.pf		= NFPROTO_IPV4,
		.hooknum	= NF_INET_FORWARD,
		.priority	= 99,
	},
	/* After packet filtering, change source only for VS/NAT */
	{
		.hook		= ip_vs_out_hook,
		.pf		= NFPROTO_IPV4,
		.hooknum	= NF_INET_FORWARD,
		.priority	= 100,
	},
};

#ifdef CONFIG_IP_VS_IPV6
static const struct nf_hook_ops ip_vs_ops6[] = {
	/* After packet filtering, change source only for VS/NAT */
	{
		.hook		= ip_vs_out_hook,
		.pf		= NFPROTO_IPV6,
		.hooknum	= NF_INET_LOCAL_IN,
		.priority	= NF_IP6_PRI_NAT_SRC - 2,
	},
	/* After packet filtering, forward packet through VS/DR, VS/TUN,
	 * or VS/NAT(change destination), so that filtering rules can be
	 * applied to IPVS. */
	{
		.hook		= ip_vs_in_hook,
		.pf		= NFPROTO_IPV6,
		.hooknum	= NF_INET_LOCAL_IN,
		.priority	= NF_IP6_PRI_NAT_SRC - 1,
	},
	/* Before ip_vs_in, change source only for VS/NAT */
	{
		.hook		= ip_vs_out_hook,
		.pf		= NFPROTO_IPV6,
		.hooknum	= NF_INET_LOCAL_OUT,
		.priority	= NF_IP6_PRI_NAT_DST + 1,
	},
	/* After mangle, schedule and forward local requests */
	{
		.hook		= ip_vs_in_hook,
		.pf		= NFPROTO_IPV6,
		.hooknum	= NF_INET_LOCAL_OUT,
		.priority	= NF_IP6_PRI_NAT_DST + 2,
	},
	/* After packet filtering (but before ip_vs_out_icmp), catch icmp
	 * destined for 0.0.0.0/0, which is for incoming IPVS connections */
	{
		.hook		= ip_vs_forward_icmp,
		.pf		= NFPROTO_IPV6,
		.hooknum	= NF_INET_FORWARD,
		.priority	= 99,
	},
	/* After packet filtering, change source only for VS/NAT */
	{
		.hook		= ip_vs_out_hook,
		.pf		= NFPROTO_IPV6,
		.hooknum	= NF_INET_FORWARD,
		.priority	= 100,
	},
};
#endif

int ip_vs_register_hooks(struct netns_ipvs *ipvs, unsigned int af)
{
	const struct nf_hook_ops *ops;
	unsigned int count;
	unsigned int afmask;
	int ret = 0;

	if (af == AF_INET6) {
#ifdef CONFIG_IP_VS_IPV6
		ops = ip_vs_ops6;
		count = ARRAY_SIZE(ip_vs_ops6);
		afmask = 2;
#else
		return -EINVAL;
#endif
	} else {
		ops = ip_vs_ops4;
		count = ARRAY_SIZE(ip_vs_ops4);
		afmask = 1;
	}

	if (!(ipvs->hooks_afmask & afmask)) {
		ret = nf_register_net_hooks(ipvs->net, ops, count);
		if (ret >= 0)
			ipvs->hooks_afmask |= afmask;
	}
	return ret;
}

void ip_vs_unregister_hooks(struct netns_ipvs *ipvs, unsigned int af)
{
	const struct nf_hook_ops *ops;
	unsigned int count;
	unsigned int afmask;

	if (af == AF_INET6) {
#ifdef CONFIG_IP_VS_IPV6
		ops = ip_vs_ops6;
		count = ARRAY_SIZE(ip_vs_ops6);
		afmask = 2;
#else
		return;
#endif
	} else {
		ops = ip_vs_ops4;
		count = ARRAY_SIZE(ip_vs_ops4);
		afmask = 1;
	}

	if (ipvs->hooks_afmask & afmask) {
		nf_unregister_net_hooks(ipvs->net, ops, count);
		ipvs->hooks_afmask &= ~afmask;
	}
}

/*
 *	Initialize IP Virtual Server netns mem.
 */
static int __net_init __ip_vs_init(struct net *net)
{
	struct netns_ipvs *ipvs;

	ipvs = net_generic(net, ip_vs_net_id);
	if (ipvs == NULL)
		return -ENOMEM;

	/* Hold the beast until a service is registered */
	WRITE_ONCE(ipvs->enable, 0);
	ipvs->net = net;
	/* Counters used for creating unique names */
	ipvs->gen = atomic_read(&ipvs_netns_cnt);
	atomic_inc(&ipvs_netns_cnt);
	net->ipvs = ipvs;

	if (ip_vs_estimator_net_init(ipvs) < 0)
		goto estimator_fail;

	if (ip_vs_control_net_init(ipvs) < 0)
		goto control_fail;

	if (ip_vs_protocol_net_init(ipvs) < 0)
		goto protocol_fail;

	if (ip_vs_app_net_init(ipvs) < 0)
		goto app_fail;

	if (ip_vs_conn_net_init(ipvs) < 0)
		goto conn_fail;

	if (ip_vs_sync_net_init(ipvs) < 0)
		goto sync_fail;

	return 0;
/*
 * Error handling
 */

sync_fail:
	ip_vs_conn_net_cleanup(ipvs);
conn_fail:
	ip_vs_app_net_cleanup(ipvs);
app_fail:
	ip_vs_protocol_net_cleanup(ipvs);
protocol_fail:
	ip_vs_control_net_cleanup(ipvs);
control_fail:
	ip_vs_estimator_net_cleanup(ipvs);
estimator_fail:
	net->ipvs = NULL;
	return -ENOMEM;
}

static void __net_exit __ip_vs_cleanup_batch(struct list_head *net_list)
{
	struct netns_ipvs *ipvs;
	struct net *net;

	ip_vs_service_nets_cleanup(net_list);	/* ip_vs_flush() with locks */
	list_for_each_entry(net, net_list, exit_list) {
		ipvs = net_ipvs(net);
		ip_vs_conn_net_cleanup(ipvs);
		ip_vs_app_net_cleanup(ipvs);
		ip_vs_protocol_net_cleanup(ipvs);
		ip_vs_control_net_cleanup(ipvs);
		ip_vs_estimator_net_cleanup(ipvs);
		IP_VS_DBG(2, "ipvs netns %d released\n", ipvs->gen);
		net->ipvs = NULL;
	}
}

static void __net_exit __ip_vs_dev_cleanup_batch(struct list_head *net_list)
{
	struct netns_ipvs *ipvs;
	struct net *net;

	list_for_each_entry(net, net_list, exit_list) {
		ipvs = net_ipvs(net);
		ip_vs_unregister_hooks(ipvs, AF_INET);
		ip_vs_unregister_hooks(ipvs, AF_INET6);
		WRITE_ONCE(ipvs->enable, 0);	/* Disable packet reception */
		smp_wmb();
		ip_vs_sync_net_cleanup(ipvs);
	}
}

static struct pernet_operations ipvs_core_ops = {
	.init = __ip_vs_init,
	.exit_batch = __ip_vs_cleanup_batch,
	.id   = &ip_vs_net_id,
	.size = sizeof(struct netns_ipvs),
};

static struct pernet_operations ipvs_core_dev_ops = {
	.exit_batch = __ip_vs_dev_cleanup_batch,
};

/*
 *	Initialize IP Virtual Server
 */
static int __init ip_vs_init(void)
{
	int ret;

	ret = ip_vs_control_init();
	if (ret < 0) {
		pr_err("can't setup control.\n");
		goto exit;
	}

	ip_vs_protocol_init();

	ret = ip_vs_conn_init();
	if (ret < 0) {
		pr_err("can't setup connection table.\n");
		goto cleanup_protocol;
	}

	ret = register_pernet_subsys(&ipvs_core_ops);	/* Alloc ip_vs struct */
	if (ret < 0)
		goto cleanup_conn;

	ret = register_pernet_device(&ipvs_core_dev_ops);
	if (ret < 0)
		goto cleanup_sub;

	ret = ip_vs_register_nl_ioctl();
	if (ret < 0) {
		pr_err("can't register netlink/ioctl.\n");
		goto cleanup_dev;
	}

	pr_info("ipvs loaded.\n");

	return ret;

cleanup_dev:
	unregister_pernet_device(&ipvs_core_dev_ops);
cleanup_sub:
	unregister_pernet_subsys(&ipvs_core_ops);
cleanup_conn:
	ip_vs_conn_cleanup();
cleanup_protocol:
	ip_vs_protocol_cleanup();
	ip_vs_control_cleanup();
exit:
	return ret;
}

static void __exit ip_vs_cleanup(void)
{
	ip_vs_unregister_nl_ioctl();
	unregister_pernet_device(&ipvs_core_dev_ops);
	unregister_pernet_subsys(&ipvs_core_ops);	/* free ip_vs struct */
	ip_vs_conn_cleanup();
	ip_vs_protocol_cleanup();
	ip_vs_control_cleanup();
	/* common rcu_barrier() used by:
	 * - ip_vs_control_cleanup()
	 */
	rcu_barrier();
	pr_info("ipvs unloaded.\n");
}

module_init(ip_vs_init);
module_exit(ip_vs_cleanup);
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("IP Virtual Server");