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

/*
 * IPVS:        Source Hashing scheduling module
 *
 * Authors:     Wensong Zhang <wensong@gnuchina.org>
 *
 *              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:
 *
 */

/*
 * The sh algorithm is to select server by the hash key of source IP
 * address. The pseudo code is as follows:
 *
 *       n <- servernode[src_ip];
 *       if (n is dead) OR
 *          (n is overloaded) or (n.weight <= 0) then
 *                 return NULL;
 *
 *       return n;
 *
 * Notes that servernode is a 256-bucket hash table that maps the hash
 * index derived from packet source IP address to the current server
 * array. If the sh scheduler is used in cache cluster, it is good to
 * combine it with cache_bypass feature. When the statically assigned
 * server is dead or overloaded, the load balancer can bypass the cache
 * server and send requests to the original server directly.
 *
 * The weight destination attribute can be used to control the
 * distribution of connections to the destinations in servernode. The
 * greater the weight, the more connections the destination
 * will receive.
 *
 */

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

#include <linux/ip.h>
#include <linux/slab.h>
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/skbuff.h>

#include <net/ip_vs.h>

#include <net/tcp.h>
#include <linux/udp.h>
#include <linux/sctp.h>


/*
 *      IPVS SH bucket
 */
struct ip_vs_sh_bucket {
	struct ip_vs_dest __rcu	*dest;	/* real server (cache) */
};

/*
 *     for IPVS SH entry hash table
 */
#ifndef CONFIG_IP_VS_SH_TAB_BITS
#define CONFIG_IP_VS_SH_TAB_BITS        8
#endif
#define IP_VS_SH_TAB_BITS               CONFIG_IP_VS_SH_TAB_BITS
#define IP_VS_SH_TAB_SIZE               (1 << IP_VS_SH_TAB_BITS)
#define IP_VS_SH_TAB_MASK               (IP_VS_SH_TAB_SIZE - 1)

struct ip_vs_sh_state {
	struct rcu_head			rcu_head;
	struct ip_vs_sh_bucket		buckets[IP_VS_SH_TAB_SIZE];
};

/* Helper function to determine if server is unavailable */
static inline bool is_unavailable(struct ip_vs_dest *dest)
{
	return atomic_read(&dest->weight) <= 0 ||
	       dest->flags & IP_VS_DEST_F_OVERLOAD;
}

/*
 *	Returns hash value for IPVS SH entry
 */
static inline unsigned int
ip_vs_sh_hashkey(int af, const union nf_inet_addr *addr,
		 __be16 port, unsigned int offset)
{
	__be32 addr_fold = addr->ip;

#ifdef CONFIG_IP_VS_IPV6
	if (af == AF_INET6)
		addr_fold = addr->ip6[0]^addr->ip6[1]^
			    addr->ip6[2]^addr->ip6[3];
#endif
	return (offset + (ntohs(port) + ntohl(addr_fold))*2654435761UL) &
		IP_VS_SH_TAB_MASK;
}


/*
 *      Get ip_vs_dest associated with supplied parameters.
 */
static inline struct ip_vs_dest *
ip_vs_sh_get(struct ip_vs_service *svc, struct ip_vs_sh_state *s,
	     const union nf_inet_addr *addr, __be16 port)
{
	unsigned int hash = ip_vs_sh_hashkey(svc->af, addr, port, 0);
	struct ip_vs_dest *dest = rcu_dereference(s->buckets[hash].dest);

	return (!dest || is_unavailable(dest)) ? NULL : dest;
}


/* As ip_vs_sh_get, but with fallback if selected server is unavailable
 *
 * The fallback strategy loops around the table starting from a "random"
 * point (in fact, it is chosen to be the original hash value to make the
 * algorithm deterministic) to find a new server.
 */
static inline struct ip_vs_dest *
ip_vs_sh_get_fallback(struct ip_vs_service *svc, struct ip_vs_sh_state *s,
		      const union nf_inet_addr *addr, __be16 port)
{
	unsigned int offset, roffset;
	unsigned int hash, ihash;
	struct ip_vs_dest *dest;

	/* first try the dest it's supposed to go to */
	ihash = ip_vs_sh_hashkey(svc->af, addr, port, 0);
	dest = rcu_dereference(s->buckets[ihash].dest);
	if (!dest)
		return NULL;
	if (!is_unavailable(dest))
		return dest;

	IP_VS_DBG_BUF(6, "SH: selected unavailable server %s:%d, reselecting",
		      IP_VS_DBG_ADDR(dest->af, &dest->addr), ntohs(dest->port));

	/* if the original dest is unavailable, loop around the table
	 * starting from ihash to find a new dest
	 */
	for (offset = 0; offset < IP_VS_SH_TAB_SIZE; offset++) {
		roffset = (offset + ihash) % IP_VS_SH_TAB_SIZE;
		hash = ip_vs_sh_hashkey(svc->af, addr, port, roffset);
		dest = rcu_dereference(s->buckets[hash].dest);
		if (!dest)
			break;
		if (!is_unavailable(dest))
			return dest;
		IP_VS_DBG_BUF(6, "SH: selected unavailable "
			      "server %s:%d (offset %d), reselecting",
			      IP_VS_DBG_ADDR(dest->af, &dest->addr),
			      ntohs(dest->port), roffset);
	}

	return NULL;
}

/*
 *      Assign all the hash buckets of the specified table with the service.
 */
static int
ip_vs_sh_reassign(struct ip_vs_sh_state *s, struct ip_vs_service *svc)
{
	int i;
	struct ip_vs_sh_bucket *b;
	struct list_head *p;
	struct ip_vs_dest *dest;
	int d_count;
	bool empty;

	b = &s->buckets[0];
	p = &svc->destinations;
	empty = list_empty(p);
	d_count = 0;
	for (i=0; i<IP_VS_SH_TAB_SIZE; i++) {
		dest = rcu_dereference_protected(b->dest, 1);
		if (dest)
			ip_vs_dest_put(dest);
		if (empty)
			RCU_INIT_POINTER(b->dest, NULL);
		else {
			if (p == &svc->destinations)
				p = p->next;

			dest = list_entry(p, struct ip_vs_dest, n_list);
			ip_vs_dest_hold(dest);
			RCU_INIT_POINTER(b->dest, dest);

			IP_VS_DBG_BUF(6, "assigned i: %d dest: %s weight: %d\n",
				      i, IP_VS_DBG_ADDR(dest->af, &dest->addr),
				      atomic_read(&dest->weight));

			/* Don't move to next dest until filling weight */
			if (++d_count >= atomic_read(&dest->weight)) {
				p = p->next;
				d_count = 0;
			}

		}
		b++;
	}
	return 0;
}


/*
 *      Flush all the hash buckets of the specified table.
 */
static void ip_vs_sh_flush(struct ip_vs_sh_state *s)
{
	int i;
	struct ip_vs_sh_bucket *b;
	struct ip_vs_dest *dest;

	b = &s->buckets[0];
	for (i=0; i<IP_VS_SH_TAB_SIZE; i++) {
		dest = rcu_dereference_protected(b->dest, 1);
		if (dest) {
			ip_vs_dest_put(dest);
			RCU_INIT_POINTER(b->dest, NULL);
		}
		b++;
	}
}


static int ip_vs_sh_init_svc(struct ip_vs_service *svc)
{
	struct ip_vs_sh_state *s;

	/* allocate the SH table for this service */
	s = kzalloc(sizeof(struct ip_vs_sh_state), GFP_KERNEL);
	if (s == NULL)
		return -ENOMEM;

	svc->sched_data = s;
	IP_VS_DBG(6, "SH hash table (memory=%Zdbytes) allocated for "
		  "current service\n",
		  sizeof(struct ip_vs_sh_bucket)*IP_VS_SH_TAB_SIZE);

	/* assign the hash buckets with current dests */
	ip_vs_sh_reassign(s, svc);

	return 0;
}


static void ip_vs_sh_done_svc(struct ip_vs_service *svc)
{
	struct ip_vs_sh_state *s = svc->sched_data;

	/* got to clean up hash buckets here */
	ip_vs_sh_flush(s);

	/* release the table itself */
	kfree_rcu(s, rcu_head);
	IP_VS_DBG(6, "SH hash table (memory=%Zdbytes) released\n",
		  sizeof(struct ip_vs_sh_bucket)*IP_VS_SH_TAB_SIZE);
}


static int ip_vs_sh_dest_changed(struct ip_vs_service *svc,
				 struct ip_vs_dest *dest)
{
	struct ip_vs_sh_state *s = svc->sched_data;

	/* assign the hash buckets with the updated service */
	ip_vs_sh_reassign(s, svc);

	return 0;
}


/* Helper function to get port number */
static inline __be16
ip_vs_sh_get_port(const struct sk_buff *skb, struct ip_vs_iphdr *iph)
{
	__be16 _ports[2], *ports;

	/* At this point we know that we have a valid packet of some kind.
	 * Because ICMP packets are only guaranteed to have the first 8
	 * bytes, let's just grab the ports.  Fortunately they're in the
	 * same position for all three of the protocols we care about.
	 */
	switch (iph->protocol) {
	case IPPROTO_TCP:
	case IPPROTO_UDP:
	case IPPROTO_SCTP:
		ports = skb_header_pointer(skb, iph->len, sizeof(_ports),
					   &_ports);
		if (unlikely(!ports))
			return 0;

		if (likely(!ip_vs_iph_inverse(iph)))
			return ports[0];
		else
			return ports[1];
	default:
		return 0;
	}
}


/*
 *      Source Hashing scheduling
 */
static struct ip_vs_dest *
ip_vs_sh_schedule(struct ip_vs_service *svc, const struct sk_buff *skb,
		  struct ip_vs_iphdr *iph)
{
	struct ip_vs_dest *dest;
	struct ip_vs_sh_state *s;
	__be16 port = 0;
	const union nf_inet_addr *hash_addr;

	hash_addr = ip_vs_iph_inverse(iph) ? &iph->daddr : &iph->saddr;

	IP_VS_DBG(6, "ip_vs_sh_schedule(): Scheduling...\n");

	if (svc->flags & IP_VS_SVC_F_SCHED_SH_PORT)
		port = ip_vs_sh_get_port(skb, iph);

	s = (struct ip_vs_sh_state *) svc->sched_data;

	if (svc->flags & IP_VS_SVC_F_SCHED_SH_FALLBACK)
		dest = ip_vs_sh_get_fallback(svc, s, hash_addr, port);
	else
		dest = ip_vs_sh_get(svc, s, hash_addr, port);

	if (!dest) {
		ip_vs_scheduler_err(svc, "no destination available");
		return NULL;
	}

	IP_VS_DBG_BUF(6, "SH: source IP address %s --> server %s:%d\n",
		      IP_VS_DBG_ADDR(svc->af, hash_addr),
		      IP_VS_DBG_ADDR(dest->af, &dest->addr),
		      ntohs(dest->port));

	return dest;
}


/*
 *      IPVS SH Scheduler structure
 */
static struct ip_vs_scheduler ip_vs_sh_scheduler =
{
	.name =			"sh",
	.refcnt =		ATOMIC_INIT(0),
	.module =		THIS_MODULE,
	.n_list	 =		LIST_HEAD_INIT(ip_vs_sh_scheduler.n_list),
	.init_service =		ip_vs_sh_init_svc,
	.done_service =		ip_vs_sh_done_svc,
	.add_dest =		ip_vs_sh_dest_changed,
	.del_dest =		ip_vs_sh_dest_changed,
	.upd_dest =		ip_vs_sh_dest_changed,
	.schedule =		ip_vs_sh_schedule,
};


static int __init ip_vs_sh_init(void)
{
	return register_ip_vs_scheduler(&ip_vs_sh_scheduler);
}


static void __exit ip_vs_sh_cleanup(void)
{
	unregister_ip_vs_scheduler(&ip_vs_sh_scheduler);
	synchronize_rcu();
}


module_init(ip_vs_sh_init);
module_exit(ip_vs_sh_cleanup);
MODULE_LICENSE("GPL");
Commit	Line	Data
1da177e4 LT	1	/*
	2	* IPVS: Source Hashing scheduling module
	3	*
1da177e4 LT	4	* Authors: Wensong Zhang <wensong@gnuchina.org>
	5	*
	6	* This program is free software; you can redistribute it and/or
	7	* modify it under the terms of the GNU General Public License
	8	* as published by the Free Software Foundation; either version
	9	* 2 of the License, or (at your option) any later version.
	10	*
	11	* Changes:
	12	*
	13	*/
	14
	15	/*
	16	* The sh algorithm is to select server by the hash key of source IP
	17	* address. The pseudo code is as follows:
	18	*
	19	* n <- servernode[src_ip];
	20	* if (n is dead) OR
	21	* (n is overloaded) or (n.weight <= 0) then
	22	* return NULL;
	23	*
	24	* return n;
	25	*
	26	* Notes that servernode is a 256-bucket hash table that maps the hash
	27	* index derived from packet source IP address to the current server
	28	* array. If the sh scheduler is used in cache cluster, it is good to
	29	* combine it with cache_bypass feature. When the statically assigned
	30	* server is dead or overloaded, the load balancer can bypass the cache
	31	* server and send requests to the original server directly.
	32	*
76ad94fc MM	33	* The weight destination attribute can be used to control the
	34	* distribution of connections to the destinations in servernode. The
	35	* greater the weight, the more connections the destination
	36	* will receive.
	37	*
1da177e4 LT	38	*/
1da177e4 LT	39
9aada7ac HE	40	#define KMSG_COMPONENT "IPVS"
	41	#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
	42
14c85021	43	#include <linux/ip.h>
5a0e3ad6	44	#include <linux/slab.h>
1da177e4 LT	45	#include <linux/module.h>
1da177e4 LT	46	#include <linux/kernel.h>
14c85021	47	#include <linux/skbuff.h>
1da177e4 LT	48
	49	#include <net/ip_vs.h>
	50
eba3b5a7 AF	51	#include <net/tcp.h>
	52	#include <linux/udp.h>
	53	#include <linux/sctp.h>
	54
1da177e4 LT	55
	56	/*
	57	* IPVS SH bucket
	58	*/
	59	struct ip_vs_sh_bucket {
1acb7f67	60	struct ip_vs_dest __rcu dest; / real server (cache) */
1da177e4 LT	61	};
	62
	63	/*
	64	* for IPVS SH entry hash table
	65	*/
	66	#ifndef CONFIG_IP_VS_SH_TAB_BITS
	67	#define CONFIG_IP_VS_SH_TAB_BITS 8
	68	#endif
	69	#define IP_VS_SH_TAB_BITS CONFIG_IP_VS_SH_TAB_BITS
	70	#define IP_VS_SH_TAB_SIZE (1 << IP_VS_SH_TAB_BITS)
	71	#define IP_VS_SH_TAB_MASK (IP_VS_SH_TAB_SIZE - 1)
	72
1acb7f67	73	struct ip_vs_sh_state {
1acb7f67	74	struct rcu_head rcu_head;
a70b9641	75	struct ip_vs_sh_bucket buckets[IP_VS_SH_TAB_SIZE];
1acb7f67	76	};
1da177e4	77
eba3b5a7 AF	78	/* Helper function to determine if server is unavailable */
	79	static inline bool is_unavailable(struct ip_vs_dest *dest)
	80	{
	81	return atomic_read(&dest->weight) <= 0 \|\|
	82	dest->flags & IP_VS_DEST_F_OVERLOAD;
	83	}
	84
1da177e4 LT	85	/*
	86	* Returns hash value for IPVS SH entry
	87	*/
eba3b5a7 AF	88	static inline unsigned int
	89	ip_vs_sh_hashkey(int af, const union nf_inet_addr *addr,
	90	__be16 port, unsigned int offset)
1da177e4	91	{
20971a0a JV	92	__be32 addr_fold = addr->ip;
	93
	94	#ifdef CONFIG_IP_VS_IPV6
	95	if (af == AF_INET6)
	96	addr_fold = addr->ip6[0]^addr->ip6[1]^
	97	addr->ip6[2]^addr->ip6[3];
	98	#endif
eba3b5a7 AF	99	return (offset + (ntohs(port) + ntohl(addr_fold))*2654435761UL) &
eba3b5a7 AF	100	IP_VS_SH_TAB_MASK;
1da177e4 LT	101	}
	102
	103
	104	/*
	105	* Get ip_vs_dest associated with supplied parameters.
	106	*/
	107	static inline struct ip_vs_dest *
eba3b5a7 AF	108	ip_vs_sh_get(struct ip_vs_service svc, struct ip_vs_sh_state s,
eba3b5a7 AF	109	const union nf_inet_addr *addr, __be16 port)
1da177e4	110	{
eba3b5a7 AF	111	unsigned int hash = ip_vs_sh_hashkey(svc->af, addr, port, 0);
	112	struct ip_vs_dest *dest = rcu_dereference(s->buckets[hash].dest);
	113
	114	return (!dest \|\| is_unavailable(dest)) ? NULL : dest;
1da177e4 LT	115	}
	116
	117
1255ce5f AF	118	/* As ip_vs_sh_get, but with fallback if selected server is unavailable
	119	*
	120	* The fallback strategy loops around the table starting from a "random"
	121	* point (in fact, it is chosen to be the original hash value to make the
	122	* algorithm deterministic) to find a new server.
	123	*/
eba3b5a7 AF	124	static inline struct ip_vs_dest *
	125	ip_vs_sh_get_fallback(struct ip_vs_service svc, struct ip_vs_sh_state s,
	126	const union nf_inet_addr *addr, __be16 port)
	127	{
1255ce5f AF	128	unsigned int offset, roffset;
1255ce5f AF	129	unsigned int hash, ihash;
eba3b5a7 AF	130	struct ip_vs_dest *dest;
eba3b5a7 AF	131
1255ce5f AF	132	/* first try the dest it's supposed to go to */
	133	ihash = ip_vs_sh_hashkey(svc->af, addr, port, 0);
	134	dest = rcu_dereference(s->buckets[ihash].dest);
	135	if (!dest)
	136	return NULL;
	137	if (!is_unavailable(dest))
	138	return dest;
	139
	140	IP_VS_DBG_BUF(6, "SH: selected unavailable server %s:%d, reselecting",
4d316f3f	141	IP_VS_DBG_ADDR(dest->af, &dest->addr), ntohs(dest->port));
1255ce5f AF	142
	143	/* if the original dest is unavailable, loop around the table
	144	* starting from ihash to find a new dest
	145	*/
eba3b5a7	146	for (offset = 0; offset < IP_VS_SH_TAB_SIZE; offset++) {
1255ce5f AF	147	roffset = (offset + ihash) % IP_VS_SH_TAB_SIZE;
1255ce5f AF	148	hash = ip_vs_sh_hashkey(svc->af, addr, port, roffset);
eba3b5a7 AF	149	dest = rcu_dereference(s->buckets[hash].dest);
	150	if (!dest)
	151	break;
1255ce5f	152	if (!is_unavailable(dest))
eba3b5a7	153	return dest;
1255ce5f AF	154	IP_VS_DBG_BUF(6, "SH: selected unavailable "
1255ce5f AF	155	"server %s:%d (offset %d), reselecting",
4d316f3f	156	IP_VS_DBG_ADDR(dest->af, &dest->addr),
1255ce5f	157	ntohs(dest->port), roffset);
eba3b5a7 AF	158	}
	159
	160	return NULL;
	161	}
	162
1da177e4 LT	163	/*
	164	* Assign all the hash buckets of the specified table with the service.
	165	*/
	166	static int
1acb7f67	167	ip_vs_sh_reassign(struct ip_vs_sh_state s, struct ip_vs_service svc)
1da177e4 LT	168	{
	169	int i;
	170	struct ip_vs_sh_bucket *b;
	171	struct list_head *p;
	172	struct ip_vs_dest *dest;
76ad94fc	173	int d_count;
1acb7f67	174	bool empty;
1da177e4	175
1acb7f67	176	b = &s->buckets[0];
1da177e4	177	p = &svc->destinations;
1acb7f67	178	empty = list_empty(p);
76ad94fc	179	d_count = 0;
1da177e4	180	for (i=0; i<IP_VS_SH_TAB_SIZE; i++) {
1acb7f67 JA	181	dest = rcu_dereference_protected(b->dest, 1);
	182	if (dest)
	183	ip_vs_dest_put(dest);
	184	if (empty)
	185	RCU_INIT_POINTER(b->dest, NULL);
	186	else {
1da177e4 LT	187	if (p == &svc->destinations)
	188	p = p->next;
	189
	190	dest = list_entry(p, struct ip_vs_dest, n_list);
1acb7f67 JA	191	ip_vs_dest_hold(dest);
1acb7f67 JA	192	RCU_INIT_POINTER(b->dest, dest);
1da177e4	193
76ad94fc	194	IP_VS_DBG_BUF(6, "assigned i: %d dest: %s weight: %d\n",
4d316f3f	195	i, IP_VS_DBG_ADDR(dest->af, &dest->addr),
76ad94fc MM	196	atomic_read(&dest->weight));
	197
	198	/* Don't move to next dest until filling weight */
	199	if (++d_count >= atomic_read(&dest->weight)) {
	200	p = p->next;
	201	d_count = 0;
	202	}
	203
1da177e4 LT	204	}
	205	b++;
	206	}
	207	return 0;
	208	}
	209
	210
	211	/*
	212	* Flush all the hash buckets of the specified table.
	213	*/
1acb7f67	214	static void ip_vs_sh_flush(struct ip_vs_sh_state *s)
1da177e4 LT	215	{
	216	int i;
	217	struct ip_vs_sh_bucket *b;
1acb7f67	218	struct ip_vs_dest *dest;
1da177e4	219
1acb7f67	220	b = &s->buckets[0];
1da177e4	221	for (i=0; i<IP_VS_SH_TAB_SIZE; i++) {
1acb7f67 JA	222	dest = rcu_dereference_protected(b->dest, 1);
	223	if (dest) {
	224	ip_vs_dest_put(dest);
	225	RCU_INIT_POINTER(b->dest, NULL);
1da177e4 LT	226	}
	227	b++;
	228	}
	229	}
	230
	231
	232	static int ip_vs_sh_init_svc(struct ip_vs_service *svc)
	233	{
1acb7f67	234	struct ip_vs_sh_state *s;
1da177e4 LT	235
1da177e4 LT	236	/* allocate the SH table for this service */
1acb7f67 JA	237	s = kzalloc(sizeof(struct ip_vs_sh_state), GFP_KERNEL);
1acb7f67 JA	238	if (s == NULL)
1da177e4	239	return -ENOMEM;
0a9ee813	240
1acb7f67	241	svc->sched_data = s;
1da177e4 LT	242	IP_VS_DBG(6, "SH hash table (memory=%Zdbytes) allocated for "
	243	"current service\n",
	244	sizeof(struct ip_vs_sh_bucket)*IP_VS_SH_TAB_SIZE);
	245
1acb7f67 JA	246	/* assign the hash buckets with current dests */
1acb7f67 JA	247	ip_vs_sh_reassign(s, svc);
1da177e4 LT	248
	249	return 0;
	250	}
	251
	252
ed3ffc4e	253	static void ip_vs_sh_done_svc(struct ip_vs_service *svc)
1da177e4	254	{
1acb7f67	255	struct ip_vs_sh_state *s = svc->sched_data;
1da177e4 LT	256
1da177e4 LT	257	/* got to clean up hash buckets here */
1acb7f67	258	ip_vs_sh_flush(s);
1da177e4 LT	259
1da177e4 LT	260	/* release the table itself */
1acb7f67	261	kfree_rcu(s, rcu_head);
1da177e4 LT	262	IP_VS_DBG(6, "SH hash table (memory=%Zdbytes) released\n",
1da177e4 LT	263	sizeof(struct ip_vs_sh_bucket)*IP_VS_SH_TAB_SIZE);
1da177e4 LT	264	}
	265
	266
1acb7f67 JA	267	static int ip_vs_sh_dest_changed(struct ip_vs_service *svc,
1acb7f67 JA	268	struct ip_vs_dest *dest)
1da177e4	269	{
1acb7f67	270	struct ip_vs_sh_state *s = svc->sched_data;
1da177e4 LT	271
1da177e4 LT	272	/* assign the hash buckets with the updated service */
1acb7f67	273	ip_vs_sh_reassign(s, svc);
1da177e4 LT	274
	275	return 0;
	276	}
	277
	278
eba3b5a7 AF	279	/* Helper function to get port number */
	280	static inline __be16
	281	ip_vs_sh_get_port(const struct sk_buff skb, struct ip_vs_iphdr iph)
1da177e4	282	{
1471f35e	283	__be16 _ports[2], *ports;
eba3b5a7	284
1471f35e AG	285	/* At this point we know that we have a valid packet of some kind.
	286	* Because ICMP packets are only guaranteed to have the first 8
	287	* bytes, let's just grab the ports. Fortunately they're in the
	288	* same position for all three of the protocols we care about.
	289	*/
eba3b5a7 AF	290	switch (iph->protocol) {
eba3b5a7 AF	291	case IPPROTO_TCP:
eba3b5a7	292	case IPPROTO_UDP:
eba3b5a7	293	case IPPROTO_SCTP:
1471f35e AG	294	ports = skb_header_pointer(skb, iph->len, sizeof(_ports),
	295	&_ports);
	296	if (unlikely(!ports))
54e35cc5	297	return 0;
1471f35e AG	298
	299	if (likely(!ip_vs_iph_inverse(iph)))
	300	return ports[0];
	301	else
	302	return ports[1];
eba3b5a7	303	default:
1471f35e	304	return 0;
eba3b5a7	305	}
1da177e4 LT	306	}
	307
	308
	309	/*
	310	* Source Hashing scheduling
	311	*/
	312	static struct ip_vs_dest *
bba54de5 JA	313	ip_vs_sh_schedule(struct ip_vs_service svc, const struct sk_buff skb,
bba54de5 JA	314	struct ip_vs_iphdr *iph)
1da177e4 LT	315	{
1da177e4 LT	316	struct ip_vs_dest *dest;
1acb7f67	317	struct ip_vs_sh_state *s;
eba3b5a7	318	__be16 port = 0;
1471f35e AG	319	const union nf_inet_addr *hash_addr;
	320
	321	hash_addr = ip_vs_iph_inverse(iph) ? &iph->daddr : &iph->saddr;
1da177e4 LT	322
	323	IP_VS_DBG(6, "ip_vs_sh_schedule(): Scheduling...\n");
	324
eba3b5a7 AF	325	if (svc->flags & IP_VS_SVC_F_SCHED_SH_PORT)
	326	port = ip_vs_sh_get_port(skb, iph);
	327
1acb7f67	328	s = (struct ip_vs_sh_state *) svc->sched_data;
eba3b5a7 AF	329
eba3b5a7 AF	330	if (svc->flags & IP_VS_SVC_F_SCHED_SH_FALLBACK)
1471f35e	331	dest = ip_vs_sh_get_fallback(svc, s, hash_addr, port);
eba3b5a7	332	else
1471f35e	333	dest = ip_vs_sh_get(svc, s, hash_addr, port);
eba3b5a7 AF	334
eba3b5a7 AF	335	if (!dest) {
41ac51ee	336	ip_vs_scheduler_err(svc, "no destination available");
1da177e4 LT	337	return NULL;
	338	}
	339
20971a0a	340	IP_VS_DBG_BUF(6, "SH: source IP address %s --> server %s:%d\n",
1471f35e	341	IP_VS_DBG_ADDR(svc->af, hash_addr),
4d316f3f	342	IP_VS_DBG_ADDR(dest->af, &dest->addr),
20971a0a	343	ntohs(dest->port));
1da177e4 LT	344
	345	return dest;
	346	}
	347
	348
	349	/*
	350	* IPVS SH Scheduler structure
	351	*/
	352	static struct ip_vs_scheduler ip_vs_sh_scheduler =
	353	{
	354	.name = "sh",
	355	.refcnt = ATOMIC_INIT(0),
	356	.module = THIS_MODULE,
d149ccc9	357	.n_list = LIST_HEAD_INIT(ip_vs_sh_scheduler.n_list),
1da177e4 LT	358	.init_service = ip_vs_sh_init_svc,
1da177e4 LT	359	.done_service = ip_vs_sh_done_svc,
1acb7f67 JA	360	.add_dest = ip_vs_sh_dest_changed,
	361	.del_dest = ip_vs_sh_dest_changed,
	362	.upd_dest = ip_vs_sh_dest_changed,
1da177e4 LT	363	.schedule = ip_vs_sh_schedule,
	364	};
	365
	366
	367	static int __init ip_vs_sh_init(void)
	368	{
1da177e4 LT	369	return register_ip_vs_scheduler(&ip_vs_sh_scheduler);
	370	}
	371
	372
	373	static void __exit ip_vs_sh_cleanup(void)
	374	{
	375	unregister_ip_vs_scheduler(&ip_vs_sh_scheduler);
ceec4c38	376	synchronize_rcu();
1da177e4 LT	377	}
	378
	379
	380	module_init(ip_vs_sh_init);
	381	module_exit(ip_vs_sh_cleanup);
	382	MODULE_LICENSE("GPL");