[GitHub/mt8127/android_kernel_alcatel_ttab.git] / net / netfilter / ipvs / ip_vs_dh.c

/*
 * IPVS:        Destination Hashing scheduling module
 *
 * Authors:     Wensong Zhang <wensong@gnuchina.org>
 *
 *              Inspired by the consistent hashing scheduler patch from
 *              Thomas Proell <proellt@gmx.de>
 *
 *              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 dh algorithm is to select server by the hash key of destination IP
 * address. The pseudo code is as follows:
 *
 *       n <- servernode[dest_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 destination IP address to the current server
 * array. If the dh 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.
 *
 */

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

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

#include <net/ip_vs.h>


/*
 *      IPVS DH bucket
 */
struct ip_vs_dh_bucket {
	struct ip_vs_dest       *dest;          /* real server (cache) */
};

/*
 *     for IPVS DH entry hash table
 */
#ifndef CONFIG_IP_VS_DH_TAB_BITS
#define CONFIG_IP_VS_DH_TAB_BITS        8
#endif
#define IP_VS_DH_TAB_BITS               CONFIG_IP_VS_DH_TAB_BITS
#define IP_VS_DH_TAB_SIZE               (1 << IP_VS_DH_TAB_BITS)
#define IP_VS_DH_TAB_MASK               (IP_VS_DH_TAB_SIZE - 1)


/*
 *	Returns hash value for IPVS DH entry
 */
static inline unsigned ip_vs_dh_hashkey(int af, const union nf_inet_addr *addr)
{
	__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 (ntohl(addr_fold)*2654435761UL) & IP_VS_DH_TAB_MASK;
}


/*
 *      Get ip_vs_dest associated with supplied parameters.
 */
static inline struct ip_vs_dest *
ip_vs_dh_get(int af, struct ip_vs_dh_bucket *tbl,
	     const union nf_inet_addr *addr)
{
	return (tbl[ip_vs_dh_hashkey(af, addr)]).dest;
}


/*
 *      Assign all the hash buckets of the specified table with the service.
 */
static int
ip_vs_dh_assign(struct ip_vs_dh_bucket *tbl, struct ip_vs_service *svc)
{
	int i;
	struct ip_vs_dh_bucket *b;
	struct list_head *p;
	struct ip_vs_dest *dest;

	b = tbl;
	p = &svc->destinations;
	for (i=0; i<IP_VS_DH_TAB_SIZE; i++) {
		if (list_empty(p)) {
			b->dest = NULL;
		} else {
			if (p == &svc->destinations)
				p = p->next;

			dest = list_entry(p, struct ip_vs_dest, n_list);
			atomic_inc(&dest->refcnt);
			b->dest = dest;

			p = p->next;
		}
		b++;
	}
	return 0;
}


/*
 *      Flush all the hash buckets of the specified table.
 */
static void ip_vs_dh_flush(struct ip_vs_dh_bucket *tbl)
{
	int i;
	struct ip_vs_dh_bucket *b;

	b = tbl;
	for (i=0; i<IP_VS_DH_TAB_SIZE; i++) {
		if (b->dest) {
			atomic_dec(&b->dest->refcnt);
			b->dest = NULL;
		}
		b++;
	}
}


static int ip_vs_dh_init_svc(struct ip_vs_service *svc)
{
	struct ip_vs_dh_bucket *tbl;

	/* allocate the DH table for this service */
	tbl = kmalloc(sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE,
		      GFP_ATOMIC);
	if (tbl == NULL) {
		IP_VS_ERR("ip_vs_dh_init_svc(): no memory\n");
		return -ENOMEM;
	}
	svc->sched_data = tbl;
	IP_VS_DBG(6, "DH hash table (memory=%Zdbytes) allocated for "
		  "current service\n",
		  sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);

	/* assign the hash buckets with the updated service */
	ip_vs_dh_assign(tbl, svc);

	return 0;
}


static int ip_vs_dh_done_svc(struct ip_vs_service *svc)
{
	struct ip_vs_dh_bucket *tbl = svc->sched_data;

	/* got to clean up hash buckets here */
	ip_vs_dh_flush(tbl);

	/* release the table itself */
	kfree(svc->sched_data);
	IP_VS_DBG(6, "DH hash table (memory=%Zdbytes) released\n",
		  sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);

	return 0;
}


static int ip_vs_dh_update_svc(struct ip_vs_service *svc)
{
	struct ip_vs_dh_bucket *tbl = svc->sched_data;

	/* got to clean up hash buckets here */
	ip_vs_dh_flush(tbl);

	/* assign the hash buckets with the updated service */
	ip_vs_dh_assign(tbl, svc);

	return 0;
}


/*
 *      If the dest flags is set with IP_VS_DEST_F_OVERLOAD,
 *      consider that the server is overloaded here.
 */
static inline int is_overloaded(struct ip_vs_dest *dest)
{
	return dest->flags & IP_VS_DEST_F_OVERLOAD;
}


/*
 *      Destination hashing scheduling
 */
static struct ip_vs_dest *
ip_vs_dh_schedule(struct ip_vs_service *svc, const struct sk_buff *skb)
{
	struct ip_vs_dest *dest;
	struct ip_vs_dh_bucket *tbl;
	struct ip_vs_iphdr iph;

	ip_vs_fill_iphdr(svc->af, skb_network_header(skb), &iph);

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

	tbl = (struct ip_vs_dh_bucket *)svc->sched_data;
	dest = ip_vs_dh_get(svc->af, tbl, &iph.daddr);
	if (!dest
	    || !(dest->flags & IP_VS_DEST_F_AVAILABLE)
	    || atomic_read(&dest->weight) <= 0
	    || is_overloaded(dest)) {
		return NULL;
	}

	IP_VS_DBG_BUF(6, "DH: destination IP address %s --> server %s:%d\n",
		      IP_VS_DBG_ADDR(svc->af, &iph.daddr),
		      IP_VS_DBG_ADDR(svc->af, &dest->addr),
		      ntohs(dest->port));

	return dest;
}


/*
 *      IPVS DH Scheduler structure
 */
static struct ip_vs_scheduler ip_vs_dh_scheduler =
{
	.name =			"dh",
	.refcnt =		ATOMIC_INIT(0),
	.module =		THIS_MODULE,
	.n_list =		LIST_HEAD_INIT(ip_vs_dh_scheduler.n_list),
	.init_service =		ip_vs_dh_init_svc,
	.done_service =		ip_vs_dh_done_svc,
	.update_service =	ip_vs_dh_update_svc,
	.schedule =		ip_vs_dh_schedule,
};


static int __init ip_vs_dh_init(void)
{
	return register_ip_vs_scheduler(&ip_vs_dh_scheduler);
}


static void __exit ip_vs_dh_cleanup(void)
{
	unregister_ip_vs_scheduler(&ip_vs_dh_scheduler);
}


module_init(ip_vs_dh_init);
module_exit(ip_vs_dh_cleanup);
MODULE_LICENSE("GPL");
Commit	Line	Data
cb7f6a7b JV	1	/*
	2	* IPVS: Destination Hashing scheduling module
	3	*
	4	* Authors: Wensong Zhang <wensong@gnuchina.org>
	5	*
	6	* Inspired by the consistent hashing scheduler patch from
	7	* Thomas Proell <proellt@gmx.de>
	8	*
	9	* This program is free software; you can redistribute it and/or
	10	* modify it under the terms of the GNU General Public License
	11	* as published by the Free Software Foundation; either version
	12	* 2 of the License, or (at your option) any later version.
	13	*
	14	* Changes:
	15	*
	16	*/
	17
	18	/*
	19	* The dh algorithm is to select server by the hash key of destination IP
	20	* address. The pseudo code is as follows:
	21	*
	22	* n <- servernode[dest_ip];
	23	* if (n is dead) OR
	24	* (n is overloaded) OR (n.weight <= 0) then
	25	* return NULL;
	26	*
	27	* return n;
	28	*
	29	* Notes that servernode is a 256-bucket hash table that maps the hash
	30	* index derived from packet destination IP address to the current server
	31	* array. If the dh scheduler is used in cache cluster, it is good to
	32	* combine it with cache_bypass feature. When the statically assigned
	33	* server is dead or overloaded, the load balancer can bypass the cache
	34	* server and send requests to the original server directly.
	35	*
	36	*/
	37
9aada7ac HE	38	#define KMSG_COMPONENT "IPVS"
	39	#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
	40
cb7f6a7b JV	41	#include <linux/ip.h>
	42	#include <linux/module.h>
	43	#include <linux/kernel.h>
	44	#include <linux/skbuff.h>
	45
	46	#include <net/ip_vs.h>
	47
	48
	49	/*
	50	* IPVS DH bucket
	51	*/
	52	struct ip_vs_dh_bucket {
	53	struct ip_vs_dest dest; / real server (cache) */
	54	};
	55
	56	/*
	57	* for IPVS DH entry hash table
	58	*/
	59	#ifndef CONFIG_IP_VS_DH_TAB_BITS
	60	#define CONFIG_IP_VS_DH_TAB_BITS 8
	61	#endif
	62	#define IP_VS_DH_TAB_BITS CONFIG_IP_VS_DH_TAB_BITS
	63	#define IP_VS_DH_TAB_SIZE (1 << IP_VS_DH_TAB_BITS)
	64	#define IP_VS_DH_TAB_MASK (IP_VS_DH_TAB_SIZE - 1)
	65
	66
	67	/*
	68	* Returns hash value for IPVS DH entry
	69	*/
20971a0a	70	static inline unsigned ip_vs_dh_hashkey(int af, const union nf_inet_addr *addr)
cb7f6a7b	71	{
20971a0a JV	72	__be32 addr_fold = addr->ip;
	73
	74	#ifdef CONFIG_IP_VS_IPV6
	75	if (af == AF_INET6)
	76	addr_fold = addr->ip6[0]^addr->ip6[1]^
	77	addr->ip6[2]^addr->ip6[3];
	78	#endif
	79	return (ntohl(addr_fold)*2654435761UL) & IP_VS_DH_TAB_MASK;
cb7f6a7b JV	80	}
	81
	82
	83	/*
	84	* Get ip_vs_dest associated with supplied parameters.
	85	*/
	86	static inline struct ip_vs_dest *
20971a0a JV	87	ip_vs_dh_get(int af, struct ip_vs_dh_bucket *tbl,
20971a0a JV	88	const union nf_inet_addr *addr)
cb7f6a7b	89	{
20971a0a	90	return (tbl[ip_vs_dh_hashkey(af, addr)]).dest;
cb7f6a7b JV	91	}
	92
	93
	94	/*
	95	* Assign all the hash buckets of the specified table with the service.
	96	*/
	97	static int
	98	ip_vs_dh_assign(struct ip_vs_dh_bucket tbl, struct ip_vs_service svc)
	99	{
	100	int i;
	101	struct ip_vs_dh_bucket *b;
	102	struct list_head *p;
	103	struct ip_vs_dest *dest;
	104
	105	b = tbl;
	106	p = &svc->destinations;
	107	for (i=0; i<IP_VS_DH_TAB_SIZE; i++) {
	108	if (list_empty(p)) {
	109	b->dest = NULL;
	110	} else {
	111	if (p == &svc->destinations)
	112	p = p->next;
	113
	114	dest = list_entry(p, struct ip_vs_dest, n_list);
	115	atomic_inc(&dest->refcnt);
	116	b->dest = dest;
	117
	118	p = p->next;
	119	}
	120	b++;
	121	}
	122	return 0;
	123	}
	124
	125
	126	/*
	127	* Flush all the hash buckets of the specified table.
	128	*/
	129	static void ip_vs_dh_flush(struct ip_vs_dh_bucket *tbl)
	130	{
	131	int i;
	132	struct ip_vs_dh_bucket *b;
	133
	134	b = tbl;
	135	for (i=0; i<IP_VS_DH_TAB_SIZE; i++) {
	136	if (b->dest) {
	137	atomic_dec(&b->dest->refcnt);
	138	b->dest = NULL;
	139	}
	140	b++;
	141	}
	142	}
	143
	144
	145	static int ip_vs_dh_init_svc(struct ip_vs_service *svc)
	146	{
	147	struct ip_vs_dh_bucket *tbl;
	148
	149	/* allocate the DH table for this service */
	150	tbl = kmalloc(sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE,
	151	GFP_ATOMIC);
	152	if (tbl == NULL) {
	153	IP_VS_ERR("ip_vs_dh_init_svc(): no memory\n");
	154	return -ENOMEM;
155	}
156	svc->sched_data = tbl;
157	IP_VS_DBG(6, "DH hash table (memory=%Zdbytes) allocated for "
158	"current service\n",
159	sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);
160
161	/* assign the hash buckets with the updated service */
162	ip_vs_dh_assign(tbl, svc);
163
164	return 0;
165	}
166
167
168	static int ip_vs_dh_done_svc(struct ip_vs_service *svc)
169	{
170	struct ip_vs_dh_bucket *tbl = svc->sched_data;
171
172	/* got to clean up hash buckets here */
173	ip_vs_dh_flush(tbl);
174
175	/* release the table itself */
176	kfree(svc->sched_data);
177	IP_VS_DBG(6, "DH hash table (memory=%Zdbytes) released\n",
178	sizeof(struct ip_vs_dh_bucket)*IP_VS_DH_TAB_SIZE);
179
180	return 0;
181	}
182
183
184	static int ip_vs_dh_update_svc(struct ip_vs_service *svc)
185	{
186	struct ip_vs_dh_bucket *tbl = svc->sched_data;
187
188	/* got to clean up hash buckets here */
189	ip_vs_dh_flush(tbl);
190
191	/* assign the hash buckets with the updated service */
192	ip_vs_dh_assign(tbl, svc);
193
194	return 0;
195	}
196
197
198	/*
199	* If the dest flags is set with IP_VS_DEST_F_OVERLOAD,
200	* consider that the server is overloaded here.
201	*/
202	static inline int is_overloaded(struct ip_vs_dest *dest)
203	{
204	return dest->flags & IP_VS_DEST_F_OVERLOAD;
205	}
206
207
208	/*
209	* Destination hashing scheduling
210	*/
211	static struct ip_vs_dest *
212	ip_vs_dh_schedule(struct ip_vs_service svc, const struct sk_buff skb)
213	{
214	struct ip_vs_dest *dest;
215	struct ip_vs_dh_bucket *tbl;
20971a0a JV	216	struct ip_vs_iphdr iph;
	217
	218	ip_vs_fill_iphdr(svc->af, skb_network_header(skb), &iph);
cb7f6a7b JV	219
	220	IP_VS_DBG(6, "ip_vs_dh_schedule(): Scheduling...\n");
	221
	222	tbl = (struct ip_vs_dh_bucket *)svc->sched_data;
20971a0a	223	dest = ip_vs_dh_get(svc->af, tbl, &iph.daddr);
cb7f6a7b JV	224	if (!dest
	225	\|\| !(dest->flags & IP_VS_DEST_F_AVAILABLE)
	226	\|\| atomic_read(&dest->weight) <= 0
	227	\|\| is_overloaded(dest)) {
	228	return NULL;
	229	}
	230
20971a0a JV	231	IP_VS_DBG_BUF(6, "DH: destination IP address %s --> server %s:%d\n",
	232	IP_VS_DBG_ADDR(svc->af, &iph.daddr),
	233	IP_VS_DBG_ADDR(svc->af, &dest->addr),
	234	ntohs(dest->port));
cb7f6a7b JV	235
	236	return dest;
	237	}
	238
	239
	240	/*
	241	* IPVS DH Scheduler structure
	242	*/
	243	static struct ip_vs_scheduler ip_vs_dh_scheduler =
	244	{
	245	.name = "dh",
	246	.refcnt = ATOMIC_INIT(0),
	247	.module = THIS_MODULE,
	248	.n_list = LIST_HEAD_INIT(ip_vs_dh_scheduler.n_list),
cb7f6a7b JV	249	.init_service = ip_vs_dh_init_svc,
	250	.done_service = ip_vs_dh_done_svc,
	251	.update_service = ip_vs_dh_update_svc,
	252	.schedule = ip_vs_dh_schedule,
	253	};
	254
	255
	256	static int __init ip_vs_dh_init(void)
	257	{
	258	return register_ip_vs_scheduler(&ip_vs_dh_scheduler);
	259	}
	260
	261
	262	static void __exit ip_vs_dh_cleanup(void)
	263	{
	264	unregister_ip_vs_scheduler(&ip_vs_dh_scheduler);
	265	}
	266
	267
	268	module_init(ip_vs_dh_init);
	269	module_exit(ip_vs_dh_cleanup);
	270	MODULE_LICENSE("GPL");