[linux-2.6-block.git] / net / netfilter / nft_rbtree.c

/*
 * Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 * Development of this code funded by Astaro AG (http://www.astaro.com/)
 */

#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/list.h>
#include <linux/rbtree.h>
#include <linux/netlink.h>
#include <linux/netfilter.h>
#include <linux/netfilter/nf_tables.h>
#include <net/netfilter/nf_tables.h>

static DEFINE_SPINLOCK(nft_rbtree_lock);

struct nft_rbtree {
	struct rb_root		root;
};

struct nft_rbtree_elem {
	struct rb_node		node;
	u16			flags;
	struct nft_data		key;
	struct nft_data		data[];
};

static bool nft_rbtree_lookup(const struct nft_set *set,
			      const struct nft_data *key,
			      struct nft_data *data)
{
	const struct nft_rbtree *priv = nft_set_priv(set);
	const struct nft_rbtree_elem *rbe, *interval = NULL;
	const struct rb_node *parent = priv->root.rb_node;
	int d;

	spin_lock_bh(&nft_rbtree_lock);
	while (parent != NULL) {
		rbe = rb_entry(parent, struct nft_rbtree_elem, node);

		d = nft_data_cmp(&rbe->key, key, set->klen);
		if (d < 0) {
			parent = parent->rb_left;
			interval = rbe;
		} else if (d > 0)
			parent = parent->rb_right;
		else {
found:
			if (rbe->flags & NFT_SET_ELEM_INTERVAL_END)
				goto out;
			if (set->flags & NFT_SET_MAP)
				nft_data_copy(data, rbe->data);

			spin_unlock_bh(&nft_rbtree_lock);
			return true;
		}
	}

	if (set->flags & NFT_SET_INTERVAL && interval != NULL) {
		rbe = interval;
		goto found;
	}
out:
	spin_unlock_bh(&nft_rbtree_lock);
	return false;
}

static void nft_rbtree_elem_destroy(const struct nft_set *set,
				    struct nft_rbtree_elem *rbe)
{
	nft_data_uninit(&rbe->key, NFT_DATA_VALUE);
	if (set->flags & NFT_SET_MAP &&
	    !(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
		nft_data_uninit(rbe->data, set->dtype);

	kfree(rbe);
}

static int __nft_rbtree_insert(const struct nft_set *set,
			       struct nft_rbtree_elem *new)
{
	struct nft_rbtree *priv = nft_set_priv(set);
	struct nft_rbtree_elem *rbe;
	struct rb_node *parent, **p;
	int d;

	parent = NULL;
	p = &priv->root.rb_node;
	while (*p != NULL) {
		parent = *p;
		rbe = rb_entry(parent, struct nft_rbtree_elem, node);
		d = nft_data_cmp(&rbe->key, &new->key, set->klen);
		if (d < 0)
			p = &parent->rb_left;
		else if (d > 0)
			p = &parent->rb_right;
		else
			return -EEXIST;
	}
	rb_link_node(&new->node, parent, p);
	rb_insert_color(&new->node, &priv->root);
	return 0;
}

static int nft_rbtree_insert(const struct nft_set *set,
			     const struct nft_set_elem *elem)
{
	struct nft_rbtree_elem *rbe;
	unsigned int size;
	int err;

	size = sizeof(*rbe);
	if (set->flags & NFT_SET_MAP &&
	    !(elem->flags & NFT_SET_ELEM_INTERVAL_END))
		size += sizeof(rbe->data[0]);

	rbe = kzalloc(size, GFP_KERNEL);
	if (rbe == NULL)
		return -ENOMEM;

	rbe->flags = elem->flags;
	nft_data_copy(&rbe->key, &elem->key);
	if (set->flags & NFT_SET_MAP &&
	    !(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
		nft_data_copy(rbe->data, &elem->data);

	spin_lock_bh(&nft_rbtree_lock);
	err = __nft_rbtree_insert(set, rbe);
	if (err < 0)
		kfree(rbe);

	spin_unlock_bh(&nft_rbtree_lock);
	return err;
}

static void nft_rbtree_remove(const struct nft_set *set,
			      const struct nft_set_elem *elem)
{
	struct nft_rbtree *priv = nft_set_priv(set);
	struct nft_rbtree_elem *rbe = elem->cookie;

	spin_lock_bh(&nft_rbtree_lock);
	rb_erase(&rbe->node, &priv->root);
	spin_unlock_bh(&nft_rbtree_lock);
	kfree(rbe);
}

static int nft_rbtree_get(const struct nft_set *set, struct nft_set_elem *elem)
{
	const struct nft_rbtree *priv = nft_set_priv(set);
	const struct rb_node *parent = priv->root.rb_node;
	struct nft_rbtree_elem *rbe;
	int d;

	spin_lock_bh(&nft_rbtree_lock);
	while (parent != NULL) {
		rbe = rb_entry(parent, struct nft_rbtree_elem, node);

		d = nft_data_cmp(&rbe->key, &elem->key, set->klen);
		if (d < 0)
			parent = parent->rb_left;
		else if (d > 0)
			parent = parent->rb_right;
		else {
			elem->cookie = rbe;
			if (set->flags & NFT_SET_MAP &&
			    !(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
				nft_data_copy(&elem->data, rbe->data);
			elem->flags = rbe->flags;
			spin_unlock_bh(&nft_rbtree_lock);
			return 0;
		}
	}
	spin_unlock_bh(&nft_rbtree_lock);
	return -ENOENT;
}

static void nft_rbtree_walk(const struct nft_ctx *ctx,
			    const struct nft_set *set,
			    struct nft_set_iter *iter)
{
	const struct nft_rbtree *priv = nft_set_priv(set);
	const struct nft_rbtree_elem *rbe;
	struct nft_set_elem elem;
	struct rb_node *node;

	spin_lock_bh(&nft_rbtree_lock);
	for (node = rb_first(&priv->root); node != NULL; node = rb_next(node)) {
		if (iter->count < iter->skip)
			goto cont;

		rbe = rb_entry(node, struct nft_rbtree_elem, node);
		nft_data_copy(&elem.key, &rbe->key);
		if (set->flags & NFT_SET_MAP &&
		    !(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
			nft_data_copy(&elem.data, rbe->data);
		elem.flags = rbe->flags;

		iter->err = iter->fn(ctx, set, iter, &elem);
		if (iter->err < 0) {
			spin_unlock_bh(&nft_rbtree_lock);
			return;
		}
cont:
		iter->count++;
	}
	spin_unlock_bh(&nft_rbtree_lock);
}

static unsigned int nft_rbtree_privsize(const struct nlattr * const nla[])
{
	return sizeof(struct nft_rbtree);
}

static int nft_rbtree_init(const struct nft_set *set,
			   const struct nft_set_desc *desc,
			   const struct nlattr * const nla[])
{
	struct nft_rbtree *priv = nft_set_priv(set);

	priv->root = RB_ROOT;
	return 0;
}

static void nft_rbtree_destroy(const struct nft_set *set)
{
	struct nft_rbtree *priv = nft_set_priv(set);
	struct nft_rbtree_elem *rbe;
	struct rb_node *node;

	spin_lock_bh(&nft_rbtree_lock);
	while ((node = priv->root.rb_node) != NULL) {
		rb_erase(node, &priv->root);
		rbe = rb_entry(node, struct nft_rbtree_elem, node);
		nft_rbtree_elem_destroy(set, rbe);
	}
	spin_unlock_bh(&nft_rbtree_lock);
}

static bool nft_rbtree_estimate(const struct nft_set_desc *desc, u32 features,
				struct nft_set_estimate *est)
{
	unsigned int nsize;

	nsize = sizeof(struct nft_rbtree_elem);
	if (features & NFT_SET_MAP)
		nsize += FIELD_SIZEOF(struct nft_rbtree_elem, data[0]);

	if (desc->size)
		est->size = sizeof(struct nft_rbtree) + desc->size * nsize;
	else
		est->size = nsize;

	est->class = NFT_SET_CLASS_O_LOG_N;

	return true;
}

static struct nft_set_ops nft_rbtree_ops __read_mostly = {
	.privsize	= nft_rbtree_privsize,
	.estimate	= nft_rbtree_estimate,
	.init		= nft_rbtree_init,
	.destroy	= nft_rbtree_destroy,
	.insert		= nft_rbtree_insert,
	.remove		= nft_rbtree_remove,
	.get		= nft_rbtree_get,
	.lookup		= nft_rbtree_lookup,
	.walk		= nft_rbtree_walk,
	.features	= NFT_SET_INTERVAL | NFT_SET_MAP,
	.owner		= THIS_MODULE,
};

static int __init nft_rbtree_module_init(void)
{
	return nft_register_set(&nft_rbtree_ops);
}

static void __exit nft_rbtree_module_exit(void)
{
	nft_unregister_set(&nft_rbtree_ops);
}

module_init(nft_rbtree_module_init);
module_exit(nft_rbtree_module_exit);

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
MODULE_ALIAS_NFT_SET();
Commit	Line	Data
20a69341 PM	1	/*
	2	* Copyright (c) 2008-2009 Patrick McHardy <kaber@trash.net>
	3	*
	4	* This program is free software; you can redistribute it and/or modify
	5	* it under the terms of the GNU General Public License version 2 as
	6	* published by the Free Software Foundation.
	7	*
	8	* Development of this code funded by Astaro AG (http://www.astaro.com/)
	9	*/
	10
	11	#include <linux/kernel.h>
	12	#include <linux/init.h>
	13	#include <linux/module.h>
	14	#include <linux/list.h>
	15	#include <linux/rbtree.h>
	16	#include <linux/netlink.h>
	17	#include <linux/netfilter.h>
	18	#include <linux/netfilter/nf_tables.h>
	19	#include <net/netfilter/nf_tables.h>
	20
7632667d PNA	21	static DEFINE_SPINLOCK(nft_rbtree_lock);
7632667d PNA	22
20a69341 PM	23	struct nft_rbtree {
	24	struct rb_root root;
	25	};
	26
	27	struct nft_rbtree_elem {
	28	struct rb_node node;
	29	u16 flags;
	30	struct nft_data key;
	31	struct nft_data data[];
	32	};
	33
	34	static bool nft_rbtree_lookup(const struct nft_set *set,
	35	const struct nft_data *key,
	36	struct nft_data *data)
	37	{
	38	const struct nft_rbtree *priv = nft_set_priv(set);
	39	const struct nft_rbtree_elem rbe, interval = NULL;
	40	const struct rb_node *parent = priv->root.rb_node;
	41	int d;
	42
7632667d	43	spin_lock_bh(&nft_rbtree_lock);
20a69341 PM	44	while (parent != NULL) {
	45	rbe = rb_entry(parent, struct nft_rbtree_elem, node);
	46
	47	d = nft_data_cmp(&rbe->key, key, set->klen);
	48	if (d < 0) {
	49	parent = parent->rb_left;
	50	interval = rbe;
	51	} else if (d > 0)
	52	parent = parent->rb_right;
	53	else {
	54	found:
	55	if (rbe->flags & NFT_SET_ELEM_INTERVAL_END)
	56	goto out;
	57	if (set->flags & NFT_SET_MAP)
	58	nft_data_copy(data, rbe->data);
7632667d PNA	59
7632667d PNA	60	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	61	return true;
	62	}
	63	}
	64
	65	if (set->flags & NFT_SET_INTERVAL && interval != NULL) {
	66	rbe = interval;
	67	goto found;
	68	}
	69	out:
7632667d	70	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	71	return false;
	72	}
	73
	74	static void nft_rbtree_elem_destroy(const struct nft_set *set,
	75	struct nft_rbtree_elem *rbe)
	76	{
	77	nft_data_uninit(&rbe->key, NFT_DATA_VALUE);
2fb91ddb PNA	78	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	79	!(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341	80	nft_data_uninit(rbe->data, set->dtype);
2fb91ddb	81
20a69341 PM	82	kfree(rbe);
	83	}
	84
	85	static int __nft_rbtree_insert(const struct nft_set *set,
	86	struct nft_rbtree_elem *new)
	87	{
	88	struct nft_rbtree *priv = nft_set_priv(set);
	89	struct nft_rbtree_elem *rbe;
	90	struct rb_node parent, *p;
	91	int d;
	92
	93	parent = NULL;
	94	p = &priv->root.rb_node;
	95	while (*p != NULL) {
	96	parent = *p;
	97	rbe = rb_entry(parent, struct nft_rbtree_elem, node);
	98	d = nft_data_cmp(&rbe->key, &new->key, set->klen);
	99	if (d < 0)
	100	p = &parent->rb_left;
	101	else if (d > 0)
	102	p = &parent->rb_right;
	103	else
	104	return -EEXIST;
	105	}
	106	rb_link_node(&new->node, parent, p);
	107	rb_insert_color(&new->node, &priv->root);
	108	return 0;
	109	}
	110
	111	static int nft_rbtree_insert(const struct nft_set *set,
	112	const struct nft_set_elem *elem)
	113	{
	114	struct nft_rbtree_elem *rbe;
	115	unsigned int size;
	116	int err;
	117
	118	size = sizeof(*rbe);
2fb91ddb PNA	119	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	120	!(elem->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341 PM	121	size += sizeof(rbe->data[0]);
	122
	123	rbe = kzalloc(size, GFP_KERNEL);
	124	if (rbe == NULL)
	125	return -ENOMEM;
	126
	127	rbe->flags = elem->flags;
	128	nft_data_copy(&rbe->key, &elem->key);
2fb91ddb PNA	129	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	130	!(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341 PM	131	nft_data_copy(rbe->data, &elem->data);
20a69341 PM	132
7632667d	133	spin_lock_bh(&nft_rbtree_lock);
20a69341 PM	134	err = __nft_rbtree_insert(set, rbe);
	135	if (err < 0)
	136	kfree(rbe);
7632667d PNA	137
7632667d PNA	138	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	139	return err;
	140	}
	141
	142	static void nft_rbtree_remove(const struct nft_set *set,
	143	const struct nft_set_elem *elem)
	144	{
	145	struct nft_rbtree *priv = nft_set_priv(set);
	146	struct nft_rbtree_elem *rbe = elem->cookie;
	147
7632667d	148	spin_lock_bh(&nft_rbtree_lock);
20a69341	149	rb_erase(&rbe->node, &priv->root);
7632667d	150	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	151	kfree(rbe);
	152	}
	153
	154	static int nft_rbtree_get(const struct nft_set set, struct nft_set_elem elem)
	155	{
	156	const struct nft_rbtree *priv = nft_set_priv(set);
	157	const struct rb_node *parent = priv->root.rb_node;
	158	struct nft_rbtree_elem *rbe;
	159	int d;
	160
7632667d	161	spin_lock_bh(&nft_rbtree_lock);
20a69341 PM	162	while (parent != NULL) {
	163	rbe = rb_entry(parent, struct nft_rbtree_elem, node);
	164
	165	d = nft_data_cmp(&rbe->key, &elem->key, set->klen);
	166	if (d < 0)
	167	parent = parent->rb_left;
	168	else if (d > 0)
	169	parent = parent->rb_right;
	170	else {
	171	elem->cookie = rbe;
2fb91ddb PNA	172	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	173	!(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341 PM	174	nft_data_copy(&elem->data, rbe->data);
20a69341 PM	175	elem->flags = rbe->flags;
7632667d	176	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	177	return 0;
	178	}
	179	}
7632667d	180	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	181	return -ENOENT;
	182	}
	183
	184	static void nft_rbtree_walk(const struct nft_ctx *ctx,
	185	const struct nft_set *set,
	186	struct nft_set_iter *iter)
	187	{
	188	const struct nft_rbtree *priv = nft_set_priv(set);
	189	const struct nft_rbtree_elem *rbe;
	190	struct nft_set_elem elem;
	191	struct rb_node *node;
	192
7632667d	193	spin_lock_bh(&nft_rbtree_lock);
20a69341 PM	194	for (node = rb_first(&priv->root); node != NULL; node = rb_next(node)) {
	195	if (iter->count < iter->skip)
	196	goto cont;
	197
	198	rbe = rb_entry(node, struct nft_rbtree_elem, node);
	199	nft_data_copy(&elem.key, &rbe->key);
2fb91ddb PNA	200	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	201	!(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341 PM	202	nft_data_copy(&elem.data, rbe->data);
	203	elem.flags = rbe->flags;
	204
	205	iter->err = iter->fn(ctx, set, iter, &elem);
7632667d PNA	206	if (iter->err < 0) {
7632667d PNA	207	spin_unlock_bh(&nft_rbtree_lock);
20a69341	208	return;
7632667d	209	}
20a69341 PM	210	cont:
	211	iter->count++;
	212	}
7632667d	213	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	214	}
	215
	216	static unsigned int nft_rbtree_privsize(const struct nlattr * const nla[])
	217	{
	218	return sizeof(struct nft_rbtree);
	219	}
	220
	221	static int nft_rbtree_init(const struct nft_set *set,
c50b960c	222	const struct nft_set_desc *desc,
20a69341 PM	223	const struct nlattr * const nla[])
	224	{
	225	struct nft_rbtree *priv = nft_set_priv(set);
	226
	227	priv->root = RB_ROOT;
	228	return 0;
	229	}
	230
	231	static void nft_rbtree_destroy(const struct nft_set *set)
	232	{
	233	struct nft_rbtree *priv = nft_set_priv(set);
	234	struct nft_rbtree_elem *rbe;
	235	struct rb_node *node;
	236
7632667d	237	spin_lock_bh(&nft_rbtree_lock);
20a69341 PM	238	while ((node = priv->root.rb_node) != NULL) {
	239	rb_erase(node, &priv->root);
	240	rbe = rb_entry(node, struct nft_rbtree_elem, node);
	241	nft_rbtree_elem_destroy(set, rbe);
	242	}
7632667d	243	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	244	}
20a69341 PM	245
c50b960c PM	246	static bool nft_rbtree_estimate(const struct nft_set_desc *desc, u32 features,
	247	struct nft_set_estimate *est)
	248	{
	249	unsigned int nsize;
	250
	251	nsize = sizeof(struct nft_rbtree_elem);
	252	if (features & NFT_SET_MAP)
	253	nsize += FIELD_SIZEOF(struct nft_rbtree_elem, data[0]);
	254
	255	if (desc->size)
	256	est->size = sizeof(struct nft_rbtree) + desc->size * nsize;
	257	else
	258	est->size = nsize;
	259
	260	est->class = NFT_SET_CLASS_O_LOG_N;
	261
	262	return true;
	263	}
	264
20a69341 PM	265	static struct nft_set_ops nft_rbtree_ops __read_mostly = {
20a69341 PM	266	.privsize = nft_rbtree_privsize,
c50b960c	267	.estimate = nft_rbtree_estimate,
20a69341 PM	268	.init = nft_rbtree_init,
	269	.destroy = nft_rbtree_destroy,
	270	.insert = nft_rbtree_insert,
	271	.remove = nft_rbtree_remove,
	272	.get = nft_rbtree_get,
	273	.lookup = nft_rbtree_lookup,
	274	.walk = nft_rbtree_walk,
	275	.features = NFT_SET_INTERVAL \| NFT_SET_MAP,
	276	.owner = THIS_MODULE,
	277	};
	278
	279	static int __init nft_rbtree_module_init(void)
	280	{
	281	return nft_register_set(&nft_rbtree_ops);
	282	}
	283
	284	static void __exit nft_rbtree_module_exit(void)
	285	{
	286	nft_unregister_set(&nft_rbtree_ops);
	287	}
	288
	289	module_init(nft_rbtree_module_init);
	290	module_exit(nft_rbtree_module_exit);
	291
	292	MODULE_LICENSE("GPL");
	293	MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
	294	MODULE_ALIAS_NFT_SET();