[deliverable/linux.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;
	int d;

	spin_lock_bh(&nft_rbtree_lock);
	parent = priv->root.rb_node;
	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;

	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;
			return 0;
		}
	}
	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;

	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);
	}
}

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;
16c45eda	40	const struct rb_node *parent;
20a69341 PM	41	int d;
20a69341 PM	42
7632667d	43	spin_lock_bh(&nft_rbtree_lock);
16c45eda	44	parent = priv->root.rb_node;
20a69341 PM	45	while (parent != NULL) {
	46	rbe = rb_entry(parent, struct nft_rbtree_elem, node);
	47
	48	d = nft_data_cmp(&rbe->key, key, set->klen);
	49	if (d < 0) {
	50	parent = parent->rb_left;
	51	interval = rbe;
	52	} else if (d > 0)
	53	parent = parent->rb_right;
	54	else {
	55	found:
	56	if (rbe->flags & NFT_SET_ELEM_INTERVAL_END)
	57	goto out;
	58	if (set->flags & NFT_SET_MAP)
	59	nft_data_copy(data, rbe->data);
7632667d PNA	60
7632667d PNA	61	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	62	return true;
	63	}
	64	}
	65
	66	if (set->flags & NFT_SET_INTERVAL && interval != NULL) {
	67	rbe = interval;
	68	goto found;
	69	}
	70	out:
7632667d	71	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	72	return false;
	73	}
	74
	75	static void nft_rbtree_elem_destroy(const struct nft_set *set,
	76	struct nft_rbtree_elem *rbe)
	77	{
	78	nft_data_uninit(&rbe->key, NFT_DATA_VALUE);
2fb91ddb PNA	79	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	80	!(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341	81	nft_data_uninit(rbe->data, set->dtype);
2fb91ddb	82
20a69341 PM	83	kfree(rbe);
	84	}
	85
	86	static int __nft_rbtree_insert(const struct nft_set *set,
	87	struct nft_rbtree_elem *new)
	88	{
	89	struct nft_rbtree *priv = nft_set_priv(set);
	90	struct nft_rbtree_elem *rbe;
	91	struct rb_node parent, *p;
	92	int d;
	93
	94	parent = NULL;
	95	p = &priv->root.rb_node;
	96	while (*p != NULL) {
	97	parent = *p;
	98	rbe = rb_entry(parent, struct nft_rbtree_elem, node);
	99	d = nft_data_cmp(&rbe->key, &new->key, set->klen);
	100	if (d < 0)
	101	p = &parent->rb_left;
	102	else if (d > 0)
	103	p = &parent->rb_right;
	104	else
	105	return -EEXIST;
	106	}
	107	rb_link_node(&new->node, parent, p);
	108	rb_insert_color(&new->node, &priv->root);
	109	return 0;
	110	}
	111
	112	static int nft_rbtree_insert(const struct nft_set *set,
	113	const struct nft_set_elem *elem)
	114	{
	115	struct nft_rbtree_elem *rbe;
	116	unsigned int size;
	117	int err;
	118
	119	size = sizeof(*rbe);
2fb91ddb PNA	120	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	121	!(elem->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341 PM	122	size += sizeof(rbe->data[0]);
	123
	124	rbe = kzalloc(size, GFP_KERNEL);
	125	if (rbe == NULL)
	126	return -ENOMEM;
	127
	128	rbe->flags = elem->flags;
	129	nft_data_copy(&rbe->key, &elem->key);
2fb91ddb PNA	130	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	131	!(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341 PM	132	nft_data_copy(rbe->data, &elem->data);
20a69341 PM	133
7632667d	134	spin_lock_bh(&nft_rbtree_lock);
20a69341 PM	135	err = __nft_rbtree_insert(set, rbe);
	136	if (err < 0)
	137	kfree(rbe);
7632667d PNA	138
7632667d PNA	139	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	140	return err;
	141	}
	142
	143	static void nft_rbtree_remove(const struct nft_set *set,
	144	const struct nft_set_elem *elem)
	145	{
	146	struct nft_rbtree *priv = nft_set_priv(set);
	147	struct nft_rbtree_elem *rbe = elem->cookie;
	148
7632667d	149	spin_lock_bh(&nft_rbtree_lock);
20a69341	150	rb_erase(&rbe->node, &priv->root);
7632667d	151	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	152	kfree(rbe);
	153	}
	154
	155	static int nft_rbtree_get(const struct nft_set set, struct nft_set_elem elem)
	156	{
	157	const struct nft_rbtree *priv = nft_set_priv(set);
	158	const struct rb_node *parent = priv->root.rb_node;
	159	struct nft_rbtree_elem *rbe;
	160	int d;
	161
	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);
	175	elem->flags = rbe->flags;
	176	return 0;
	177	}
	178	}
	179	return -ENOENT;
	180	}
	181
	182	static void nft_rbtree_walk(const struct nft_ctx *ctx,
	183	const struct nft_set *set,
	184	struct nft_set_iter *iter)
	185	{
	186	const struct nft_rbtree *priv = nft_set_priv(set);
	187	const struct nft_rbtree_elem *rbe;
	188	struct nft_set_elem elem;
	189	struct rb_node *node;
	190
7632667d	191	spin_lock_bh(&nft_rbtree_lock);
20a69341 PM	192	for (node = rb_first(&priv->root); node != NULL; node = rb_next(node)) {
	193	if (iter->count < iter->skip)
	194	goto cont;
	195
	196	rbe = rb_entry(node, struct nft_rbtree_elem, node);
	197	nft_data_copy(&elem.key, &rbe->key);
2fb91ddb PNA	198	if (set->flags & NFT_SET_MAP &&
2fb91ddb PNA	199	!(rbe->flags & NFT_SET_ELEM_INTERVAL_END))
20a69341 PM	200	nft_data_copy(&elem.data, rbe->data);
	201	elem.flags = rbe->flags;
	202
	203	iter->err = iter->fn(ctx, set, iter, &elem);
7632667d PNA	204	if (iter->err < 0) {
7632667d PNA	205	spin_unlock_bh(&nft_rbtree_lock);
20a69341	206	return;
7632667d	207	}
20a69341 PM	208	cont:
	209	iter->count++;
	210	}
7632667d	211	spin_unlock_bh(&nft_rbtree_lock);
20a69341 PM	212	}
	213
	214	static unsigned int nft_rbtree_privsize(const struct nlattr * const nla[])
	215	{
	216	return sizeof(struct nft_rbtree);
	217	}
	218
	219	static int nft_rbtree_init(const struct nft_set *set,
c50b960c	220	const struct nft_set_desc *desc,
20a69341 PM	221	const struct nlattr * const nla[])
	222	{
	223	struct nft_rbtree *priv = nft_set_priv(set);
	224
	225	priv->root = RB_ROOT;
	226	return 0;
	227	}
	228
	229	static void nft_rbtree_destroy(const struct nft_set *set)
	230	{
	231	struct nft_rbtree *priv = nft_set_priv(set);
	232	struct nft_rbtree_elem *rbe;
	233	struct rb_node *node;
	234
	235	while ((node = priv->root.rb_node) != NULL) {
	236	rb_erase(node, &priv->root);
	237	rbe = rb_entry(node, struct nft_rbtree_elem, node);
	238	nft_rbtree_elem_destroy(set, rbe);
	239	}
	240	}
	241
c50b960c PM	242	static bool nft_rbtree_estimate(const struct nft_set_desc *desc, u32 features,
	243	struct nft_set_estimate *est)
	244	{
	245	unsigned int nsize;
	246
	247	nsize = sizeof(struct nft_rbtree_elem);
	248	if (features & NFT_SET_MAP)
	249	nsize += FIELD_SIZEOF(struct nft_rbtree_elem, data[0]);
	250
	251	if (desc->size)
	252	est->size = sizeof(struct nft_rbtree) + desc->size * nsize;
	253	else
	254	est->size = nsize;
	255
	256	est->class = NFT_SET_CLASS_O_LOG_N;
	257
	258	return true;
	259	}
	260
20a69341 PM	261	static struct nft_set_ops nft_rbtree_ops __read_mostly = {
20a69341 PM	262	.privsize = nft_rbtree_privsize,
c50b960c	263	.estimate = nft_rbtree_estimate,
20a69341 PM	264	.init = nft_rbtree_init,
	265	.destroy = nft_rbtree_destroy,
	266	.insert = nft_rbtree_insert,
	267	.remove = nft_rbtree_remove,
	268	.get = nft_rbtree_get,
	269	.lookup = nft_rbtree_lookup,
	270	.walk = nft_rbtree_walk,
	271	.features = NFT_SET_INTERVAL \| NFT_SET_MAP,
	272	.owner = THIS_MODULE,
	273	};
	274
	275	static int __init nft_rbtree_module_init(void)
	276	{
	277	return nft_register_set(&nft_rbtree_ops);
	278	}
	279
	280	static void __exit nft_rbtree_module_exit(void)
	281	{
	282	nft_unregister_set(&nft_rbtree_ops);
	283	}
	284
	285	module_init(nft_rbtree_module_init);
	286	module_exit(nft_rbtree_module_exit);
	287
	288	MODULE_LICENSE("GPL");
	289	MODULE_AUTHOR("Patrick McHardy <kaber@trash.net>");
	290	MODULE_ALIAS_NFT_SET();