[mirror_ubuntu-bionic-kernel.git] / net / netfilter / xt_connlimit.c

/*
 * netfilter module to limit the number of parallel tcp
 * connections per IP address.
 *   (c) 2000 Gerd Knorr <kraxel@bytesex.org>
 *   Nov 2002: Martin Bene <martin.bene@icomedias.com>:
 *		only ignore TIME_WAIT or gone connections
 *   (C) CC Computer Consultants GmbH, 2007
 *
 * based on ...
 *
 * Kernel module to match connection tracking information.
 * GPL (C) 1999  Rusty Russell (rusty@rustcorp.com.au).
 */
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/in.h>
#include <linux/in6.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/jhash.h>
#include <linux/slab.h>
#include <linux/list.h>
#include <linux/rbtree.h>
#include <linux/module.h>
#include <linux/random.h>
#include <linux/skbuff.h>
#include <linux/spinlock.h>
#include <linux/netfilter/nf_conntrack_tcp.h>
#include <linux/netfilter/x_tables.h>
#include <linux/netfilter/xt_connlimit.h>
#include <net/netfilter/nf_conntrack.h>
#include <net/netfilter/nf_conntrack_core.h>
#include <net/netfilter/nf_conntrack_tuple.h>
#include <net/netfilter/nf_conntrack_zones.h>

#define CONNLIMIT_SLOTS		256U

#ifdef CONFIG_LOCKDEP
#define CONNLIMIT_LOCK_SLOTS	8U
#else
#define CONNLIMIT_LOCK_SLOTS	256U
#endif

#define CONNLIMIT_GC_MAX_NODES	8

/* we will save the tuples of all connections we care about */
struct xt_connlimit_conn {
	struct hlist_node		node;
	struct nf_conntrack_tuple	tuple;
};

struct xt_connlimit_rb {
	struct rb_node node;
	struct hlist_head hhead; /* connections/hosts in same subnet */
	union nf_inet_addr addr; /* search key */
};

static spinlock_t xt_connlimit_locks[CONNLIMIT_LOCK_SLOTS] __cacheline_aligned_in_smp;

struct xt_connlimit_data {
	struct rb_root climit_root[CONNLIMIT_SLOTS];
};

static u_int32_t connlimit_rnd __read_mostly;
static struct kmem_cache *connlimit_rb_cachep __read_mostly;
static struct kmem_cache *connlimit_conn_cachep __read_mostly;

static inline unsigned int connlimit_iphash(__be32 addr)
{
	return jhash_1word((__force __u32)addr,
			    connlimit_rnd) % CONNLIMIT_SLOTS;
}

static inline unsigned int
connlimit_iphash6(const union nf_inet_addr *addr)
{
	return jhash2((u32 *)addr->ip6, ARRAY_SIZE(addr->ip6),
		       connlimit_rnd) % CONNLIMIT_SLOTS;
}

static inline bool already_closed(const struct nf_conn *conn)
{
	if (nf_ct_protonum(conn) == IPPROTO_TCP)
		return conn->proto.tcp.state == TCP_CONNTRACK_TIME_WAIT ||
		       conn->proto.tcp.state == TCP_CONNTRACK_CLOSE;
	else
		return 0;
}

static int
same_source(const union nf_inet_addr *addr,
	    const union nf_inet_addr *u3, u_int8_t family)
{
	if (family == NFPROTO_IPV4)
		return ntohl(addr->ip) - ntohl(u3->ip);

	return memcmp(addr->ip6, u3->ip6, sizeof(addr->ip6));
}

static bool add_hlist(struct hlist_head *head,
		      const struct nf_conntrack_tuple *tuple,
		      const union nf_inet_addr *addr)
{
	struct xt_connlimit_conn *conn;

	conn = kmem_cache_alloc(connlimit_conn_cachep, GFP_ATOMIC);
	if (conn == NULL)
		return false;
	conn->tuple = *tuple;
	hlist_add_head(&conn->node, head);
	return true;
}

static unsigned int check_hlist(struct net *net,
				struct hlist_head *head,
				const struct nf_conntrack_tuple *tuple,
				const struct nf_conntrack_zone *zone,
				bool *addit)
{
	const struct nf_conntrack_tuple_hash *found;
	struct xt_connlimit_conn *conn;
	struct hlist_node *n;
	struct nf_conn *found_ct;
	unsigned int length = 0;

	*addit = true;

	/* check the saved connections */
	hlist_for_each_entry_safe(conn, n, head, node) {
		found = nf_conntrack_find_get(net, zone, &conn->tuple);
		if (found == NULL) {
			hlist_del(&conn->node);
			kmem_cache_free(connlimit_conn_cachep, conn);
			continue;
		}

		found_ct = nf_ct_tuplehash_to_ctrack(found);

		if (nf_ct_tuple_equal(&conn->tuple, tuple)) {
			/*
			 * Just to be sure we have it only once in the list.
			 * We should not see tuples twice unless someone hooks
			 * this into a table without "-p tcp --syn".
			 */
			*addit = false;
		} else if (already_closed(found_ct)) {
			/*
			 * we do not care about connections which are
			 * closed already -> ditch it
			 */
			nf_ct_put(found_ct);
			hlist_del(&conn->node);
			kmem_cache_free(connlimit_conn_cachep, conn);
			continue;
		}

		nf_ct_put(found_ct);
		length++;
	}

	return length;
}

static void tree_nodes_free(struct rb_root *root,
			    struct xt_connlimit_rb *gc_nodes[],
			    unsigned int gc_count)
{
	struct xt_connlimit_rb *rbconn;

	while (gc_count) {
		rbconn = gc_nodes[--gc_count];
		rb_erase(&rbconn->node, root);
		kmem_cache_free(connlimit_rb_cachep, rbconn);
	}
}

static unsigned int
count_tree(struct net *net, struct rb_root *root,
	   const struct nf_conntrack_tuple *tuple,
	   const union nf_inet_addr *addr,
	   u8 family, const struct nf_conntrack_zone *zone)
{
	struct xt_connlimit_rb *gc_nodes[CONNLIMIT_GC_MAX_NODES];
	struct rb_node **rbnode, *parent;
	struct xt_connlimit_rb *rbconn;
	struct xt_connlimit_conn *conn;
	unsigned int gc_count;
	bool no_gc = false;

 restart:
	gc_count = 0;
	parent = NULL;
	rbnode = &(root->rb_node);
	while (*rbnode) {
		int diff;
		bool addit;

		rbconn = rb_entry(*rbnode, struct xt_connlimit_rb, node);

		parent = *rbnode;
		diff = same_source(addr, &rbconn->addr, family);
		if (diff < 0) {
			rbnode = &((*rbnode)->rb_left);
		} else if (diff > 0) {
			rbnode = &((*rbnode)->rb_right);
		} else {
			/* same source network -> be counted! */
			unsigned int count;
			count = check_hlist(net, &rbconn->hhead, tuple, zone, &addit);

			tree_nodes_free(root, gc_nodes, gc_count);
			if (!addit)
				return count;

			if (!add_hlist(&rbconn->hhead, tuple, addr))
				return 0; /* hotdrop */

			return count + 1;
		}

		if (no_gc || gc_count >= ARRAY_SIZE(gc_nodes))
			continue;

		/* only used for GC on hhead, retval and 'addit' ignored */
		check_hlist(net, &rbconn->hhead, tuple, zone, &addit);
		if (hlist_empty(&rbconn->hhead))
			gc_nodes[gc_count++] = rbconn;
	}

	if (gc_count) {
		no_gc = true;
		tree_nodes_free(root, gc_nodes, gc_count);
		/* tree_node_free before new allocation permits
		 * allocator to re-use newly free'd object.
		 *
		 * This is a rare event; in most cases we will find
		 * existing node to re-use. (or gc_count is 0).
		 */
		goto restart;
	}

	/* no match, need to insert new node */
	rbconn = kmem_cache_alloc(connlimit_rb_cachep, GFP_ATOMIC);
	if (rbconn == NULL)
		return 0;

	conn = kmem_cache_alloc(connlimit_conn_cachep, GFP_ATOMIC);
	if (conn == NULL) {
		kmem_cache_free(connlimit_rb_cachep, rbconn);
		return 0;
	}

	conn->tuple = *tuple;
	rbconn->addr = *addr;

	INIT_HLIST_HEAD(&rbconn->hhead);
	hlist_add_head(&conn->node, &rbconn->hhead);

	rb_link_node(&rbconn->node, parent, rbnode);
	rb_insert_color(&rbconn->node, root);
	return 1;
}

static int count_them(struct net *net,
		      struct xt_connlimit_data *data,
		      const struct nf_conntrack_tuple *tuple,
		      const union nf_inet_addr *addr,
		      u_int8_t family,
		      const struct nf_conntrack_zone *zone)
{
	struct rb_root *root;
	int count;
	u32 hash;

	if (family == NFPROTO_IPV6)
		hash = connlimit_iphash6(addr);
	else
		hash = connlimit_iphash(addr->ip);
	root = &data->climit_root[hash];

	spin_lock_bh(&xt_connlimit_locks[hash % CONNLIMIT_LOCK_SLOTS]);

	count = count_tree(net, root, tuple, addr, family, zone);

	spin_unlock_bh(&xt_connlimit_locks[hash % CONNLIMIT_LOCK_SLOTS]);

	return count;
}

static bool
connlimit_mt(const struct sk_buff *skb, struct xt_action_param *par)
{
	struct net *net = xt_net(par);
	const struct xt_connlimit_info *info = par->matchinfo;
	union nf_inet_addr addr;
	struct nf_conntrack_tuple tuple;
	const struct nf_conntrack_tuple *tuple_ptr = &tuple;
	const struct nf_conntrack_zone *zone = &nf_ct_zone_dflt;
	enum ip_conntrack_info ctinfo;
	const struct nf_conn *ct;
	unsigned int connections;

	ct = nf_ct_get(skb, &ctinfo);
	if (ct != NULL) {
		tuple_ptr = &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple;
		zone = nf_ct_zone(ct);
	} else if (!nf_ct_get_tuplepr(skb, skb_network_offset(skb),
				      xt_family(par), net, &tuple)) {
		goto hotdrop;
	}

	if (xt_family(par) == NFPROTO_IPV6) {
		const struct ipv6hdr *iph = ipv6_hdr(skb);
		unsigned int i;

		memcpy(&addr.ip6, (info->flags & XT_CONNLIMIT_DADDR) ?
		       &iph->daddr : &iph->saddr, sizeof(addr.ip6));

		for (i = 0; i < ARRAY_SIZE(addr.ip6); ++i)
			addr.ip6[i] &= info->mask.ip6[i];
	} else {
		const struct iphdr *iph = ip_hdr(skb);
		addr.ip = (info->flags & XT_CONNLIMIT_DADDR) ?
			  iph->daddr : iph->saddr;

		addr.ip &= info->mask.ip;
	}

	connections = count_them(net, info->data, tuple_ptr, &addr,
				 xt_family(par), zone);
	if (connections == 0)
		/* kmalloc failed, drop it entirely */
		goto hotdrop;

	return (connections > info->limit) ^
	       !!(info->flags & XT_CONNLIMIT_INVERT);

 hotdrop:
	par->hotdrop = true;
	return false;
}

static int connlimit_mt_check(const struct xt_mtchk_param *par)
{
	struct xt_connlimit_info *info = par->matchinfo;
	unsigned int i;
	int ret;

	net_get_random_once(&connlimit_rnd, sizeof(connlimit_rnd));

	ret = nf_ct_netns_get(par->net, par->family);
	if (ret < 0) {
		pr_info("cannot load conntrack support for "
			"address family %u\n", par->family);
		return ret;
	}

	/* init private data */
	info->data = kmalloc(sizeof(struct xt_connlimit_data), GFP_KERNEL);
	if (info->data == NULL) {
		nf_ct_netns_put(par->net, par->family);
		return -ENOMEM;
	}

	for (i = 0; i < ARRAY_SIZE(info->data->climit_root); ++i)
		info->data->climit_root[i] = RB_ROOT;

	return 0;
}

static void destroy_tree(struct rb_root *r)
{
	struct xt_connlimit_conn *conn;
	struct xt_connlimit_rb *rbconn;
	struct hlist_node *n;
	struct rb_node *node;

	while ((node = rb_first(r)) != NULL) {
		rbconn = rb_entry(node, struct xt_connlimit_rb, node);

		rb_erase(node, r);

		hlist_for_each_entry_safe(conn, n, &rbconn->hhead, node)
			kmem_cache_free(connlimit_conn_cachep, conn);

		kmem_cache_free(connlimit_rb_cachep, rbconn);
	}
}

static void connlimit_mt_destroy(const struct xt_mtdtor_param *par)
{
	const struct xt_connlimit_info *info = par->matchinfo;
	unsigned int i;

	nf_ct_netns_put(par->net, par->family);

	for (i = 0; i < ARRAY_SIZE(info->data->climit_root); ++i)
		destroy_tree(&info->data->climit_root[i]);

	kfree(info->data);
}

static struct xt_match connlimit_mt_reg __read_mostly = {
	.name       = "connlimit",
	.revision   = 1,
	.family     = NFPROTO_UNSPEC,
	.checkentry = connlimit_mt_check,
	.match      = connlimit_mt,
	.matchsize  = sizeof(struct xt_connlimit_info),
	.usersize   = offsetof(struct xt_connlimit_info, data),
	.destroy    = connlimit_mt_destroy,
	.me         = THIS_MODULE,
};

static int __init connlimit_mt_init(void)
{
	int ret, i;

	BUILD_BUG_ON(CONNLIMIT_LOCK_SLOTS > CONNLIMIT_SLOTS);
	BUILD_BUG_ON((CONNLIMIT_SLOTS % CONNLIMIT_LOCK_SLOTS) != 0);

	for (i = 0; i < CONNLIMIT_LOCK_SLOTS; ++i)
		spin_lock_init(&xt_connlimit_locks[i]);

	connlimit_conn_cachep = kmem_cache_create("xt_connlimit_conn",
					   sizeof(struct xt_connlimit_conn),
					   0, 0, NULL);
	if (!connlimit_conn_cachep)
		return -ENOMEM;

	connlimit_rb_cachep = kmem_cache_create("xt_connlimit_rb",
					   sizeof(struct xt_connlimit_rb),
					   0, 0, NULL);
	if (!connlimit_rb_cachep) {
		kmem_cache_destroy(connlimit_conn_cachep);
		return -ENOMEM;
	}
	ret = xt_register_match(&connlimit_mt_reg);
	if (ret != 0) {
		kmem_cache_destroy(connlimit_conn_cachep);
		kmem_cache_destroy(connlimit_rb_cachep);
	}
	return ret;
}

static void __exit connlimit_mt_exit(void)
{
	xt_unregister_match(&connlimit_mt_reg);
	kmem_cache_destroy(connlimit_conn_cachep);
	kmem_cache_destroy(connlimit_rb_cachep);
}

module_init(connlimit_mt_init);
module_exit(connlimit_mt_exit);
MODULE_AUTHOR("Jan Engelhardt <jengelh@medozas.de>");
MODULE_DESCRIPTION("Xtables: Number of connections matching");
MODULE_LICENSE("GPL");
MODULE_ALIAS("ipt_connlimit");
MODULE_ALIAS("ip6t_connlimit");
Commit	Line	Data
370786f9 JE	1	/*
	2	* netfilter module to limit the number of parallel tcp
	3	* connections per IP address.
	4	* (c) 2000 Gerd Knorr <kraxel@bytesex.org>
	5	* Nov 2002: Martin Bene <martin.bene@icomedias.com>:
	6	* only ignore TIME_WAIT or gone connections
ba5dc275	7	* (C) CC Computer Consultants GmbH, 2007
370786f9 JE	8	*
	9	* based on ...
	10	*
	11	* Kernel module to match connection tracking information.
	12	* GPL (C) 1999 Rusty Russell (rusty@rustcorp.com.au).
	13	*/
8bee4bad	14	#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
370786f9 JE	15	#include <linux/in.h>
	16	#include <linux/in6.h>
	17	#include <linux/ip.h>
	18	#include <linux/ipv6.h>
	19	#include <linux/jhash.h>
5a0e3ad6	20	#include <linux/slab.h>
370786f9	21	#include <linux/list.h>
7d084877	22	#include <linux/rbtree.h>
370786f9 JE	23	#include <linux/module.h>
	24	#include <linux/random.h>
	25	#include <linux/skbuff.h>
	26	#include <linux/spinlock.h>
	27	#include <linux/netfilter/nf_conntrack_tcp.h>
	28	#include <linux/netfilter/x_tables.h>
	29	#include <linux/netfilter/xt_connlimit.h>
	30	#include <net/netfilter/nf_conntrack.h>
	31	#include <net/netfilter/nf_conntrack_core.h>
	32	#include <net/netfilter/nf_conntrack_tuple.h>
5d0aa2cc	33	#include <net/netfilter/nf_conntrack_zones.h>
370786f9	34
e00b437b FW	35	#define CONNLIMIT_SLOTS 256U
	36
	37	#ifdef CONFIG_LOCKDEP
	38	#define CONNLIMIT_LOCK_SLOTS 8U
	39	#else
	40	#define CONNLIMIT_LOCK_SLOTS 256U
	41	#endif
	42
7d084877	43	#define CONNLIMIT_GC_MAX_NODES 8
1442e750	44
370786f9 JE	45	/* we will save the tuples of all connections we care about */
370786f9 JE	46	struct xt_connlimit_conn {
3e0d5149	47	struct hlist_node node;
8183e3a8	48	struct nf_conntrack_tuple tuple;
370786f9 JE	49	};
370786f9 JE	50
7d084877 FW	51	struct xt_connlimit_rb {
	52	struct rb_node node;
	53	struct hlist_head hhead; /* connections/hosts in same subnet */
	54	union nf_inet_addr addr; /* search key */
	55	};
	56
e00b437b FW	57	static spinlock_t xt_connlimit_locks[CONNLIMIT_LOCK_SLOTS] __cacheline_aligned_in_smp;
e00b437b FW	58
370786f9	59	struct xt_connlimit_data {
a2acc543	60	struct rb_root climit_root[CONNLIMIT_SLOTS];
370786f9 JE	61	};
370786f9 JE	62
294188ae	63	static u_int32_t connlimit_rnd __read_mostly;
7d084877	64	static struct kmem_cache *connlimit_rb_cachep __read_mostly;
14e1a977	65	static struct kmem_cache *connlimit_conn_cachep __read_mostly;
370786f9	66
a34c4589	67	static inline unsigned int connlimit_iphash(__be32 addr)
370786f9	68	{
1442e750 FW	69	return jhash_1word((__force __u32)addr,
1442e750 FW	70	connlimit_rnd) % CONNLIMIT_SLOTS;
370786f9 JE	71	}
	72
	73	static inline unsigned int
b1fc1372	74	connlimit_iphash6(const union nf_inet_addr *addr)
370786f9	75	{
b1fc1372	76	return jhash2((u32 *)addr->ip6, ARRAY_SIZE(addr->ip6),
1442e750	77	connlimit_rnd) % CONNLIMIT_SLOTS;
370786f9 JE	78	}
	79
	80	static inline bool already_closed(const struct nf_conn *conn)
	81	{
5e8fbe2a	82	if (nf_ct_protonum(conn) == IPPROTO_TCP)
d2ee3f2c DW	83	return conn->proto.tcp.state == TCP_CONNTRACK_TIME_WAIT \|\|
d2ee3f2c DW	84	conn->proto.tcp.state == TCP_CONNTRACK_CLOSE;
370786f9 JE	85	else
	86	return 0;
	87	}
	88
50e0e9b1	89	static int
b1fc1372 FW	90	same_source(const union nf_inet_addr *addr,
b1fc1372 FW	91	const union nf_inet_addr *u3, u_int8_t family)
370786f9	92	{
b1fc1372 FW	93	if (family == NFPROTO_IPV4)
b1fc1372 FW	94	return ntohl(addr->ip) - ntohl(u3->ip);
370786f9	95
b1fc1372	96	return memcmp(addr->ip6, u3->ip6, sizeof(addr->ip6));
370786f9 JE	97	}
370786f9 JE	98
7d084877 FW	99	static bool add_hlist(struct hlist_head *head,
	100	const struct nf_conntrack_tuple *tuple,
	101	const union nf_inet_addr *addr)
	102	{
	103	struct xt_connlimit_conn *conn;
	104
	105	conn = kmem_cache_alloc(connlimit_conn_cachep, GFP_ATOMIC);
	106	if (conn == NULL)
	107	return false;
	108	conn->tuple = *tuple;
7d084877 FW	109	hlist_add_head(&conn->node, head);
	110	return true;
	111	}
	112
	113	static unsigned int check_hlist(struct net *net,
	114	struct hlist_head *head,
	115	const struct nf_conntrack_tuple *tuple,
308ac914	116	const struct nf_conntrack_zone *zone,
7d084877	117	bool *addit)
370786f9	118	{
3cf93c96	119	const struct nf_conntrack_tuple_hash *found;
370786f9	120	struct xt_connlimit_conn *conn;
b67bfe0d	121	struct hlist_node *n;
ea781f19	122	struct nf_conn *found_ct;
7d084877	123	unsigned int length = 0;
370786f9	124
7d084877	125	*addit = true;
370786f9 JE	126
370786f9 JE	127	/* check the saved connections */
15cfd528	128	hlist_for_each_entry_safe(conn, n, head, node) {
e59ea3df	129	found = nf_conntrack_find_get(net, zone, &conn->tuple);
d9ec4f1e FW	130	if (found == NULL) {
d9ec4f1e FW	131	hlist_del(&conn->node);
14e1a977	132	kmem_cache_free(connlimit_conn_cachep, conn);
d9ec4f1e FW	133	continue;
d9ec4f1e FW	134	}
370786f9	135
d9ec4f1e	136	found_ct = nf_ct_tuplehash_to_ctrack(found);
370786f9	137
d9ec4f1e	138	if (nf_ct_tuple_equal(&conn->tuple, tuple)) {
370786f9 JE	139	/*
	140	* Just to be sure we have it only once in the list.
	141	* We should not see tuples twice unless someone hooks
	142	* this into a table without "-p tcp --syn".
	143	*/
3bcc5fdf	144	*addit = false;
d9ec4f1e	145	} else if (already_closed(found_ct)) {
370786f9 JE	146	/*
	147	* we do not care about connections which are
	148	* closed already -> ditch it
	149	*/
ea781f19	150	nf_ct_put(found_ct);
3e0d5149	151	hlist_del(&conn->node);
14e1a977	152	kmem_cache_free(connlimit_conn_cachep, conn);
370786f9 JE	153	continue;
	154	}
	155
ea781f19	156	nf_ct_put(found_ct);
7d084877	157	length++;
370786f9 JE	158	}
370786f9 JE	159
7d084877	160	return length;
370786f9 JE	161	}
370786f9 JE	162
7d084877 FW	163	static void tree_nodes_free(struct rb_root *root,
	164	struct xt_connlimit_rb *gc_nodes[],
	165	unsigned int gc_count)
	166	{
	167	struct xt_connlimit_rb *rbconn;
	168
	169	while (gc_count) {
	170	rbconn = gc_nodes[--gc_count];
	171	rb_erase(&rbconn->node, root);
	172	kmem_cache_free(connlimit_rb_cachep, rbconn);
	173	}
	174	}
	175
	176	static unsigned int
	177	count_tree(struct net net, struct rb_root root,
	178	const struct nf_conntrack_tuple *tuple,
b1fc1372	179	const union nf_inet_addr *addr,
308ac914	180	u8 family, const struct nf_conntrack_zone *zone)
3bcc5fdf	181	{
7d084877 FW	182	struct xt_connlimit_rb *gc_nodes[CONNLIMIT_GC_MAX_NODES];
	183	struct rb_node *rbnode, parent;
	184	struct xt_connlimit_rb *rbconn;
14e1a977	185	struct xt_connlimit_conn *conn;
7d084877 FW	186	unsigned int gc_count;
	187	bool no_gc = false;
	188
	189	restart:
	190	gc_count = 0;
	191	parent = NULL;
	192	rbnode = &(root->rb_node);
	193	while (*rbnode) {
	194	int diff;
	195	bool addit;
	196
4cc4b72c	197	rbconn = rb_entry(*rbnode, struct xt_connlimit_rb, node);
7d084877 FW	198
7d084877 FW	199	parent = *rbnode;
b1fc1372	200	diff = same_source(addr, &rbconn->addr, family);
7d084877 FW	201	if (diff < 0) {
	202	rbnode = &((*rbnode)->rb_left);
	203	} else if (diff > 0) {
	204	rbnode = &((*rbnode)->rb_right);
	205	} else {
	206	/* same source network -> be counted! */
	207	unsigned int count;
e59ea3df	208	count = check_hlist(net, &rbconn->hhead, tuple, zone, &addit);
7d084877 FW	209
	210	tree_nodes_free(root, gc_nodes, gc_count);
	211	if (!addit)
	212	return count;
	213
	214	if (!add_hlist(&rbconn->hhead, tuple, addr))
	215	return 0; /* hotdrop */
	216
	217	return count + 1;
	218	}
	219
	220	if (no_gc \|\| gc_count >= ARRAY_SIZE(gc_nodes))
	221	continue;
	222
	223	/* only used for GC on hhead, retval and 'addit' ignored */
e59ea3df	224	check_hlist(net, &rbconn->hhead, tuple, zone, &addit);
7d084877 FW	225	if (hlist_empty(&rbconn->hhead))
	226	gc_nodes[gc_count++] = rbconn;
	227	}
	228
	229	if (gc_count) {
	230	no_gc = true;
	231	tree_nodes_free(root, gc_nodes, gc_count);
	232	/* tree_node_free before new allocation permits
	233	* allocator to re-use newly free'd object.
	234	*
	235	* This is a rare event; in most cases we will find
	236	* existing node to re-use. (or gc_count is 0).
	237	*/
	238	goto restart;
	239	}
	240
	241	/* no match, need to insert new node */
	242	rbconn = kmem_cache_alloc(connlimit_rb_cachep, GFP_ATOMIC);
	243	if (rbconn == NULL)
	244	return 0;
14e1a977 FW	245
14e1a977 FW	246	conn = kmem_cache_alloc(connlimit_conn_cachep, GFP_ATOMIC);
7d084877 FW	247	if (conn == NULL) {
	248	kmem_cache_free(connlimit_rb_cachep, rbconn);
	249	return 0;
	250	}
	251
3bcc5fdf	252	conn->tuple = *tuple;
7d084877 FW	253	rbconn->addr = *addr;
	254
	255	INIT_HLIST_HEAD(&rbconn->hhead);
	256	hlist_add_head(&conn->node, &rbconn->hhead);
	257
	258	rb_link_node(&rbconn->node, parent, rbnode);
	259	rb_insert_color(&rbconn->node, root);
	260	return 1;
3bcc5fdf FW	261	}
3bcc5fdf FW	262
15cfd528 FW	263	static int count_them(struct net *net,
	264	struct xt_connlimit_data *data,
	265	const struct nf_conntrack_tuple *tuple,
	266	const union nf_inet_addr *addr,
308ac914 DB	267	u_int8_t family,
308ac914 DB	268	const struct nf_conntrack_zone *zone)
15cfd528	269	{
7d084877	270	struct rb_root *root;
15cfd528 FW	271	int count;
	272	u32 hash;
	273
a2acc543	274	if (family == NFPROTO_IPV6)
b1fc1372	275	hash = connlimit_iphash6(addr);
a2acc543	276	else
b1fc1372	277	hash = connlimit_iphash(addr->ip);
a2acc543	278	root = &data->climit_root[hash];
15cfd528	279
e00b437b	280	spin_lock_bh(&xt_connlimit_locks[hash % CONNLIMIT_LOCK_SLOTS]);
7d084877	281
b1fc1372	282	count = count_tree(net, root, tuple, addr, family, zone);
7d084877	283
e00b437b	284	spin_unlock_bh(&xt_connlimit_locks[hash % CONNLIMIT_LOCK_SLOTS]);
15cfd528 FW	285
	286	return count;
	287	}
	288
d3c5ee6d	289	static bool
62fc8051	290	connlimit_mt(const struct sk_buff skb, struct xt_action_param par)
370786f9	291	{
613dbd95	292	struct net *net = xt_net(par);
f7108a20	293	const struct xt_connlimit_info *info = par->matchinfo;
22c2d8bc	294	union nf_inet_addr addr;
370786f9 JE	295	struct nf_conntrack_tuple tuple;
370786f9 JE	296	const struct nf_conntrack_tuple *tuple_ptr = &tuple;
308ac914	297	const struct nf_conntrack_zone *zone = &nf_ct_zone_dflt;
370786f9 JE	298	enum ip_conntrack_info ctinfo;
370786f9 JE	299	const struct nf_conn *ct;
7d084877	300	unsigned int connections;
370786f9 JE	301
370786f9 JE	302	ct = nf_ct_get(skb, &ctinfo);
e59ea3df	303	if (ct != NULL) {
8183e3a8	304	tuple_ptr = &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple;
e59ea3df FW	305	zone = nf_ct_zone(ct);
e59ea3df FW	306	} else if (!nf_ct_get_tuplepr(skb, skb_network_offset(skb),
613dbd95	307	xt_family(par), net, &tuple)) {
370786f9	308	goto hotdrop;
e59ea3df	309	}
370786f9	310
613dbd95	311	if (xt_family(par) == NFPROTO_IPV6) {
370786f9	312	const struct ipv6hdr *iph = ipv6_hdr(skb);
b1fc1372 FW	313	unsigned int i;
b1fc1372 FW	314
cc4fc022 JE	315	memcpy(&addr.ip6, (info->flags & XT_CONNLIMIT_DADDR) ?
cc4fc022 JE	316	&iph->daddr : &iph->saddr, sizeof(addr.ip6));
b1fc1372 FW	317
	318	for (i = 0; i < ARRAY_SIZE(addr.ip6); ++i)
	319	addr.ip6[i] &= info->mask.ip6[i];
370786f9 JE	320	} else {
370786f9 JE	321	const struct iphdr *iph = ip_hdr(skb);
cc4fc022 JE	322	addr.ip = (info->flags & XT_CONNLIMIT_DADDR) ?
cc4fc022 JE	323	iph->daddr : iph->saddr;
b1fc1372 FW	324
b1fc1372 FW	325	addr.ip &= info->mask.ip;
370786f9 JE	326	}
370786f9 JE	327
83fc8102	328	connections = count_them(net, info->data, tuple_ptr, &addr,
b1fc1372	329	xt_family(par), zone);
7d084877	330	if (connections == 0)
370786f9	331	/* kmalloc failed, drop it entirely */
1cc34c30	332	goto hotdrop;
370786f9	333
cc4fc022 JE	334	return (connections > info->limit) ^
cc4fc022 JE	335	!!(info->flags & XT_CONNLIMIT_INVERT);
370786f9 JE	336
370786f9 JE	337	hotdrop:
b4ba2611	338	par->hotdrop = true;
370786f9 JE	339	return false;
	340	}
	341
b0f38452	342	static int connlimit_mt_check(const struct xt_mtchk_param *par)
370786f9	343	{
9b4fce7a	344	struct xt_connlimit_info *info = par->matchinfo;
370786f9	345	unsigned int i;
4a5a5c73	346	int ret;
370786f9	347
7bdc6624	348	net_get_random_once(&connlimit_rnd, sizeof(connlimit_rnd));
4656c4d6	349
ecb2421b	350	ret = nf_ct_netns_get(par->net, par->family);
4a5a5c73	351	if (ret < 0) {
8bee4bad JE	352	pr_info("cannot load conntrack support for "
8bee4bad JE	353	"address family %u\n", par->family);
4a5a5c73	354	return ret;
370786f9 JE	355	}
	356
	357	/* init private data */
	358	info->data = kmalloc(sizeof(struct xt_connlimit_data), GFP_KERNEL);
	359	if (info->data == NULL) {
ecb2421b	360	nf_ct_netns_put(par->net, par->family);
4a5a5c73	361	return -ENOMEM;
370786f9 JE	362	}
370786f9 JE	363
a2acc543 TY	364	for (i = 0; i < ARRAY_SIZE(info->data->climit_root); ++i)
a2acc543 TY	365	info->data->climit_root[i] = RB_ROOT;
370786f9	366
bd414ee6	367	return 0;
370786f9 JE	368	}
370786f9 JE	369
7d084877	370	static void destroy_tree(struct rb_root *r)
370786f9	371	{
370786f9	372	struct xt_connlimit_conn *conn;
7d084877	373	struct xt_connlimit_rb *rbconn;
b67bfe0d	374	struct hlist_node *n;
7d084877	375	struct rb_node *node;
370786f9	376
7d084877	377	while ((node = rb_first(r)) != NULL) {
4cc4b72c	378	rbconn = rb_entry(node, struct xt_connlimit_rb, node);
370786f9	379
7d084877 FW	380	rb_erase(node, r);
	381
	382	hlist_for_each_entry_safe(conn, n, &rbconn->hhead, node)
14e1a977	383	kmem_cache_free(connlimit_conn_cachep, conn);
7d084877 FW	384
7d084877 FW	385	kmem_cache_free(connlimit_rb_cachep, rbconn);
370786f9	386	}
7d084877 FW	387	}
	388
	389	static void connlimit_mt_destroy(const struct xt_mtdtor_param *par)
	390	{
	391	const struct xt_connlimit_info *info = par->matchinfo;
	392	unsigned int i;
	393
ecb2421b	394	nf_ct_netns_put(par->net, par->family);
7d084877	395
a2acc543 TY	396	for (i = 0; i < ARRAY_SIZE(info->data->climit_root); ++i)
a2acc543 TY	397	destroy_tree(&info->data->climit_root[i]);
370786f9 JE	398
	399	kfree(info->data);
	400	}
	401
68c07cb6 CW	402	static struct xt_match connlimit_mt_reg __read_mostly = {
	403	.name = "connlimit",
	404	.revision = 1,
	405	.family = NFPROTO_UNSPEC,
	406	.checkentry = connlimit_mt_check,
	407	.match = connlimit_mt,
	408	.matchsize = sizeof(struct xt_connlimit_info),
ec231890	409	.usersize = offsetof(struct xt_connlimit_info, data),
68c07cb6 CW	410	.destroy = connlimit_mt_destroy,
68c07cb6 CW	411	.me = THIS_MODULE,
370786f9 JE	412	};
370786f9 JE	413
d3c5ee6d	414	static int __init connlimit_mt_init(void)
370786f9	415	{
e00b437b	416	int ret, i;
1442e750 FW	417
	418	BUILD_BUG_ON(CONNLIMIT_LOCK_SLOTS > CONNLIMIT_SLOTS);
	419	BUILD_BUG_ON((CONNLIMIT_SLOTS % CONNLIMIT_LOCK_SLOTS) != 0);
	420
e00b437b FW	421	for (i = 0; i < CONNLIMIT_LOCK_SLOTS; ++i)
	422	spin_lock_init(&xt_connlimit_locks[i]);
	423
14e1a977 FW	424	connlimit_conn_cachep = kmem_cache_create("xt_connlimit_conn",
	425	sizeof(struct xt_connlimit_conn),
	426	0, 0, NULL);
	427	if (!connlimit_conn_cachep)
	428	return -ENOMEM;
	429
7d084877 FW	430	connlimit_rb_cachep = kmem_cache_create("xt_connlimit_rb",
	431	sizeof(struct xt_connlimit_rb),
	432	0, 0, NULL);
	433	if (!connlimit_rb_cachep) {
	434	kmem_cache_destroy(connlimit_conn_cachep);
	435	return -ENOMEM;
	436	}
14e1a977	437	ret = xt_register_match(&connlimit_mt_reg);
7d084877	438	if (ret != 0) {
14e1a977	439	kmem_cache_destroy(connlimit_conn_cachep);
7d084877 FW	440	kmem_cache_destroy(connlimit_rb_cachep);
7d084877 FW	441	}
14e1a977	442	return ret;
370786f9 JE	443	}
370786f9 JE	444
d3c5ee6d	445	static void __exit connlimit_mt_exit(void)
370786f9	446	{
68c07cb6	447	xt_unregister_match(&connlimit_mt_reg);
14e1a977	448	kmem_cache_destroy(connlimit_conn_cachep);
7d084877	449	kmem_cache_destroy(connlimit_rb_cachep);
370786f9 JE	450	}
370786f9 JE	451
d3c5ee6d JE	452	module_init(connlimit_mt_init);
d3c5ee6d JE	453	module_exit(connlimit_mt_exit);
92f3b2b1	454	MODULE_AUTHOR("Jan Engelhardt <jengelh@medozas.de>");
2ae15b64	455	MODULE_DESCRIPTION("Xtables: Number of connections matching");
370786f9 JE	456	MODULE_LICENSE("GPL");
	457	MODULE_ALIAS("ipt_connlimit");
	458	MODULE_ALIAS("ip6t_connlimit");