* 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.
- *
- * 23 Apr 2001: Harald Welte <laforge@gnumonks.org>
- * - new API and handling of conntrack/nat helpers
- * - now capable of multiple expectations for one master
- * 16 Jul 2002: Harald Welte <laforge@gnumonks.org>
- * - add usage/reference counts to ip_conntrack_expect
- * - export ip_conntrack[_expect]_{find_get,put} functions
- * 16 Dec 2003: Yasuyuki Kozakai @USAGI <yasuyuki.kozakai@toshiba.co.jp>
- * - generalize L3 protocol denendent part.
- * 23 Mar 2004: Yasuyuki Kozakai @USAGI <yasuyuki.kozakai@toshiba.co.jp>
- * - add support various size of conntrack structures.
- * 26 Jan 2006: Harald Welte <laforge@netfilter.org>
- * - restructure nf_conn (introduce nf_conn_help)
- * - redesign 'features' how they were originally intended
- * 26 Feb 2006: Pablo Neira Ayuso <pablo@eurodev.net>
- * - add support for L3 protocol module load on demand.
- *
- * Derived from net/ipv4/netfilter/ip_conntrack_core.c
*/
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/netdevice.h>
#include <linux/socket.h>
-
-/* This rwlock protects the main hash table, protocol/helper/expected
- registrations, conntrack timers*/
-#define ASSERT_READ_LOCK(x)
-#define ASSERT_WRITE_LOCK(x)
+#include <linux/mm.h>
#include <net/netfilter/nf_conntrack.h>
#include <net/netfilter/nf_conntrack_l3proto.h>
-#include <net/netfilter/nf_conntrack_protocol.h>
+#include <net/netfilter/nf_conntrack_l4proto.h>
#include <net/netfilter/nf_conntrack_expect.h>
#include <net/netfilter/nf_conntrack_helper.h>
#include <net/netfilter/nf_conntrack_core.h>
+#include <net/netfilter/nf_conntrack_extend.h>
#define NF_CONNTRACK_VERSION "0.5.0"
-#if 0
-#define DEBUGP printk
-#else
-#define DEBUGP(format, args...)
-#endif
-
-DEFINE_RWLOCK(nf_conntrack_lock);
+DEFINE_SPINLOCK(nf_conntrack_lock);
+EXPORT_SYMBOL_GPL(nf_conntrack_lock);
/* nf_conntrack_standalone needs this */
atomic_t nf_conntrack_count = ATOMIC_INIT(0);
+EXPORT_SYMBOL_GPL(nf_conntrack_count);
+
+unsigned int nf_conntrack_htable_size __read_mostly;
+EXPORT_SYMBOL_GPL(nf_conntrack_htable_size);
-void (*nf_conntrack_destroyed)(struct nf_conn *conntrack) = NULL;
-unsigned int nf_conntrack_htable_size __read_mostly = 0;
int nf_conntrack_max __read_mostly;
-struct list_head *nf_conntrack_hash __read_mostly;
-struct nf_conn nf_conntrack_untracked;
+EXPORT_SYMBOL_GPL(nf_conntrack_max);
+
+struct hlist_head *nf_conntrack_hash __read_mostly;
+EXPORT_SYMBOL_GPL(nf_conntrack_hash);
+
+struct nf_conn nf_conntrack_untracked __read_mostly;
+EXPORT_SYMBOL_GPL(nf_conntrack_untracked);
+
unsigned int nf_ct_log_invalid __read_mostly;
-LIST_HEAD(unconfirmed);
+HLIST_HEAD(unconfirmed);
static int nf_conntrack_vmalloc __read_mostly;
-
-static unsigned int nf_conntrack_next_id;
+static struct kmem_cache *nf_conntrack_cachep __read_mostly;
DEFINE_PER_CPU(struct ip_conntrack_stat, nf_conntrack_stat);
EXPORT_PER_CPU_SYMBOL(nf_conntrack_stat);
-/*
- * This scheme offers various size of "struct nf_conn" dependent on
- * features(helper, nat, ...)
- */
-
-#define NF_CT_FEATURES_NAMELEN 256
-static struct {
- /* name of slab cache. printed in /proc/slabinfo */
- char *name;
-
- /* size of slab cache */
- size_t size;
-
- /* slab cache pointer */
- kmem_cache_t *cachep;
-
- /* allocated slab cache + modules which uses this slab cache */
- int use;
-
-} nf_ct_cache[NF_CT_F_NUM];
-
-/* protect members of nf_ct_cache except of "use" */
-DEFINE_RWLOCK(nf_ct_cache_lock);
-
-/* This avoids calling kmem_cache_create() with same name simultaneously */
-static DEFINE_MUTEX(nf_ct_cache_mutex);
-
static int nf_conntrack_hash_rnd_initted;
static unsigned int nf_conntrack_hash_rnd;
static u_int32_t __hash_conntrack(const struct nf_conntrack_tuple *tuple,
unsigned int size, unsigned int rnd)
{
- unsigned int a, b;
- a = jhash((void *)tuple->src.u3.all, sizeof(tuple->src.u3.all),
- ((tuple->src.l3num) << 16) | tuple->dst.protonum);
- b = jhash((void *)tuple->dst.u3.all, sizeof(tuple->dst.u3.all),
- (tuple->src.u.all << 16) | tuple->dst.u.all);
+ unsigned int n;
+ u_int32_t h;
+
+ /* The direction must be ignored, so we hash everything up to the
+ * destination ports (which is a multiple of 4) and treat the last
+ * three bytes manually.
+ */
+ n = (sizeof(tuple->src) + sizeof(tuple->dst.u3)) / sizeof(u32);
+ h = jhash2((u32 *)tuple, n,
+ rnd ^ (((__force __u16)tuple->dst.u.all << 16) |
+ tuple->dst.protonum));
- return jhash_2words(a, b, rnd) % size;
+ return ((u64)h * size) >> 32;
}
static inline u_int32_t hash_conntrack(const struct nf_conntrack_tuple *tuple)
nf_conntrack_hash_rnd);
}
-int nf_conntrack_register_cache(u_int32_t features, const char *name,
- size_t size)
-{
- int ret = 0;
- char *cache_name;
- kmem_cache_t *cachep;
-
- DEBUGP("nf_conntrack_register_cache: features=0x%x, name=%s, size=%d\n",
- features, name, size);
-
- if (features < NF_CT_F_BASIC || features >= NF_CT_F_NUM) {
- DEBUGP("nf_conntrack_register_cache: invalid features.: 0x%x\n",
- features);
- return -EINVAL;
- }
-
- mutex_lock(&nf_ct_cache_mutex);
-
- write_lock_bh(&nf_ct_cache_lock);
- /* e.g: multiple helpers are loaded */
- if (nf_ct_cache[features].use > 0) {
- DEBUGP("nf_conntrack_register_cache: already resisterd.\n");
- if ((!strncmp(nf_ct_cache[features].name, name,
- NF_CT_FEATURES_NAMELEN))
- && nf_ct_cache[features].size == size) {
- DEBUGP("nf_conntrack_register_cache: reusing.\n");
- nf_ct_cache[features].use++;
- ret = 0;
- } else
- ret = -EBUSY;
-
- write_unlock_bh(&nf_ct_cache_lock);
- mutex_unlock(&nf_ct_cache_mutex);
- return ret;
- }
- write_unlock_bh(&nf_ct_cache_lock);
-
- /*
- * The memory space for name of slab cache must be alive until
- * cache is destroyed.
- */
- cache_name = kmalloc(sizeof(char)*NF_CT_FEATURES_NAMELEN, GFP_ATOMIC);
- if (cache_name == NULL) {
- DEBUGP("nf_conntrack_register_cache: can't alloc cache_name\n");
- ret = -ENOMEM;
- goto out_up_mutex;
- }
-
- if (strlcpy(cache_name, name, NF_CT_FEATURES_NAMELEN)
- >= NF_CT_FEATURES_NAMELEN) {
- printk("nf_conntrack_register_cache: name too long\n");
- ret = -EINVAL;
- goto out_free_name;
- }
-
- cachep = kmem_cache_create(cache_name, size, 0, 0,
- NULL, NULL);
- if (!cachep) {
- printk("nf_conntrack_register_cache: Can't create slab cache "
- "for the features = 0x%x\n", features);
- ret = -ENOMEM;
- goto out_free_name;
- }
-
- write_lock_bh(&nf_ct_cache_lock);
- nf_ct_cache[features].use = 1;
- nf_ct_cache[features].size = size;
- nf_ct_cache[features].cachep = cachep;
- nf_ct_cache[features].name = cache_name;
- write_unlock_bh(&nf_ct_cache_lock);
-
- goto out_up_mutex;
-
-out_free_name:
- kfree(cache_name);
-out_up_mutex:
- mutex_unlock(&nf_ct_cache_mutex);
- return ret;
-}
-
-/* FIXME: In the current, only nf_conntrack_cleanup() can call this function. */
-void nf_conntrack_unregister_cache(u_int32_t features)
-{
- kmem_cache_t *cachep;
- char *name;
-
- /*
- * This assures that kmem_cache_create() isn't called before destroying
- * slab cache.
- */
- DEBUGP("nf_conntrack_unregister_cache: 0x%04x\n", features);
- mutex_lock(&nf_ct_cache_mutex);
-
- write_lock_bh(&nf_ct_cache_lock);
- if (--nf_ct_cache[features].use > 0) {
- write_unlock_bh(&nf_ct_cache_lock);
- mutex_unlock(&nf_ct_cache_mutex);
- return;
- }
- cachep = nf_ct_cache[features].cachep;
- name = nf_ct_cache[features].name;
- nf_ct_cache[features].cachep = NULL;
- nf_ct_cache[features].name = NULL;
- nf_ct_cache[features].size = 0;
- write_unlock_bh(&nf_ct_cache_lock);
-
- synchronize_net();
-
- kmem_cache_destroy(cachep);
- kfree(name);
-
- mutex_unlock(&nf_ct_cache_mutex);
-}
-
-int
+bool
nf_ct_get_tuple(const struct sk_buff *skb,
unsigned int nhoff,
unsigned int dataoff,
u_int8_t protonum,
struct nf_conntrack_tuple *tuple,
const struct nf_conntrack_l3proto *l3proto,
- const struct nf_conntrack_protocol *protocol)
+ const struct nf_conntrack_l4proto *l4proto)
{
- NF_CT_TUPLE_U_BLANK(tuple);
+ memset(tuple, 0, sizeof(*tuple));
tuple->src.l3num = l3num;
if (l3proto->pkt_to_tuple(skb, nhoff, tuple) == 0)
- return 0;
+ return false;
tuple->dst.protonum = protonum;
tuple->dst.dir = IP_CT_DIR_ORIGINAL;
- return protocol->pkt_to_tuple(skb, dataoff, tuple);
+ return l4proto->pkt_to_tuple(skb, dataoff, tuple);
}
+EXPORT_SYMBOL_GPL(nf_ct_get_tuple);
-int
+bool nf_ct_get_tuplepr(const struct sk_buff *skb, unsigned int nhoff,
+ u_int16_t l3num, struct nf_conntrack_tuple *tuple)
+{
+ struct nf_conntrack_l3proto *l3proto;
+ struct nf_conntrack_l4proto *l4proto;
+ unsigned int protoff;
+ u_int8_t protonum;
+ int ret;
+
+ rcu_read_lock();
+
+ l3proto = __nf_ct_l3proto_find(l3num);
+ ret = l3proto->get_l4proto(skb, nhoff, &protoff, &protonum);
+ if (ret != NF_ACCEPT) {
+ rcu_read_unlock();
+ return false;
+ }
+
+ l4proto = __nf_ct_l4proto_find(l3num, protonum);
+
+ ret = nf_ct_get_tuple(skb, nhoff, protoff, l3num, protonum, tuple,
+ l3proto, l4proto);
+
+ rcu_read_unlock();
+ return ret;
+}
+EXPORT_SYMBOL_GPL(nf_ct_get_tuplepr);
+
+bool
nf_ct_invert_tuple(struct nf_conntrack_tuple *inverse,
const struct nf_conntrack_tuple *orig,
const struct nf_conntrack_l3proto *l3proto,
- const struct nf_conntrack_protocol *protocol)
+ const struct nf_conntrack_l4proto *l4proto)
{
- NF_CT_TUPLE_U_BLANK(inverse);
+ memset(inverse, 0, sizeof(*inverse));
inverse->src.l3num = orig->src.l3num;
if (l3proto->invert_tuple(inverse, orig) == 0)
- return 0;
+ return false;
inverse->dst.dir = !orig->dst.dir;
inverse->dst.protonum = orig->dst.protonum;
- return protocol->invert_tuple(inverse, orig);
+ return l4proto->invert_tuple(inverse, orig);
}
+EXPORT_SYMBOL_GPL(nf_ct_invert_tuple);
static void
clean_from_lists(struct nf_conn *ct)
{
- DEBUGP("clean_from_lists(%p)\n", ct);
- ASSERT_WRITE_LOCK(&nf_conntrack_lock);
- list_del(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list);
- list_del(&ct->tuplehash[IP_CT_DIR_REPLY].list);
+ pr_debug("clean_from_lists(%p)\n", ct);
+ hlist_del_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnode);
+ hlist_del_rcu(&ct->tuplehash[IP_CT_DIR_REPLY].hnode);
/* Destroy all pending expectations */
nf_ct_remove_expectations(ct);
destroy_conntrack(struct nf_conntrack *nfct)
{
struct nf_conn *ct = (struct nf_conn *)nfct;
- struct nf_conntrack_l3proto *l3proto;
- struct nf_conntrack_protocol *proto;
+ struct nf_conntrack_l4proto *l4proto;
- DEBUGP("destroy_conntrack(%p)\n", ct);
+ pr_debug("destroy_conntrack(%p)\n", ct);
NF_CT_ASSERT(atomic_read(&nfct->use) == 0);
NF_CT_ASSERT(!timer_pending(&ct->timeout));
/* To make sure we don't get any weird locking issues here:
* destroy_conntrack() MUST NOT be called with a write lock
* to nf_conntrack_lock!!! -HW */
- l3proto = __nf_ct_l3proto_find(ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.l3num);
- if (l3proto && l3proto->destroy)
- l3proto->destroy(ct);
-
- proto = __nf_ct_proto_find(ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.l3num, ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.protonum);
- if (proto && proto->destroy)
- proto->destroy(ct);
+ rcu_read_lock();
+ l4proto = __nf_ct_l4proto_find(nf_ct_l3num(ct), nf_ct_protonum(ct));
+ if (l4proto && l4proto->destroy)
+ l4proto->destroy(ct);
- if (nf_conntrack_destroyed)
- nf_conntrack_destroyed(ct);
+ rcu_read_unlock();
- write_lock_bh(&nf_conntrack_lock);
+ spin_lock_bh(&nf_conntrack_lock);
/* Expectations will have been removed in clean_from_lists,
* except TFTP can create an expectation on the first packet,
* before connection is in the list, so we need to clean here,
/* We overload first tuple to link into unconfirmed list. */
if (!nf_ct_is_confirmed(ct)) {
- BUG_ON(list_empty(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list));
- list_del(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list);
+ BUG_ON(hlist_unhashed(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnode));
+ hlist_del(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnode);
}
NF_CT_STAT_INC(delete);
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
if (ct->master)
nf_ct_put(ct->master);
- DEBUGP("destroy_conntrack: returning ct=%p to slab\n", ct);
+ pr_debug("destroy_conntrack: returning ct=%p to slab\n", ct);
nf_conntrack_free(ct);
}
static void death_by_timeout(unsigned long ul_conntrack)
{
struct nf_conn *ct = (void *)ul_conntrack;
+ struct nf_conn_help *help = nfct_help(ct);
+ struct nf_conntrack_helper *helper;
- write_lock_bh(&nf_conntrack_lock);
+ if (help) {
+ rcu_read_lock();
+ helper = rcu_dereference(help->helper);
+ if (helper && helper->destroy)
+ helper->destroy(ct);
+ rcu_read_unlock();
+ }
+
+ spin_lock_bh(&nf_conntrack_lock);
/* Inside lock so preempt is disabled on module removal path.
* Otherwise we can get spurious warnings. */
NF_CT_STAT_INC(delete_list);
clean_from_lists(ct);
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
nf_ct_put(ct);
}
struct nf_conntrack_tuple_hash *
-__nf_conntrack_find(const struct nf_conntrack_tuple *tuple,
- const struct nf_conn *ignored_conntrack)
+__nf_conntrack_find(const struct nf_conntrack_tuple *tuple)
{
struct nf_conntrack_tuple_hash *h;
+ struct hlist_node *n;
unsigned int hash = hash_conntrack(tuple);
- ASSERT_READ_LOCK(&nf_conntrack_lock);
- list_for_each_entry(h, &nf_conntrack_hash[hash], list) {
- if (nf_ct_tuplehash_to_ctrack(h) != ignored_conntrack &&
- nf_ct_tuple_equal(tuple, &h->tuple)) {
+ /* Disable BHs the entire time since we normally need to disable them
+ * at least once for the stats anyway.
+ */
+ local_bh_disable();
+ hlist_for_each_entry_rcu(h, n, &nf_conntrack_hash[hash], hnode) {
+ if (nf_ct_tuple_equal(tuple, &h->tuple)) {
NF_CT_STAT_INC(found);
+ local_bh_enable();
return h;
}
NF_CT_STAT_INC(searched);
}
+ local_bh_enable();
return NULL;
}
+EXPORT_SYMBOL_GPL(__nf_conntrack_find);
/* Find a connection corresponding to a tuple. */
struct nf_conntrack_tuple_hash *
-nf_conntrack_find_get(const struct nf_conntrack_tuple *tuple,
- const struct nf_conn *ignored_conntrack)
+nf_conntrack_find_get(const struct nf_conntrack_tuple *tuple)
{
struct nf_conntrack_tuple_hash *h;
+ struct nf_conn *ct;
- read_lock_bh(&nf_conntrack_lock);
- h = __nf_conntrack_find(tuple, ignored_conntrack);
- if (h)
- atomic_inc(&nf_ct_tuplehash_to_ctrack(h)->ct_general.use);
- read_unlock_bh(&nf_conntrack_lock);
+ rcu_read_lock();
+ h = __nf_conntrack_find(tuple);
+ if (h) {
+ ct = nf_ct_tuplehash_to_ctrack(h);
+ if (unlikely(!atomic_inc_not_zero(&ct->ct_general.use)))
+ h = NULL;
+ }
+ rcu_read_unlock();
return h;
}
+EXPORT_SYMBOL_GPL(nf_conntrack_find_get);
static void __nf_conntrack_hash_insert(struct nf_conn *ct,
unsigned int hash,
- unsigned int repl_hash)
+ unsigned int repl_hash)
{
- ct->id = ++nf_conntrack_next_id;
- list_add(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list,
- &nf_conntrack_hash[hash]);
- list_add(&ct->tuplehash[IP_CT_DIR_REPLY].list,
- &nf_conntrack_hash[repl_hash]);
+ hlist_add_head_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnode,
+ &nf_conntrack_hash[hash]);
+ hlist_add_head_rcu(&ct->tuplehash[IP_CT_DIR_REPLY].hnode,
+ &nf_conntrack_hash[repl_hash]);
}
void nf_conntrack_hash_insert(struct nf_conn *ct)
hash = hash_conntrack(&ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple);
repl_hash = hash_conntrack(&ct->tuplehash[IP_CT_DIR_REPLY].tuple);
- write_lock_bh(&nf_conntrack_lock);
+ spin_lock_bh(&nf_conntrack_lock);
__nf_conntrack_hash_insert(ct, hash, repl_hash);
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
}
+EXPORT_SYMBOL_GPL(nf_conntrack_hash_insert);
/* Confirm a connection given skb; places it in hash table */
int
-__nf_conntrack_confirm(struct sk_buff **pskb)
+__nf_conntrack_confirm(struct sk_buff *skb)
{
unsigned int hash, repl_hash;
struct nf_conntrack_tuple_hash *h;
struct nf_conn *ct;
struct nf_conn_help *help;
+ struct hlist_node *n;
enum ip_conntrack_info ctinfo;
- ct = nf_ct_get(*pskb, &ctinfo);
+ ct = nf_ct_get(skb, &ctinfo);
/* ipt_REJECT uses nf_conntrack_attach to attach related
ICMP/TCP RST packets in other direction. Actual packet
/* No external references means noone else could have
confirmed us. */
NF_CT_ASSERT(!nf_ct_is_confirmed(ct));
- DEBUGP("Confirming conntrack %p\n", ct);
+ pr_debug("Confirming conntrack %p\n", ct);
- write_lock_bh(&nf_conntrack_lock);
+ spin_lock_bh(&nf_conntrack_lock);
/* See if there's one in the list already, including reverse:
NAT could have grabbed it without realizing, since we're
not in the hash. If there is, we lost race. */
- list_for_each_entry(h, &nf_conntrack_hash[hash], list)
+ hlist_for_each_entry(h, n, &nf_conntrack_hash[hash], hnode)
if (nf_ct_tuple_equal(&ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple,
&h->tuple))
goto out;
- list_for_each_entry(h, &nf_conntrack_hash[repl_hash], list)
+ hlist_for_each_entry(h, n, &nf_conntrack_hash[repl_hash], hnode)
if (nf_ct_tuple_equal(&ct->tuplehash[IP_CT_DIR_REPLY].tuple,
&h->tuple))
goto out;
/* Remove from unconfirmed list */
- list_del(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list);
+ hlist_del(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnode);
__nf_conntrack_hash_insert(ct, hash, repl_hash);
/* Timer relative to confirmation time, not original
atomic_inc(&ct->ct_general.use);
set_bit(IPS_CONFIRMED_BIT, &ct->status);
NF_CT_STAT_INC(insert);
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
help = nfct_help(ct);
if (help && help->helper)
- nf_conntrack_event_cache(IPCT_HELPER, *pskb);
+ nf_conntrack_event_cache(IPCT_HELPER, skb);
#ifdef CONFIG_NF_NAT_NEEDED
if (test_bit(IPS_SRC_NAT_DONE_BIT, &ct->status) ||
test_bit(IPS_DST_NAT_DONE_BIT, &ct->status))
- nf_conntrack_event_cache(IPCT_NATINFO, *pskb);
+ nf_conntrack_event_cache(IPCT_NATINFO, skb);
#endif
nf_conntrack_event_cache(master_ct(ct) ?
- IPCT_RELATED : IPCT_NEW, *pskb);
+ IPCT_RELATED : IPCT_NEW, skb);
return NF_ACCEPT;
out:
NF_CT_STAT_INC(insert_failed);
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
return NF_DROP;
}
+EXPORT_SYMBOL_GPL(__nf_conntrack_confirm);
/* Returns true if a connection correspondings to the tuple (required
for NAT). */
const struct nf_conn *ignored_conntrack)
{
struct nf_conntrack_tuple_hash *h;
+ struct hlist_node *n;
+ unsigned int hash = hash_conntrack(tuple);
- read_lock_bh(&nf_conntrack_lock);
- h = __nf_conntrack_find(tuple, ignored_conntrack);
- read_unlock_bh(&nf_conntrack_lock);
+ /* Disable BHs the entire time since we need to disable them at
+ * least once for the stats anyway.
+ */
+ rcu_read_lock_bh();
+ hlist_for_each_entry_rcu(h, n, &nf_conntrack_hash[hash], hnode) {
+ if (nf_ct_tuplehash_to_ctrack(h) != ignored_conntrack &&
+ nf_ct_tuple_equal(tuple, &h->tuple)) {
+ NF_CT_STAT_INC(found);
+ rcu_read_unlock_bh();
+ return 1;
+ }
+ NF_CT_STAT_INC(searched);
+ }
+ rcu_read_unlock_bh();
- return h != NULL;
+ return 0;
}
+EXPORT_SYMBOL_GPL(nf_conntrack_tuple_taken);
+
+#define NF_CT_EVICTION_RANGE 8
/* There's a small race here where we may free a just-assured
connection. Too bad: we're in trouble anyway. */
-static int early_drop(struct list_head *chain)
+static noinline int early_drop(unsigned int hash)
{
- /* Traverse backwards: gives us oldest, which is roughly LRU */
+ /* Use oldest entry, which is roughly LRU */
struct nf_conntrack_tuple_hash *h;
struct nf_conn *ct = NULL, *tmp;
+ struct hlist_node *n;
+ unsigned int i, cnt = 0;
int dropped = 0;
- read_lock_bh(&nf_conntrack_lock);
- list_for_each_entry_reverse(h, chain, list) {
- tmp = nf_ct_tuplehash_to_ctrack(h);
- if (!test_bit(IPS_ASSURED_BIT, &tmp->status)) {
- ct = tmp;
- atomic_inc(&ct->ct_general.use);
- break;
+ rcu_read_lock();
+ for (i = 0; i < nf_conntrack_htable_size; i++) {
+ hlist_for_each_entry_rcu(h, n, &nf_conntrack_hash[hash],
+ hnode) {
+ tmp = nf_ct_tuplehash_to_ctrack(h);
+ if (!test_bit(IPS_ASSURED_BIT, &tmp->status))
+ ct = tmp;
+ cnt++;
}
+
+ if (ct && unlikely(!atomic_inc_not_zero(&ct->ct_general.use)))
+ ct = NULL;
+ if (ct || cnt >= NF_CT_EVICTION_RANGE)
+ break;
+ hash = (hash + 1) % nf_conntrack_htable_size;
}
- read_unlock_bh(&nf_conntrack_lock);
+ rcu_read_unlock();
if (!ct)
return dropped;
if (del_timer(&ct->timeout)) {
death_by_timeout((unsigned long)ct);
dropped = 1;
- NF_CT_STAT_INC(early_drop);
+ NF_CT_STAT_INC_ATOMIC(early_drop);
}
nf_ct_put(ct);
return dropped;
}
-static struct nf_conn *
-__nf_conntrack_alloc(const struct nf_conntrack_tuple *orig,
- const struct nf_conntrack_tuple *repl,
- const struct nf_conntrack_l3proto *l3proto)
+struct nf_conn *nf_conntrack_alloc(const struct nf_conntrack_tuple *orig,
+ const struct nf_conntrack_tuple *repl)
{
- struct nf_conn *conntrack = NULL;
- u_int32_t features = 0;
- struct nf_conntrack_helper *helper;
+ struct nf_conn *ct = NULL;
if (unlikely(!nf_conntrack_hash_rnd_initted)) {
get_random_bytes(&nf_conntrack_hash_rnd, 4);
/* We don't want any race condition at early drop stage */
atomic_inc(&nf_conntrack_count);
- if (nf_conntrack_max
- && atomic_read(&nf_conntrack_count) > nf_conntrack_max) {
+ if (nf_conntrack_max &&
+ unlikely(atomic_read(&nf_conntrack_count) > nf_conntrack_max)) {
unsigned int hash = hash_conntrack(orig);
- /* Try dropping from this hash chain. */
- if (!early_drop(&nf_conntrack_hash[hash])) {
+ if (!early_drop(hash)) {
atomic_dec(&nf_conntrack_count);
if (net_ratelimit())
printk(KERN_WARNING
}
}
- /* find features needed by this conntrack. */
- features = l3proto->get_features(orig);
-
- /* FIXME: protect helper list per RCU */
- read_lock_bh(&nf_conntrack_lock);
- helper = __nf_ct_helper_find(repl);
- if (helper)
- features |= NF_CT_F_HELP;
- read_unlock_bh(&nf_conntrack_lock);
-
- DEBUGP("nf_conntrack_alloc: features=0x%x\n", features);
-
- read_lock_bh(&nf_ct_cache_lock);
-
- if (unlikely(!nf_ct_cache[features].use)) {
- DEBUGP("nf_conntrack_alloc: not supported features = 0x%x\n",
- features);
- goto out;
- }
-
- conntrack = kmem_cache_alloc(nf_ct_cache[features].cachep, GFP_ATOMIC);
- if (conntrack == NULL) {
- DEBUGP("nf_conntrack_alloc: Can't alloc conntrack from cache\n");
- goto out;
+ ct = kmem_cache_zalloc(nf_conntrack_cachep, GFP_ATOMIC);
+ if (ct == NULL) {
+ pr_debug("nf_conntrack_alloc: Can't alloc conntrack.\n");
+ atomic_dec(&nf_conntrack_count);
+ return ERR_PTR(-ENOMEM);
}
- memset(conntrack, 0, nf_ct_cache[features].size);
- conntrack->features = features;
- atomic_set(&conntrack->ct_general.use, 1);
- conntrack->ct_general.destroy = destroy_conntrack;
- conntrack->tuplehash[IP_CT_DIR_ORIGINAL].tuple = *orig;
- conntrack->tuplehash[IP_CT_DIR_REPLY].tuple = *repl;
+ atomic_set(&ct->ct_general.use, 1);
+ ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple = *orig;
+ ct->tuplehash[IP_CT_DIR_REPLY].tuple = *repl;
/* Don't set timer yet: wait for confirmation */
- init_timer(&conntrack->timeout);
- conntrack->timeout.data = (unsigned long)conntrack;
- conntrack->timeout.function = death_by_timeout;
- read_unlock_bh(&nf_ct_cache_lock);
+ setup_timer(&ct->timeout, death_by_timeout, (unsigned long)ct);
+ INIT_RCU_HEAD(&ct->rcu);
- return conntrack;
-out:
- read_unlock_bh(&nf_ct_cache_lock);
- atomic_dec(&nf_conntrack_count);
- return conntrack;
+ return ct;
}
+EXPORT_SYMBOL_GPL(nf_conntrack_alloc);
-struct nf_conn *nf_conntrack_alloc(const struct nf_conntrack_tuple *orig,
- const struct nf_conntrack_tuple *repl)
+static void nf_conntrack_free_rcu(struct rcu_head *head)
{
- struct nf_conntrack_l3proto *l3proto;
+ struct nf_conn *ct = container_of(head, struct nf_conn, rcu);
- l3proto = __nf_ct_l3proto_find(orig->src.l3num);
- return __nf_conntrack_alloc(orig, repl, l3proto);
+ nf_ct_ext_free(ct);
+ kmem_cache_free(nf_conntrack_cachep, ct);
+ atomic_dec(&nf_conntrack_count);
}
-void nf_conntrack_free(struct nf_conn *conntrack)
+void nf_conntrack_free(struct nf_conn *ct)
{
- u_int32_t features = conntrack->features;
- NF_CT_ASSERT(features >= NF_CT_F_BASIC && features < NF_CT_F_NUM);
- DEBUGP("nf_conntrack_free: features = 0x%x, conntrack=%p\n", features,
- conntrack);
- kmem_cache_free(nf_ct_cache[features].cachep, conntrack);
- atomic_dec(&nf_conntrack_count);
+ nf_ct_ext_destroy(ct);
+ call_rcu(&ct->rcu, nf_conntrack_free_rcu);
}
+EXPORT_SYMBOL_GPL(nf_conntrack_free);
/* Allocate a new conntrack: we return -ENOMEM if classification
failed due to stress. Otherwise it really is unclassifiable. */
static struct nf_conntrack_tuple_hash *
init_conntrack(const struct nf_conntrack_tuple *tuple,
struct nf_conntrack_l3proto *l3proto,
- struct nf_conntrack_protocol *protocol,
+ struct nf_conntrack_l4proto *l4proto,
struct sk_buff *skb,
unsigned int dataoff)
{
- struct nf_conn *conntrack;
+ struct nf_conn *ct;
+ struct nf_conn_help *help;
struct nf_conntrack_tuple repl_tuple;
struct nf_conntrack_expect *exp;
- if (!nf_ct_invert_tuple(&repl_tuple, tuple, l3proto, protocol)) {
- DEBUGP("Can't invert tuple.\n");
+ if (!nf_ct_invert_tuple(&repl_tuple, tuple, l3proto, l4proto)) {
+ pr_debug("Can't invert tuple.\n");
return NULL;
}
- conntrack = __nf_conntrack_alloc(tuple, &repl_tuple, l3proto);
- if (conntrack == NULL || IS_ERR(conntrack)) {
- DEBUGP("Can't allocate conntrack.\n");
- return (struct nf_conntrack_tuple_hash *)conntrack;
+ ct = nf_conntrack_alloc(tuple, &repl_tuple);
+ if (ct == NULL || IS_ERR(ct)) {
+ pr_debug("Can't allocate conntrack.\n");
+ return (struct nf_conntrack_tuple_hash *)ct;
}
- if (!protocol->new(conntrack, skb, dataoff)) {
- nf_conntrack_free(conntrack);
- DEBUGP("init conntrack: can't track with proto module\n");
+ if (!l4proto->new(ct, skb, dataoff)) {
+ nf_conntrack_free(ct);
+ pr_debug("init conntrack: can't track with proto module\n");
return NULL;
}
- write_lock_bh(&nf_conntrack_lock);
- exp = find_expectation(tuple);
-
+ spin_lock_bh(&nf_conntrack_lock);
+ exp = nf_ct_find_expectation(tuple);
if (exp) {
- DEBUGP("conntrack: expectation arrives ct=%p exp=%p\n",
- conntrack, exp);
+ pr_debug("conntrack: expectation arrives ct=%p exp=%p\n",
+ ct, exp);
/* Welcome, Mr. Bond. We've been expecting you... */
- __set_bit(IPS_EXPECTED_BIT, &conntrack->status);
- conntrack->master = exp->master;
+ __set_bit(IPS_EXPECTED_BIT, &ct->status);
+ ct->master = exp->master;
+ if (exp->helper) {
+ help = nf_ct_helper_ext_add(ct, GFP_ATOMIC);
+ if (help)
+ rcu_assign_pointer(help->helper, exp->helper);
+ }
+
#ifdef CONFIG_NF_CONNTRACK_MARK
- conntrack->mark = exp->master->mark;
+ ct->mark = exp->master->mark;
#endif
#ifdef CONFIG_NF_CONNTRACK_SECMARK
- conntrack->secmark = exp->master->secmark;
+ ct->secmark = exp->master->secmark;
#endif
- nf_conntrack_get(&conntrack->master->ct_general);
+ nf_conntrack_get(&ct->master->ct_general);
NF_CT_STAT_INC(expect_new);
} else {
- struct nf_conn_help *help = nfct_help(conntrack);
+ struct nf_conntrack_helper *helper;
- if (help)
- help->helper = __nf_ct_helper_find(&repl_tuple);
+ helper = __nf_ct_helper_find(&repl_tuple);
+ if (helper) {
+ help = nf_ct_helper_ext_add(ct, GFP_ATOMIC);
+ if (help)
+ rcu_assign_pointer(help->helper, helper);
+ }
NF_CT_STAT_INC(new);
}
/* Overload tuple linked list to put us in unconfirmed list. */
- list_add(&conntrack->tuplehash[IP_CT_DIR_ORIGINAL].list, &unconfirmed);
+ hlist_add_head(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnode, &unconfirmed);
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
if (exp) {
if (exp->expectfn)
- exp->expectfn(conntrack, exp);
- nf_conntrack_expect_put(exp);
+ exp->expectfn(ct, exp);
+ nf_ct_expect_put(exp);
}
- return &conntrack->tuplehash[IP_CT_DIR_ORIGINAL];
+ return &ct->tuplehash[IP_CT_DIR_ORIGINAL];
}
/* On success, returns conntrack ptr, sets skb->nfct and ctinfo */
u_int16_t l3num,
u_int8_t protonum,
struct nf_conntrack_l3proto *l3proto,
- struct nf_conntrack_protocol *proto,
+ struct nf_conntrack_l4proto *l4proto,
int *set_reply,
enum ip_conntrack_info *ctinfo)
{
struct nf_conntrack_tuple_hash *h;
struct nf_conn *ct;
- if (!nf_ct_get_tuple(skb, (unsigned int)(skb->nh.raw - skb->data),
+ if (!nf_ct_get_tuple(skb, skb_network_offset(skb),
dataoff, l3num, protonum, &tuple, l3proto,
- proto)) {
- DEBUGP("resolve_normal_ct: Can't get tuple\n");
+ l4proto)) {
+ pr_debug("resolve_normal_ct: Can't get tuple\n");
return NULL;
}
/* look for tuple match */
- h = nf_conntrack_find_get(&tuple, NULL);
+ h = nf_conntrack_find_get(&tuple);
if (!h) {
- h = init_conntrack(&tuple, l3proto, proto, skb, dataoff);
+ h = init_conntrack(&tuple, l3proto, l4proto, skb, dataoff);
if (!h)
return NULL;
if (IS_ERR(h))
} else {
/* Once we've had two way comms, always ESTABLISHED. */
if (test_bit(IPS_SEEN_REPLY_BIT, &ct->status)) {
- DEBUGP("nf_conntrack_in: normal packet for %p\n", ct);
+ pr_debug("nf_conntrack_in: normal packet for %p\n", ct);
*ctinfo = IP_CT_ESTABLISHED;
} else if (test_bit(IPS_EXPECTED_BIT, &ct->status)) {
- DEBUGP("nf_conntrack_in: related packet for %p\n", ct);
+ pr_debug("nf_conntrack_in: related packet for %p\n",
+ ct);
*ctinfo = IP_CT_RELATED;
} else {
- DEBUGP("nf_conntrack_in: new packet for %p\n", ct);
+ pr_debug("nf_conntrack_in: new packet for %p\n", ct);
*ctinfo = IP_CT_NEW;
}
*set_reply = 0;
}
unsigned int
-nf_conntrack_in(int pf, unsigned int hooknum, struct sk_buff **pskb)
+nf_conntrack_in(int pf, unsigned int hooknum, struct sk_buff *skb)
{
struct nf_conn *ct;
enum ip_conntrack_info ctinfo;
struct nf_conntrack_l3proto *l3proto;
- struct nf_conntrack_protocol *proto;
+ struct nf_conntrack_l4proto *l4proto;
unsigned int dataoff;
u_int8_t protonum;
int set_reply = 0;
int ret;
/* Previously seen (loopback or untracked)? Ignore. */
- if ((*pskb)->nfct) {
- NF_CT_STAT_INC(ignore);
+ if (skb->nfct) {
+ NF_CT_STAT_INC_ATOMIC(ignore);
return NF_ACCEPT;
}
+ /* rcu_read_lock()ed by nf_hook_slow */
l3proto = __nf_ct_l3proto_find((u_int16_t)pf);
- if ((ret = l3proto->prepare(pskb, hooknum, &dataoff, &protonum)) <= 0) {
- DEBUGP("not prepared to track yet or error occured\n");
+ ret = l3proto->get_l4proto(skb, skb_network_offset(skb),
+ &dataoff, &protonum);
+ if (ret <= 0) {
+ pr_debug("not prepared to track yet or error occured\n");
+ NF_CT_STAT_INC_ATOMIC(error);
+ NF_CT_STAT_INC_ATOMIC(invalid);
return -ret;
}
- proto = __nf_ct_proto_find((u_int16_t)pf, protonum);
+ l4proto = __nf_ct_l4proto_find((u_int16_t)pf, protonum);
/* It may be an special packet, error, unclean...
* inverse of the return code tells to the netfilter
* core what to do with the packet. */
- if (proto->error != NULL &&
- (ret = proto->error(*pskb, dataoff, &ctinfo, pf, hooknum)) <= 0) {
- NF_CT_STAT_INC(error);
- NF_CT_STAT_INC(invalid);
+ if (l4proto->error != NULL &&
+ (ret = l4proto->error(skb, dataoff, &ctinfo, pf, hooknum)) <= 0) {
+ NF_CT_STAT_INC_ATOMIC(error);
+ NF_CT_STAT_INC_ATOMIC(invalid);
return -ret;
}
- ct = resolve_normal_ct(*pskb, dataoff, pf, protonum, l3proto, proto,
+ ct = resolve_normal_ct(skb, dataoff, pf, protonum, l3proto, l4proto,
&set_reply, &ctinfo);
if (!ct) {
/* Not valid part of a connection */
- NF_CT_STAT_INC(invalid);
+ NF_CT_STAT_INC_ATOMIC(invalid);
return NF_ACCEPT;
}
if (IS_ERR(ct)) {
/* Too stressed to deal. */
- NF_CT_STAT_INC(drop);
+ NF_CT_STAT_INC_ATOMIC(drop);
return NF_DROP;
}
- NF_CT_ASSERT((*pskb)->nfct);
+ NF_CT_ASSERT(skb->nfct);
- ret = proto->packet(ct, *pskb, dataoff, ctinfo, pf, hooknum);
+ ret = l4proto->packet(ct, skb, dataoff, ctinfo, pf, hooknum);
if (ret < 0) {
/* Invalid: inverse of the return code tells
* the netfilter core what to do */
- DEBUGP("nf_conntrack_in: Can't track with proto module\n");
- nf_conntrack_put((*pskb)->nfct);
- (*pskb)->nfct = NULL;
- NF_CT_STAT_INC(invalid);
+ pr_debug("nf_conntrack_in: Can't track with proto module\n");
+ nf_conntrack_put(skb->nfct);
+ skb->nfct = NULL;
+ NF_CT_STAT_INC_ATOMIC(invalid);
return -ret;
}
if (set_reply && !test_and_set_bit(IPS_SEEN_REPLY_BIT, &ct->status))
- nf_conntrack_event_cache(IPCT_STATUS, *pskb);
+ nf_conntrack_event_cache(IPCT_STATUS, skb);
return ret;
}
+EXPORT_SYMBOL_GPL(nf_conntrack_in);
-int nf_ct_invert_tuplepr(struct nf_conntrack_tuple *inverse,
- const struct nf_conntrack_tuple *orig)
+bool nf_ct_invert_tuplepr(struct nf_conntrack_tuple *inverse,
+ const struct nf_conntrack_tuple *orig)
{
- return nf_ct_invert_tuple(inverse, orig,
- __nf_ct_l3proto_find(orig->src.l3num),
- __nf_ct_proto_find(orig->src.l3num,
- orig->dst.protonum));
+ bool ret;
+
+ rcu_read_lock();
+ ret = nf_ct_invert_tuple(inverse, orig,
+ __nf_ct_l3proto_find(orig->src.l3num),
+ __nf_ct_l4proto_find(orig->src.l3num,
+ orig->dst.protonum));
+ rcu_read_unlock();
+ return ret;
}
+EXPORT_SYMBOL_GPL(nf_ct_invert_tuplepr);
+
+/* Alter reply tuple (maybe alter helper). This is for NAT, and is
+ implicitly racy: see __nf_conntrack_confirm */
+void nf_conntrack_alter_reply(struct nf_conn *ct,
+ const struct nf_conntrack_tuple *newreply)
+{
+ struct nf_conn_help *help = nfct_help(ct);
+ struct nf_conntrack_helper *helper;
+
+ /* Should be unconfirmed, so not in hash table yet */
+ NF_CT_ASSERT(!nf_ct_is_confirmed(ct));
+
+ pr_debug("Altering reply tuple of %p to ", ct);
+ nf_ct_dump_tuple(newreply);
+
+ ct->tuplehash[IP_CT_DIR_REPLY].tuple = *newreply;
+ if (ct->master || (help && !hlist_empty(&help->expectations)))
+ return;
+
+ rcu_read_lock();
+ helper = __nf_ct_helper_find(newreply);
+ if (helper == NULL) {
+ if (help)
+ rcu_assign_pointer(help->helper, NULL);
+ goto out;
+ }
+
+ if (help == NULL) {
+ help = nf_ct_helper_ext_add(ct, GFP_ATOMIC);
+ if (help == NULL)
+ goto out;
+ } else {
+ memset(&help->help, 0, sizeof(help->help));
+ }
+
+ rcu_assign_pointer(help->helper, helper);
+out:
+ rcu_read_unlock();
+}
+EXPORT_SYMBOL_GPL(nf_conntrack_alter_reply);
/* Refresh conntrack for this many jiffies and do accounting if do_acct is 1 */
void __nf_ct_refresh_acct(struct nf_conn *ct,
NF_CT_ASSERT(ct->timeout.data == (unsigned long)ct);
NF_CT_ASSERT(skb);
- write_lock_bh(&nf_conntrack_lock);
+ spin_lock_bh(&nf_conntrack_lock);
/* Only update if this is not a fixed timeout */
- if (test_bit(IPS_FIXED_TIMEOUT_BIT, &ct->status)) {
- write_unlock_bh(&nf_conntrack_lock);
- return;
- }
+ if (test_bit(IPS_FIXED_TIMEOUT_BIT, &ct->status))
+ goto acct;
/* If not in hash table, timer will not be active yet */
if (!nf_ct_is_confirmed(ct)) {
ct->timeout.expires = extra_jiffies;
event = IPCT_REFRESH;
} else {
- /* Need del_timer for race avoidance (may already be dying). */
- if (del_timer(&ct->timeout)) {
- ct->timeout.expires = jiffies + extra_jiffies;
+ unsigned long newtime = jiffies + extra_jiffies;
+
+ /* Only update the timeout if the new timeout is at least
+ HZ jiffies from the old timeout. Need del_timer for race
+ avoidance (may already be dying). */
+ if (newtime - ct->timeout.expires >= HZ
+ && del_timer(&ct->timeout)) {
+ ct->timeout.expires = newtime;
add_timer(&ct->timeout);
event = IPCT_REFRESH;
}
}
+acct:
#ifdef CONFIG_NF_CT_ACCT
if (do_acct) {
ct->counters[CTINFO2DIR(ctinfo)].packets++;
ct->counters[CTINFO2DIR(ctinfo)].bytes +=
- skb->len - (unsigned int)(skb->nh.raw - skb->data);
- if ((ct->counters[CTINFO2DIR(ctinfo)].packets & 0x80000000)
- || (ct->counters[CTINFO2DIR(ctinfo)].bytes & 0x80000000))
- event |= IPCT_COUNTER_FILLING;
+ skb->len - skb_network_offset(skb);
+
+ if ((ct->counters[CTINFO2DIR(ctinfo)].packets & 0x80000000)
+ || (ct->counters[CTINFO2DIR(ctinfo)].bytes & 0x80000000))
+ event |= IPCT_COUNTER_FILLING;
}
#endif
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
/* must be unlocked when calling event cache */
if (event)
nf_conntrack_event_cache(event, skb);
}
+EXPORT_SYMBOL_GPL(__nf_ct_refresh_acct);
+
+void __nf_ct_kill_acct(struct nf_conn *ct,
+ enum ip_conntrack_info ctinfo,
+ const struct sk_buff *skb,
+ int do_acct)
+{
+#ifdef CONFIG_NF_CT_ACCT
+ if (do_acct) {
+ spin_lock_bh(&nf_conntrack_lock);
+ ct->counters[CTINFO2DIR(ctinfo)].packets++;
+ ct->counters[CTINFO2DIR(ctinfo)].bytes +=
+ skb->len - skb_network_offset(skb);
+ spin_unlock_bh(&nf_conntrack_lock);
+ }
+#endif
+ if (del_timer(&ct->timeout))
+ ct->timeout.function((unsigned long)ct);
+}
+EXPORT_SYMBOL_GPL(__nf_ct_kill_acct);
-#if defined(CONFIG_NF_CT_NETLINK) || \
- defined(CONFIG_NF_CT_NETLINK_MODULE)
+#if defined(CONFIG_NF_CT_NETLINK) || defined(CONFIG_NF_CT_NETLINK_MODULE)
#include <linux/netfilter/nfnetlink.h>
#include <linux/netfilter/nfnetlink_conntrack.h>
#include <linux/mutex.h>
-
/* Generic function for tcp/udp/sctp/dccp and alike. This needs to be
* in ip_conntrack_core, since we don't want the protocols to autoload
* or depend on ctnetlink */
-int nf_ct_port_tuple_to_nfattr(struct sk_buff *skb,
+int nf_ct_port_tuple_to_nlattr(struct sk_buff *skb,
const struct nf_conntrack_tuple *tuple)
{
- NFA_PUT(skb, CTA_PROTO_SRC_PORT, sizeof(u_int16_t),
- &tuple->src.u.tcp.port);
- NFA_PUT(skb, CTA_PROTO_DST_PORT, sizeof(u_int16_t),
- &tuple->dst.u.tcp.port);
+ NLA_PUT_BE16(skb, CTA_PROTO_SRC_PORT, tuple->src.u.tcp.port);
+ NLA_PUT_BE16(skb, CTA_PROTO_DST_PORT, tuple->dst.u.tcp.port);
return 0;
-nfattr_failure:
+nla_put_failure:
return -1;
}
+EXPORT_SYMBOL_GPL(nf_ct_port_tuple_to_nlattr);
-static const size_t cta_min_proto[CTA_PROTO_MAX] = {
- [CTA_PROTO_SRC_PORT-1] = sizeof(u_int16_t),
- [CTA_PROTO_DST_PORT-1] = sizeof(u_int16_t)
+const struct nla_policy nf_ct_port_nla_policy[CTA_PROTO_MAX+1] = {
+ [CTA_PROTO_SRC_PORT] = { .type = NLA_U16 },
+ [CTA_PROTO_DST_PORT] = { .type = NLA_U16 },
};
+EXPORT_SYMBOL_GPL(nf_ct_port_nla_policy);
-int nf_ct_port_nfattr_to_tuple(struct nfattr *tb[],
+int nf_ct_port_nlattr_to_tuple(struct nlattr *tb[],
struct nf_conntrack_tuple *t)
{
- if (!tb[CTA_PROTO_SRC_PORT-1] || !tb[CTA_PROTO_DST_PORT-1])
+ if (!tb[CTA_PROTO_SRC_PORT] || !tb[CTA_PROTO_DST_PORT])
return -EINVAL;
- if (nfattr_bad_size(tb, CTA_PROTO_MAX, cta_min_proto))
- return -EINVAL;
-
- t->src.u.tcp.port =
- *(u_int16_t *)NFA_DATA(tb[CTA_PROTO_SRC_PORT-1]);
- t->dst.u.tcp.port =
- *(u_int16_t *)NFA_DATA(tb[CTA_PROTO_DST_PORT-1]);
+ t->src.u.tcp.port = nla_get_be16(tb[CTA_PROTO_SRC_PORT]);
+ t->dst.u.tcp.port = nla_get_be16(tb[CTA_PROTO_DST_PORT]);
return 0;
}
+EXPORT_SYMBOL_GPL(nf_ct_port_nlattr_to_tuple);
#endif
/* Used by ipt_REJECT and ip6t_REJECT. */
-void __nf_conntrack_attach(struct sk_buff *nskb, struct sk_buff *skb)
+static void nf_conntrack_attach(struct sk_buff *nskb, struct sk_buff *skb)
{
struct nf_conn *ct;
enum ip_conntrack_info ctinfo;
nf_conntrack_get(nskb->nfct);
}
-static inline int
-do_iter(const struct nf_conntrack_tuple_hash *i,
- int (*iter)(struct nf_conn *i, void *data),
- void *data)
-{
- return iter(nf_ct_tuplehash_to_ctrack(i), data);
-}
-
/* Bring out ya dead! */
static struct nf_conn *
get_next_corpse(int (*iter)(struct nf_conn *i, void *data),
{
struct nf_conntrack_tuple_hash *h;
struct nf_conn *ct;
+ struct hlist_node *n;
- write_lock_bh(&nf_conntrack_lock);
+ spin_lock_bh(&nf_conntrack_lock);
for (; *bucket < nf_conntrack_htable_size; (*bucket)++) {
- list_for_each_entry(h, &nf_conntrack_hash[*bucket], list) {
+ hlist_for_each_entry(h, n, &nf_conntrack_hash[*bucket], hnode) {
ct = nf_ct_tuplehash_to_ctrack(h);
if (iter(ct, data))
goto found;
}
- }
- list_for_each_entry(h, &unconfirmed, list) {
+ }
+ hlist_for_each_entry(h, n, &unconfirmed, hnode) {
ct = nf_ct_tuplehash_to_ctrack(h);
if (iter(ct, data))
- goto found;
+ set_bit(IPS_DYING_BIT, &ct->status);
}
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
return NULL;
found:
atomic_inc(&ct->ct_general.use);
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
return ct;
}
nf_ct_put(ct);
}
}
+EXPORT_SYMBOL_GPL(nf_ct_iterate_cleanup);
static int kill_all(struct nf_conn *i, void *data)
{
return 1;
}
-static void free_conntrack_hash(struct list_head *hash, int vmalloced, int size)
+void nf_ct_free_hashtable(struct hlist_head *hash, int vmalloced, unsigned int size)
{
if (vmalloced)
vfree(hash);
else
- free_pages((unsigned long)hash,
- get_order(sizeof(struct list_head) * size));
+ free_pages((unsigned long)hash,
+ get_order(sizeof(struct hlist_head) * size));
}
+EXPORT_SYMBOL_GPL(nf_ct_free_hashtable);
-void nf_conntrack_flush()
+void nf_conntrack_flush(void)
{
nf_ct_iterate_cleanup(kill_all, NULL);
}
+EXPORT_SYMBOL_GPL(nf_conntrack_flush);
/* Mishearing the voices in his head, our hero wonders how he's
supposed to kill the mall. */
void nf_conntrack_cleanup(void)
{
- int i;
-
- ip_ct_attach = NULL;
+ rcu_assign_pointer(ip_ct_attach, NULL);
/* This makes sure all current packets have passed through
netfilter framework. Roll on, two-stage module
while (atomic_read(&nf_conntrack_untracked.ct_general.use) > 1)
schedule();
- for (i = 0; i < NF_CT_F_NUM; i++) {
- if (nf_ct_cache[i].use == 0)
- continue;
+ rcu_assign_pointer(nf_ct_destroy, NULL);
- NF_CT_ASSERT(nf_ct_cache[i].use == 1);
- nf_ct_cache[i].use = 1;
- nf_conntrack_unregister_cache(i);
- }
- kmem_cache_destroy(nf_conntrack_expect_cachep);
- free_conntrack_hash(nf_conntrack_hash, nf_conntrack_vmalloc,
- nf_conntrack_htable_size);
-
- /* free l3proto protocol tables */
- for (i = 0; i < PF_MAX; i++)
- if (nf_ct_protos[i]) {
- kfree(nf_ct_protos[i]);
- nf_ct_protos[i] = NULL;
- }
+ kmem_cache_destroy(nf_conntrack_cachep);
+ nf_ct_free_hashtable(nf_conntrack_hash, nf_conntrack_vmalloc,
+ nf_conntrack_htable_size);
+
+ nf_conntrack_proto_fini();
+ nf_conntrack_helper_fini();
+ nf_conntrack_expect_fini();
}
-static struct list_head *alloc_hashtable(int size, int *vmalloced)
+struct hlist_head *nf_ct_alloc_hashtable(unsigned int *sizep, int *vmalloced)
{
- struct list_head *hash;
- unsigned int i;
+ struct hlist_head *hash;
+ unsigned int size, i;
+
+ *vmalloced = 0;
- *vmalloced = 0;
- hash = (void*)__get_free_pages(GFP_KERNEL,
- get_order(sizeof(struct list_head)
+ size = *sizep = roundup(*sizep, PAGE_SIZE / sizeof(struct hlist_head));
+ hash = (void*)__get_free_pages(GFP_KERNEL|__GFP_NOWARN,
+ get_order(sizeof(struct hlist_head)
* size));
- if (!hash) {
+ if (!hash) {
*vmalloced = 1;
printk(KERN_WARNING "nf_conntrack: falling back to vmalloc.\n");
- hash = vmalloc(sizeof(struct list_head) * size);
+ hash = vmalloc(sizeof(struct hlist_head) * size);
}
if (hash)
- for (i = 0; i < size; i++)
- INIT_LIST_HEAD(&hash[i]);
+ for (i = 0; i < size; i++)
+ INIT_HLIST_HEAD(&hash[i]);
return hash;
}
+EXPORT_SYMBOL_GPL(nf_ct_alloc_hashtable);
-int set_hashsize(const char *val, struct kernel_param *kp)
+int nf_conntrack_set_hashsize(const char *val, struct kernel_param *kp)
{
- int i, bucket, hashsize, vmalloced;
- int old_vmalloced, old_size;
+ int i, bucket, vmalloced, old_vmalloced;
+ unsigned int hashsize, old_size;
int rnd;
- struct list_head *hash, *old_hash;
+ struct hlist_head *hash, *old_hash;
struct nf_conntrack_tuple_hash *h;
/* On boot, we can set this without any fancy locking. */
if (!nf_conntrack_htable_size)
return param_set_uint(val, kp);
- hashsize = simple_strtol(val, NULL, 0);
+ hashsize = simple_strtoul(val, NULL, 0);
if (!hashsize)
return -EINVAL;
- hash = alloc_hashtable(hashsize, &vmalloced);
+ hash = nf_ct_alloc_hashtable(&hashsize, &vmalloced);
if (!hash)
return -ENOMEM;
* use a newrandom seed */
get_random_bytes(&rnd, 4);
- write_lock_bh(&nf_conntrack_lock);
+ /* Lookups in the old hash might happen in parallel, which means we
+ * might get false negatives during connection lookup. New connections
+ * created because of a false negative won't make it into the hash
+ * though since that required taking the lock.
+ */
+ spin_lock_bh(&nf_conntrack_lock);
for (i = 0; i < nf_conntrack_htable_size; i++) {
- while (!list_empty(&nf_conntrack_hash[i])) {
- h = list_entry(nf_conntrack_hash[i].next,
- struct nf_conntrack_tuple_hash, list);
- list_del(&h->list);
+ while (!hlist_empty(&nf_conntrack_hash[i])) {
+ h = hlist_entry(nf_conntrack_hash[i].first,
+ struct nf_conntrack_tuple_hash, hnode);
+ hlist_del_rcu(&h->hnode);
bucket = __hash_conntrack(&h->tuple, hashsize, rnd);
- list_add_tail(&h->list, &hash[bucket]);
+ hlist_add_head(&h->hnode, &hash[bucket]);
}
}
old_size = nf_conntrack_htable_size;
nf_conntrack_vmalloc = vmalloced;
nf_conntrack_hash = hash;
nf_conntrack_hash_rnd = rnd;
- write_unlock_bh(&nf_conntrack_lock);
+ spin_unlock_bh(&nf_conntrack_lock);
- free_conntrack_hash(old_hash, old_vmalloced, old_size);
+ nf_ct_free_hashtable(old_hash, old_vmalloced, old_size);
return 0;
}
+EXPORT_SYMBOL_GPL(nf_conntrack_set_hashsize);
-module_param_call(hashsize, set_hashsize, param_get_uint,
+module_param_call(hashsize, nf_conntrack_set_hashsize, param_get_uint,
&nf_conntrack_htable_size, 0600);
int __init nf_conntrack_init(void)
{
- unsigned int i;
+ int max_factor = 8;
int ret;
/* Idea from tcp.c: use 1/16384 of memory. On i386: 32MB
- * machine has 256 buckets. >= 1GB machines have 8192 buckets. */
+ * machine has 512 buckets. >= 1GB machines have 16384 buckets. */
if (!nf_conntrack_htable_size) {
nf_conntrack_htable_size
= (((num_physpages << PAGE_SHIFT) / 16384)
- / sizeof(struct list_head));
+ / sizeof(struct hlist_head));
if (num_physpages > (1024 * 1024 * 1024 / PAGE_SIZE))
- nf_conntrack_htable_size = 8192;
- if (nf_conntrack_htable_size < 16)
- nf_conntrack_htable_size = 16;
+ nf_conntrack_htable_size = 16384;
+ if (nf_conntrack_htable_size < 32)
+ nf_conntrack_htable_size = 32;
+
+ /* Use a max. factor of four by default to get the same max as
+ * with the old struct list_heads. When a table size is given
+ * we use the old value of 8 to avoid reducing the max.
+ * entries. */
+ max_factor = 4;
}
- nf_conntrack_max = 8 * nf_conntrack_htable_size;
-
- printk("nf_conntrack version %s (%u buckets, %d max)\n",
- NF_CONNTRACK_VERSION, nf_conntrack_htable_size,
- nf_conntrack_max);
-
- nf_conntrack_hash = alloc_hashtable(nf_conntrack_htable_size,
- &nf_conntrack_vmalloc);
+ nf_conntrack_hash = nf_ct_alloc_hashtable(&nf_conntrack_htable_size,
+ &nf_conntrack_vmalloc);
if (!nf_conntrack_hash) {
printk(KERN_ERR "Unable to create nf_conntrack_hash\n");
goto err_out;
}
- ret = nf_conntrack_register_cache(NF_CT_F_BASIC, "nf_conntrack:basic",
- sizeof(struct nf_conn));
- if (ret < 0) {
+ nf_conntrack_max = max_factor * nf_conntrack_htable_size;
+
+ printk("nf_conntrack version %s (%u buckets, %d max)\n",
+ NF_CONNTRACK_VERSION, nf_conntrack_htable_size,
+ nf_conntrack_max);
+
+ nf_conntrack_cachep = kmem_cache_create("nf_conntrack",
+ sizeof(struct nf_conn),
+ 0, 0, NULL);
+ if (!nf_conntrack_cachep) {
printk(KERN_ERR "Unable to create nf_conn slab cache\n");
goto err_free_hash;
}
- nf_conntrack_expect_cachep = kmem_cache_create("nf_conntrack_expect",
- sizeof(struct nf_conntrack_expect),
- 0, 0, NULL, NULL);
- if (!nf_conntrack_expect_cachep) {
- printk(KERN_ERR "Unable to create nf_expect slab cache\n");
+ ret = nf_conntrack_proto_init();
+ if (ret < 0)
goto err_free_conntrack_slab;
- }
- /* Don't NEED lock here, but good form anyway. */
- write_lock_bh(&nf_conntrack_lock);
- for (i = 0; i < PF_MAX; i++)
- nf_ct_l3protos[i] = &nf_conntrack_generic_l3proto;
- write_unlock_bh(&nf_conntrack_lock);
+ ret = nf_conntrack_expect_init();
+ if (ret < 0)
+ goto out_fini_proto;
+
+ ret = nf_conntrack_helper_init();
+ if (ret < 0)
+ goto out_fini_expect;
/* For use by REJECT target */
- ip_ct_attach = __nf_conntrack_attach;
+ rcu_assign_pointer(ip_ct_attach, nf_conntrack_attach);
+ rcu_assign_pointer(nf_ct_destroy, destroy_conntrack);
/* Set up fake conntrack:
- to never be deleted, not in any hashes */
return ret;
+out_fini_expect:
+ nf_conntrack_expect_fini();
+out_fini_proto:
+ nf_conntrack_proto_fini();
err_free_conntrack_slab:
- nf_conntrack_unregister_cache(NF_CT_F_BASIC);
+ kmem_cache_destroy(nf_conntrack_cachep);
err_free_hash:
- free_conntrack_hash(nf_conntrack_hash, nf_conntrack_vmalloc,
- nf_conntrack_htable_size);
+ nf_ct_free_hashtable(nf_conntrack_hash, nf_conntrack_vmalloc,
+ nf_conntrack_htable_size);
err_out:
return -ENOMEM;
}
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff