[NETFILTER]: nf_conntrack_expect: use RCU for expectation hash
Use RCU for expectation hash. This doesn't buy much for conntrack runtime
performance, but allows to reduce the use of nf_conntrack_lock for /proc
and nf_netlink_conntrack.
Signed-off-by: Patrick McHardy <kaber@xxxxxxxxx>
---
commit 504ca298fd3846f894b3979666df5bf5021707ff
tree 2b94fd833ebb414fd8305ada3295a3fd5fb5e516
parent 23cf79fc3f39291ad9e98a00584d1f0c0b8c5d25
author Patrick McHardy <kaber@xxxxxxxxx> Tue, 29 Jan 2008 16:22:11 +0100
committer Patrick McHardy <kaber@xxxxxxxxx> Wed, 30 Jan 2008 21:03:06 +0100
include/net/netfilter/nf_conntrack_expect.h | 2 +
.../netfilter/nf_conntrack_l3proto_ipv4_compat.c | 14 ++++---
net/netfilter/nf_conntrack_expect.c | 43 +++++++++++++-------
net/netfilter/nf_conntrack_netlink.c | 7 ++-
4 files changed, 41 insertions(+), 25 deletions(-)
diff --git a/include/net/netfilter/nf_conntrack_expect.h b/include/net/netfilter/nf_conntrack_expect.h
index 6c3fd25..cb608a1 100644
--- a/include/net/netfilter/nf_conntrack_expect.h
+++ b/include/net/netfilter/nf_conntrack_expect.h
@@ -49,6 +49,8 @@ struct nf_conntrack_expect
/* Direction relative to the master connection. */
enum ip_conntrack_dir dir;
#endif
+
+ struct rcu_head rcu;
};
#define NF_CT_EXPECT_PERMANENT 0x1
diff --git a/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4_compat.c b/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4_compat.c
index 543c02b..2fdcd92 100644
--- a/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4_compat.c
+++ b/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4_compat.c
@@ -191,10 +191,12 @@ struct ct_expect_iter_state {
static struct hlist_node *ct_expect_get_first(struct seq_file *seq)
{
struct ct_expect_iter_state *st = seq->private;
+ struct hlist_node *n;
for (st->bucket = 0; st->bucket < nf_ct_expect_hsize; st->bucket++) {
- if (!hlist_empty(&nf_ct_expect_hash[st->bucket]))
- return nf_ct_expect_hash[st->bucket].first;
+ n = rcu_dereference(nf_ct_expect_hash[st->bucket].first);
+ if (n)
+ return n;
}
return NULL;
}
@@ -204,11 +206,11 @@ static struct hlist_node *ct_expect_get_next(struct seq_file *seq,
{
struct ct_expect_iter_state *st = seq->private;
- head = head->next;
+ head = rcu_dereference(head->next);
while (head == NULL) {
if (++st->bucket >= nf_ct_expect_hsize)
return NULL;
- head = nf_ct_expect_hash[st->bucket].first;
+ head = rcu_dereference(nf_ct_expect_hash[st->bucket].first);
}
return head;
}
@@ -225,7 +227,7 @@ static struct hlist_node *ct_expect_get_idx(struct seq_file *seq, loff_t pos)
static void *exp_seq_start(struct seq_file *seq, loff_t *pos)
{
- read_lock_bh(&nf_conntrack_lock);
+ rcu_read_lock();
return ct_expect_get_idx(seq, *pos);
}
@@ -237,7 +239,7 @@ static void *exp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
static void exp_seq_stop(struct seq_file *seq, void *v)
{
- read_unlock_bh(&nf_conntrack_lock);
+ rcu_read_unlock();
}
static int exp_seq_show(struct seq_file *s, void *v)
diff --git a/net/netfilter/nf_conntrack_expect.c b/net/netfilter/nf_conntrack_expect.c
index e405079..a5c8ef0 100644
--- a/net/netfilter/nf_conntrack_expect.c
+++ b/net/netfilter/nf_conntrack_expect.c
@@ -50,7 +50,7 @@ void nf_ct_unlink_expect(struct nf_conntrack_expect *exp)
NF_CT_ASSERT(master_help);
NF_CT_ASSERT(!timer_pending(&exp->timeout));
- hlist_del(&exp->hnode);
+ hlist_del_rcu(&exp->hnode);
nf_ct_expect_count--;
hlist_del(&exp->lnode);
@@ -97,7 +97,7 @@ __nf_ct_expect_find(const struct nf_conntrack_tuple *tuple)
return NULL;
h = nf_ct_expect_dst_hash(tuple);
- hlist_for_each_entry(i, n, &nf_ct_expect_hash[h], hnode) {
+ hlist_for_each_entry_rcu(i, n, &nf_ct_expect_hash[h], hnode) {
if (nf_ct_tuple_mask_cmp(tuple, &i->tuple, &i->mask))
return i;
}
@@ -111,11 +111,11 @@ nf_ct_expect_find_get(const struct nf_conntrack_tuple *tuple)
{
struct nf_conntrack_expect *i;
- read_lock_bh(&nf_conntrack_lock);
+ rcu_read_lock();
i = __nf_ct_expect_find(tuple);
- if (i)
- atomic_inc(&i->use);
- read_unlock_bh(&nf_conntrack_lock);
+ if (i && !atomic_inc_not_zero(&i->use))
+ i = NULL;
+ rcu_read_unlock();
return i;
}
@@ -223,6 +223,7 @@ struct nf_conntrack_expect *nf_ct_expect_alloc(struct nf_conn *me)
new->master = me;
atomic_set(&new->use, 1);
+ INIT_RCU_HEAD(&new->rcu);
return new;
}
EXPORT_SYMBOL_GPL(nf_ct_expect_alloc);
@@ -278,10 +279,18 @@ void nf_ct_expect_init(struct nf_conntrack_expect *exp, int family,
}
EXPORT_SYMBOL_GPL(nf_ct_expect_init);
+static void nf_ct_expect_free_rcu(struct rcu_head *head)
+{
+ struct nf_conntrack_expect *exp;
+
+ exp = container_of(head, struct nf_conntrack_expect, rcu);
+ kmem_cache_free(nf_ct_expect_cachep, exp);
+}
+
void nf_ct_expect_put(struct nf_conntrack_expect *exp)
{
if (atomic_dec_and_test(&exp->use))
- kmem_cache_free(nf_ct_expect_cachep, exp);
+ call_rcu(&exp->rcu, nf_ct_expect_free_rcu);
}
EXPORT_SYMBOL_GPL(nf_ct_expect_put);
@@ -295,7 +304,7 @@ static void nf_ct_expect_insert(struct nf_conntrack_expect *exp)
hlist_add_head(&exp->lnode, &master_help->expectations);
master_help->expecting++;
- hlist_add_head(&exp->hnode, &nf_ct_expect_hash[h]);
+ hlist_add_head_rcu(&exp->hnode, &nf_ct_expect_hash[h]);
nf_ct_expect_count++;
setup_timer(&exp->timeout, nf_ct_expectation_timed_out,
@@ -394,10 +403,12 @@ struct ct_expect_iter_state {
static struct hlist_node *ct_expect_get_first(struct seq_file *seq)
{
struct ct_expect_iter_state *st = seq->private;
+ struct hlist_node *n;
for (st->bucket = 0; st->bucket < nf_ct_expect_hsize; st->bucket++) {
- if (!hlist_empty(&nf_ct_expect_hash[st->bucket]))
- return nf_ct_expect_hash[st->bucket].first;
+ n = rcu_dereference(nf_ct_expect_hash[st->bucket].first);
+ if (n)
+ return n;
}
return NULL;
}
@@ -407,11 +418,11 @@ static struct hlist_node *ct_expect_get_next(struct seq_file *seq,
{
struct ct_expect_iter_state *st = seq->private;
- head = head->next;
+ head = rcu_dereference(head->next);
while (head == NULL) {
if (++st->bucket >= nf_ct_expect_hsize)
return NULL;
- head = nf_ct_expect_hash[st->bucket].first;
+ head = rcu_dereference(nf_ct_expect_hash[st->bucket].first);
}
return head;
}
@@ -427,9 +438,9 @@ static struct hlist_node *ct_expect_get_idx(struct seq_file *seq, loff_t pos)
}
static void *exp_seq_start(struct seq_file *seq, loff_t *pos)
- __acquires(nf_conntrack_lock)
+ __acquires(RCU)
{
- read_lock_bh(&nf_conntrack_lock);
+ rcu_read_lock();
return ct_expect_get_idx(seq, *pos);
}
@@ -440,9 +451,9 @@ static void *exp_seq_next(struct seq_file *seq, void *v, loff_t *pos)
}
static void exp_seq_stop(struct seq_file *seq, void *v)
- __releases(nf_conntrack_lock)
+ __releases(RCU)
{
- read_unlock_bh(&nf_conntrack_lock);
+ rcu_read_unlock();
}
static int exp_seq_show(struct seq_file *s, void *v)
diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c
index b6c0935..557f471 100644
--- a/net/netfilter/nf_conntrack_netlink.c
+++ b/net/netfilter/nf_conntrack_netlink.c
@@ -1471,7 +1471,7 @@ ctnetlink_exp_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
struct hlist_node *n;
u_int8_t l3proto = nfmsg->nfgen_family;
- read_lock_bh(&nf_conntrack_lock);
+ rcu_read_lock();
last = (struct nf_conntrack_expect *)cb->args[1];
for (; cb->args[0] < nf_ct_expect_hsize; cb->args[0]++) {
restart:
@@ -1488,7 +1488,8 @@ restart:
cb->nlh->nlmsg_seq,
IPCTNL_MSG_EXP_NEW,
1, exp) < 0) {
- atomic_inc(&exp->use);
+ if (!atomic_inc_not_zero(&exp->use))
+ continue;
cb->args[1] = (unsigned long)exp;
goto out;
}
@@ -1499,7 +1500,7 @@ restart:
}
}
out:
- read_unlock_bh(&nf_conntrack_lock);
+ rcu_read_unlock();
if (last)
nf_ct_expect_put(last);
-
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
