mirror of
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git
synced 2024-09-04 01:49:36 +00:00
netfilter: nf_conntrack: use safer way to lock all buckets
When we need to lock all buckets in the connection hashtable we'd attempt to lock 1024 spinlocks, which is way more preemption levels than supported by the kernel. Furthermore, this behavior was hidden by checking if lockdep is enabled, and if it was - use only 8 buckets(!). Fix this by using a global lock and synchronize all buckets on it when we need to lock them all. This is pretty heavyweight, but is only done when we need to resize the hashtable, and that doesn't happen often enough (or at all). Signed-off-by: Sasha Levin <sasha.levin@oracle.com> Acked-by: Jesper Dangaard Brouer <brouer@redhat.com> Reviewed-by: Florian Westphal <fw@strlen.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This commit is contained in:
parent
35b815392a
commit
b16c29191d
5 changed files with 35 additions and 19 deletions
|
@ -79,12 +79,10 @@ print_tuple(struct seq_file *s, const struct nf_conntrack_tuple *tuple,
|
||||||
const struct nf_conntrack_l3proto *l3proto,
|
const struct nf_conntrack_l3proto *l3proto,
|
||||||
const struct nf_conntrack_l4proto *proto);
|
const struct nf_conntrack_l4proto *proto);
|
||||||
|
|
||||||
#ifdef CONFIG_LOCKDEP
|
#define CONNTRACK_LOCKS 1024
|
||||||
# define CONNTRACK_LOCKS 8
|
|
||||||
#else
|
|
||||||
# define CONNTRACK_LOCKS 1024
|
|
||||||
#endif
|
|
||||||
extern spinlock_t nf_conntrack_locks[CONNTRACK_LOCKS];
|
extern spinlock_t nf_conntrack_locks[CONNTRACK_LOCKS];
|
||||||
|
void nf_conntrack_lock(spinlock_t *lock);
|
||||||
|
|
||||||
extern spinlock_t nf_conntrack_expect_lock;
|
extern spinlock_t nf_conntrack_expect_lock;
|
||||||
|
|
||||||
|
|
|
@ -66,6 +66,21 @@ EXPORT_SYMBOL_GPL(nf_conntrack_locks);
|
||||||
__cacheline_aligned_in_smp DEFINE_SPINLOCK(nf_conntrack_expect_lock);
|
__cacheline_aligned_in_smp DEFINE_SPINLOCK(nf_conntrack_expect_lock);
|
||||||
EXPORT_SYMBOL_GPL(nf_conntrack_expect_lock);
|
EXPORT_SYMBOL_GPL(nf_conntrack_expect_lock);
|
||||||
|
|
||||||
|
static __read_mostly spinlock_t nf_conntrack_locks_all_lock;
|
||||||
|
static __read_mostly bool nf_conntrack_locks_all;
|
||||||
|
|
||||||
|
void nf_conntrack_lock(spinlock_t *lock) __acquires(lock)
|
||||||
|
{
|
||||||
|
spin_lock(lock);
|
||||||
|
while (unlikely(nf_conntrack_locks_all)) {
|
||||||
|
spin_unlock(lock);
|
||||||
|
spin_lock(&nf_conntrack_locks_all_lock);
|
||||||
|
spin_unlock(&nf_conntrack_locks_all_lock);
|
||||||
|
spin_lock(lock);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
EXPORT_SYMBOL_GPL(nf_conntrack_lock);
|
||||||
|
|
||||||
static void nf_conntrack_double_unlock(unsigned int h1, unsigned int h2)
|
static void nf_conntrack_double_unlock(unsigned int h1, unsigned int h2)
|
||||||
{
|
{
|
||||||
h1 %= CONNTRACK_LOCKS;
|
h1 %= CONNTRACK_LOCKS;
|
||||||
|
@ -82,12 +97,12 @@ static bool nf_conntrack_double_lock(struct net *net, unsigned int h1,
|
||||||
h1 %= CONNTRACK_LOCKS;
|
h1 %= CONNTRACK_LOCKS;
|
||||||
h2 %= CONNTRACK_LOCKS;
|
h2 %= CONNTRACK_LOCKS;
|
||||||
if (h1 <= h2) {
|
if (h1 <= h2) {
|
||||||
spin_lock(&nf_conntrack_locks[h1]);
|
nf_conntrack_lock(&nf_conntrack_locks[h1]);
|
||||||
if (h1 != h2)
|
if (h1 != h2)
|
||||||
spin_lock_nested(&nf_conntrack_locks[h2],
|
spin_lock_nested(&nf_conntrack_locks[h2],
|
||||||
SINGLE_DEPTH_NESTING);
|
SINGLE_DEPTH_NESTING);
|
||||||
} else {
|
} else {
|
||||||
spin_lock(&nf_conntrack_locks[h2]);
|
nf_conntrack_lock(&nf_conntrack_locks[h2]);
|
||||||
spin_lock_nested(&nf_conntrack_locks[h1],
|
spin_lock_nested(&nf_conntrack_locks[h1],
|
||||||
SINGLE_DEPTH_NESTING);
|
SINGLE_DEPTH_NESTING);
|
||||||
}
|
}
|
||||||
|
@ -102,16 +117,19 @@ static void nf_conntrack_all_lock(void)
|
||||||
{
|
{
|
||||||
int i;
|
int i;
|
||||||
|
|
||||||
for (i = 0; i < CONNTRACK_LOCKS; i++)
|
spin_lock(&nf_conntrack_locks_all_lock);
|
||||||
spin_lock_nested(&nf_conntrack_locks[i], i);
|
nf_conntrack_locks_all = true;
|
||||||
|
|
||||||
|
for (i = 0; i < CONNTRACK_LOCKS; i++) {
|
||||||
|
spin_lock(&nf_conntrack_locks[i]);
|
||||||
|
spin_unlock(&nf_conntrack_locks[i]);
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
static void nf_conntrack_all_unlock(void)
|
static void nf_conntrack_all_unlock(void)
|
||||||
{
|
{
|
||||||
int i;
|
nf_conntrack_locks_all = false;
|
||||||
|
spin_unlock(&nf_conntrack_locks_all_lock);
|
||||||
for (i = 0; i < CONNTRACK_LOCKS; i++)
|
|
||||||
spin_unlock(&nf_conntrack_locks[i]);
|
|
||||||
}
|
}
|
||||||
|
|
||||||
unsigned int nf_conntrack_htable_size __read_mostly;
|
unsigned int nf_conntrack_htable_size __read_mostly;
|
||||||
|
@ -757,7 +775,7 @@ static noinline int early_drop(struct net *net, unsigned int _hash)
|
||||||
hash = hash_bucket(_hash, net);
|
hash = hash_bucket(_hash, net);
|
||||||
for (; i < net->ct.htable_size; i++) {
|
for (; i < net->ct.htable_size; i++) {
|
||||||
lockp = &nf_conntrack_locks[hash % CONNTRACK_LOCKS];
|
lockp = &nf_conntrack_locks[hash % CONNTRACK_LOCKS];
|
||||||
spin_lock(lockp);
|
nf_conntrack_lock(lockp);
|
||||||
if (read_seqcount_retry(&net->ct.generation, sequence)) {
|
if (read_seqcount_retry(&net->ct.generation, sequence)) {
|
||||||
spin_unlock(lockp);
|
spin_unlock(lockp);
|
||||||
goto restart;
|
goto restart;
|
||||||
|
@ -1382,7 +1400,7 @@ get_next_corpse(struct net *net, int (*iter)(struct nf_conn *i, void *data),
|
||||||
for (; *bucket < net->ct.htable_size; (*bucket)++) {
|
for (; *bucket < net->ct.htable_size; (*bucket)++) {
|
||||||
lockp = &nf_conntrack_locks[*bucket % CONNTRACK_LOCKS];
|
lockp = &nf_conntrack_locks[*bucket % CONNTRACK_LOCKS];
|
||||||
local_bh_disable();
|
local_bh_disable();
|
||||||
spin_lock(lockp);
|
nf_conntrack_lock(lockp);
|
||||||
if (*bucket < net->ct.htable_size) {
|
if (*bucket < net->ct.htable_size) {
|
||||||
hlist_nulls_for_each_entry(h, n, &net->ct.hash[*bucket], hnnode) {
|
hlist_nulls_for_each_entry(h, n, &net->ct.hash[*bucket], hnnode) {
|
||||||
if (NF_CT_DIRECTION(h) != IP_CT_DIR_ORIGINAL)
|
if (NF_CT_DIRECTION(h) != IP_CT_DIR_ORIGINAL)
|
||||||
|
|
|
@ -425,7 +425,7 @@ static void __nf_conntrack_helper_unregister(struct nf_conntrack_helper *me,
|
||||||
}
|
}
|
||||||
local_bh_disable();
|
local_bh_disable();
|
||||||
for (i = 0; i < net->ct.htable_size; i++) {
|
for (i = 0; i < net->ct.htable_size; i++) {
|
||||||
spin_lock(&nf_conntrack_locks[i % CONNTRACK_LOCKS]);
|
nf_conntrack_lock(&nf_conntrack_locks[i % CONNTRACK_LOCKS]);
|
||||||
if (i < net->ct.htable_size) {
|
if (i < net->ct.htable_size) {
|
||||||
hlist_nulls_for_each_entry(h, nn, &net->ct.hash[i], hnnode)
|
hlist_nulls_for_each_entry(h, nn, &net->ct.hash[i], hnnode)
|
||||||
unhelp(h, me);
|
unhelp(h, me);
|
||||||
|
|
|
@ -840,7 +840,7 @@ ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
|
||||||
for (; cb->args[0] < net->ct.htable_size; cb->args[0]++) {
|
for (; cb->args[0] < net->ct.htable_size; cb->args[0]++) {
|
||||||
restart:
|
restart:
|
||||||
lockp = &nf_conntrack_locks[cb->args[0] % CONNTRACK_LOCKS];
|
lockp = &nf_conntrack_locks[cb->args[0] % CONNTRACK_LOCKS];
|
||||||
spin_lock(lockp);
|
nf_conntrack_lock(lockp);
|
||||||
if (cb->args[0] >= net->ct.htable_size) {
|
if (cb->args[0] >= net->ct.htable_size) {
|
||||||
spin_unlock(lockp);
|
spin_unlock(lockp);
|
||||||
goto out;
|
goto out;
|
||||||
|
|
|
@ -307,12 +307,12 @@ static void ctnl_untimeout(struct net *net, struct ctnl_timeout *timeout)
|
||||||
|
|
||||||
local_bh_disable();
|
local_bh_disable();
|
||||||
for (i = 0; i < net->ct.htable_size; i++) {
|
for (i = 0; i < net->ct.htable_size; i++) {
|
||||||
spin_lock(&nf_conntrack_locks[i % CONNTRACK_LOCKS]);
|
nf_conntrack_lock(&nf_conntrack_locks[i % CONNTRACK_LOCKS]);
|
||||||
if (i < net->ct.htable_size) {
|
if (i < net->ct.htable_size) {
|
||||||
hlist_nulls_for_each_entry(h, nn, &net->ct.hash[i], hnnode)
|
hlist_nulls_for_each_entry(h, nn, &net->ct.hash[i], hnnode)
|
||||||
untimeout(h, timeout);
|
untimeout(h, timeout);
|
||||||
}
|
}
|
||||||
spin_unlock(&nf_conntrack_locks[i % CONNTRACK_LOCKS]);
|
nf_conntrack_lock(&nf_conntrack_locks[i % CONNTRACK_LOCKS]);
|
||||||
}
|
}
|
||||||
local_bh_enable();
|
local_bh_enable();
|
||||||
}
|
}
|
||||||
|
|
Loading…
Reference in a new issue