struct conntrack_gc_work {
struct delayed_work dwork;
u32 next_bucket;
+ u32 avg_timeout;
+ u32 start_time;
bool exiting;
bool early_drop;
};
static DEFINE_SPINLOCK(nf_conntrack_locks_all_lock);
static __read_mostly bool nf_conntrack_locks_all;
-#define GC_SCAN_INTERVAL (120u * HZ)
+/* serialize hash resizes and nf_ct_iterate_cleanup */
+static DEFINE_MUTEX(nf_conntrack_mutex);
+
+#define GC_SCAN_INTERVAL_MAX (60ul * HZ)
+#define GC_SCAN_INTERVAL_MIN (1ul * HZ)
+
+/* clamp timeouts to this value (TCP unacked) */
+#define GC_SCAN_INTERVAL_CLAMP (300ul * HZ)
+
+/* large initial bias so that we don't scan often just because we have
+ * three entries with a 1s timeout.
+ */
+#define GC_SCAN_INTERVAL_INIT INT_MAX
+
#define GC_SCAN_MAX_DURATION msecs_to_jiffies(10)
+#define GC_SCAN_EXPIRED_MAX (64000u / HZ)
-#define MAX_CHAINLEN 64u
+#define MIN_CHAINLEN 8u
+#define MAX_CHAINLEN (32u - MIN_CHAINLEN)
static struct conntrack_gc_work conntrack_gc_work;
static siphash_key_t nf_conntrack_hash_rnd __read_mostly;
static u32 hash_conntrack_raw(const struct nf_conntrack_tuple *tuple,
+ unsigned int zoneid,
const struct net *net)
{
struct {
struct nf_conntrack_man src;
union nf_inet_addr dst_addr;
+ unsigned int zone;
u32 net_mix;
u16 dport;
u16 proto;
/* The direction must be ignored, so handle usable members manually. */
combined.src = tuple->src;
combined.dst_addr = tuple->dst.u3;
+ combined.zone = zoneid;
combined.net_mix = net_hash_mix(net);
combined.dport = (__force __u16)tuple->dst.u.all;
combined.proto = tuple->dst.protonum;
static u32 __hash_conntrack(const struct net *net,
const struct nf_conntrack_tuple *tuple,
+ unsigned int zoneid,
unsigned int size)
{
- return reciprocal_scale(hash_conntrack_raw(tuple, net), size);
+ return reciprocal_scale(hash_conntrack_raw(tuple, zoneid, net), size);
}
static u32 hash_conntrack(const struct net *net,
- const struct nf_conntrack_tuple *tuple)
+ const struct nf_conntrack_tuple *tuple,
+ unsigned int zoneid)
{
- return scale_hash(hash_conntrack_raw(tuple, net));
+ return scale_hash(hash_conntrack_raw(tuple, zoneid, net));
}
static bool nf_ct_get_tuple_ports(const struct sk_buff *skb,
#define NFCT_ALIGN(len) (((len) + NFCT_INFOMASK) & ~NFCT_INFOMASK)
-/* Released via destroy_conntrack() */
+/* Released via nf_ct_destroy() */
struct nf_conn *nf_ct_tmpl_alloc(struct net *net,
const struct nf_conntrack_zone *zone,
gfp_t flags)
tmpl->status = IPS_TEMPLATE;
write_pnet(&tmpl->ct_net, net);
nf_ct_zone_add(tmpl, zone);
- atomic_set(&tmpl->ct_general.use, 0);
+ refcount_set(&tmpl->ct_general.use, 1);
return tmpl;
}
#endif
}
-static void
-destroy_conntrack(struct nf_conntrack *nfct)
+void nf_ct_destroy(struct nf_conntrack *nfct)
{
struct nf_conn *ct = (struct nf_conn *)nfct;
- pr_debug("destroy_conntrack(%p)\n", ct);
- WARN_ON(atomic_read(&nfct->use) != 0);
+ pr_debug("%s(%p)\n", __func__, ct);
+ WARN_ON(refcount_read(&nfct->use) != 0);
if (unlikely(nf_ct_is_template(ct))) {
nf_ct_tmpl_free(ct);
if (ct->master)
nf_ct_put(ct->master);
- pr_debug("destroy_conntrack: returning ct=%p to slab\n", ct);
+ pr_debug("%s: returning ct=%p to slab\n", __func__, ct);
nf_conntrack_free(ct);
}
+EXPORT_SYMBOL(nf_ct_destroy);
static void nf_ct_delete_from_lists(struct nf_conn *ct)
{
do {
sequence = read_seqcount_begin(&nf_conntrack_generation);
hash = hash_conntrack(net,
- &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple);
+ &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple,
+ nf_ct_zone_id(nf_ct_zone(ct), IP_CT_DIR_ORIGINAL));
reply_hash = hash_conntrack(net,
- &ct->tuplehash[IP_CT_DIR_REPLY].tuple);
+ &ct->tuplehash[IP_CT_DIR_REPLY].tuple,
+ nf_ct_zone_id(nf_ct_zone(ct), IP_CT_DIR_REPLY));
} while (nf_conntrack_double_lock(net, hash, reply_hash, sequence));
clean_from_lists(ct);
tstamp = nf_conn_tstamp_find(ct);
if (tstamp) {
- s32 timeout = ct->timeout - nfct_time_stamp;
+ s32 timeout = READ_ONCE(ct->timeout) - nfct_time_stamp;
tstamp->stop = ktime_get_real_ns();
if (timeout < 0)
/* caller must hold rcu readlock and none of the nf_conntrack_locks */
static void nf_ct_gc_expired(struct nf_conn *ct)
{
- if (!atomic_inc_not_zero(&ct->ct_general.use))
+ if (!refcount_inc_not_zero(&ct->ct_general.use))
return;
if (nf_ct_should_gc(ct))
* in, try to obtain a reference and re-check tuple
*/
ct = nf_ct_tuplehash_to_ctrack(h);
- if (likely(atomic_inc_not_zero(&ct->ct_general.use))) {
+ if (likely(refcount_inc_not_zero(&ct->ct_general.use))) {
if (likely(nf_ct_key_equal(h, tuple, zone, net)))
goto found;
nf_conntrack_find_get(struct net *net, const struct nf_conntrack_zone *zone,
const struct nf_conntrack_tuple *tuple)
{
- return __nf_conntrack_find_get(net, zone, tuple,
- hash_conntrack_raw(tuple, net));
+ unsigned int rid, zone_id = nf_ct_zone_id(zone, IP_CT_DIR_ORIGINAL);
+ struct nf_conntrack_tuple_hash *thash;
+
+ thash = __nf_conntrack_find_get(net, zone, tuple,
+ hash_conntrack_raw(tuple, zone_id, net));
+
+ if (thash)
+ return thash;
+
+ rid = nf_ct_zone_id(zone, IP_CT_DIR_REPLY);
+ if (rid != zone_id)
+ return __nf_conntrack_find_get(net, zone, tuple,
+ hash_conntrack_raw(tuple, rid, net));
+ return thash;
}
EXPORT_SYMBOL_GPL(nf_conntrack_find_get);
unsigned int hash, reply_hash;
struct nf_conntrack_tuple_hash *h;
struct hlist_nulls_node *n;
+ unsigned int max_chainlen;
unsigned int chainlen = 0;
unsigned int sequence;
int err = -EEXIST;
do {
sequence = read_seqcount_begin(&nf_conntrack_generation);
hash = hash_conntrack(net,
- &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple);
+ &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple,
+ nf_ct_zone_id(nf_ct_zone(ct), IP_CT_DIR_ORIGINAL));
reply_hash = hash_conntrack(net,
- &ct->tuplehash[IP_CT_DIR_REPLY].tuple);
+ &ct->tuplehash[IP_CT_DIR_REPLY].tuple,
+ nf_ct_zone_id(nf_ct_zone(ct), IP_CT_DIR_REPLY));
} while (nf_conntrack_double_lock(net, hash, reply_hash, sequence));
+ max_chainlen = MIN_CHAINLEN + prandom_u32_max(MAX_CHAINLEN);
+
/* See if there's one in the list already, including reverse */
hlist_nulls_for_each_entry(h, n, &nf_conntrack_hash[hash], hnnode) {
if (nf_ct_key_equal(h, &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple,
zone, net))
goto out;
- if (chainlen++ > MAX_CHAINLEN)
+ if (chainlen++ > max_chainlen)
goto chaintoolong;
}
if (nf_ct_key_equal(h, &ct->tuplehash[IP_CT_DIR_REPLY].tuple,
zone, net))
goto out;
- if (chainlen++ > MAX_CHAINLEN)
+ if (chainlen++ > max_chainlen)
goto chaintoolong;
}
smp_wmb();
/* The caller holds a reference to this object */
- atomic_set(&ct->ct_general.use, 2);
+ refcount_set(&ct->ct_general.use, 2);
__nf_conntrack_hash_insert(ct, hash, reply_hash);
nf_conntrack_double_unlock(hash, reply_hash);
NF_CT_STAT_INC(net, insert);
{
struct nf_conn_tstamp *tstamp;
- atomic_inc(&ct->ct_general.use);
+ refcount_inc(&ct->ct_general.use);
ct->status |= IPS_CONFIRMED;
/* set conntrack timestamp, if enabled. */
nf_ct_acct_merge(ct, ctinfo, loser_ct);
nf_ct_add_to_dying_list(loser_ct);
- nf_conntrack_put(&loser_ct->ct_general);
+ nf_ct_put(loser_ct);
nf_ct_set(skb, ct, ctinfo);
NF_CT_STAT_INC(net, clash_resolve);
}
/* We want the clashing entry to go away real soon: 1 second timeout. */
- loser_ct->timeout = nfct_time_stamp + HZ;
+ WRITE_ONCE(loser_ct->timeout, nfct_time_stamp + HZ);
/* IPS_NAT_CLASH removes the entry automatically on the first
* reply. Also prevents UDP tracker from moving the entry to
int
__nf_conntrack_confirm(struct sk_buff *skb)
{
+ unsigned int chainlen = 0, sequence, max_chainlen;
const struct nf_conntrack_zone *zone;
- unsigned int chainlen = 0, sequence;
unsigned int hash, reply_hash;
struct nf_conntrack_tuple_hash *h;
struct nf_conn *ct;
hash = *(unsigned long *)&ct->tuplehash[IP_CT_DIR_REPLY].hnnode.pprev;
hash = scale_hash(hash);
reply_hash = hash_conntrack(net,
- &ct->tuplehash[IP_CT_DIR_REPLY].tuple);
-
+ &ct->tuplehash[IP_CT_DIR_REPLY].tuple,
+ nf_ct_zone_id(nf_ct_zone(ct), IP_CT_DIR_REPLY));
} while (nf_conntrack_double_lock(net, hash, reply_hash, sequence));
/* We're not in hash table, and we refuse to set up related
goto dying;
}
+ max_chainlen = MIN_CHAINLEN + prandom_u32_max(MAX_CHAINLEN);
/* 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. */
if (nf_ct_key_equal(h, &ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple,
zone, net))
goto out;
- if (chainlen++ > MAX_CHAINLEN)
+ if (chainlen++ > max_chainlen)
goto chaintoolong;
}
if (nf_ct_key_equal(h, &ct->tuplehash[IP_CT_DIR_REPLY].tuple,
zone, net))
goto out;
- if (chainlen++ > MAX_CHAINLEN) {
+ if (chainlen++ > max_chainlen) {
chaintoolong:
nf_ct_add_to_dying_list(ct);
NF_CT_STAT_INC(net, chaintoolong);
rcu_read_lock();
begin:
nf_conntrack_get_ht(&ct_hash, &hsize);
- hash = __hash_conntrack(net, tuple, hsize);
+ hash = __hash_conntrack(net, tuple, nf_ct_zone_id(zone, IP_CT_DIR_REPLY), hsize);
hlist_nulls_for_each_entry_rcu(h, n, &ct_hash[hash], hnnode) {
ct = nf_ct_tuplehash_to_ctrack(h);
nf_ct_is_dying(tmp))
continue;
- if (!atomic_inc_not_zero(&tmp->ct_general.use))
+ if (!refcount_inc_not_zero(&tmp->ct_general.use))
continue;
/* kill only if still in same netns -- might have moved due to
static void gc_worker(struct work_struct *work)
{
- unsigned long end_time = jiffies + GC_SCAN_MAX_DURATION;
unsigned int i, hashsz, nf_conntrack_max95 = 0;
- unsigned long next_run = GC_SCAN_INTERVAL;
+ u32 end_time, start_time = nfct_time_stamp;
struct conntrack_gc_work *gc_work;
+ unsigned int expired_count = 0;
+ unsigned long next_run;
+ s32 delta_time;
+
gc_work = container_of(work, struct conntrack_gc_work, dwork.work);
i = gc_work->next_bucket;
if (gc_work->early_drop)
nf_conntrack_max95 = nf_conntrack_max / 100u * 95u;
+ if (i == 0) {
+ gc_work->avg_timeout = GC_SCAN_INTERVAL_INIT;
+ gc_work->start_time = start_time;
+ }
+
+ next_run = gc_work->avg_timeout;
+
+ end_time = start_time + GC_SCAN_MAX_DURATION;
+
do {
struct nf_conntrack_tuple_hash *h;
struct hlist_nulls_head *ct_hash;
hlist_nulls_for_each_entry_rcu(h, n, &ct_hash[i], hnnode) {
struct nf_conntrack_net *cnet;
+ unsigned long expires;
struct net *net;
tmp = nf_ct_tuplehash_to_ctrack(h);
continue;
}
+ if (expired_count > GC_SCAN_EXPIRED_MAX) {
+ rcu_read_unlock();
+
+ gc_work->next_bucket = i;
+ gc_work->avg_timeout = next_run;
+
+ delta_time = nfct_time_stamp - gc_work->start_time;
+
+ /* re-sched immediately if total cycle time is exceeded */
+ next_run = delta_time < (s32)GC_SCAN_INTERVAL_MAX;
+ goto early_exit;
+ }
+
if (nf_ct_is_expired(tmp)) {
nf_ct_gc_expired(tmp);
+ expired_count++;
continue;
}
+ expires = clamp(nf_ct_expires(tmp), GC_SCAN_INTERVAL_MIN, GC_SCAN_INTERVAL_CLAMP);
+ next_run += expires;
+ next_run /= 2u;
+
if (nf_conntrack_max95 == 0 || gc_worker_skip_ct(tmp))
continue;
continue;
/* need to take reference to avoid possible races */
- if (!atomic_inc_not_zero(&tmp->ct_general.use))
+ if (!refcount_inc_not_zero(&tmp->ct_general.use))
continue;
if (gc_worker_skip_ct(tmp)) {
continue;
}
- if (gc_worker_can_early_drop(tmp))
+ if (gc_worker_can_early_drop(tmp)) {
nf_ct_kill(tmp);
+ expired_count++;
+ }
nf_ct_put(tmp);
}
cond_resched();
i++;
- if (time_after(jiffies, end_time) && i < hashsz) {
+ delta_time = nfct_time_stamp - end_time;
+ if (delta_time > 0 && i < hashsz) {
+ gc_work->avg_timeout = next_run;
gc_work->next_bucket = i;
next_run = 0;
- break;
+ goto early_exit;
}
} while (i < hashsz);
+ gc_work->next_bucket = 0;
+
+ next_run = clamp(next_run, GC_SCAN_INTERVAL_MIN, GC_SCAN_INTERVAL_MAX);
+
+ delta_time = max_t(s32, nfct_time_stamp - gc_work->start_time, 1);
+ if (next_run > (unsigned long)delta_time)
+ next_run -= delta_time;
+ else
+ next_run = 1;
+
+early_exit:
if (gc_work->exiting)
return;
- /*
- * Eviction will normally happen from the packet path, and not
- * from this gc worker.
- *
- * This worker is only here to reap expired entries when system went
- * idle after a busy period.
- */
- if (next_run) {
+ if (next_run)
gc_work->early_drop = false;
- gc_work->next_bucket = 0;
- }
+
queue_delayed_work(system_power_efficient_wq, &gc_work->dwork, next_run);
}
static void conntrack_gc_work_init(struct conntrack_gc_work *gc_work)
{
- INIT_DEFERRABLE_WORK(&gc_work->dwork, gc_worker);
+ INIT_DELAYED_WORK(&gc_work->dwork, gc_worker);
gc_work->exiting = false;
}
/* save hash for reusing when confirming */
*(unsigned long *)(&ct->tuplehash[IP_CT_DIR_REPLY].hnnode.pprev) = hash;
ct->status = 0;
- ct->timeout = 0;
+ WRITE_ONCE(ct->timeout, 0);
write_pnet(&ct->ct_net, net);
memset(&ct->__nfct_init_offset, 0,
offsetof(struct nf_conn, proto) -
/* Because we use RCU lookups, we set ct_general.use to zero before
* this is inserted in any list.
*/
- atomic_set(&ct->ct_general.use, 0);
+ refcount_set(&ct->ct_general.use, 0);
return ct;
out:
atomic_dec(&cnet->count);
/* A freed object has refcnt == 0, that's
* the golden rule for SLAB_TYPESAFE_BY_RCU
*/
- WARN_ON(atomic_read(&ct->ct_general.use) != 0);
+ WARN_ON(refcount_read(&ct->ct_general.use) != 0);
nf_ct_ext_destroy(ct);
kmem_cache_free(nf_conntrack_cachep, ct);
if (!exp)
__nf_ct_try_assign_helper(ct, tmpl, GFP_ATOMIC);
- /* Now it is inserted into the unconfirmed list, bump refcount */
- nf_conntrack_get(&ct->ct_general);
+ /* Now it is inserted into the unconfirmed list, set refcount to 1. */
+ refcount_set(&ct->ct_general.use, 1);
nf_ct_add_to_unconfirmed_list(ct);
local_bh_enable();
struct nf_conntrack_tuple_hash *h;
enum ip_conntrack_info ctinfo;
struct nf_conntrack_zone tmp;
+ u32 hash, zone_id, rid;
struct nf_conn *ct;
- u32 hash;
if (!nf_ct_get_tuple(skb, skb_network_offset(skb),
dataoff, state->pf, protonum, state->net,
/* look for tuple match */
zone = nf_ct_zone_tmpl(tmpl, skb, &tmp);
- hash = hash_conntrack_raw(&tuple, state->net);
+
+ zone_id = nf_ct_zone_id(zone, IP_CT_DIR_ORIGINAL);
+ hash = hash_conntrack_raw(&tuple, zone_id, state->net);
h = __nf_conntrack_find_get(state->net, zone, &tuple, hash);
+
+ if (!h) {
+ rid = nf_ct_zone_id(zone, IP_CT_DIR_REPLY);
+ if (zone_id != rid) {
+ u32 tmp = hash_conntrack_raw(&tuple, rid, state->net);
+
+ h = __nf_conntrack_find_get(state->net, zone, &tuple, tmp);
+ }
+ }
+
if (!h) {
h = init_conntrack(state->net, tmpl, &tuple,
skb, dataoff, hash);
/* Invalid: inverse of the return code tells
* the netfilter core what to do */
pr_debug("nf_conntrack_in: Can't track with proto module\n");
- nf_conntrack_put(&ct->ct_general);
+ nf_ct_put(ct);
skb->_nfct = 0;
- NF_CT_STAT_INC_ATOMIC(state->net, invalid);
- if (ret == -NF_DROP)
- NF_CT_STAT_INC_ATOMIC(state->net, drop);
/* Special case: TCP tracker reports an attempt to reopen a
* closed/aborted connection. We have to go back and create a
* fresh conntrack.
*/
if (ret == -NF_REPEAT)
goto repeat;
+
+ NF_CT_STAT_INC_ATOMIC(state->net, invalid);
+ if (ret == -NF_DROP)
+ NF_CT_STAT_INC_ATOMIC(state->net, drop);
+
ret = -ret;
goto out;
}
spinlock_t *lockp;
for (; *bucket < nf_conntrack_htable_size; (*bucket)++) {
+ struct hlist_nulls_head *hslot = &nf_conntrack_hash[*bucket];
+
+ if (hlist_nulls_empty(hslot))
+ continue;
+
lockp = &nf_conntrack_locks[*bucket % CONNTRACK_LOCKS];
local_bh_disable();
nf_conntrack_lock(lockp);
- if (*bucket < nf_conntrack_htable_size) {
- hlist_nulls_for_each_entry(h, n, &nf_conntrack_hash[*bucket], hnnode) {
- if (NF_CT_DIRECTION(h) != IP_CT_DIR_REPLY)
- continue;
- /* All nf_conn objects are added to hash table twice, one
- * for original direction tuple, once for the reply tuple.
- *
- * Exception: In the IPS_NAT_CLASH case, only the reply
- * tuple is added (the original tuple already existed for
- * a different object).
- *
- * We only need to call the iterator once for each
- * conntrack, so we just use the 'reply' direction
- * tuple while iterating.
- */
- ct = nf_ct_tuplehash_to_ctrack(h);
- if (iter(ct, data))
- goto found;
- }
+ hlist_nulls_for_each_entry(h, n, hslot, hnnode) {
+ if (NF_CT_DIRECTION(h) != IP_CT_DIR_REPLY)
+ continue;
+ /* All nf_conn objects are added to hash table twice, one
+ * for original direction tuple, once for the reply tuple.
+ *
+ * Exception: In the IPS_NAT_CLASH case, only the reply
+ * tuple is added (the original tuple already existed for
+ * a different object).
+ *
+ * We only need to call the iterator once for each
+ * conntrack, so we just use the 'reply' direction
+ * tuple while iterating.
+ */
+ ct = nf_ct_tuplehash_to_ctrack(h);
+ if (iter(ct, data))
+ goto found;
}
spin_unlock(lockp);
local_bh_enable();
return NULL;
found:
- atomic_inc(&ct->ct_general.use);
+ refcount_inc(&ct->ct_general.use);
spin_unlock(lockp);
local_bh_enable();
return ct;
static void nf_ct_iterate_cleanup(int (*iter)(struct nf_conn *i, void *data),
void *data, u32 portid, int report)
{
- unsigned int bucket = 0, sequence;
+ unsigned int bucket = 0;
struct nf_conn *ct;
might_sleep();
- for (;;) {
- sequence = read_seqcount_begin(&nf_conntrack_generation);
-
- while ((ct = get_next_corpse(iter, data, &bucket)) != NULL) {
- /* Time to push up daises... */
+ mutex_lock(&nf_conntrack_mutex);
+ while ((ct = get_next_corpse(iter, data, &bucket)) != NULL) {
+ /* Time to push up daises... */
- nf_ct_delete(ct, portid, report);
- nf_ct_put(ct);
- cond_resched();
- }
-
- if (!read_seqcount_retry(&nf_conntrack_generation, sequence))
- break;
- bucket = 0;
+ nf_ct_delete(ct, portid, report);
+ nf_ct_put(ct);
+ cond_resched();
}
+ mutex_unlock(&nf_conntrack_mutex);
}
struct iter_data {
if (!hash)
return -ENOMEM;
+ mutex_lock(&nf_conntrack_mutex);
old_size = nf_conntrack_htable_size;
if (old_size == hashsize) {
+ mutex_unlock(&nf_conntrack_mutex);
kvfree(hash);
return 0;
}
for (i = 0; i < nf_conntrack_htable_size; i++) {
while (!hlist_nulls_empty(&nf_conntrack_hash[i])) {
+ unsigned int zone_id;
+
h = hlist_nulls_entry(nf_conntrack_hash[i].first,
struct nf_conntrack_tuple_hash, hnnode);
ct = nf_ct_tuplehash_to_ctrack(h);
hlist_nulls_del_rcu(&h->hnnode);
+
+ zone_id = nf_ct_zone_id(nf_ct_zone(ct), NF_CT_DIRECTION(h));
bucket = __hash_conntrack(nf_ct_net(ct),
- &h->tuple, hashsize);
+ &h->tuple, zone_id, hashsize);
hlist_nulls_add_head_rcu(&h->hnnode, &hash[bucket]);
}
}
nf_conntrack_all_unlock();
local_bh_enable();
+ mutex_unlock(&nf_conntrack_mutex);
+
synchronize_net();
kvfree(old_hash);
return 0;
static struct nf_ct_hook nf_conntrack_hook = {
.update = nf_conntrack_update,
- .destroy = destroy_conntrack,
+ .destroy = nf_ct_destroy,
.get_tuple_skb = nf_conntrack_get_tuple_skb,
};