1 files changed, 133 insertions, 171 deletions

diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
index 0164e5f522e8..082a2fd8d85b 100644
--- a/net/netfilter/nf_conntrack_core.c
+++ b/net/netfilter/nf_conntrack_core.c
@@ -525,50 +525,6 @@ clean_from_lists(struct nf_conn *ct)
 	nf_ct_remove_expectations(ct);
 }
 
-/* must be called with local_bh_disable */
-static void nf_ct_add_to_dying_list(struct nf_conn *ct)
-{
-	struct ct_pcpu *pcpu;
-
-	/* add this conntrack to the (per cpu) dying list */
-	ct->cpu = smp_processor_id();
-	pcpu = per_cpu_ptr(nf_ct_net(ct)->ct.pcpu_lists, ct->cpu);
-
-	spin_lock(&pcpu->lock);
-	hlist_nulls_add_head(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
-			     &pcpu->dying);
-	spin_unlock(&pcpu->lock);
-}
-
-/* must be called with local_bh_disable */
-static void nf_ct_add_to_unconfirmed_list(struct nf_conn *ct)
-{
-	struct ct_pcpu *pcpu;
-
-	/* add this conntrack to the (per cpu) unconfirmed list */
-	ct->cpu = smp_processor_id();
-	pcpu = per_cpu_ptr(nf_ct_net(ct)->ct.pcpu_lists, ct->cpu);
-
-	spin_lock(&pcpu->lock);
-	hlist_nulls_add_head(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
-			     &pcpu->unconfirmed);
-	spin_unlock(&pcpu->lock);
-}
-
-/* must be called with local_bh_disable */
-static void nf_ct_del_from_dying_or_unconfirmed_list(struct nf_conn *ct)
-{
-	struct ct_pcpu *pcpu;
-
-	/* We overload first tuple to link into unconfirmed or dying list.*/
-	pcpu = per_cpu_ptr(nf_ct_net(ct)->ct.pcpu_lists, ct->cpu);
-
-	spin_lock(&pcpu->lock);
-	BUG_ON(hlist_nulls_unhashed(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode));
-	hlist_nulls_del_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode);
-	spin_unlock(&pcpu->lock);
-}
-
 #define NFCT_ALIGN(len)	(((len) + NFCT_INFOMASK) & ~NFCT_INFOMASK)
 
 /* Released via nf_ct_destroy() */
@@ -640,7 +596,6 @@ void nf_ct_destroy(struct nf_conntrack *nfct)
 	if (unlikely(nf_ct_protonum(ct) == IPPROTO_GRE))
 		destroy_gre_conntrack(ct);
 
-	local_bh_disable();
 	/* 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,
@@ -648,10 +603,6 @@ void nf_ct_destroy(struct nf_conntrack *nfct)
 	 */
 	nf_ct_remove_expectations(ct);
 
-	nf_ct_del_from_dying_or_unconfirmed_list(ct);
-
-	local_bh_enable();
-
 	if (ct->master)
 		nf_ct_put(ct->master);
 
@@ -660,15 +611,12 @@ void nf_ct_destroy(struct nf_conntrack *nfct)
 }
 EXPORT_SYMBOL(nf_ct_destroy);
 
-static void nf_ct_delete_from_lists(struct nf_conn *ct)
+static void __nf_ct_delete_from_lists(struct nf_conn *ct)
 {
 	struct net *net = nf_ct_net(ct);
 	unsigned int hash, reply_hash;
 	unsigned int sequence;
 
-	nf_ct_helper_destroy(ct);
-
-	local_bh_disable();
 	do {
 		sequence = read_seqcount_begin(&nf_conntrack_generation);
 		hash = hash_conntrack(net,
@@ -681,12 +629,30 @@ static void nf_ct_delete_from_lists(struct nf_conn *ct)
 
 	clean_from_lists(ct);
 	nf_conntrack_double_unlock(hash, reply_hash);
+}
+
+static void nf_ct_delete_from_lists(struct nf_conn *ct)
+{
+	nf_ct_helper_destroy(ct);
+	local_bh_disable();
 
-	nf_ct_add_to_dying_list(ct);
+	__nf_ct_delete_from_lists(ct);
 
 	local_bh_enable();
 }
 
+static void nf_ct_add_to_ecache_list(struct nf_conn *ct)
+{
+#ifdef CONFIG_NF_CONNTRACK_EVENTS
+	struct nf_conntrack_net *cnet = nf_ct_pernet(nf_ct_net(ct));
+
+	spin_lock(&cnet->ecache.dying_lock);
+	hlist_nulls_add_head_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode,
+				 &cnet->ecache.dying_list);
+	spin_unlock(&cnet->ecache.dying_lock);
+#endif
+}
+
 bool nf_ct_delete(struct nf_conn *ct, u32 portid, int report)
 {
 	struct nf_conn_tstamp *tstamp;
@@ -709,7 +675,12 @@ bool nf_ct_delete(struct nf_conn *ct, u32 portid, int report)
 		/* destroy event was not delivered. nf_ct_put will
 		 * be done by event cache worker on redelivery.
 		 */
-		nf_ct_delete_from_lists(ct);
+		nf_ct_helper_destroy(ct);
+		local_bh_disable();
+		__nf_ct_delete_from_lists(ct);
+		nf_ct_add_to_ecache_list(ct);
+		local_bh_enable();
+
 		nf_conntrack_ecache_work(nf_ct_net(ct), NFCT_ECACHE_DESTROY_FAIL);
 		return false;
 	}
@@ -870,6 +841,33 @@ static void __nf_conntrack_hash_insert(struct nf_conn *ct,
 			   &nf_conntrack_hash[reply_hash]);
 }
 
+static bool nf_ct_ext_valid_pre(const struct nf_ct_ext *ext)
+{
+	/* if ext->gen_id is not equal to nf_conntrack_ext_genid, some extensions
+	 * may contain stale pointers to e.g. helper that has been removed.
+	 *
+	 * The helper can't clear this because the nf_conn object isn't in
+	 * any hash and synchronize_rcu() isn't enough because associated skb
+	 * might sit in a queue.
+	 */
+	return !ext || ext->gen_id == atomic_read(&nf_conntrack_ext_genid);
+}
+
+static bool nf_ct_ext_valid_post(struct nf_ct_ext *ext)
+{
+	if (!ext)
+		return true;
+
+	if (ext->gen_id != atomic_read(&nf_conntrack_ext_genid))
+		return false;
+
+	/* inserted into conntrack table, nf_ct_iterate_cleanup()
+	 * will find it.  Disable nf_ct_ext_find() id check.
+	 */
+	WRITE_ONCE(ext->gen_id, 0);
+	return true;
+}
+
 int
 nf_conntrack_hash_check_insert(struct nf_conn *ct)
 {
@@ -885,6 +883,11 @@ nf_conntrack_hash_check_insert(struct nf_conn *ct)
 
 	zone = nf_ct_zone(ct);
 
+	if (!nf_ct_ext_valid_pre(ct->ext)) {
+		NF_CT_STAT_INC(net, insert_failed);
+		return -ETIMEDOUT;
+	}
+
 	local_bh_disable();
 	do {
 		sequence = read_seqcount_begin(&nf_conntrack_generation);
@@ -925,6 +928,13 @@ nf_conntrack_hash_check_insert(struct nf_conn *ct)
 	nf_conntrack_double_unlock(hash, reply_hash);
 	NF_CT_STAT_INC(net, insert);
 	local_bh_enable();
+
+	if (!nf_ct_ext_valid_post(ct->ext)) {
+		nf_ct_kill(ct);
+		NF_CT_STAT_INC(net, drop);
+		return -ETIMEDOUT;
+	}
+
 	return 0;
 chaintoolong:
 	NF_CT_STAT_INC(net, chaintoolong);
@@ -972,7 +982,6 @@ static void __nf_conntrack_insert_prepare(struct nf_conn *ct)
 	struct nf_conn_tstamp *tstamp;
 
 	refcount_inc(&ct->ct_general.use);
-	ct->status |= IPS_CONFIRMED;
 
 	/* set conntrack timestamp, if enabled. */
 	tstamp = nf_conn_tstamp_find(ct);
@@ -1001,7 +1010,6 @@ static int __nf_ct_resolve_clash(struct sk_buff *skb,
 		nf_conntrack_get(&ct->ct_general);
 
 		nf_ct_acct_merge(ct, ctinfo, loser_ct);
-		nf_ct_add_to_dying_list(loser_ct);
 		nf_ct_put(loser_ct);
 		nf_ct_set(skb, ct, ctinfo);
 
@@ -1134,7 +1142,6 @@ nf_ct_resolve_clash(struct sk_buff *skb, struct nf_conntrack_tuple_hash *h,
 		return ret;
 
 drop:
-	nf_ct_add_to_dying_list(loser_ct);
 	NF_CT_STAT_INC(net, drop);
 	NF_CT_STAT_INC(net, insert_failed);
 	return NF_DROP;
@@ -1195,16 +1202,20 @@ __nf_conntrack_confirm(struct sk_buff *skb)
 		return NF_DROP;
 	}
 
+	if (!nf_ct_ext_valid_pre(ct->ext)) {
+		NF_CT_STAT_INC(net, insert_failed);
+		goto dying;
+	}
+
 	pr_debug("Confirming conntrack %p\n", ct);
 	/* We have to check the DYING flag after unlink to prevent
 	 * a race against nf_ct_get_next_corpse() possibly called from
 	 * user context, else we insert an already 'dead' hash, blocking
 	 * further use of that particular connection -JM.
 	 */
-	nf_ct_del_from_dying_or_unconfirmed_list(ct);
+	ct->status |= IPS_CONFIRMED;
 
 	if (unlikely(nf_ct_is_dying(ct))) {
-		nf_ct_add_to_dying_list(ct);
 		NF_CT_STAT_INC(net, insert_failed);
 		goto dying;
 	}
@@ -1228,7 +1239,6 @@ __nf_conntrack_confirm(struct sk_buff *skb)
 			goto out;
 		if (chainlen++ > max_chainlen) {
 chaintoolong:
-			nf_ct_add_to_dying_list(ct);
 			NF_CT_STAT_INC(net, chaintoolong);
 			NF_CT_STAT_INC(net, insert_failed);
 			ret = NF_DROP;
@@ -1252,6 +1262,16 @@ chaintoolong:
 	nf_conntrack_double_unlock(hash, reply_hash);
 	local_bh_enable();
 
+	/* ext area is still valid (rcu read lock is held,
+	 * but will go out of scope soon, we need to remove
+	 * this conntrack again.
+	 */
+	if (!nf_ct_ext_valid_post(ct->ext)) {
+		nf_ct_kill(ct);
+		NF_CT_STAT_INC(net, drop);
+		return NF_DROP;
+	}
+
 	help = nfct_help(ct);
 	if (help && help->helper)
 		nf_conntrack_event_cache(IPCT_HELPER, ct);
@@ -1678,7 +1698,9 @@ init_conntrack(struct net *net, struct nf_conn *tmpl,
 	struct nf_conn *ct;
 	struct nf_conn_help *help;
 	struct nf_conntrack_tuple repl_tuple;
+#ifdef CONFIG_NF_CONNTRACK_EVENTS
 	struct nf_conntrack_ecache *ecache;
+#endif
 	struct nf_conntrack_expect *exp = NULL;
 	const struct nf_conntrack_zone *zone;
 	struct nf_conn_timeout *timeout_ext;
@@ -1711,15 +1733,21 @@ init_conntrack(struct net *net, struct nf_conn *tmpl,
 	nf_ct_tstamp_ext_add(ct, GFP_ATOMIC);
 	nf_ct_labels_ext_add(ct);
 
+#ifdef CONFIG_NF_CONNTRACK_EVENTS
 	ecache = tmpl ? nf_ct_ecache_find(tmpl) : NULL;
-	nf_ct_ecache_ext_add(ct, ecache ? ecache->ctmask : 0,
-				 ecache ? ecache->expmask : 0,
-			     GFP_ATOMIC);
 
-	local_bh_disable();
+	if ((ecache || net->ct.sysctl_events) &&
+	    !nf_ct_ecache_ext_add(ct, ecache ? ecache->ctmask : 0,
+				  ecache ? ecache->expmask : 0,
+				  GFP_ATOMIC)) {
+		nf_conntrack_free(ct);
+		return ERR_PTR(-ENOMEM);
+	}
+#endif
+
 	cnet = nf_ct_pernet(net);
 	if (cnet->expect_count) {
-		spin_lock(&nf_conntrack_expect_lock);
+		spin_lock_bh(&nf_conntrack_expect_lock);
 		exp = nf_ct_find_expectation(net, zone, tuple);
 		if (exp) {
 			pr_debug("expectation arrives ct=%p exp=%p\n",
@@ -1742,16 +1770,13 @@ init_conntrack(struct net *net, struct nf_conn *tmpl,
 #endif
 			NF_CT_STAT_INC(net, expect_new);
 		}
-		spin_unlock(&nf_conntrack_expect_lock);
+		spin_unlock_bh(&nf_conntrack_expect_lock);
 	}
 	if (!exp)
 		__nf_ct_try_assign_helper(ct, tmpl, GFP_ATOMIC);
 
-	/* Now it is inserted into the unconfirmed list, set refcount to 1. */
+	/* Now it is going to be associated with an sk_buff, set refcount to 1. */
 	refcount_set(&ct->ct_general.use, 1);
-	nf_ct_add_to_unconfirmed_list(ct);
-
-	local_bh_enable();
 
 	if (exp) {
 		if (exp->expectfn)
@@ -2319,7 +2344,7 @@ static bool nf_conntrack_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple,
 /* Bring out ya dead! */
 static struct nf_conn *
 get_next_corpse(int (*iter)(struct nf_conn *i, void *data),
-		void *data, unsigned int *bucket)
+		const struct nf_ct_iter_data *iter_data, unsigned int *bucket)
 {
 	struct nf_conntrack_tuple_hash *h;
 	struct nf_conn *ct;
@@ -2350,7 +2375,12 @@ get_next_corpse(int (*iter)(struct nf_conn *i, void *data),
 			 * tuple while iterating.
 			 */
 			ct = nf_ct_tuplehash_to_ctrack(h);
-			if (iter(ct, data))
+
+			if (iter_data->net &&
+			    !net_eq(iter_data->net, nf_ct_net(ct)))
+				continue;
+
+			if (iter(ct, iter_data->data))
 				goto found;
 		}
 		spin_unlock(lockp);
@@ -2367,7 +2397,7 @@ found:
 }
 
 static void nf_ct_iterate_cleanup(int (*iter)(struct nf_conn *i, void *data),
-				  void *data, u32 portid, int report)
+				  const struct nf_ct_iter_data *iter_data)
 {
 	unsigned int bucket = 0;
 	struct nf_conn *ct;
@@ -2375,91 +2405,28 @@ static void nf_ct_iterate_cleanup(int (*iter)(struct nf_conn *i, void *data),
 	might_sleep();
 
 	mutex_lock(&nf_conntrack_mutex);
-	while ((ct = get_next_corpse(iter, data, &bucket)) != NULL) {
+	while ((ct = get_next_corpse(iter, iter_data, &bucket)) != NULL) {
 		/* Time to push up daises... */
 
-		nf_ct_delete(ct, portid, report);
+		nf_ct_delete(ct, iter_data->portid, iter_data->report);
 		nf_ct_put(ct);
 		cond_resched();
 	}
 	mutex_unlock(&nf_conntrack_mutex);
 }
 
-struct iter_data {
-	int (*iter)(struct nf_conn *i, void *data);
-	void *data;
-	struct net *net;
-};
-
-static int iter_net_only(struct nf_conn *i, void *data)
-{
-	struct iter_data *d = data;
-
-	if (!net_eq(d->net, nf_ct_net(i)))
-		return 0;
-
-	return d->iter(i, d->data);
-}
-
-static void
-__nf_ct_unconfirmed_destroy(struct net *net)
-{
-	int cpu;
-
-	for_each_possible_cpu(cpu) {
-		struct nf_conntrack_tuple_hash *h;
-		struct hlist_nulls_node *n;
-		struct ct_pcpu *pcpu;
-
-		pcpu = per_cpu_ptr(net->ct.pcpu_lists, cpu);
-
-		spin_lock_bh(&pcpu->lock);
-		hlist_nulls_for_each_entry(h, n, &pcpu->unconfirmed, hnnode) {
-			struct nf_conn *ct;
-
-			ct = nf_ct_tuplehash_to_ctrack(h);
-
-			/* we cannot call iter() on unconfirmed list, the
-			 * owning cpu can reallocate ct->ext at any time.
-			 */
-			set_bit(IPS_DYING_BIT, &ct->status);
-		}
-		spin_unlock_bh(&pcpu->lock);
-		cond_resched();
-	}
-}
-
-void nf_ct_unconfirmed_destroy(struct net *net)
+void nf_ct_iterate_cleanup_net(int (*iter)(struct nf_conn *i, void *data),
+			       const struct nf_ct_iter_data *iter_data)
 {
+	struct net *net = iter_data->net;
 	struct nf_conntrack_net *cnet = nf_ct_pernet(net);
 
 	might_sleep();
 
-	if (atomic_read(&cnet->count) > 0) {
-		__nf_ct_unconfirmed_destroy(net);
-		nf_queue_nf_hook_drop(net);
-		synchronize_net();
-	}
-}
-EXPORT_SYMBOL_GPL(nf_ct_unconfirmed_destroy);
-
-void nf_ct_iterate_cleanup_net(struct net *net,
-			       int (*iter)(struct nf_conn *i, void *data),
-			       void *data, u32 portid, int report)
-{
-	struct nf_conntrack_net *cnet = nf_ct_pernet(net);
-	struct iter_data d;
-
-	might_sleep();
-
 	if (atomic_read(&cnet->count) == 0)
 		return;
 
-	d.iter = iter;
-	d.data = data;
-	d.net = net;
-
-	nf_ct_iterate_cleanup(iter_net_only, &d, portid, report);
+	nf_ct_iterate_cleanup(iter, iter_data);
 }
 EXPORT_SYMBOL_GPL(nf_ct_iterate_cleanup_net);
 
@@ -2477,6 +2444,7 @@ EXPORT_SYMBOL_GPL(nf_ct_iterate_cleanup_net);
 void
 nf_ct_iterate_destroy(int (*iter)(struct nf_conn *i, void *data), void *data)
 {
+	struct nf_ct_iter_data iter_data = {};
 	struct net *net;
 
 	down_read(&net_rwsem);
@@ -2485,31 +2453,41 @@ nf_ct_iterate_destroy(int (*iter)(struct nf_conn *i, void *data), void *data)
 
 		if (atomic_read(&cnet->count) == 0)
 			continue;
-		__nf_ct_unconfirmed_destroy(net);
 		nf_queue_nf_hook_drop(net);
 	}
 	up_read(&net_rwsem);
 
 	/* Need to wait for netns cleanup worker to finish, if its
 	 * running -- it might have deleted a net namespace from
-	 * the global list, so our __nf_ct_unconfirmed_destroy() might
-	 * not have affected all namespaces.
+	 * the global list, so hook drop above might not have
+	 * affected all namespaces.
 	 */
 	net_ns_barrier();
 
-	/* a conntrack could have been unlinked from unconfirmed list
-	 * before we grabbed pcpu lock in __nf_ct_unconfirmed_destroy().
+	/* a skb w. unconfirmed conntrack could have been reinjected just
+	 * before we called nf_queue_nf_hook_drop().
+	 *
 	 * This makes sure its inserted into conntrack table.
 	 */
 	synchronize_net();
 
-	nf_ct_iterate_cleanup(iter, data, 0, 0);
+	nf_ct_ext_bump_genid();
+	iter_data.data = data;
+	nf_ct_iterate_cleanup(iter, &iter_data);
+
+	/* Another cpu might be in a rcu read section with
+	 * rcu protected pointer cleared in iter callback
+	 * or hidden via nf_ct_ext_bump_genid() above.
+	 *
+	 * Wait until those are done.
+	 */
+	synchronize_rcu();
 }
 EXPORT_SYMBOL_GPL(nf_ct_iterate_destroy);
 
 static int kill_all(struct nf_conn *i, void *data)
 {
-	return net_eq(nf_ct_net(i), data);
+	return 1;
 }
 
 void nf_conntrack_cleanup_start(void)
@@ -2544,8 +2522,9 @@ void nf_conntrack_cleanup_net(struct net *net)
 
 void nf_conntrack_cleanup_net_list(struct list_head *net_exit_list)
 {
-	int busy;
+	struct nf_ct_iter_data iter_data = {};
 	struct net *net;
+	int busy;
 
 	/*
 	 * This makes sure all current packets have passed through
@@ -2558,7 +2537,8 @@ i_see_dead_people:
 	list_for_each_entry(net, net_exit_list, exit_list) {
 		struct nf_conntrack_net *cnet = nf_ct_pernet(net);
 
-		nf_ct_iterate_cleanup(kill_all, net, 0, 0);
+		iter_data.net = net;
+		nf_ct_iterate_cleanup_net(kill_all, &iter_data);
 		if (atomic_read(&cnet->count) != 0)
 			busy = 1;
 	}
@@ -2571,7 +2551,6 @@ i_see_dead_people:
 		nf_conntrack_ecache_pernet_fini(net);
 		nf_conntrack_expect_pernet_fini(net);
 		free_percpu(net->ct.stat);
-		free_percpu(net->ct.pcpu_lists);
 	}
 }
 
@@ -2777,33 +2756,19 @@ void nf_conntrack_init_end(void)
  * We need to use special "null" values, not used in hash table
  */
 #define UNCONFIRMED_NULLS_VAL	((1<<30)+0)
-#define DYING_NULLS_VAL		((1<<30)+1)
 
 int nf_conntrack_init_net(struct net *net)
 {
 	struct nf_conntrack_net *cnet = nf_ct_pernet(net);
 	int ret = -ENOMEM;
-	int cpu;
 
 	BUILD_BUG_ON(IP_CT_UNTRACKED == IP_CT_NUMBER);
 	BUILD_BUG_ON_NOT_POWER_OF_2(CONNTRACK_LOCKS);
 	atomic_set(&cnet->count, 0);
 
-	net->ct.pcpu_lists = alloc_percpu(struct ct_pcpu);
-	if (!net->ct.pcpu_lists)
-		goto err_stat;
-
-	for_each_possible_cpu(cpu) {
-		struct ct_pcpu *pcpu = per_cpu_ptr(net->ct.pcpu_lists, cpu);
-
-		spin_lock_init(&pcpu->lock);
-		INIT_HLIST_NULLS_HEAD(&pcpu->unconfirmed, UNCONFIRMED_NULLS_VAL);
-		INIT_HLIST_NULLS_HEAD(&pcpu->dying, DYING_NULLS_VAL);
-	}
-
 	net->ct.stat = alloc_percpu(struct ip_conntrack_stat);
 	if (!net->ct.stat)
-		goto err_pcpu_lists;
+		return ret;
 
 	ret = nf_conntrack_expect_pernet_init(net);
 	if (ret < 0)
@@ -2819,8 +2784,5 @@ int nf_conntrack_init_net(struct net *net)
 
 err_expect:
 	free_percpu(net->ct.stat);
-err_pcpu_lists:
-	free_percpu(net->ct.pcpu_lists);
-err_stat:
 	return ret;
 }