Merge git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf-next

Pablo Neira Ayuso says:

====================
Netfilter updates for net-next

This is v2 including deadlock fix in conntrack ecache rework
reported by Jakub Kicinski.

The following patchset contains Netfilter updates for net-next,
mostly updates to conntrack from Florian Westphal.

1) Add a dedicated list for conntrack event redelivery.

2) Include event redelivery list in conntrack dumps of dying type.

3) Remove per-cpu dying list for event redelivery, not used anymore.

4) Add netns .pre_exit to cttimeout to zap timeout objects before
   synchronize_rcu() call.

5) Remove nf_ct_unconfirmed_destroy.

6) Add generation id for conntrack extensions for conntrack
   timeout and helpers.

7) Detach timeout policy from conntrack on cttimeout module removal.

8) Remove __nf_ct_unconfirmed_destroy.

9) Remove unconfirmed list.

10) Remove unconditional local_bh_disable in init_conntrack().

11) Consolidate conntrack iterator nf_ct_iterate_cleanup().

12) Detect if ctnetlink listeners exist to short-circuit event
    path early.

13) Un-inline nf_ct_ecache_ext_add().

14) Add nf_conntrack_events autodetect ctnetlink listener mode
    and make it default.

15) Add nf_ct_ecache_exist() to check for event cache extension.

16) Extend flowtable reverse route lookup to include source, iif,
    tos and mark, from Sven Auhagen.

17) Do not verify zero checksum UDP packets in nf_reject,
    from Kevin Mitchell.

====================

Signed-off-by: David S. Miller <davem@davemloft.net>

14 files changed, 420 insertions, 305 deletions

diff --git a/net/ipv4/netfilter/nf_reject_ipv4.c b/net/ipv4/netfilter/nf_reject_ipv4.c
index 4eed5afca392..918c61fda0f3 100644
--- a/net/ipv4/netfilter/nf_reject_ipv4.c
+++ b/net/ipv4/netfilter/nf_reject_ipv4.c
@@ -80,6 +80,7 @@ struct sk_buff *nf_reject_skb_v4_unreach(struct net *net,
 	struct iphdr *niph;
 	struct icmphdr *icmph;
 	unsigned int len;
+	int dataoff;
 	__wsum csum;
 	u8 proto;
 
@@ -99,10 +100,11 @@ struct sk_buff *nf_reject_skb_v4_unreach(struct net *net,
 	if (pskb_trim_rcsum(oldskb, ntohs(ip_hdr(oldskb)->tot_len)))
 		return NULL;
 
+	dataoff = ip_hdrlen(oldskb);
 	proto = ip_hdr(oldskb)->protocol;
 
 	if (!skb_csum_unnecessary(oldskb) &&
-	    nf_reject_verify_csum(proto) &&
+	    nf_reject_verify_csum(oldskb, dataoff, proto) &&
 	    nf_ip_checksum(oldskb, hook, ip_hdrlen(oldskb), proto))
 		return NULL;
 
@@ -311,6 +313,7 @@ EXPORT_SYMBOL_GPL(nf_send_reset);
 void nf_send_unreach(struct sk_buff *skb_in, int code, int hook)
 {
 	struct iphdr *iph = ip_hdr(skb_in);
+	int dataoff = ip_hdrlen(skb_in);
 	u8 proto = iph->protocol;
 
 	if (iph->frag_off & htons(IP_OFFSET))
@@ -320,12 +323,13 @@ void nf_send_unreach(struct sk_buff *skb_in, int code, int hook)
 	    nf_reject_fill_skb_dst(skb_in) < 0)
 		return;
 
-	if (skb_csum_unnecessary(skb_in) || !nf_reject_verify_csum(proto)) {
+	if (skb_csum_unnecessary(skb_in) ||
+	    !nf_reject_verify_csum(skb_in, dataoff, proto)) {
 		icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
 		return;
 	}
 
-	if (nf_ip_checksum(skb_in, hook, ip_hdrlen(skb_in), proto) == 0)
+	if (nf_ip_checksum(skb_in, hook, dataoff, proto) == 0)
 		icmp_send(skb_in, ICMP_DEST_UNREACH, code, 0);
 }
 EXPORT_SYMBOL_GPL(nf_send_unreach);
diff --git a/net/ipv6/netfilter/nf_reject_ipv6.c b/net/ipv6/netfilter/nf_reject_ipv6.c
index dffeaaaadcde..f61d4f18e1cf 100644
--- a/net/ipv6/netfilter/nf_reject_ipv6.c
+++ b/net/ipv6/netfilter/nf_reject_ipv6.c
@@ -31,7 +31,7 @@ static bool nf_reject_v6_csum_ok(struct sk_buff *skb, int hook)
 	if (thoff < 0 || thoff >= skb->len || (fo & htons(~0x7)) != 0)
 		return false;
 
-	if (!nf_reject_verify_csum(proto))
+	if (!nf_reject_verify_csum(skb, thoff, proto))
 		return true;
 
 	return nf_ip6_checksum(skb, hook, thoff, proto) == 0;
@@ -388,7 +388,7 @@ static bool reject6_csum_ok(struct sk_buff *skb, int hook)
 	if (thoff < 0 || thoff >= skb->len || (fo & htons(~0x7)) != 0)
 		return false;
 
-	if (!nf_reject_verify_csum(proto))
+	if (!nf_reject_verify_csum(skb, thoff, proto))
 		return true;
 
 	return nf_ip6_checksum(skb, hook, thoff, proto) == 0;
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;
 }
diff --git a/net/netfilter/nf_conntrack_ecache.c b/net/netfilter/nf_conntrack_ecache.c
index 0cb2da0a759a..8698b3424646 100644
--- a/net/netfilter/nf_conntrack_ecache.c
+++ b/net/netfilter/nf_conntrack_ecache.c
@@ -16,7 +16,6 @@
 #include <linux/vmalloc.h>
 #include <linux/stddef.h>
 #include <linux/err.h>
-#include <linux/percpu.h>
 #include <linux/kernel.h>
 #include <linux/netdevice.h>
 #include <linux/slab.h>
@@ -29,8 +28,9 @@
 
 static DEFINE_MUTEX(nf_ct_ecache_mutex);
 
-#define ECACHE_RETRY_WAIT (HZ/10)
-#define ECACHE_STACK_ALLOC (256 / sizeof(void *))
+#define DYING_NULLS_VAL			((1 << 30) + 1)
+#define ECACHE_MAX_JIFFIES		msecs_to_jiffies(10)
+#define ECACHE_RETRY_JIFFIES		msecs_to_jiffies(10)
 
 enum retry_state {
 	STATE_CONGESTED,
@@ -38,58 +38,67 @@ enum retry_state {
 	STATE_DONE,
 };
 
-static enum retry_state ecache_work_evict_list(struct ct_pcpu *pcpu)
+struct nf_conntrack_net_ecache *nf_conn_pernet_ecache(const struct net *net)
 {
-	struct nf_conn *refs[ECACHE_STACK_ALLOC];
+	struct nf_conntrack_net *cnet = nf_ct_pernet(net);
+
+	return &cnet->ecache;
+}
+#if IS_MODULE(CONFIG_NF_CT_NETLINK)
+EXPORT_SYMBOL_GPL(nf_conn_pernet_ecache);
+#endif
+
+static enum retry_state ecache_work_evict_list(struct nf_conntrack_net *cnet)
+{
+	unsigned long stop = jiffies + ECACHE_MAX_JIFFIES;
+	struct hlist_nulls_head evicted_list;
 	enum retry_state ret = STATE_DONE;
 	struct nf_conntrack_tuple_hash *h;
 	struct hlist_nulls_node *n;
-	unsigned int evicted = 0;
+	unsigned int sent;
 
-	spin_lock(&pcpu->lock);
+	INIT_HLIST_NULLS_HEAD(&evicted_list, DYING_NULLS_VAL);
 
-	hlist_nulls_for_each_entry(h, n, &pcpu->dying, hnnode) {
+next:
+	sent = 0;
+	spin_lock_bh(&cnet->ecache.dying_lock);
+
+	hlist_nulls_for_each_entry_safe(h, n, &cnet->ecache.dying_list, hnnode) {
 		struct nf_conn *ct = nf_ct_tuplehash_to_ctrack(h);
-		struct nf_conntrack_ecache *e;
-
-		if (!nf_ct_is_confirmed(ct))
-			continue;
-
-		/* This ecache access is safe because the ct is on the
-		 * pcpu dying list and we hold the spinlock -- the entry
-		 * cannot be free'd until after the lock is released.
-		 *
-		 * This is true even if ct has a refcount of 0: the
-		 * cpu that is about to free the entry must remove it
-		 * from the dying list and needs the lock to do so.
-		 */
-		e = nf_ct_ecache_find(ct);
-		if (!e || e->state != NFCT_ECACHE_DESTROY_FAIL)
-			continue;
 
-		/* ct is in NFCT_ECACHE_DESTROY_FAIL state, this means
-		 * the worker owns this entry: the ct will remain valid
-		 * until the worker puts its ct reference.
+		/* The worker owns all entries, ct remains valid until nf_ct_put
+		 * in the loop below.
 		 */
 		if (nf_conntrack_event(IPCT_DESTROY, ct)) {
 			ret = STATE_CONGESTED;
 			break;
 		}
 
-		e->state = NFCT_ECACHE_DESTROY_SENT;
-		refs[evicted] = ct;
+		hlist_nulls_del_rcu(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode);
+		hlist_nulls_add_head(&ct->tuplehash[IP_CT_DIR_REPLY].hnnode, &evicted_list);
 
-		if (++evicted >= ARRAY_SIZE(refs)) {
+		if (time_after(stop, jiffies)) {
 			ret = STATE_RESTART;
 			break;
 		}
+
+		if (sent++ > 16) {
+			spin_unlock_bh(&cnet->ecache.dying_lock);
+			cond_resched();
+			goto next;
+		}
 	}
 
-	spin_unlock(&pcpu->lock);
+	spin_unlock_bh(&cnet->ecache.dying_lock);
 
-	/* can't _put while holding lock */
-	while (evicted)
-		nf_ct_put(refs[--evicted]);
+	hlist_nulls_for_each_entry_safe(h, n, &evicted_list, hnnode) {
+		struct nf_conn *ct = nf_ct_tuplehash_to_ctrack(h);
+
+		hlist_nulls_del_rcu(&ct->tuplehash[IP_CT_DIR_REPLY].hnnode);
+		nf_ct_put(ct);
+
+		cond_resched();
+	}
 
 	return ret;
 }
@@ -97,35 +106,20 @@ static enum retry_state ecache_work_evict_list(struct ct_pcpu *pcpu)
 static void ecache_work(struct work_struct *work)
 {
 	struct nf_conntrack_net *cnet = container_of(work, struct nf_conntrack_net, ecache.dwork.work);
-	struct netns_ct *ctnet = cnet->ecache.ct_net;
-	int cpu, delay = -1;
-	struct ct_pcpu *pcpu;
-
-	local_bh_disable();
-
-	for_each_possible_cpu(cpu) {
-		enum retry_state ret;
-
-		pcpu = per_cpu_ptr(ctnet->pcpu_lists, cpu);
-
-		ret = ecache_work_evict_list(pcpu);
-
-		switch (ret) {
-		case STATE_CONGESTED:
-			delay = ECACHE_RETRY_WAIT;
-			goto out;
-		case STATE_RESTART:
-			delay = 0;
-			break;
-		case STATE_DONE:
-			break;
-		}
+	int ret, delay = -1;
+
+	ret = ecache_work_evict_list(cnet);
+	switch (ret) {
+	case STATE_CONGESTED:
+		delay = ECACHE_RETRY_JIFFIES;
+		break;
+	case STATE_RESTART:
+		delay = 0;
+		break;
+	case STATE_DONE:
+		break;
 	}
 
- out:
-	local_bh_enable();
-
-	ctnet->ecache_dwork_pending = delay > 0;
 	if (delay >= 0)
 		schedule_delayed_work(&cnet->ecache.dwork, delay);
 }
@@ -199,7 +193,6 @@ int nf_conntrack_eventmask_report(unsigned int events, struct nf_conn *ct,
 		 */
 		if (e->portid == 0 && portid != 0)
 			e->portid = portid;
-		e->state = NFCT_ECACHE_DESTROY_FAIL;
 	}
 
 	return ret;
@@ -297,12 +290,51 @@ void nf_conntrack_ecache_work(struct net *net, enum nf_ct_ecache_state state)
 		schedule_delayed_work(&cnet->ecache.dwork, HZ);
 		net->ct.ecache_dwork_pending = true;
 	} else if (state == NFCT_ECACHE_DESTROY_SENT) {
-		net->ct.ecache_dwork_pending = false;
-		mod_delayed_work(system_wq, &cnet->ecache.dwork, 0);
+		if (!hlist_nulls_empty(&cnet->ecache.dying_list))
+			mod_delayed_work(system_wq, &cnet->ecache.dwork, 0);
+		else
+			net->ct.ecache_dwork_pending = false;
 	}
 }
 
-#define NF_CT_EVENTS_DEFAULT 1
+bool nf_ct_ecache_ext_add(struct nf_conn *ct, u16 ctmask, u16 expmask, gfp_t gfp)
+{
+	struct net *net = nf_ct_net(ct);
+	struct nf_conntrack_ecache *e;
+
+	switch (net->ct.sysctl_events) {
+	case 0:
+		 /* assignment via template / ruleset? ignore sysctl. */
+		if (ctmask || expmask)
+			break;
+		return true;
+	case 2: /* autodetect: no event listener, don't allocate extension. */
+		if (!READ_ONCE(net->ct.ctnetlink_has_listener))
+			return true;
+		fallthrough;
+	case 1:
+		/* always allocate an extension. */
+		if (!ctmask && !expmask) {
+			ctmask = ~0;
+			expmask = ~0;
+		}
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return true;
+	}
+
+	e = nf_ct_ext_add(ct, NF_CT_EXT_ECACHE, gfp);
+	if (e) {
+		e->ctmask  = ctmask;
+		e->expmask = expmask;
+	}
+
+	return e != NULL;
+}
+EXPORT_SYMBOL_GPL(nf_ct_ecache_ext_add);
+
+#define NF_CT_EVENTS_DEFAULT 2
 static int nf_ct_events __read_mostly = NF_CT_EVENTS_DEFAULT;
 
 void nf_conntrack_ecache_pernet_init(struct net *net)
@@ -311,8 +343,9 @@ void nf_conntrack_ecache_pernet_init(struct net *net)
 
 	net->ct.sysctl_events = nf_ct_events;
 
-	cnet->ecache.ct_net = &net->ct;
 	INIT_DELAYED_WORK(&cnet->ecache.dwork, ecache_work);
+	INIT_HLIST_NULLS_HEAD(&cnet->ecache.dying_list, DYING_NULLS_VAL);
+	spin_lock_init(&cnet->ecache.dying_lock);
 
 	BUILD_BUG_ON(__IPCT_MAX >= 16);	/* e->ctmask is u16 */
 }
diff --git a/net/netfilter/nf_conntrack_extend.c b/net/netfilter/nf_conntrack_extend.c
index 1296fda54ac6..0b513f7bf9f3 100644
--- a/net/netfilter/nf_conntrack_extend.c
+++ b/net/netfilter/nf_conntrack_extend.c
@@ -27,6 +27,8 @@
 
 #define NF_CT_EXT_PREALLOC	128u /* conntrack events are on by default */
 
+atomic_t nf_conntrack_ext_genid __read_mostly = ATOMIC_INIT(1);
+
 static const u8 nf_ct_ext_type_len[NF_CT_EXT_NUM] = {
 	[NF_CT_EXT_HELPER] = sizeof(struct nf_conn_help),
 #if IS_ENABLED(CONFIG_NF_NAT)
@@ -116,8 +118,10 @@ void *nf_ct_ext_add(struct nf_conn *ct, enum nf_ct_ext_id id, gfp_t gfp)
 	if (!new)
 		return NULL;
 
-	if (!ct->ext)
+	if (!ct->ext) {
 		memset(new->offset, 0, sizeof(new->offset));
+		new->gen_id = atomic_read(&nf_conntrack_ext_genid);
+	}
 
 	new->offset[id] = newoff;
 	new->len = newlen;
@@ -127,3 +131,29 @@ void *nf_ct_ext_add(struct nf_conn *ct, enum nf_ct_ext_id id, gfp_t gfp)
 	return (void *)new + newoff;
 }
 EXPORT_SYMBOL(nf_ct_ext_add);
+
+/* Use nf_ct_ext_find wrapper. This is only useful for unconfirmed entries. */
+void *__nf_ct_ext_find(const struct nf_ct_ext *ext, u8 id)
+{
+	unsigned int gen_id = atomic_read(&nf_conntrack_ext_genid);
+	unsigned int this_id = READ_ONCE(ext->gen_id);
+
+	if (!__nf_ct_ext_exist(ext, id))
+		return NULL;
+
+	if (this_id == 0 || ext->gen_id == gen_id)
+		return (void *)ext + ext->offset[id];
+
+	return NULL;
+}
+EXPORT_SYMBOL(__nf_ct_ext_find);
+
+void nf_ct_ext_bump_genid(void)
+{
+	unsigned int value = atomic_inc_return(&nf_conntrack_ext_genid);
+
+	if (value == UINT_MAX)
+		atomic_set(&nf_conntrack_ext_genid, 1);
+
+	msleep(HZ);
+}
diff --git a/net/netfilter/nf_conntrack_helper.c b/net/netfilter/nf_conntrack_helper.c
index 8dec42ec603e..c12a87ebc3ee 100644
--- a/net/netfilter/nf_conntrack_helper.c
+++ b/net/netfilter/nf_conntrack_helper.c
@@ -468,11 +468,6 @@ void nf_conntrack_helper_unregister(struct nf_conntrack_helper *me)
 
 	nf_ct_expect_iterate_destroy(expect_iter_me, NULL);
 	nf_ct_iterate_destroy(unhelp, me);
-
-	/* Maybe someone has gotten the helper already when unhelp above.
-	 * So need to wait it.
-	 */
-	synchronize_rcu();
 }
 EXPORT_SYMBOL_GPL(nf_conntrack_helper_unregister);
 
diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c
index 924d766e6c53..e768f59741a6 100644
--- a/net/netfilter/nf_conntrack_netlink.c
+++ b/net/netfilter/nf_conntrack_netlink.c
@@ -1559,6 +1559,11 @@ static int ctnetlink_flush_conntrack(struct net *net,
 				     u32 portid, int report, u8 family)
 {
 	struct ctnetlink_filter *filter = NULL;
+	struct nf_ct_iter_data iter = {
+		.net		= net,
+		.portid		= portid,
+		.report		= report,
+	};
 
 	if (ctnetlink_needs_filter(family, cda)) {
 		if (cda[CTA_FILTER])
@@ -1567,10 +1572,11 @@ static int ctnetlink_flush_conntrack(struct net *net,
 		filter = ctnetlink_alloc_filter(cda, family);
 		if (IS_ERR(filter))
 			return PTR_ERR(filter);
+
+		iter.data = filter;
 	}
 
-	nf_ct_iterate_cleanup_net(net, ctnetlink_flush_iterate, filter,
-				  portid, report);
+	nf_ct_iterate_cleanup_net(ctnetlink_flush_iterate, &iter);
 	kfree(filter);
 
 	return 0;
@@ -1750,61 +1756,59 @@ static int ctnetlink_dump_one_entry(struct sk_buff *skb,
 }
 
 static int
-ctnetlink_dump_list(struct sk_buff *skb, struct netlink_callback *cb, bool dying)
+ctnetlink_dump_unconfirmed(struct sk_buff *skb, struct netlink_callback *cb)
+{
+	return 0;
+}
+
+static int
+ctnetlink_dump_dying(struct sk_buff *skb, struct netlink_callback *cb)
 {
 	struct ctnetlink_list_dump_ctx *ctx = (void *)cb->ctx;
-	struct nf_conn *ct, *last;
+	struct nf_conn *last = ctx->last;
+#ifdef CONFIG_NF_CONNTRACK_EVENTS
+	const struct net *net = sock_net(skb->sk);
+	struct nf_conntrack_net_ecache *ecache_net;
 	struct nf_conntrack_tuple_hash *h;
 	struct hlist_nulls_node *n;
-	struct hlist_nulls_head *list;
-	struct net *net = sock_net(skb->sk);
-	int res, cpu;
+#endif
 
 	if (ctx->done)
 		return 0;
 
-	last = ctx->last;
+	ctx->last = NULL;
 
-	for (cpu = ctx->cpu; cpu < nr_cpu_ids; cpu++) {
-		struct ct_pcpu *pcpu;
+#ifdef CONFIG_NF_CONNTRACK_EVENTS
+	ecache_net = nf_conn_pernet_ecache(net);
+	spin_lock_bh(&ecache_net->dying_lock);
 
-		if (!cpu_possible(cpu))
-			continue;
+	hlist_nulls_for_each_entry(h, n, &ecache_net->dying_list, hnnode) {
+		struct nf_conn *ct;
+		int res;
 
-		pcpu = per_cpu_ptr(net->ct.pcpu_lists, cpu);
-		spin_lock_bh(&pcpu->lock);
-		list = dying ? &pcpu->dying : &pcpu->unconfirmed;
-restart:
-		hlist_nulls_for_each_entry(h, n, list, hnnode) {
-			ct = nf_ct_tuplehash_to_ctrack(h);
+		ct = nf_ct_tuplehash_to_ctrack(h);
+		if (last && last != ct)
+			continue;
 
-			res = ctnetlink_dump_one_entry(skb, cb, ct, dying);
-			if (res < 0) {
-				ctx->cpu = cpu;
-				spin_unlock_bh(&pcpu->lock);
-				goto out;
-			}
-		}
-		if (ctx->last) {
-			ctx->last = NULL;
-			goto restart;
+		res = ctnetlink_dump_one_entry(skb, cb, ct, true);
+		if (res < 0) {
+			spin_unlock_bh(&ecache_net->dying_lock);
+			nf_ct_put(last);
+			return skb->len;
 		}
-		spin_unlock_bh(&pcpu->lock);
+
+		nf_ct_put(last);
+		last = NULL;
 	}
+
+	spin_unlock_bh(&ecache_net->dying_lock);
+#endif
 	ctx->done = true;
-out:
-	if (last)
-		nf_ct_put(last);
+	nf_ct_put(last);
 
 	return skb->len;
 }
 
-static int
-ctnetlink_dump_dying(struct sk_buff *skb, struct netlink_callback *cb)
-{
-	return ctnetlink_dump_list(skb, cb, true);
-}
-
 static int ctnetlink_get_ct_dying(struct sk_buff *skb,
 				  const struct nfnl_info *info,
 				  const struct nlattr * const cda[])
@@ -1820,12 +1824,6 @@ static int ctnetlink_get_ct_dying(struct sk_buff *skb,
 	return -EOPNOTSUPP;
 }
 
-static int
-ctnetlink_dump_unconfirmed(struct sk_buff *skb, struct netlink_callback *cb)
-{
-	return ctnetlink_dump_list(skb, cb, false);
-}
-
 static int ctnetlink_get_ct_unconfirmed(struct sk_buff *skb,
 					const struct nfnl_info *info,
 					const struct nlattr * const cda[])
diff --git a/net/netfilter/nf_conntrack_proto.c b/net/netfilter/nf_conntrack_proto.c
index d1f2d3c8d2b1..895b09cbd7cf 100644
--- a/net/netfilter/nf_conntrack_proto.c
+++ b/net/netfilter/nf_conntrack_proto.c
@@ -538,9 +538,13 @@ retry:
  out_unlock:
 	mutex_unlock(&nf_ct_proto_mutex);
 
-	if (fixup_needed)
-		nf_ct_iterate_cleanup_net(net, nf_ct_tcp_fixup,
-					  (void *)(unsigned long)nfproto, 0, 0);
+	if (fixup_needed) {
+		struct nf_ct_iter_data iter_data = {
+			.net	= net,
+			.data	= (void *)(unsigned long)nfproto,
+		};
+		nf_ct_iterate_cleanup_net(nf_ct_tcp_fixup, &iter_data);
+	}
 
 	return err;
 }
diff --git a/net/netfilter/nf_conntrack_standalone.c b/net/netfilter/nf_conntrack_standalone.c
index 55aa55b252b2..6ad7bbc90d38 100644
--- a/net/netfilter/nf_conntrack_standalone.c
+++ b/net/netfilter/nf_conntrack_standalone.c
@@ -693,7 +693,7 @@ static struct ctl_table nf_ct_sysctl_table[] = {
 		.mode		= 0644,
 		.proc_handler	= proc_dou8vec_minmax,
 		.extra1 	= SYSCTL_ZERO,
-		.extra2 	= SYSCTL_ONE,
+		.extra2		= SYSCTL_TWO,
 	},
 #endif
 #ifdef CONFIG_NF_CONNTRACK_TIMESTAMP
diff --git a/net/netfilter/nf_conntrack_timeout.c b/net/netfilter/nf_conntrack_timeout.c
index cec166ecba77..0f828d05ea60 100644
--- a/net/netfilter/nf_conntrack_timeout.c
+++ b/net/netfilter/nf_conntrack_timeout.c
@@ -38,7 +38,12 @@ static int untimeout(struct nf_conn *ct, void *timeout)
 
 void nf_ct_untimeout(struct net *net, struct nf_ct_timeout *timeout)
 {
-	nf_ct_iterate_cleanup_net(net, untimeout, timeout, 0, 0);
+	struct nf_ct_iter_data iter_data = {
+		.net	= net,
+		.data	= timeout,
+	};
+
+	nf_ct_iterate_cleanup_net(untimeout, &iter_data);
 }
 EXPORT_SYMBOL_GPL(nf_ct_untimeout);
 
diff --git a/net/netfilter/nf_nat_masquerade.c b/net/netfilter/nf_nat_masquerade.c
index e32fac374608..1a506b0c6511 100644
--- a/net/netfilter/nf_nat_masquerade.c
+++ b/net/netfilter/nf_nat_masquerade.c
@@ -77,11 +77,14 @@ EXPORT_SYMBOL_GPL(nf_nat_masquerade_ipv4);
 
 static void iterate_cleanup_work(struct work_struct *work)
 {
+	struct nf_ct_iter_data iter_data = {};
 	struct masq_dev_work *w;
 
 	w = container_of(work, struct masq_dev_work, work);
 
-	nf_ct_iterate_cleanup_net(w->net, w->iter, (void *)w, 0, 0);
+	iter_data.net = w->net;
+	iter_data.data = (void *)w;
+	nf_ct_iterate_cleanup_net(w->iter, &iter_data);
 
 	put_net_track(w->net, &w->ns_tracker);
 	kfree(w);
diff --git a/net/netfilter/nfnetlink.c b/net/netfilter/nfnetlink.c
index 7e2c8dd01408..ad3bbe34ca88 100644
--- a/net/netfilter/nfnetlink.c
+++ b/net/netfilter/nfnetlink.c
@@ -45,6 +45,7 @@ MODULE_DESCRIPTION("Netfilter messages via netlink socket");
 static unsigned int nfnetlink_pernet_id __read_mostly;
 
 struct nfnl_net {
+	unsigned int ctnetlink_listeners;
 	struct sock *nfnl;
 };
 
@@ -654,7 +655,6 @@ static void nfnetlink_rcv(struct sk_buff *skb)
 		netlink_rcv_skb(skb, nfnetlink_rcv_msg);
 }
 
-#ifdef CONFIG_MODULES
 static int nfnetlink_bind(struct net *net, int group)
 {
 	const struct nfnetlink_subsystem *ss;
@@ -670,9 +670,44 @@ static int nfnetlink_bind(struct net *net, int group)
 	rcu_read_unlock();
 	if (!ss)
 		request_module_nowait("nfnetlink-subsys-%d", type);
+
+#ifdef CONFIG_NF_CONNTRACK_EVENTS
+	if (type == NFNL_SUBSYS_CTNETLINK) {
+		struct nfnl_net *nfnlnet = nfnl_pernet(net);
+
+		nfnl_lock(NFNL_SUBSYS_CTNETLINK);
+
+		if (WARN_ON_ONCE(nfnlnet->ctnetlink_listeners == UINT_MAX)) {
+			nfnl_unlock(NFNL_SUBSYS_CTNETLINK);
+			return -EOVERFLOW;
+		}
+
+		nfnlnet->ctnetlink_listeners++;
+		if (nfnlnet->ctnetlink_listeners == 1)
+			WRITE_ONCE(net->ct.ctnetlink_has_listener, true);
+		nfnl_unlock(NFNL_SUBSYS_CTNETLINK);
+	}
+#endif
 	return 0;
 }
+
+static void nfnetlink_unbind(struct net *net, int group)
+{
+#ifdef CONFIG_NF_CONNTRACK_EVENTS
+	int type = nfnl_group2type[group];
+
+	if (type == NFNL_SUBSYS_CTNETLINK) {
+		struct nfnl_net *nfnlnet = nfnl_pernet(net);
+
+		nfnl_lock(NFNL_SUBSYS_CTNETLINK);
+		WARN_ON_ONCE(nfnlnet->ctnetlink_listeners == 0);
+		nfnlnet->ctnetlink_listeners--;
+		if (nfnlnet->ctnetlink_listeners == 0)
+			WRITE_ONCE(net->ct.ctnetlink_has_listener, false);
+		nfnl_unlock(NFNL_SUBSYS_CTNETLINK);
+	}
 #endif
+}
 
 static int __net_init nfnetlink_net_init(struct net *net)
 {
@@ -680,9 +715,8 @@ static int __net_init nfnetlink_net_init(struct net *net)
 	struct netlink_kernel_cfg cfg = {
 		.groups	= NFNLGRP_MAX,
 		.input	= nfnetlink_rcv,
-#ifdef CONFIG_MODULES
 		.bind	= nfnetlink_bind,
-#endif
+		.unbind	= nfnetlink_unbind,
 	};
 
 	nfnlnet->nfnl = netlink_kernel_create(net, NETLINK_NETFILTER, &cfg);
diff --git a/net/netfilter/nfnetlink_cttimeout.c b/net/netfilter/nfnetlink_cttimeout.c
index eea486f32971..f069c24c6146 100644
--- a/net/netfilter/nfnetlink_cttimeout.c
+++ b/net/netfilter/nfnetlink_cttimeout.c
@@ -33,8 +33,19 @@
 
 static unsigned int nfct_timeout_id __read_mostly;
 
+struct ctnl_timeout {
+	struct list_head	head;
+	struct rcu_head		rcu_head;
+	refcount_t		refcnt;
+	char			name[CTNL_TIMEOUT_NAME_MAX];
+	struct nf_ct_timeout	timeout;
+
+	struct list_head	free_head;
+};
+
 struct nfct_timeout_pernet {
 	struct list_head	nfct_timeout_list;
+	struct list_head	nfct_timeout_freelist;
 };
 
 MODULE_LICENSE("GPL");
@@ -574,20 +585,36 @@ static int __net_init cttimeout_net_init(struct net *net)
 	struct nfct_timeout_pernet *pernet = nfct_timeout_pernet(net);
 
 	INIT_LIST_HEAD(&pernet->nfct_timeout_list);
+	INIT_LIST_HEAD(&pernet->nfct_timeout_freelist);
 
 	return 0;
 }
 
+static void __net_exit cttimeout_net_pre_exit(struct net *net)
+{
+	struct nfct_timeout_pernet *pernet = nfct_timeout_pernet(net);
+	struct ctnl_timeout *cur, *tmp;
+
+	list_for_each_entry_safe(cur, tmp, &pernet->nfct_timeout_list, head) {
+		list_del_rcu(&cur->head);
+		list_add(&cur->free_head, &pernet->nfct_timeout_freelist);
+	}
+
+	/* core calls synchronize_rcu() after this */
+}
+
 static void __net_exit cttimeout_net_exit(struct net *net)
 {
 	struct nfct_timeout_pernet *pernet = nfct_timeout_pernet(net);
 	struct ctnl_timeout *cur, *tmp;
 
-	nf_ct_unconfirmed_destroy(net);
+	if (list_empty(&pernet->nfct_timeout_freelist))
+		return;
+
 	nf_ct_untimeout(net, NULL);
 
-	list_for_each_entry_safe(cur, tmp, &pernet->nfct_timeout_list, head) {
-		list_del_rcu(&cur->head);
+	list_for_each_entry_safe(cur, tmp, &pernet->nfct_timeout_freelist, head) {
+		list_del(&cur->free_head);
 
 		if (refcount_dec_and_test(&cur->refcnt))
 			kfree_rcu(cur, rcu_head);
@@ -596,6 +623,7 @@ static void __net_exit cttimeout_net_exit(struct net *net)
 
 static struct pernet_operations cttimeout_ops = {
 	.init	= cttimeout_net_init,
+	.pre_exit = cttimeout_net_pre_exit,
 	.exit	= cttimeout_net_exit,
 	.id     = &nfct_timeout_id,
 	.size   = sizeof(struct nfct_timeout_pernet),
@@ -628,13 +656,24 @@ err_out:
 	return ret;
 }
 
+static int untimeout(struct nf_conn *ct, void *timeout)
+{
+	struct nf_conn_timeout *timeout_ext = nf_ct_timeout_find(ct);
+
+	if (timeout_ext)
+		RCU_INIT_POINTER(timeout_ext->timeout, NULL);
+
+	return 0;
+}
+
 static void __exit cttimeout_exit(void)
 {
 	nfnetlink_subsys_unregister(&cttimeout_subsys);
 
 	unregister_pernet_subsys(&cttimeout_ops);
 	RCU_INIT_POINTER(nf_ct_timeout_hook, NULL);
-	synchronize_rcu();
+
+	nf_ct_iterate_destroy(untimeout, NULL);
 }
 
 module_init(cttimeout_init);
diff --git a/net/netfilter/nft_flow_offload.c b/net/netfilter/nft_flow_offload.c
index 900d48c810a1..50991ab1e2d2 100644
--- a/net/netfilter/nft_flow_offload.c
+++ b/net/netfilter/nft_flow_offload.c
@@ -227,11 +227,19 @@ static int nft_flow_route(const struct nft_pktinfo *pkt,
 	switch (nft_pf(pkt)) {
 	case NFPROTO_IPV4:
 		fl.u.ip4.daddr = ct->tuplehash[dir].tuple.src.u3.ip;
+		fl.u.ip4.saddr = ct->tuplehash[dir].tuple.dst.u3.ip;
 		fl.u.ip4.flowi4_oif = nft_in(pkt)->ifindex;
+		fl.u.ip4.flowi4_iif = this_dst->dev->ifindex;
+		fl.u.ip4.flowi4_tos = RT_TOS(ip_hdr(pkt->skb)->tos);
+		fl.u.ip4.flowi4_mark = pkt->skb->mark;
 		break;
 	case NFPROTO_IPV6:
 		fl.u.ip6.daddr = ct->tuplehash[dir].tuple.src.u3.in6;
+		fl.u.ip6.saddr = ct->tuplehash[dir].tuple.dst.u3.in6;
 		fl.u.ip6.flowi6_oif = nft_in(pkt)->ifindex;
+		fl.u.ip6.flowi6_iif = this_dst->dev->ifindex;
+		fl.u.ip6.flowlabel = ip6_flowinfo(ipv6_hdr(pkt->skb));
+		fl.u.ip6.flowi6_mark = pkt->skb->mark;
 		break;
 	}