Merge git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf

Pablo Neira Ayuso says:

====================
Netfilter/IPVS fixes for net

The following patchset contains Netfilter/IPVS fixes for your net tree:

1) Fix crash when dumping rules after conversion to RCU,
   from Florian Westphal.

2) Fix incorrect hook reinjection from nf_queue in case NF_REPEAT,
   from Jagdish Motwani.

3) Fix check for route existence in fib extension, from Phil Sutter.

4) Fix use after free in ip_vs_in() hook, from YueHaibing.

5) Check for veth existence from netfilter selftests,
   from Jeffrin Jose T.

6) Checksum corruption in UDP NAT helpers due to typo,
   from Florian Westphal.

7) Pass up packets to classic forwarding path regardless of
   IPv4 DF bit, patch for the flowtable infrastructure from Florian.

8) Set liberal TCP tracking for flows that are placed in the
   flowtable, in case they need to go back to classic forwarding path,
   also from Florian.

9) Don't add flow with sequence adjustment to flowtable, from Florian.

10) Skip IPv4 options from IPv6 datapath in flowtable, from Florian.

11) Add selftest for the flowtable infrastructure, from Florian.
====================

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

13 files changed, 375 insertions, 63 deletions

diff --git a/include/net/netfilter/nft_fib.h b/include/net/netfilter/nft_fib.h
index a88f92737308..e4c4d8eaca8c 100644
--- a/include/net/netfilter/nft_fib.h
+++ b/include/net/netfilter/nft_fib.h
@@ -34,5 +34,5 @@ void nft_fib6_eval(const struct nft_expr *expr, struct nft_regs *regs,
 		   const struct nft_pktinfo *pkt);
 
 void nft_fib_store_result(void *reg, const struct nft_fib *priv,
-			  const struct nft_pktinfo *pkt, int index);
+			  const struct net_device *dev);
 #endif
diff --git a/net/ipv4/netfilter/nft_fib_ipv4.c b/net/ipv4/netfilter/nft_fib_ipv4.c
index 94eb25bc8d7e..c8888e52591f 100644
--- a/net/ipv4/netfilter/nft_fib_ipv4.c
+++ b/net/ipv4/netfilter/nft_fib_ipv4.c
@@ -58,11 +58,6 @@ void nft_fib4_eval_type(const struct nft_expr *expr, struct nft_regs *regs,
 }
 EXPORT_SYMBOL_GPL(nft_fib4_eval_type);
 
-static int get_ifindex(const struct net_device *dev)
-{
-	return dev ? dev->ifindex : 0;
-}
-
 void nft_fib4_eval(const struct nft_expr *expr, struct nft_regs *regs,
 		   const struct nft_pktinfo *pkt)
 {
@@ -94,8 +89,7 @@ void nft_fib4_eval(const struct nft_expr *expr, struct nft_regs *regs,
 
 	if (nft_hook(pkt) == NF_INET_PRE_ROUTING &&
 	    nft_fib_is_loopback(pkt->skb, nft_in(pkt))) {
-		nft_fib_store_result(dest, priv, pkt,
-				     nft_in(pkt)->ifindex);
+		nft_fib_store_result(dest, priv, nft_in(pkt));
 		return;
 	}
 
@@ -108,8 +102,7 @@ void nft_fib4_eval(const struct nft_expr *expr, struct nft_regs *regs,
 	if (ipv4_is_zeronet(iph->saddr)) {
 		if (ipv4_is_lbcast(iph->daddr) ||
 		    ipv4_is_local_multicast(iph->daddr)) {
-			nft_fib_store_result(dest, priv, pkt,
-					     get_ifindex(pkt->skb->dev));
+			nft_fib_store_result(dest, priv, pkt->skb->dev);
 			return;
 		}
 	}
@@ -150,17 +143,7 @@ void nft_fib4_eval(const struct nft_expr *expr, struct nft_regs *regs,
 		found = oif;
 	}
 
-	switch (priv->result) {
-	case NFT_FIB_RESULT_OIF:
-		*dest = found->ifindex;
-		break;
-	case NFT_FIB_RESULT_OIFNAME:
-		strncpy((char *)dest, found->name, IFNAMSIZ);
-		break;
-	default:
-		WARN_ON_ONCE(1);
-		break;
-	}
+	nft_fib_store_result(dest, priv, found);
 }
 EXPORT_SYMBOL_GPL(nft_fib4_eval);
 
diff --git a/net/ipv6/netfilter/nft_fib_ipv6.c b/net/ipv6/netfilter/nft_fib_ipv6.c
index 73cdc0bc63f7..ec068b0cffca 100644
--- a/net/ipv6/netfilter/nft_fib_ipv6.c
+++ b/net/ipv6/netfilter/nft_fib_ipv6.c
@@ -169,8 +169,7 @@ void nft_fib6_eval(const struct nft_expr *expr, struct nft_regs *regs,
 
 	if (nft_hook(pkt) == NF_INET_PRE_ROUTING &&
 	    nft_fib_is_loopback(pkt->skb, nft_in(pkt))) {
-		nft_fib_store_result(dest, priv, pkt,
-				     nft_in(pkt)->ifindex);
+		nft_fib_store_result(dest, priv, nft_in(pkt));
 		return;
 	}
 
@@ -187,18 +186,7 @@ void nft_fib6_eval(const struct nft_expr *expr, struct nft_regs *regs,
 	if (oif && oif != rt->rt6i_idev->dev)
 		goto put_rt_err;
 
-	switch (priv->result) {
-	case NFT_FIB_RESULT_OIF:
-		*dest = rt->rt6i_idev->dev->ifindex;
-		break;
-	case NFT_FIB_RESULT_OIFNAME:
-		strncpy((char *)dest, rt->rt6i_idev->dev->name, IFNAMSIZ);
-		break;
-	default:
-		WARN_ON_ONCE(1);
-		break;
-	}
-
+	nft_fib_store_result(dest, priv, rt->rt6i_idev->dev);
  put_rt_err:
 	ip6_rt_put(rt);
 }
diff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c
index 14457551bcb4..8ebf21149ec3 100644
--- a/net/netfilter/ipvs/ip_vs_core.c
+++ b/net/netfilter/ipvs/ip_vs_core.c
@@ -2312,7 +2312,6 @@ static void __net_exit __ip_vs_cleanup(struct net *net)
 {
 	struct netns_ipvs *ipvs = net_ipvs(net);
 
-	nf_unregister_net_hooks(net, ip_vs_ops, ARRAY_SIZE(ip_vs_ops));
 	ip_vs_service_net_cleanup(ipvs);	/* ip_vs_flush() with locks */
 	ip_vs_conn_net_cleanup(ipvs);
 	ip_vs_app_net_cleanup(ipvs);
@@ -2327,6 +2326,7 @@ static void __net_exit __ip_vs_dev_cleanup(struct net *net)
 {
 	struct netns_ipvs *ipvs = net_ipvs(net);
 	EnterFunction(2);
+	nf_unregister_net_hooks(net, ip_vs_ops, ARRAY_SIZE(ip_vs_ops));
 	ipvs->enable = 0;	/* Disable packet reception */
 	smp_wmb();
 	ip_vs_sync_net_cleanup(ipvs);
diff --git a/net/netfilter/nf_flow_table_ip.c b/net/netfilter/nf_flow_table_ip.c
index 96825e20368f..241317473114 100644
--- a/net/netfilter/nf_flow_table_ip.c
+++ b/net/netfilter/nf_flow_table_ip.c
@@ -244,8 +244,7 @@ nf_flow_offload_ip_hook(void *priv, struct sk_buff *skb,
 	rt = (struct rtable *)flow->tuplehash[dir].tuple.dst_cache;
 	outdev = rt->dst.dev;
 
-	if (unlikely(nf_flow_exceeds_mtu(skb, flow->tuplehash[dir].tuple.mtu)) &&
-	    (ip_hdr(skb)->frag_off & htons(IP_DF)) != 0)
+	if (unlikely(nf_flow_exceeds_mtu(skb, flow->tuplehash[dir].tuple.mtu)))
 		return NF_ACCEPT;
 
 	if (skb_try_make_writable(skb, sizeof(*iph)))
diff --git a/net/netfilter/nf_nat_helper.c b/net/netfilter/nf_nat_helper.c
index ccc06f7539d7..53aeb12b70fb 100644
--- a/net/netfilter/nf_nat_helper.c
+++ b/net/netfilter/nf_nat_helper.c
@@ -170,7 +170,7 @@ nf_nat_mangle_udp_packet(struct sk_buff *skb,
 	if (!udph->check && skb->ip_summed != CHECKSUM_PARTIAL)
 		return true;
 
-	nf_nat_csum_recalc(skb, nf_ct_l3num(ct), IPPROTO_TCP,
+	nf_nat_csum_recalc(skb, nf_ct_l3num(ct), IPPROTO_UDP,
 			   udph, &udph->check, datalen, oldlen);
 
 	return true;
diff --git a/net/netfilter/nf_queue.c b/net/netfilter/nf_queue.c
index 9dc1d6e04946..b5b2be55ca82 100644
--- a/net/netfilter/nf_queue.c
+++ b/net/netfilter/nf_queue.c
@@ -255,6 +255,7 @@ static unsigned int nf_iterate(struct sk_buff *skb,
 repeat:
 		verdict = nf_hook_entry_hookfn(hook, skb, state);
 		if (verdict != NF_ACCEPT) {
+			*index = i;
 			if (verdict != NF_REPEAT)
 				return verdict;
 			goto repeat;
diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 28241e82fd15..4b5159936034 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -2270,13 +2270,13 @@ static int nf_tables_fill_rule_info(struct sk_buff *skb, struct net *net,
 				    u32 flags, int family,
 				    const struct nft_table *table,
 				    const struct nft_chain *chain,
-				    const struct nft_rule *rule)
+				    const struct nft_rule *rule,
+				    const struct nft_rule *prule)
 {
 	struct nlmsghdr *nlh;
 	struct nfgenmsg *nfmsg;
 	const struct nft_expr *expr, *next;
 	struct nlattr *list;
-	const struct nft_rule *prule;
 	u16 type = nfnl_msg_type(NFNL_SUBSYS_NFTABLES, event);
 
 	nlh = nlmsg_put(skb, portid, seq, type, sizeof(struct nfgenmsg), flags);
@@ -2296,8 +2296,7 @@ static int nf_tables_fill_rule_info(struct sk_buff *skb, struct net *net,
 			 NFTA_RULE_PAD))
 		goto nla_put_failure;
 
-	if ((event != NFT_MSG_DELRULE) && (rule->list.prev != &chain->rules)) {
-		prule = list_prev_entry(rule, list);
+	if (event != NFT_MSG_DELRULE && prule) {
 		if (nla_put_be64(skb, NFTA_RULE_POSITION,
 				 cpu_to_be64(prule->handle),
 				 NFTA_RULE_PAD))
@@ -2344,7 +2343,7 @@ static void nf_tables_rule_notify(const struct nft_ctx *ctx,
 
 	err = nf_tables_fill_rule_info(skb, ctx->net, ctx->portid, ctx->seq,
 				       event, 0, ctx->family, ctx->table,
-				       ctx->chain, rule);
+				       ctx->chain, rule, NULL);
 	if (err < 0) {
 		kfree_skb(skb);
 		goto err;
@@ -2369,12 +2368,13 @@ static int __nf_tables_dump_rules(struct sk_buff *skb,
 				  const struct nft_chain *chain)
 {
 	struct net *net = sock_net(skb->sk);
+	const struct nft_rule *rule, *prule;
 	unsigned int s_idx = cb->args[0];
-	const struct nft_rule *rule;
 
+	prule = NULL;
 	list_for_each_entry_rcu(rule, &chain->rules, list) {
 		if (!nft_is_active(net, rule))
-			goto cont;
+			goto cont_skip;
 		if (*idx < s_idx)
 			goto cont;
 		if (*idx > s_idx) {
@@ -2386,11 +2386,13 @@ static int __nf_tables_dump_rules(struct sk_buff *skb,
 					NFT_MSG_NEWRULE,
 					NLM_F_MULTI | NLM_F_APPEND,
 					table->family,
-					table, chain, rule) < 0)
+					table, chain, rule, prule) < 0)
 			return 1;
 
 		nl_dump_check_consistent(cb, nlmsg_hdr(skb));
 cont:
+		prule = rule;
+cont_skip:
 		(*idx)++;
 	}
 	return 0;
@@ -2546,7 +2548,7 @@ static int nf_tables_getrule(struct net *net, struct sock *nlsk,
 
 	err = nf_tables_fill_rule_info(skb2, net, NETLINK_CB(skb).portid,
 				       nlh->nlmsg_seq, NFT_MSG_NEWRULE, 0,
-				       family, table, chain, rule);
+				       family, table, chain, rule, NULL);
 	if (err < 0)
 		goto err;
 
diff --git a/net/netfilter/nft_fib.c b/net/netfilter/nft_fib.c
index 21df8cccea65..77f00a99dfab 100644
--- a/net/netfilter/nft_fib.c
+++ b/net/netfilter/nft_fib.c
@@ -135,17 +135,17 @@ int nft_fib_dump(struct sk_buff *skb, const struct nft_expr *expr)
 EXPORT_SYMBOL_GPL(nft_fib_dump);
 
 void nft_fib_store_result(void *reg, const struct nft_fib *priv,
-			  const struct nft_pktinfo *pkt, int index)
+			  const struct net_device *dev)
 {
-	struct net_device *dev;
 	u32 *dreg = reg;
+	int index;
 
 	switch (priv->result) {
 	case NFT_FIB_RESULT_OIF:
+		index = dev ? dev->ifindex : 0;
 		*dreg = (priv->flags & NFTA_FIB_F_PRESENT) ? !!index : index;
 		break;
 	case NFT_FIB_RESULT_OIFNAME:
-		dev = dev_get_by_index_rcu(nft_net(pkt), index);
 		if (priv->flags & NFTA_FIB_F_PRESENT)
 			*dreg = !!dev;
 		else
diff --git a/net/netfilter/nft_flow_offload.c b/net/netfilter/nft_flow_offload.c
index ffb25d5e8dbe..aa5f571d4361 100644
--- a/net/netfilter/nft_flow_offload.c
+++ b/net/netfilter/nft_flow_offload.c
@@ -13,7 +13,6 @@
 #include <net/netfilter/nf_conntrack_core.h>
 #include <linux/netfilter/nf_conntrack_common.h>
 #include <net/netfilter/nf_flow_table.h>
-#include <net/netfilter/nf_conntrack_helper.h>
 
 struct nft_flow_offload {
 	struct nft_flowtable	*flowtable;
@@ -50,15 +49,20 @@ static int nft_flow_route(const struct nft_pktinfo *pkt,
 	return 0;
 }
 
-static bool nft_flow_offload_skip(struct sk_buff *skb)
+static bool nft_flow_offload_skip(struct sk_buff *skb, int family)
 {
-	struct ip_options *opt  = &(IPCB(skb)->opt);
-
-	if (unlikely(opt->optlen))
-		return true;
 	if (skb_sec_path(skb))
 		return true;
 
+	if (family == NFPROTO_IPV4) {
+		const struct ip_options *opt;
+
+		opt = &(IPCB(skb)->opt);
+
+		if (unlikely(opt->optlen))
+			return true;
+	}
+
 	return false;
 }
 
@@ -68,15 +72,15 @@ static void nft_flow_offload_eval(const struct nft_expr *expr,
 {
 	struct nft_flow_offload *priv = nft_expr_priv(expr);
 	struct nf_flowtable *flowtable = &priv->flowtable->data;
-	const struct nf_conn_help *help;
 	enum ip_conntrack_info ctinfo;
 	struct nf_flow_route route;
 	struct flow_offload *flow;
 	enum ip_conntrack_dir dir;
+	bool is_tcp = false;
 	struct nf_conn *ct;
 	int ret;
 
-	if (nft_flow_offload_skip(pkt->skb))
+	if (nft_flow_offload_skip(pkt->skb, nft_pf(pkt)))
 		goto out;
 
 	ct = nf_ct_get(pkt->skb, &ctinfo);
@@ -85,14 +89,16 @@ static void nft_flow_offload_eval(const struct nft_expr *expr,
 
 	switch (ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.dst.protonum) {
 	case IPPROTO_TCP:
+		is_tcp = true;
+		break;
 	case IPPROTO_UDP:
 		break;
 	default:
 		goto out;
 	}
 
-	help = nfct_help(ct);
-	if (help)
+	if (nf_ct_ext_exist(ct, NF_CT_EXT_HELPER) ||
+	    ct->status & IPS_SEQ_ADJUST)
 		goto out;
 
 	if (!nf_ct_is_confirmed(ct))
@@ -109,6 +115,11 @@ static void nft_flow_offload_eval(const struct nft_expr *expr,
 	if (!flow)
 		goto err_flow_alloc;
 
+	if (is_tcp) {
+		ct->proto.tcp.seen[0].flags |= IP_CT_TCP_FLAG_BE_LIBERAL;
+		ct->proto.tcp.seen[1].flags |= IP_CT_TCP_FLAG_BE_LIBERAL;
+	}
+
 	ret = flow_offload_add(flowtable, flow);
 	if (ret < 0)
 		goto err_flow_add;
diff --git a/tools/testing/selftests/netfilter/Makefile b/tools/testing/selftests/netfilter/Makefile
index 3e6d1bcc2894..4144984ebee5 100644
--- a/tools/testing/selftests/netfilter/Makefile
+++ b/tools/testing/selftests/netfilter/Makefile
@@ -2,6 +2,6 @@
 # Makefile for netfilter selftests
 
 TEST_PROGS := nft_trans_stress.sh nft_nat.sh bridge_brouter.sh \
-	conntrack_icmp_related.sh
+	conntrack_icmp_related.sh nft_flowtable.sh
 
 include ../lib.mk
diff --git a/tools/testing/selftests/netfilter/nft_flowtable.sh b/tools/testing/selftests/netfilter/nft_flowtable.sh
new file mode 100755
index 000000000000..fe52488a6f72
--- /dev/null
+++ b/tools/testing/selftests/netfilter/nft_flowtable.sh
@@ -0,0 +1,324 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+#
+# This tests basic flowtable functionality.
+# Creates following topology:
+#
+# Originator (MTU 9000) <-Router1-> MTU 1500 <-Router2-> Responder (MTU 2000)
+# Router1 is the one doing flow offloading, Router2 has no special
+# purpose other than having a link that is smaller than either Originator
+# and responder, i.e. TCPMSS announced values are too large and will still
+# result in fragmentation and/or PMTU discovery.
+
+# Kselftest framework requirement - SKIP code is 4.
+ksft_skip=4
+ret=0
+
+ns1in=""
+ns2in=""
+ns1out=""
+ns2out=""
+
+log_netns=$(sysctl -n net.netfilter.nf_log_all_netns)
+
+nft --version > /dev/null 2>&1
+if [ $? -ne 0 ];then
+	echo "SKIP: Could not run test without nft tool"
+	exit $ksft_skip
+fi
+
+ip -Version > /dev/null 2>&1
+if [ $? -ne 0 ];then
+	echo "SKIP: Could not run test without ip tool"
+	exit $ksft_skip
+fi
+
+which nc > /dev/null 2>&1
+if [ $? -ne 0 ];then
+	echo "SKIP: Could not run test without nc (netcat)"
+	exit $ksft_skip
+fi
+
+ip netns add nsr1
+if [ $? -ne 0 ];then
+	echo "SKIP: Could not create net namespace"
+	exit $ksft_skip
+fi
+
+ip netns add ns1
+ip netns add ns2
+
+ip netns add nsr2
+
+cleanup() {
+	for i in 1 2; do
+		ip netns del ns$i
+		ip netns del nsr$i
+	done
+
+	rm -f "$ns1in" "$ns1out"
+	rm -f "$ns2in" "$ns2out"
+
+	[ $log_netns -eq 0 ] && sysctl -q net.netfilter.nf_log_all_netns=$log_netns
+}
+
+trap cleanup EXIT
+
+sysctl -q net.netfilter.nf_log_all_netns=1
+
+ip link add veth0 netns nsr1 type veth peer name eth0 netns ns1
+ip link add veth1 netns nsr1 type veth peer name veth0 netns nsr2
+
+ip link add veth1 netns nsr2 type veth peer name eth0 netns ns2
+
+for dev in lo veth0 veth1; do
+  for i in 1 2; do
+    ip -net nsr$i link set $dev up
+  done
+done
+
+ip -net nsr1 addr add 10.0.1.1/24 dev veth0
+ip -net nsr1 addr add dead:1::1/64 dev veth0
+
+ip -net nsr2 addr add 10.0.2.1/24 dev veth1
+ip -net nsr2 addr add dead:2::1/64 dev veth1
+
+# set different MTUs so we need to push packets coming from ns1 (large MTU)
+# to ns2 (smaller MTU) to stack either to perform fragmentation (ip_no_pmtu_disc=1),
+# or to do PTMU discovery (send ICMP error back to originator).
+# ns2 is going via nsr2 with a smaller mtu, so that TCPMSS announced by both peers
+# is NOT the lowest link mtu.
+
+ip -net nsr1 link set veth0 mtu 9000
+ip -net ns1 link set eth0 mtu 9000
+
+ip -net nsr2 link set veth1 mtu 2000
+ip -net ns2 link set eth0 mtu 2000
+
+# transfer-net between nsr1 and nsr2.
+# these addresses are not used for connections.
+ip -net nsr1 addr add 192.168.10.1/24 dev veth1
+ip -net nsr1 addr add fee1:2::1/64 dev veth1
+
+ip -net nsr2 addr add 192.168.10.2/24 dev veth0
+ip -net nsr2 addr add fee1:2::2/64 dev veth0
+
+for i in 1 2; do
+  ip netns exec nsr$i sysctl net.ipv4.conf.veth0.forwarding=1 > /dev/null
+  ip netns exec nsr$i sysctl net.ipv4.conf.veth1.forwarding=1 > /dev/null
+
+  ip -net ns$i link set lo up
+  ip -net ns$i link set eth0 up
+  ip -net ns$i addr add 10.0.$i.99/24 dev eth0
+  ip -net ns$i route add default via 10.0.$i.1
+  ip -net ns$i addr add dead:$i::99/64 dev eth0
+  ip -net ns$i route add default via dead:$i::1
+  ip netns exec ns$i sysctl net.ipv4.tcp_no_metrics_save=1 > /dev/null
+
+  # don't set ip DF bit for first two tests
+  ip netns exec ns$i sysctl net.ipv4.ip_no_pmtu_disc=1 > /dev/null
+done
+
+ip -net nsr1 route add default via 192.168.10.2
+ip -net nsr2 route add default via 192.168.10.1
+
+ip netns exec nsr1 nft -f - <<EOF
+table inet filter {
+  flowtable f1 {
+     hook ingress priority 0
+     devices = { veth0, veth1 }
+   }
+
+   chain forward {
+      type filter hook forward priority 0; policy drop;
+
+      # flow offloaded? Tag ct with mark 1, so we can detect when it fails.
+      meta oif "veth1" tcp dport 12345 flow offload @f1 counter
+
+      # use packet size to trigger 'should be offloaded by now'.
+      # otherwise, if 'flow offload' expression never offloads, the
+      # test will pass.
+      tcp dport 12345 meta length gt 200 ct mark set 1 counter
+
+      # this turns off flow offloading internally, so expect packets again
+      tcp flags fin,rst ct mark set 0 accept
+
+      # this allows large packets from responder, we need this as long
+      # as PMTUd is off.
+      # This rule is deleted for the last test, when we expect PMTUd
+      # to kick in and ensure all packets meet mtu requirements.
+      meta length gt 1500 accept comment something-to-grep-for
+
+      # next line blocks connection w.o. working offload.
+      # we only do this for reverse dir, because we expect packets to
+      # enter slow path due to MTU mismatch of veth0 and veth1.
+      tcp sport 12345 ct mark 1 counter log prefix "mark failure " drop
+
+      ct state established,related accept
+
+      # for packets that we can't offload yet, i.e. SYN (any ct that is not confirmed)
+      meta length lt 200 oif "veth1" tcp dport 12345 counter accept
+
+      meta nfproto ipv4 meta l4proto icmp accept
+      meta nfproto ipv6 meta l4proto icmpv6 accept
+   }
+}
+EOF
+
+if [ $? -ne 0 ]; then
+	echo "SKIP: Could not load nft ruleset"
+	exit $ksft_skip
+fi
+
+# test basic connectivity
+ip netns exec ns1 ping -c 1 -q 10.0.2.99 > /dev/null
+if [ $? -ne 0 ];then
+  echo "ERROR: ns1 cannot reach ns2" 1>&2
+  bash
+  exit 1
+fi
+
+ip netns exec ns2 ping -c 1 -q 10.0.1.99 > /dev/null
+if [ $? -ne 0 ];then
+  echo "ERROR: ns2 cannot reach ns1" 1>&2
+  exit 1
+fi
+
+if [ $ret -eq 0 ];then
+	echo "PASS: netns routing/connectivity: ns1 can reach ns2"
+fi
+
+ns1in=$(mktemp)
+ns1out=$(mktemp)
+ns2in=$(mktemp)
+ns2out=$(mktemp)
+
+make_file()
+{
+	name=$1
+	who=$2
+
+	SIZE=$((RANDOM % (1024 * 8)))
+	TSIZE=$((SIZE * 1024))
+
+	dd if=/dev/urandom of="$name" bs=1024 count=$SIZE 2> /dev/null
+
+	SIZE=$((RANDOM % 1024))
+	SIZE=$((SIZE + 128))
+	TSIZE=$((TSIZE + SIZE))
+	dd if=/dev/urandom conf=notrunc of="$name" bs=1 count=$SIZE 2> /dev/null
+}
+
+check_transfer()
+{
+	in=$1
+	out=$2
+	what=$3
+
+	cmp "$in" "$out" > /dev/null 2>&1
+	if [ $? -ne 0 ] ;then
+		echo "FAIL: file mismatch for $what" 1>&2
+		ls -l "$in"
+		ls -l "$out"
+		return 1
+	fi
+
+	return 0
+}
+
+test_tcp_forwarding()
+{
+	local nsa=$1
+	local nsb=$2
+	local lret=0
+
+	ip netns exec $nsb nc -w 5 -l -p 12345 < "$ns2in" > "$ns2out" &
+	lpid=$!
+
+	sleep 1
+	ip netns exec $nsa nc -w 4 10.0.2.99 12345 < "$ns1in" > "$ns1out" &
+	cpid=$!
+
+	sleep 3
+
+	kill $lpid
+	kill $cpid
+	wait
+
+	check_transfer "$ns1in" "$ns2out" "ns1 -> ns2"
+	if [ $? -ne 0 ];then
+		lret=1
+	fi
+
+	check_transfer "$ns2in" "$ns1out" "ns1 <- ns2"
+	if [ $? -ne 0 ];then
+		lret=1
+	fi
+
+	return $lret
+}
+
+make_file "$ns1in" "ns1"
+make_file "$ns2in" "ns2"
+
+# First test:
+# No PMTU discovery, nsr1 is expected to fragment packets from ns1 to ns2 as needed.
+test_tcp_forwarding ns1 ns2
+if [ $? -eq 0 ] ;then
+	echo "PASS: flow offloaded for ns1/ns2"
+else
+	echo "FAIL: flow offload for ns1/ns2:" 1>&2
+	ip netns exec nsr1 nft list ruleset
+	ret=1
+fi
+
+# delete default route, i.e. ns2 won't be able to reach ns1 and
+# will depend on ns1 being masqueraded in nsr1.
+# expect ns1 has nsr1 address.
+ip -net ns2 route del default via 10.0.2.1
+ip -net ns2 route del default via dead:2::1
+ip -net ns2 route add 192.168.10.1 via 10.0.2.1
+
+# Second test:
+# Same, but with NAT enabled.
+ip netns exec nsr1 nft -f - <<EOF
+table ip nat {
+   chain postrouting {
+      type nat hook postrouting priority 0; policy accept;
+      meta oifname "veth1" masquerade
+   }
+}
+EOF
+
+test_tcp_forwarding ns1 ns2
+
+if [ $? -eq 0 ] ;then
+	echo "PASS: flow offloaded for ns1/ns2 with NAT"
+else
+	echo "FAIL: flow offload for ns1/ns2 with NAT" 1>&2
+	ip netns exec nsr1 nft list ruleset
+	ret=1
+fi
+
+# Third test:
+# Same as second test, but with PMTU discovery enabled.
+handle=$(ip netns exec nsr1 nft -a list table inet filter | grep something-to-grep-for | cut -d \# -f 2)
+
+ip netns exec nsr1 nft delete rule inet filter forward $handle
+if [ $? -ne 0 ] ;then
+	echo "FAIL: Could not delete large-packet accept rule"
+	exit 1
+fi
+
+ip netns exec ns1 sysctl net.ipv4.ip_no_pmtu_disc=0 > /dev/null
+ip netns exec ns2 sysctl net.ipv4.ip_no_pmtu_disc=0 > /dev/null
+
+test_tcp_forwarding ns1 ns2
+if [ $? -eq 0 ] ;then
+	echo "PASS: flow offloaded for ns1/ns2 with NAT and pmtu discovery"
+else
+	echo "FAIL: flow offload for ns1/ns2 with NAT and pmtu discovery" 1>&2
+	ip netns exec nsr1 nft list ruleset
+fi
+
+exit $ret
diff --git a/tools/testing/selftests/netfilter/nft_nat.sh b/tools/testing/selftests/netfilter/nft_nat.sh
index 14fcf3104c77..1be55e705780 100755
--- a/tools/testing/selftests/netfilter/nft_nat.sh
+++ b/tools/testing/selftests/netfilter/nft_nat.sh
@@ -36,7 +36,11 @@ trap cleanup EXIT
 ip netns add ns1
 ip netns add ns2
 
-ip link add veth0 netns ns0 type veth peer name eth0 netns ns1
+ip link add veth0 netns ns0 type veth peer name eth0 netns ns1 > /dev/null 2>&1
+if [ $? -ne 0 ];then
+    echo "SKIP: No virtual ethernet pair device support in kernel"
+    exit $ksft_skip
+fi
 ip link add veth1 netns ns0 type veth peer name eth0 netns ns2
 
 ip -net ns0 link set lo up