[SRU][F][PATCH 1/1] ipv4: Fix device used for dst_alloc with local routes

Luke Nowakowski-Krijger luke.nowakowskikrijger at canonical.com
Mon Oct 4 18:41:09 UTC 2021

From: David Ahern <dsahern at kernel.org>

Oliver reported a use case where deleting a VRF device can hang
waiting for the refcnt to drop to 0. The root cause is that the dst
is allocated against the VRF device but cached on the loopback

The use case (added to the selftests) has an implicit VRF crossing
due to the ordering of the FIB rules (lookup local is before the
l3mdev rule, but the problem occurs even if the FIB rules are
re-ordered with local after l3mdev because the VRF table does not
have a default route to terminate the lookup). The end result is
is that the FIB lookup returns the loopback device as the nexthop,
but the ingress device is in a VRF. The mismatch causes the dst
alloc against the VRF device but then cached on the loopback.

The fix is to bring the trick used for IPv6 (see ip6_rt_get_dev_rcu):
pick the dst alloc device based the fib lookup result but with checks
that the result has a nexthop device (e.g., not an unreachable or
prohibit entry).

Fixes: f5a0aab84b74 ("net: ipv4: dst for local input routes should use l3mdev if relevant")
Reported-by: Oliver Herms <oliver.peter.herms at gmail.com>
Signed-off-by: David Ahern <dsahern at kernel.org>
Signed-off-by: David S. Miller <davem at davemloft.net>
(cherry picked from commit b87b04f5019e821c8c6c7761f258402e43500a1f)
Signed-off-by: Nicolas Dichtel <nicolas.dichtel at 6wind.com>
Signed-off-by: Luke Nowakowski-Krijger <luke.nowakowskikrijger at canonical.com>
 net/ipv4/route.c                         | 16 ++++++++++++++-
 tools/testing/selftests/net/fib_tests.sh | 25 ++++++++++++++++++++++++
 2 files changed, 40 insertions(+), 1 deletion(-)

diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 0e976848d4bb..b60510ff336f 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -2036,6 +2036,20 @@ static int ip_mkroute_input(struct sk_buff *skb,
 	return __mkroute_input(skb, res, in_dev, daddr, saddr, tos);
+/* get device for dst_alloc with local routes */
+static struct net_device *ip_rt_get_dev(struct net *net,
+					const struct fib_result *res)
+	struct fib_nh_common *nhc = res->fi ? res->nhc : NULL;
+	struct net_device *dev = NULL;
+	if (nhc)
+		dev = l3mdev_master_dev_rcu(nhc->nhc_dev);
+	return dev ? : net->loopback_dev;
  *	NOTE. We drop all the packets that has local source
  *	addresses, because every properly looped back packet
@@ -2190,7 +2204,7 @@ out:	return err;
-	rth = rt_dst_alloc(l3mdev_master_dev_rcu(dev) ? : net->loopback_dev,
+	rth = rt_dst_alloc(ip_rt_get_dev(net, res),
 			   flags | RTCF_LOCAL, res->type,
 			   IN_DEV_CONF_GET(in_dev, NOPOLICY), false, do_cache);
 	if (!rth)
diff --git a/tools/testing/selftests/net/fib_tests.sh b/tools/testing/selftests/net/fib_tests.sh
index a8cc3a3f8971..14ae81592bda 100755
--- a/tools/testing/selftests/net/fib_tests.sh
+++ b/tools/testing/selftests/net/fib_tests.sh
@@ -1385,12 +1385,37 @@ ipv4_rt_replace()
+# checks that cached input route on VRF port is deleted
+# when VRF is deleted
+	run_cmd "ip addr add dev lo"
+	run_cmd "ip netns add test-ns"
+	run_cmd "ip link add veth-outside type veth peer name veth-inside"
+	run_cmd "ip link add vrf-100 type vrf table 1100"
+	run_cmd "ip link set veth-outside master vrf-100"
+	run_cmd "ip link set veth-inside netns test-ns"
+	run_cmd "ip link set veth-outside up"
+	run_cmd "ip link set vrf-100 up"
+	run_cmd "ip route add dev veth-outside table 1100"
+	run_cmd "ip netns exec test-ns ip link set veth-inside up"
+	run_cmd "ip netns exec test-ns ip addr add dev veth-inside"
+	run_cmd "ip netns exec test-ns ip route add dev veth-inside"
+	run_cmd "ip netns exec test-ns ip route add default via"
+	run_cmd "ip netns exec test-ns ping -c 1 -i 1"
+	run_cmd "ip link delete vrf-100"
+	# if we do not hang test is a success
+	log_test $? 0 "Cached route removed from VRF port device"
+	ipv4_local_rt_cache

More information about the kernel-team mailing list