net/sched: act_police: more accurate MTU policing

in current Linux, MTU policing does not take into account that packets at
the TC ingress have the L2 header pulled. Thus, the same TC police action
(with the same value of tcfp_mtu) behaves differently for ingress/egress.
In addition, the full GSO size is compared to tcfp_mtu: as a consequence,
the policer drops GSO packets even when individual segments have the L2 +
L3 + L4 + payload length below the configured valued of tcfp_mtu.

Improve the accuracy of MTU policing as follows:
- account for mac_len for non-GSO packets at TC ingress.
- compare MTU threshold with the segmented size for GSO packets.
Also, add a kselftest that verifies the correct behavior.

Signed-off-by: Davide Caratti <[e-mail]>
Reviewed-by: Marcelo Ricardo Leitner <[e-mail]>
Signed-off-by: David S. Miller <[e-mail]>

net/sched/act_police.c

···
 	return err;
 }
 
+static bool tcf_police_mtu_check(struct sk_buff *skb, u32 limit)
+{
+	u32 len;
+
+	if (skb_is_gso(skb))
+		return skb_gso_validate_mac_len(skb, limit);
+
+	len = qdisc_pkt_len(skb);
+	if (skb_at_tc_ingress(skb))
+		len += skb->mac_len;
+
+	return len <= limit;
+}
+
 static int tcf_police_act(struct sk_buff *skb, const struct tc_action *a,
 			  struct tcf_result *res)
 {
···
 			goto inc_overlimits;
 	}
 
-	if (qdisc_pkt_len(skb) <= p->tcfp_mtu) {
+	if (tcf_police_mtu_check(skb, p->tcfp_mtu)) {
 		if (!p->rate_present && !p->pps_present) {
 			ret = p->tcfp_result;
 			goto end;

tools/testing/selftests/net/forwarding/tc_police.sh

···
 	police_tx_mirror_test
 	police_pps_rx_test
 	police_pps_tx_test
+	police_mtu_rx_test
+	police_mtu_tx_test
 "
 NUM_NETIFS=6
 source tc_common.sh
···
 	police_pps_common_test "police pps on tx"
 
 	tc filter del dev $rp2 egress protocol ip pref 1 handle 101 flower
+}
+
+police_mtu_common_test() {
+	RET=0
+
+	local test_name=$1; shift
+	local dev=$1; shift
+	local direction=$1; shift
+
+	tc filter add dev $dev $direction protocol ip pref 1 handle 101 flower \
+		dst_ip 198.51.100.1 ip_proto udp dst_port 54321 \
+		action police mtu 1042 conform-exceed drop/ok
+
+	# to count "conform" packets
+	tc filter add dev $h2 ingress protocol ip pref 1 handle 101 flower \
+		dst_ip 198.51.100.1 ip_proto udp dst_port 54321 \
+		action drop
+
+	mausezahn $h1 -a own -b $(mac_get $rp1) -A 192.0.2.1 -B 198.51.100.1 \
+		-t udp sp=12345,dp=54321 -p 1001 -c 10 -q
+
+	mausezahn $h1 -a own -b $(mac_get $rp1) -A 192.0.2.1 -B 198.51.100.1 \
+		-t udp sp=12345,dp=54321 -p 1000 -c 3 -q
+
+	tc_check_packets "dev $dev $direction" 101 13
+	check_err $? "wrong packet counter"
+
+	# "exceed" packets
+	local overlimits_t0=$(tc_rule_stats_get ${dev} 1 ${direction} .overlimits)
+	test ${overlimits_t0} = 10
+	check_err $? "wrong overlimits, expected 10 got ${overlimits_t0}"
+
+	# "conform" packets
+	tc_check_packets "dev $h2 ingress" 101 3
+	check_err $? "forwarding error"
+
+	tc filter del dev $h2 ingress protocol ip pref 1 handle 101 flower
+	tc filter del dev $dev $direction protocol ip pref 1 handle 101 flower
+
+	log_test "$test_name"
+}
+
+police_mtu_rx_test()
+{
+	police_mtu_common_test "police mtu (rx)" $rp1 ingress
+}
+
+police_mtu_tx_test()
+{
+	police_mtu_common_test "police mtu (tx)" $rp2 egress
 }
 
 setup_prepare()