selftests: net: verify fq per-band packet limit

Commit 29f834aa32 ("net_sched: sch_fq: add 3 bands and WRR
scheduling") introduces multiple traffic bands, and per-band maximum
packet count.

Per-band limits ensures that packets in one class cannot fill the
entire qdisc and so cause DoS to the traffic in the other classes.

Verify this behavior:
  1. set the limit to 10 per band
  2. send 20 pkts on band A: verify that 10 are queued, 10 dropped
  3. send 20 pkts on band A: verify that  0 are queued, 20 dropped
  4. send 20 pkts on band B: verify that 10 are queued, 10 dropped

Packets must remain queued for a period to trigger this behavior.
Use SO_TXTIME to store packets for 100 msec.

The test reuses existing upstream test infra. The script is a fork of
cmsg_time.sh. The scripts call cmsg_sender.

The test extends cmsg_sender with two arguments:

* '-P' SO_PRIORITY
  There is a subtle difference between IPv4 and IPv6 stack behavior:
  PF_INET/IP_TOS        sets IP header bits and sk_priority
  PF_INET6/IPV6_TCLASS  sets IP header bits BUT NOT sk_priority

* '-n' num pkts
  Send multiple packets in quick succession.
  I first attempted a for loop in the script, but this is too slow in
  virtualized environments, causing flakiness as the 100ms timeout is
  reached and packets are dequeued.

Also do not wait for timestamps to be queued unless timestamps are
requested.

Signed-off-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20231116203449.2627525-1-willemdebruijn.kernel@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

tools/testing/selftests/net/Makefile

@@ -91,6 +91,7 @@ TEST_PROGS += test_bridge_neigh_suppress.sh
 TEST_PROGS += test_vxlan_nolocalbypass.sh
 TEST_PROGS += test_bridge_backup_port.sh
 TEST_PROGS += fdb_flush.sh
+TEST_PROGS += fq_band_pktlimit.sh
 
 TEST_FILES := settings
 

tools/testing/selftests/net/cmsg_sender.c

@@ -45,11 +45,13 @@ struct options {
 	const char *host;
 	const char *service;
 	unsigned int size;
+	unsigned int num_pkt;
 	struct {
 		unsigned int mark;
 		unsigned int dontfrag;
 		unsigned int tclass;
 		unsigned int hlimit;
+		unsigned int priority;
 	} sockopt;
 	struct {
 		unsigned int family;
@@ -72,6 +74,7 @@ struct options {
 	} v6;
 } opt = {
 	.size = 13,
+	.num_pkt = 1,
 	.sock = {
 		.family	= AF_UNSPEC,
 		.type	= SOCK_DGRAM,
@@ -112,7 +115,7 @@ static void cs_parse_args(int argc, char *argv[])
 {
 	int o;
 
-	while ((o = getopt(argc, argv, "46sS:p:m:M:d:tf:F:c:C:l:L:H:")) != -1) {
+	while ((o = getopt(argc, argv, "46sS:p:P:m:M:n:d:tf:F:c:C:l:L:H:")) != -1) {
 		switch (o) {
 		case 's':
 			opt.silent_send = true;
@@ -138,7 +141,9 @@ static void cs_parse_args(int argc, char *argv[])
 				cs_usage(argv[0]);
 			}
 			break;
-
+		case 'P':
+			opt.sockopt.priority = atoi(optarg);
+			break;
 		case 'm':
 			opt.mark.ena = true;
 			opt.mark.val = atoi(optarg);
@@ -146,6 +151,9 @@ static void cs_parse_args(int argc, char *argv[])
 		case 'M':
 			opt.sockopt.mark = atoi(optarg);
 			break;
+		case 'n':
+			opt.num_pkt = atoi(optarg);
+			break;
 		case 'd':
 			opt.txtime.ena = true;
 			opt.txtime.delay = atoi(optarg);
@@ -410,6 +418,10 @@ static void ca_set_sockopts(int fd)
 	    setsockopt(fd, SOL_IPV6, IPV6_UNICAST_HOPS,
 		       &opt.sockopt.hlimit, sizeof(opt.sockopt.hlimit)))
 		error(ERN_SOCKOPT, errno, "setsockopt IPV6_HOPLIMIT");
+	if (opt.sockopt.priority &&
+	    setsockopt(fd, SOL_SOCKET, SO_PRIORITY,
+		       &opt.sockopt.priority, sizeof(opt.sockopt.priority)))
+		error(ERN_SOCKOPT, errno, "setsockopt SO_PRIORITY");
 }
 
 int main(int argc, char *argv[])
@@ -421,6 +433,7 @@ int main(int argc, char *argv[])
 	char *buf;
 	int err;
 	int fd;
+	int i;
 
 	cs_parse_args(argc, argv);
 
@@ -480,24 +493,27 @@ int main(int argc, char *argv[])
 
 	cs_write_cmsg(fd, &msg, cbuf, sizeof(cbuf));
 
-	err = sendmsg(fd, &msg, 0);
-	if (err < 0) {
-		if (!opt.silent_send)
-			fprintf(stderr, "send failed: %s\n", strerror(errno));
-		err = ERN_SEND;
-		goto err_out;
-	} else if (err != (int)opt.size) {
-		fprintf(stderr, "short send\n");
-		err = ERN_SEND_SHORT;
-		goto err_out;
-	} else {
-		err = ERN_SUCCESS;
+	for (i = 0; i < opt.num_pkt; i++) {
+		err = sendmsg(fd, &msg, 0);
+		if (err < 0) {
+			if (!opt.silent_send)
+				fprintf(stderr, "send failed: %s\n", strerror(errno));
+			err = ERN_SEND;
+			goto err_out;
+		} else if (err != (int)opt.size) {
+			fprintf(stderr, "short send\n");
+			err = ERN_SEND_SHORT;
+			goto err_out;
+		}
 	}
+	err = ERN_SUCCESS;
 
-	/* Make sure all timestamps have time to loop back */
-	usleep(opt.txtime.delay);
+	if (opt.ts.ena) {
+		/* Make sure all timestamps have time to loop back */
+		usleep(opt.txtime.delay);
 
-	cs_read_cmsg(fd, &msg, cbuf, sizeof(cbuf));
+		cs_read_cmsg(fd, &msg, cbuf, sizeof(cbuf));
+	}
 
 err_out:
 	close(fd);

tools/testing/selftests/net/fq_band_pktlimit.sh

@@ -0,0 +1,57 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+#
+# Verify that FQ has a packet limit per band:
+#
+# 1. set the limit to 10 per band
+# 2. send 20 pkts on band A: verify that 10 are queued, 10 dropped
+# 3. send 20 pkts on band A: verify that  0 are queued, 20 dropped
+# 4. send 20 pkts on band B: verify that 10 are queued, 10 dropped
+#
+# Send packets with a 100ms delay to ensure that previously sent
+# packets are still queued when later ones are sent.
+# Use SO_TXTIME for this.
+
+die() {
+	echo "$1"
+	exit 1
+}
+
+# run inside private netns
+if [[ $# -eq 0 ]]; then
+	./in_netns.sh "$0" __subprocess
+	exit
+fi
+
+ip link add type dummy
+ip link set dev dummy0 up
+ip -6 addr add fdaa::1/128 dev dummy0
+ip -6 route add fdaa::/64 dev dummy0
+tc qdisc replace dev dummy0 root handle 1: fq quantum 1514 initial_quantum 1514 limit 10
+
+./cmsg_sender -6 -p u -d 100000 -n 20 fdaa::2 8000
+OUT1="$(tc -s qdisc show dev dummy0 | grep '^\ Sent')"
+
+./cmsg_sender -6 -p u -d 100000 -n 20 fdaa::2 8000
+OUT2="$(tc -s qdisc show dev dummy0 | grep '^\ Sent')"
+
+./cmsg_sender -6 -p u -d 100000 -n 20 -P 7 fdaa::2 8000
+OUT3="$(tc -s qdisc show dev dummy0 | grep '^\ Sent')"
+
+# Initial stats will report zero sent, as all packets are still
+# queued in FQ. Sleep for the delay period (100ms) and see that
+# twenty are now sent.
+sleep 0.1
+OUT4="$(tc -s qdisc show dev dummy0 | grep '^\ Sent')"
+
+# Log the output after the test
+echo "${OUT1}"
+echo "${OUT2}"
+echo "${OUT3}"
+echo "${OUT4}"
+
+# Test the output for expected values
+echo "${OUT1}" | grep -q '0\ pkt\ (dropped\ 10'  || die "unexpected drop count at 1"
+echo "${OUT2}" | grep -q '0\ pkt\ (dropped\ 30'  || die "unexpected drop count at 2"
+echo "${OUT3}" | grep -q '0\ pkt\ (dropped\ 40'  || die "unexpected drop count at 3"
+echo "${OUT4}" | grep -q '20\ pkt\ (dropped\ 40' || die "unexpected accept count at 4"