1#!/bin/bash
2# SPDX-License-Identifier: GPL-2.0
3
4# Double quotes to prevent globbing and word splitting is recommended in new
5# code but we accept it, especially because there were too many before having
6# address all other issues detected by shellcheck.
7#shellcheck disable=SC2086
8
9. "$(dirname "${0}")/mptcp_lib.sh"
10
11ns1=""
12ns2=""
13ns3=""
14capture=false
15timeout_poll=30
16timeout_test=$((timeout_poll * 2 + 1))
17# a bit more space: because we have more to display
18MPTCP_LIB_TEST_FORMAT="%02u %-60s"
19ret=0
20bail=0
21slack=50
22large=""
23small=""
24sout=""
25cout=""
26capout=""
27size=0
28
29usage() {
30	echo "Usage: $0 [ -b ] [ -c ] [ -d ] [ -i]"
31	echo -e "\t-b: bail out after first error, otherwise runs al testcases"
32	echo -e "\t-c: capture packets for each test using tcpdump (default: no capture)"
33	echo -e "\t-d: debug this script"
34	echo -e "\t-i: use 'ip mptcp' instead of 'pm_nl_ctl'"
35}
36
37# This function is used in the cleanup trap
38#shellcheck disable=SC2317
39cleanup()
40{
41	rm -f "$cout" "$sout"
42	rm -f "$large" "$small"
43	rm -f "$capout"
44
45	mptcp_lib_ns_exit "${ns1}" "${ns2}" "${ns3}"
46}
47
48mptcp_lib_check_mptcp
49mptcp_lib_check_tools ip tc
50
51#  "$ns1"              ns2                    ns3
52#     ns1eth1    ns2eth1   ns2eth3      ns3eth1
53#            netem
54#     ns1eth2    ns2eth2
55#            netem
56
57setup()
58{
59	large=$(mktemp)
60	small=$(mktemp)
61	sout=$(mktemp)
62	cout=$(mktemp)
63	capout=$(mktemp)
64	size=$((2 * 2048 * 4096))
65
66	dd if=/dev/zero of=$small bs=4096 count=20 >/dev/null 2>&1
67	dd if=/dev/zero of=$large bs=4096 count=$((size / 4096)) >/dev/null 2>&1
68
69	trap cleanup EXIT
70
71	mptcp_lib_ns_init ns1 ns2 ns3
72
73	ip link add ns1eth1 netns "$ns1" type veth peer name ns2eth1 netns "$ns2"
74	ip link add ns1eth2 netns "$ns1" type veth peer name ns2eth2 netns "$ns2"
75	ip link add ns2eth3 netns "$ns2" type veth peer name ns3eth1 netns "$ns3"
76
77	ip -net "$ns1" addr add 10.0.1.1/24 dev ns1eth1
78	ip -net "$ns1" addr add dead:beef:1::1/64 dev ns1eth1 nodad
79	ip -net "$ns1" link set ns1eth1 up mtu 1500
80	ip -net "$ns1" route add default via 10.0.1.2
81	ip -net "$ns1" route add default via dead:beef:1::2
82
83	ip -net "$ns1" addr add 10.0.2.1/24 dev ns1eth2
84	ip -net "$ns1" addr add dead:beef:2::1/64 dev ns1eth2 nodad
85	ip -net "$ns1" link set ns1eth2 up mtu 1500
86	ip -net "$ns1" route add default via 10.0.2.2 metric 101
87	ip -net "$ns1" route add default via dead:beef:2::2 metric 101
88
89	mptcp_lib_pm_nl_set_limits "${ns1}" 1 1
90	mptcp_lib_pm_nl_add_endpoint "${ns1}" 10.0.2.1 dev ns1eth2 flags subflow
91
92	ip -net "$ns2" addr add 10.0.1.2/24 dev ns2eth1
93	ip -net "$ns2" addr add dead:beef:1::2/64 dev ns2eth1 nodad
94	ip -net "$ns2" link set ns2eth1 up mtu 1500
95
96	ip -net "$ns2" addr add 10.0.2.2/24 dev ns2eth2
97	ip -net "$ns2" addr add dead:beef:2::2/64 dev ns2eth2 nodad
98	ip -net "$ns2" link set ns2eth2 up mtu 1500
99
100	ip -net "$ns2" addr add 10.0.3.2/24 dev ns2eth3
101	ip -net "$ns2" addr add dead:beef:3::2/64 dev ns2eth3 nodad
102	ip -net "$ns2" link set ns2eth3 up mtu 1500
103	ip netns exec "$ns2" sysctl -q net.ipv4.ip_forward=1
104	ip netns exec "$ns2" sysctl -q net.ipv6.conf.all.forwarding=1
105
106	ip -net "$ns3" addr add 10.0.3.3/24 dev ns3eth1
107	ip -net "$ns3" addr add dead:beef:3::3/64 dev ns3eth1 nodad
108	ip -net "$ns3" link set ns3eth1 up mtu 1500
109	ip -net "$ns3" route add default via 10.0.3.2
110	ip -net "$ns3" route add default via dead:beef:3::2
111
112	mptcp_lib_pm_nl_set_limits "${ns3}" 1 1
113
114	# debug build can slow down measurably the test program
115	# we use quite tight time limit on the run-time, to ensure
116	# maximum B/W usage.
117	# Use kmemleak/lockdep/kasan/prove_locking presence as a rough
118	# estimate for this being a debug kernel and increase the
119	# maximum run-time accordingly. Observed run times for CI builds
120	# running selftests, including kbuild, were used to determine the
121	# amount of time to add.
122	grep -q ' kmemleak_init$\| lockdep_init$\| kasan_init$\| prove_locking$' /proc/kallsyms && slack=$((slack+550))
123}
124
125do_transfer()
126{
127	local cin=$1
128	local sin=$2
129	local max_time=$3
130	local port
131	port=$((10000+MPTCP_LIB_TEST_COUNTER))
132
133	:> "$cout"
134	:> "$sout"
135	:> "$capout"
136
137	if $capture; then
138		local capuser
139		local rndh="${ns1:4}"
140		if [ -z $SUDO_USER ] ; then
141			capuser=""
142		else
143			capuser="-Z $SUDO_USER"
144		fi
145
146		local capfile="${rndh}-${port}"
147		local capopt="-i any -s 65535 -B 32768 ${capuser}"
148
149		ip netns exec ${ns3}  tcpdump ${capopt} -w "${capfile}-listener.pcap"  >> "${capout}" 2>&1 &
150		local cappid_listener=$!
151
152		ip netns exec ${ns1} tcpdump ${capopt} -w "${capfile}-connector.pcap" >> "${capout}" 2>&1 &
153		local cappid_connector=$!
154
155		sleep 1
156	fi
157
158	NSTAT_HISTORY=/tmp/${ns3}.nstat ip netns exec ${ns3} \
159		nstat -n
160	NSTAT_HISTORY=/tmp/${ns1}.nstat ip netns exec ${ns1} \
161		nstat -n
162
163	timeout ${timeout_test} \
164		ip netns exec ${ns3} \
165			./mptcp_connect -jt ${timeout_poll} -l -p $port -T $max_time \
166				0.0.0.0 < "$sin" > "$sout" &
167	local spid=$!
168
169	mptcp_lib_wait_local_port_listen "${ns3}" "${port}"
170
171	timeout ${timeout_test} \
172		ip netns exec ${ns1} \
173			./mptcp_connect -jt ${timeout_poll} -p $port -T $max_time \
174				10.0.3.3 < "$cin" > "$cout" &
175	local cpid=$!
176
177	wait $cpid
178	local retc=$?
179	wait $spid
180	local rets=$?
181
182	if $capture; then
183		sleep 1
184		kill ${cappid_listener}
185		kill ${cappid_connector}
186	fi
187
188	NSTAT_HISTORY=/tmp/${ns3}.nstat ip netns exec ${ns3} \
189		nstat | grep Tcp > /tmp/${ns3}.out
190	NSTAT_HISTORY=/tmp/${ns1}.nstat ip netns exec ${ns1} \
191		nstat | grep Tcp > /tmp/${ns1}.out
192
193	cmp $sin $cout > /dev/null 2>&1
194	local cmps=$?
195	cmp $cin $sout > /dev/null 2>&1
196	local cmpc=$?
197
198	if [ $retc -eq 0 ] && [ $rets -eq 0 ] && \
199	   [ $cmpc -eq 0 ] && [ $cmps -eq 0 ]; then
200		printf "%-16s" " max $max_time "
201		mptcp_lib_pr_ok
202		cat "$capout"
203		return 0
204	fi
205
206	mptcp_lib_pr_fail "client exit code $retc, server $rets"
207	mptcp_lib_pr_err_stats "${ns3}" "${ns1}" "${port}" \
208		"/tmp/${ns3}.out" "/tmp/${ns1}.out"
209	ls -l $sin $cout
210	ls -l $cin $sout
211
212	cat "$capout"
213	return 1
214}
215
216run_test()
217{
218	local rate1=$1
219	local rate2=$2
220	local delay1=$3
221	local delay2=$4
222	local lret
223	local dev
224	shift 4
225	local msg=$*
226
227	[ $delay1 -gt 0 ] && delay1="delay ${delay1}ms" || delay1=""
228	[ $delay2 -gt 0 ] && delay2="delay ${delay2}ms" || delay2=""
229
230	for dev in ns1eth1 ns1eth2; do
231		tc -n $ns1 qdisc del dev $dev root >/dev/null 2>&1
232	done
233	for dev in ns2eth1 ns2eth2; do
234		tc -n $ns2 qdisc del dev $dev root >/dev/null 2>&1
235	done
236	tc -n $ns1 qdisc add dev ns1eth1 root netem rate ${rate1}mbit $delay1
237	tc -n $ns1 qdisc add dev ns1eth2 root netem rate ${rate2}mbit $delay2
238	tc -n $ns2 qdisc add dev ns2eth1 root netem rate ${rate1}mbit $delay1
239	tc -n $ns2 qdisc add dev ns2eth2 root netem rate ${rate2}mbit $delay2
240
241	# time is measured in ms, account for transfer size, aggregated link speed
242	# and header overhead (10%)
243	#              ms    byte -> bit   10%        mbit      -> kbit -> bit  10%
244	local time=$((1000 * size  *  8  * 10 / ((rate1 + rate2) * 1000 * 1000 * 9) ))
245
246	# mptcp_connect will do some sleeps to allow the mp_join handshake
247	# completion (see mptcp_connect): 200ms on each side, add some slack
248	time=$((time + 400 + slack))
249
250	mptcp_lib_print_title "$msg"
251	do_transfer $small $large $time
252	lret=$?
253	mptcp_lib_result_code "${lret}" "${msg}"
254	if [ $lret -ne 0 ] && ! mptcp_lib_subtest_is_flaky; then
255		ret=$lret
256		[ $bail -eq 0 ] || exit $ret
257	fi
258
259	msg+=" - reverse direction"
260	mptcp_lib_print_title "${msg}"
261	do_transfer $large $small $time
262	lret=$?
263	mptcp_lib_result_code "${lret}" "${msg}"
264	if [ $lret -ne 0 ] && ! mptcp_lib_subtest_is_flaky; then
265		ret=$lret
266		[ $bail -eq 0 ] || exit $ret
267	fi
268}
269
270while getopts "bcdhi" option;do
271	case "$option" in
272	"h")
273		usage $0
274		exit ${KSFT_PASS}
275		;;
276	"b")
277		bail=1
278		;;
279	"c")
280		capture=true
281		;;
282	"d")
283		set -x
284		;;
285	"i")
286		mptcp_lib_set_ip_mptcp
287		;;
288	"?")
289		usage $0
290		exit ${KSFT_FAIL}
291		;;
292	esac
293done
294
295setup
296mptcp_lib_subtests_last_ts_reset
297run_test 10 10 0 0 "balanced bwidth"
298run_test 10 10 1 25 "balanced bwidth with unbalanced delay"
299
300# we still need some additional infrastructure to pass the following test-cases
301MPTCP_LIB_SUBTEST_FLAKY=1 run_test 10 3 0 0 "unbalanced bwidth"
302run_test 10 3 1 25 "unbalanced bwidth with unbalanced delay"
303run_test 10 3 25 1 "unbalanced bwidth with opposed, unbalanced delay"
304
305mptcp_lib_result_print_all_tap
306exit $ret
307