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 
11 ns1=""
12 ns2=""
13 ns3=""
14 capture=false
15 timeout_poll=30
16 timeout_test=$((timeout_poll * 2 + 1))
17 # a bit more space: because we have more to display
18 MPTCP_LIB_TEST_FORMAT="%02u %-60s"
19 ret=0
20 bail=0
21 slack=50
22 large=""
23 small=""
24 sout=""
25 cout=""
26 capout=""
27 size=0
28 
29 usage() {
30 	echo "Usage: $0 [ -b ] [ -c ] [ -d ] [ -i]"
31 	echo -e "\t-b: bail out after first error, otherwise runs al testcases"
32 	echo -e "\t-c: capture packets for each test using tcpdump (default: no capture)"
33 	echo -e "\t-d: debug this script"
34 	echo -e "\t-i: use 'ip mptcp' instead of 'pm_nl_ctl'"
35 }
36 
37 # This function is used in the cleanup trap
38 #shellcheck disable=SC2317
39 cleanup()
40 {
41 	rm -f "$cout" "$sout"
42 	rm -f "$large" "$small"
43 	rm -f "$capout"
44 
45 	mptcp_lib_ns_exit "${ns1}" "${ns2}" "${ns3}"
46 }
47 
48 mptcp_lib_check_mptcp
49 mptcp_lib_check_tools ip tc
50 
51 #  "$ns1"              ns2                    ns3
52 #     ns1eth1    ns2eth1   ns2eth3      ns3eth1
53 #            netem
54 #     ns1eth2    ns2eth2
55 #            netem
56 
57 setup()
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 
125 do_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 	timeout ${timeout_test} \
159 		ip netns exec ${ns3} \
160 			./mptcp_connect -jt ${timeout_poll} -l -p $port -T $max_time \
161 				0.0.0.0 < "$sin" > "$sout" &
162 	local spid=$!
163 
164 	mptcp_lib_wait_local_port_listen "${ns3}" "${port}"
165 
166 	timeout ${timeout_test} \
167 		ip netns exec ${ns1} \
168 			./mptcp_connect -jt ${timeout_poll} -p $port -T $max_time \
169 				10.0.3.3 < "$cin" > "$cout" &
170 	local cpid=$!
171 
172 	wait $cpid
173 	local retc=$?
174 	wait $spid
175 	local rets=$?
176 
177 	if $capture; then
178 		sleep 1
179 		kill ${cappid_listener}
180 		kill ${cappid_connector}
181 	fi
182 
183 	cmp $sin $cout > /dev/null 2>&1
184 	local cmps=$?
185 	cmp $cin $sout > /dev/null 2>&1
186 	local cmpc=$?
187 
188 	printf "%-16s" " max $max_time "
189 	if [ $retc -eq 0 ] && [ $rets -eq 0 ] && \
190 	   [ $cmpc -eq 0 ] && [ $cmps -eq 0 ]; then
191 		mptcp_lib_pr_ok
192 		cat "$capout"
193 		return 0
194 	fi
195 
196 	mptcp_lib_pr_fail
197 	echo "client exit code $retc, server $rets" 1>&2
198 	echo -e "\nnetns ${ns3} socket stat for $port:" 1>&2
199 	ip netns exec ${ns3} ss -nita 1>&2 -o "sport = :$port"
200 	echo -e "\nnetns ${ns1} socket stat for $port:" 1>&2
201 	ip netns exec ${ns1} ss -nita 1>&2 -o "dport = :$port"
202 	ls -l $sin $cout
203 	ls -l $cin $sout
204 
205 	cat "$capout"
206 	return 1
207 }
208 
209 run_test()
210 {
211 	local rate1=$1
212 	local rate2=$2
213 	local delay1=$3
214 	local delay2=$4
215 	local lret
216 	local dev
217 	shift 4
218 	local msg=$*
219 
220 	[ $delay1 -gt 0 ] && delay1="delay ${delay1}ms" || delay1=""
221 	[ $delay2 -gt 0 ] && delay2="delay ${delay2}ms" || delay2=""
222 
223 	for dev in ns1eth1 ns1eth2; do
224 		tc -n $ns1 qdisc del dev $dev root >/dev/null 2>&1
225 	done
226 	for dev in ns2eth1 ns2eth2; do
227 		tc -n $ns2 qdisc del dev $dev root >/dev/null 2>&1
228 	done
229 	tc -n $ns1 qdisc add dev ns1eth1 root netem rate ${rate1}mbit $delay1
230 	tc -n $ns1 qdisc add dev ns1eth2 root netem rate ${rate2}mbit $delay2
231 	tc -n $ns2 qdisc add dev ns2eth1 root netem rate ${rate1}mbit $delay1
232 	tc -n $ns2 qdisc add dev ns2eth2 root netem rate ${rate2}mbit $delay2
233 
234 	# time is measured in ms, account for transfer size, aggregated link speed
235 	# and header overhead (10%)
236 	#              ms    byte -> bit   10%        mbit      -> kbit -> bit  10%
237 	local time=$((1000 * size  *  8  * 10 / ((rate1 + rate2) * 1000 * 1000 * 9) ))
238 
239 	# mptcp_connect will do some sleeps to allow the mp_join handshake
240 	# completion (see mptcp_connect): 200ms on each side, add some slack
241 	time=$((time + 400 + slack))
242 
243 	mptcp_lib_print_title "$msg"
244 	do_transfer $small $large $time
245 	lret=$?
246 	mptcp_lib_result_code "${lret}" "${msg}"
247 	if [ $lret -ne 0 ] && ! mptcp_lib_subtest_is_flaky; then
248 		ret=$lret
249 		[ $bail -eq 0 ] || exit $ret
250 	fi
251 
252 	msg+=" - reverse direction"
253 	mptcp_lib_print_title "${msg}"
254 	do_transfer $large $small $time
255 	lret=$?
256 	mptcp_lib_result_code "${lret}" "${msg}"
257 	if [ $lret -ne 0 ] && ! mptcp_lib_subtest_is_flaky; then
258 		ret=$lret
259 		[ $bail -eq 0 ] || exit $ret
260 	fi
261 }
262 
263 while getopts "bcdhi" option;do
264 	case "$option" in
265 	"h")
266 		usage $0
267 		exit ${KSFT_PASS}
268 		;;
269 	"b")
270 		bail=1
271 		;;
272 	"c")
273 		capture=true
274 		;;
275 	"d")
276 		set -x
277 		;;
278 	"i")
279 		mptcp_lib_set_ip_mptcp
280 		;;
281 	"?")
282 		usage $0
283 		exit ${KSFT_FAIL}
284 		;;
285 	esac
286 done
287 
288 setup
289 mptcp_lib_subtests_last_ts_reset
290 run_test 10 10 0 0 "balanced bwidth"
291 run_test 10 10 1 25 "balanced bwidth with unbalanced delay"
292 
293 # we still need some additional infrastructure to pass the following test-cases
294 MPTCP_LIB_SUBTEST_FLAKY=1 run_test 10 3 0 0 "unbalanced bwidth"
295 run_test 10 3 1 25 "unbalanced bwidth with unbalanced delay"
296 run_test 10 3 25 1 "unbalanced bwidth with opposed, unbalanced delay"
297 
298 mptcp_lib_result_print_all_tap
299 exit $ret
300