1#!/bin/bash 2# SPDX-License-Identifier: GPL-2.0 3 4. "$(dirname "${0}")/mptcp_lib.sh" 5 6rndh=$(printf %x $sec)-$(mktemp -u XXXXXX) 7ns1="ns1-$rndh" 8ns2="ns2-$rndh" 9ns3="ns3-$rndh" 10capture=false 11ksft_skip=4 12timeout_poll=30 13timeout_test=$((timeout_poll * 2 + 1)) 14test_cnt=1 15ret=0 16bail=0 17slack=50 18 19usage() { 20 echo "Usage: $0 [ -b ] [ -c ] [ -d ]" 21 echo -e "\t-b: bail out after first error, otherwise runs al testcases" 22 echo -e "\t-c: capture packets for each test using tcpdump (default: no capture)" 23 echo -e "\t-d: debug this script" 24} 25 26cleanup() 27{ 28 rm -f "$cout" "$sout" 29 rm -f "$large" "$small" 30 rm -f "$capout" 31 32 local netns 33 for netns in "$ns1" "$ns2" "$ns3";do 34 ip netns del $netns 35 done 36} 37 38mptcp_lib_check_mptcp 39 40ip -Version > /dev/null 2>&1 41if [ $? -ne 0 ];then 42 echo "SKIP: Could not run test without ip tool" 43 exit $ksft_skip 44fi 45 46# "$ns1" ns2 ns3 47# ns1eth1 ns2eth1 ns2eth3 ns3eth1 48# netem 49# ns1eth2 ns2eth2 50# netem 51 52setup() 53{ 54 large=$(mktemp) 55 small=$(mktemp) 56 sout=$(mktemp) 57 cout=$(mktemp) 58 capout=$(mktemp) 59 size=$((2 * 2048 * 4096)) 60 61 dd if=/dev/zero of=$small bs=4096 count=20 >/dev/null 2>&1 62 dd if=/dev/zero of=$large bs=4096 count=$((size / 4096)) >/dev/null 2>&1 63 64 trap cleanup EXIT 65 66 for i in "$ns1" "$ns2" "$ns3";do 67 ip netns add $i || exit $ksft_skip 68 ip -net $i link set lo up 69 ip netns exec $i sysctl -q net.ipv4.conf.all.rp_filter=0 70 ip netns exec $i sysctl -q net.ipv4.conf.default.rp_filter=0 71 done 72 73 ip link add ns1eth1 netns "$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 "$ns1" route add default via dead:beef:2::2 metric 101 88 89 ip netns exec "$ns1" ./pm_nl_ctl limits 1 1 90 ip netns exec "$ns1" ./pm_nl_ctl add 10.0.2.1 dev ns1eth2 flags subflow 91 92 ip -net "$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 "$ns3" route add default via dead:beef:3::2 111 112 ip netns exec "$ns3" ./pm_nl_ctl limits 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# $1: ns, $2: port 126wait_local_port_listen() 127{ 128 local listener_ns="${1}" 129 local port="${2}" 130 131 local port_hex i 132 133 port_hex="$(printf "%04X" "${port}")" 134 for i in $(seq 10); do 135 ip netns exec "${listener_ns}" cat /proc/net/tcp* | \ 136 awk "BEGIN {rc=1} {if (\$2 ~ /:${port_hex}\$/ && \$4 ~ /0A/) {rc=0; exit}} END {exit rc}" && 137 break 138 sleep 0.1 139 done 140} 141 142do_transfer() 143{ 144 local cin=$1 145 local sin=$2 146 local max_time=$3 147 local port 148 port=$((10000+$test_cnt)) 149 test_cnt=$((test_cnt+1)) 150 151 :> "$cout" 152 :> "$sout" 153 :> "$capout" 154 155 local addr_port 156 addr_port=$(printf "%s:%d" ${connect_addr} ${port}) 157 158 if $capture; then 159 local capuser 160 if [ -z $SUDO_USER ] ; then 161 capuser="" 162 else 163 capuser="-Z $SUDO_USER" 164 fi 165 166 local capfile="${rndh}-${port}" 167 local capopt="-i any -s 65535 -B 32768 ${capuser}" 168 169 ip netns exec ${ns3} tcpdump ${capopt} -w "${capfile}-listener.pcap" >> "${capout}" 2>&1 & 170 local cappid_listener=$! 171 172 ip netns exec ${ns1} tcpdump ${capopt} -w "${capfile}-connector.pcap" >> "${capout}" 2>&1 & 173 local cappid_connector=$! 174 175 sleep 1 176 fi 177 178 timeout ${timeout_test} \ 179 ip netns exec ${ns3} \ 180 ./mptcp_connect -jt ${timeout_poll} -l -p $port -T $time \ 181 0.0.0.0 < "$sin" > "$sout" & 182 local spid=$! 183 184 wait_local_port_listen "${ns3}" "${port}" 185 186 timeout ${timeout_test} \ 187 ip netns exec ${ns1} \ 188 ./mptcp_connect -jt ${timeout_poll} -p $port -T $time \ 189 10.0.3.3 < "$cin" > "$cout" & 190 local cpid=$! 191 192 wait $cpid 193 local retc=$? 194 wait $spid 195 local rets=$? 196 197 if $capture; then 198 sleep 1 199 kill ${cappid_listener} 200 kill ${cappid_connector} 201 fi 202 203 cmp $sin $cout > /dev/null 2>&1 204 local cmps=$? 205 cmp $cin $sout > /dev/null 2>&1 206 local cmpc=$? 207 208 printf "%-16s" " max $max_time " 209 if [ $retc -eq 0 ] && [ $rets -eq 0 ] && \ 210 [ $cmpc -eq 0 ] && [ $cmps -eq 0 ]; then 211 echo "[ OK ]" 212 cat "$capout" 213 return 0 214 fi 215 216 echo " [ fail ]" 217 echo "client exit code $retc, server $rets" 1>&2 218 echo -e "\nnetns ${ns3} socket stat for $port:" 1>&2 219 ip netns exec ${ns3} ss -nita 1>&2 -o "sport = :$port" 220 echo -e "\nnetns ${ns1} socket stat for $port:" 1>&2 221 ip netns exec ${ns1} ss -nita 1>&2 -o "dport = :$port" 222 ls -l $sin $cout 223 ls -l $cin $sout 224 225 cat "$capout" 226 return 1 227} 228 229run_test() 230{ 231 local rate1=$1 232 local rate2=$2 233 local delay1=$3 234 local delay2=$4 235 local lret 236 local dev 237 shift 4 238 local msg=$* 239 240 [ $delay1 -gt 0 ] && delay1="delay $delay1" || delay1="" 241 [ $delay2 -gt 0 ] && delay2="delay $delay2" || delay2="" 242 243 for dev in ns1eth1 ns1eth2; do 244 tc -n $ns1 qdisc del dev $dev root >/dev/null 2>&1 245 done 246 for dev in ns2eth1 ns2eth2; do 247 tc -n $ns2 qdisc del dev $dev root >/dev/null 2>&1 248 done 249 tc -n $ns1 qdisc add dev ns1eth1 root netem rate ${rate1}mbit $delay1 250 tc -n $ns1 qdisc add dev ns1eth2 root netem rate ${rate2}mbit $delay2 251 tc -n $ns2 qdisc add dev ns2eth1 root netem rate ${rate1}mbit $delay1 252 tc -n $ns2 qdisc add dev ns2eth2 root netem rate ${rate2}mbit $delay2 253 254 # time is measured in ms, account for transfer size, aggregated link speed 255 # and header overhead (10%) 256 # ms byte -> bit 10% mbit -> kbit -> bit 10% 257 local time=$((1000 * size * 8 * 10 / ((rate1 + rate2) * 1000 * 1000 * 9) )) 258 259 # mptcp_connect will do some sleeps to allow the mp_join handshake 260 # completion (see mptcp_connect): 200ms on each side, add some slack 261 time=$((time + 400 + slack)) 262 263 printf "%-60s" "$msg" 264 do_transfer $small $large $time 265 lret=$? 266 if [ $lret -ne 0 ]; then 267 ret=$lret 268 [ $bail -eq 0 ] || exit $ret 269 fi 270 271 printf "%-60s" "$msg - reverse direction" 272 do_transfer $large $small $time 273 lret=$? 274 if [ $lret -ne 0 ]; then 275 ret=$lret 276 [ $bail -eq 0 ] || exit $ret 277 fi 278} 279 280while getopts "bcdh" option;do 281 case "$option" in 282 "h") 283 usage $0 284 exit 0 285 ;; 286 "b") 287 bail=1 288 ;; 289 "c") 290 capture=true 291 ;; 292 "d") 293 set -x 294 ;; 295 "?") 296 usage $0 297 exit 1 298 ;; 299 esac 300done 301 302setup 303run_test 10 10 0 0 "balanced bwidth" 304run_test 10 10 1 50 "balanced bwidth with unbalanced delay" 305 306# we still need some additional infrastructure to pass the following test-cases 307run_test 30 10 0 0 "unbalanced bwidth" 308run_test 30 10 1 50 "unbalanced bwidth with unbalanced delay" 309run_test 30 10 50 1 "unbalanced bwidth with opposed, unbalanced delay" 310exit $ret 311