The commit
4656d72c1efa ("selftests: mptcp: userspace: validate v4-v6 subflows mix")
has been backported to v6.1.8 without any conflicts. But it looks like
it was depending on a previous one:
commit
1cc94ac1af4b ("selftests: mptcp: make evts global in userspace_pm")
Without it, the test fails with:
./userspace_pm.sh: line 788: : No such file or directory
# ADD_ADDR4 id:14 10.0.2.1 (ns1) => ns2, reuse port [FAIL]
sed: can't read : No such file or directory
This dependence refactors the way the monitoring files are being
created: only once for all the different sub-tests instead of per
sub-test.
It is probably better to avoid backporting the refactoring. That is why
the new sub-test has been adapted to work using the previous way that is
still in place here in v6.1: the monitoring is started at the beginning
of each sub-test and the created file is removed at the end.
Fixes: f59549814a64 ("selftests: mptcp: userspace: validate v4-v6 subflows mix")
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>
test_subflows_v4_v6_mix()
{
+ local client_evts
+ client_evts=$(mktemp)
+ # Capture events on the network namespace running the client
+ :>"$client_evts"
+ ip netns exec "$ns2" ./pm_nl_ctl events >> "$client_evts" 2>&1 &
+ evts_pid=$!
+ sleep 0.5
+
# Attempt to add a listener at 10.0.2.1:<subflow-port>
ip netns exec "$ns1" ./pm_nl_ctl listen 10.0.2.1\
$app6_port > /dev/null 2>&1 &
ip netns exec "$ns1" ./pm_nl_ctl rem id $server_addr_id token\
"$server6_token" > /dev/null 2>&1
sleep 0.5
+
+ kill_wait $evts_pid
+ rm -f "$client_evts"
}
test_prio()