From: Dmitry Safonov <0x7f454c46@xxxxxxxxx> On tests that are expecting failure the timeout value is TEST_RETRANSMIT_SEC == 1 second. Which is big enough for most of devices under tests. But on a particularly slow machine/VM, 1 second might be not enough for another thread to be scheduled and attempt to connect(). It is not a problem for tests that expect connect() to succeed as the timeout value for them (TEST_TIMEOUT_SEC) is intentionally bigger. One obvious way to solve this would be to increase TEST_RETRANSMIT_SEC. But as all tests would increase the timeouts, that's going to sum up. But here is less obvious way that keeps timeouts for expected connect() failures low: just synchronize the two threads, which will assure that before counter checks the other thread got a chance to run and timeout on connect(). The expected increase of the related counter for listen() socket will yet test the expected failure. Never happens on my machine, but I suppose the majority of netdev's connect-deny-* flakes [1] are caused by this. Fixes: > # selftests: net/tcp_ao: connect-deny_ipv6 > # 1..21 > # # 462[lib/setup.c:243] rand seed 1720905426 > # TAP version 13 > # ok 1 Non-AO server + AO client > # not ok 2 Non-AO server + AO client: TCPAOKeyNotFound counter did not increase: 0 <= 0 > # ok 3 AO server + Non-AO client > # ok 4 AO server + Non-AO client: counter TCPAORequired increased 0 => 1 ... [1]: https://netdev-3.bots.linux.dev/vmksft-tcp-ao/results/681741/6-connect-deny-ipv6/stdout Signed-off-by: Dmitry Safonov <0x7f454c46@xxxxxxxxx> --- tools/testing/selftests/net/tcp_ao/connect-deny.c | 3 +++ tools/testing/selftests/net/tcp_ao/restore.c | 6 ++++-- tools/testing/selftests/net/tcp_ao/seq-ext.c | 6 ++++-- tools/testing/selftests/net/tcp_ao/unsigned-md5.c | 3 +++ 4 files changed, 14 insertions(+), 4 deletions(-) diff --git a/tools/testing/selftests/net/tcp_ao/connect-deny.c b/tools/testing/selftests/net/tcp_ao/connect-deny.c index 5691f3d00603..166ad4549ef2 100644 --- a/tools/testing/selftests/net/tcp_ao/connect-deny.c +++ b/tools/testing/selftests/net/tcp_ao/connect-deny.c @@ -71,10 +71,12 @@ static void try_accept(const char *tst_name, unsigned int port, const char *pwd, } } + synchronize_threads(); /* before counter checks */ if (pwd && test_get_tcp_ao_counters(lsk, &ao_cnt2)) test_error("test_get_tcp_ao_counters()"); close(lsk); + if (pwd) test_tcp_ao_counters_cmp(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected); @@ -180,6 +182,7 @@ static void try_connect(const char *tst_name, unsigned int port, timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC; ret = _test_connect_socket(sk, this_ip_dest, port, timeout); + synchronize_threads(); /* before counter checks */ if (ret < 0) { if (fault(KEYREJECT) && ret == -EKEYREJECTED) { test_ok("%s: connect() was prevented", tst_name); diff --git a/tools/testing/selftests/net/tcp_ao/restore.c b/tools/testing/selftests/net/tcp_ao/restore.c index 7b91d7fde2bc..f6ea2190f43d 100644 --- a/tools/testing/selftests/net/tcp_ao/restore.c +++ b/tools/testing/selftests/net/tcp_ao/restore.c @@ -64,6 +64,7 @@ static void try_server_run(const char *tst_name, unsigned int port, else test_ok("%s: server alive", tst_name); } + synchronize_threads(); /* 3: counters checks */ if (test_get_tcp_ao_counters(sk, &ao2)) test_error("test_get_tcp_ao_counters()"); after_cnt = netstat_get_one(cnt_name, NULL); @@ -82,7 +83,7 @@ static void try_server_run(const char *tst_name, unsigned int port, * Before close() as that will send FIN and move the peer in TCP_CLOSE * and that will prevent reading AO counters from the peer's socket. */ - synchronize_threads(); /* 3: verified => closed */ + synchronize_threads(); /* 4: verified => closed */ out: close(sk); } @@ -176,6 +177,7 @@ static void test_sk_restore(const char *tst_name, unsigned int server_port, else test_ok("%s: post-migrate connection is alive", tst_name); } + synchronize_threads(); /* 3: counters checks */ if (test_get_tcp_ao_counters(sk, &ao2)) test_error("test_get_tcp_ao_counters()"); after_cnt = netstat_get_one(cnt_name, NULL); @@ -189,7 +191,7 @@ static void test_sk_restore(const char *tst_name, unsigned int server_port, test_ok("%s: counter %s increased %" PRIu64 " => %" PRIu64, tst_name, cnt_name, before_cnt, after_cnt); } - synchronize_threads(); /* 3: verified => closed */ + synchronize_threads(); /* 4: verified => closed */ close(sk); } diff --git a/tools/testing/selftests/net/tcp_ao/seq-ext.c b/tools/testing/selftests/net/tcp_ao/seq-ext.c index 9c7dde7fd776..885866cc193c 100644 --- a/tools/testing/selftests/net/tcp_ao/seq-ext.c +++ b/tools/testing/selftests/net/tcp_ao/seq-ext.c @@ -116,7 +116,7 @@ static void *server_fn(void *arg) sk = test_sk_restore(&img, &ao_img, &saddr, this_ip_dest, client_new_port, &ao1); - synchronize_threads(); /* 5: verify counters during SEQ-number rollover */ + synchronize_threads(); /* 5: verify the connection during SEQ-number rollover */ bytes = test_server_run(sk, quota, TEST_TIMEOUT_SEC); if (bytes != quota) { if (bytes > 0) @@ -127,6 +127,7 @@ static void *server_fn(void *arg) test_ok("server alive"); } + synchronize_threads(); /* 6: verify counters after SEQ-number rollover */ if (test_get_tcp_ao_counters(sk, &ao2)) test_error("test_get_tcp_ao_counters()"); after_good = netstat_get_one("TCPAOGood", NULL); @@ -206,12 +207,13 @@ static void *client_fn(void *arg) sk = test_sk_restore(&img, &ao_img, &saddr, this_ip_dest, test_server_port + 1, &ao1); - synchronize_threads(); /* 5: verify counters during SEQ-number rollover */ + synchronize_threads(); /* 5: verify the connection during SEQ-number rollover */ if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC)) test_fail("post-migrate verify failed"); else test_ok("post-migrate connection alive"); + synchronize_threads(); /* 5: verify counters after SEQ-number rollover */ if (test_get_tcp_ao_counters(sk, &ao2)) test_error("test_get_tcp_ao_counters()"); after_good = netstat_get_one("TCPAOGood", NULL); diff --git a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c index ec2848036341..02346b58efbd 100644 --- a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c +++ b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c @@ -70,6 +70,7 @@ static void try_accept(const char *tst_name, unsigned int port, timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC; err = test_wait_fd(lsk, timeout, 0); + synchronize_threads(); /* connect()/accept() timeouts */ if (err == -ETIMEDOUT) { if (!fault(TIMEOUT)) test_fail("timed out for accept()"); @@ -283,6 +284,7 @@ static void try_connect(const char *tst_name, unsigned int port, timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC; ret = _test_connect_socket(sk, this_ip_dest, port, timeout); + synchronize_threads(); /* connect()/accept() timeouts */ if (ret < 0) { if (fault(KEYREJECT) && ret == -EKEYREJECTED) test_ok("%s: connect() was prevented", tst_name); @@ -451,6 +453,7 @@ static void try_to_add(const char *tst_name, unsigned int port, timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC; ret = _test_connect_socket(sk, this_ip_dest, port, timeout); + synchronize_threads(); /* connect()/accept() timeouts */ if (ret <= 0) { test_error("%s: connect() returned %d", tst_name, ret); goto out; -- 2.42.2
