[PATCH 09/12] selftests/net: Add SEQ number extension test

[Dmitry Safonov <dima@arista.com>]

Check that on SEQ number wraparound there is no disruption or TCPAOBad
segments produced.

Sample of expected output:
> # ./seq-ext_ipv4
> 1..7
> # 1436[lib/setup.c:254] rand seed 1686611079
> TAP version 13
> ok 1 server alive
> ok 2 post-migrate connection alive
> ok 3 TCPAOGood counter increased 1002 => 3002
> ok 4 TCPAOGood counter increased 1003 => 3003
> ok 5 TCPAOBad counter didn't increase
> ok 6 TCPAOBad counter didn't increase
> ok 7 SEQ extension incremented: 1/1999, 1/998999
> # Totals: pass:7 fail:0 xfail:0 xpass:0 skip:0 error:0

Signed-off-by: Dmitry Safonov <dima@arista.com>
---
 tools/testing/selftests/net/tcp_ao/Makefile  |   1 +
 tools/testing/selftests/net/tcp_ao/seq-ext.c | 245 +++++++++++++++++++++++++++
 2 files changed, 246 insertions(+)

diff --git a/tools/testing/selftests/net/tcp_ao/Makefile b/tools/testing/selftests/net/tcp_ao/Makefile
index aa11a855c3e0..5408c7233460 100644
--- a/tools/testing/selftests/net/tcp_ao/Makefile
+++ b/tools/testing/selftests/net/tcp_ao/Makefile
@@ -4,6 +4,7 @@ TEST_BOTH_AF += connect
 TEST_BOTH_AF += connect-deny
 TEST_BOTH_AF += icmps-accept icmps-discard
 TEST_BOTH_AF += restore
+TEST_BOTH_AF += seq-ext
 TEST_BOTH_AF += setsockopt-closed
 TEST_BOTH_AF += unsigned-md5
 
diff --git a/tools/testing/selftests/net/tcp_ao/seq-ext.c b/tools/testing/selftests/net/tcp_ao/seq-ext.c
new file mode 100644
index 000000000000..ad4e77d6823e
--- /dev/null
+++ b/tools/testing/selftests/net/tcp_ao/seq-ext.c
@@ -0,0 +1,245 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Check that after SEQ number wrap-around:
+ * 1. SEQ-extension has upper bytes set
+ * 2. TCP conneciton is alive and no TCPAOBad segments
+ * In order to test (2), the test doesn't just adjust seq number for a queue
+ * on a connected socket, but migrates it to another sk+port number, so
+ * that there won't be any delayed packets that will fail to verify
+ * with the new SEQ numbers.
+ */
+#include <inttypes.h>
+#include "aolib.h"
+
+const unsigned int nr_packets = 1000;
+const unsigned int msg_len = 1000;
+const unsigned int quota = nr_packets * msg_len;
+unsigned int client_new_port;
+
+/* Move them closer to roll-over */
+static void test_adjust_seqs(struct tcp_sock_state *img,
+			     struct tcp_ao_repair *ao_img,
+			     bool server)
+{
+	uint32_t new_seq1, new_seq2;
+
+	/* make them roll-over during quota, but on different segments */
+	if (server) {
+		new_seq1 = ((uint32_t)-1) - msg_len;
+		new_seq2 = ((uint32_t)-1) - (quota - 2 * msg_len);
+	} else {
+		new_seq1 = ((uint32_t)-1) - (quota - 2 * msg_len);
+		new_seq2 = ((uint32_t)-1) - msg_len;
+	}
+
+	img->in.seq = new_seq1;
+	img->trw.snd_wl1 = img->in.seq - msg_len;
+	img->out.seq = new_seq2;
+	img->trw.rcv_wup = img->in.seq;
+}
+
+static int test_sk_restore(struct tcp_sock_state *img,
+			   struct tcp_ao_repair *ao_img, sockaddr_af *saddr,
+			   const union tcp_addr daddr, unsigned int dport,
+			   struct tcp_ao_counters *cnt)
+{
+	int sk;
+
+	sk = socket(test_family, SOCK_STREAM, IPPROTO_TCP);
+	if (sk < 0)
+		test_error("socket()");
+
+	test_enable_repair(sk);
+	test_sock_restore(sk, img, saddr, daddr, dport);
+	if (test_add_repaired_key(sk, DEFAULT_TEST_PASSWORD, 0, daddr, -1, 100, 100))
+		test_error("setsockopt(TCP_AO_ADD_KEY)");
+	test_ao_restore(sk, ao_img);
+
+	if (test_get_tcp_ao_counters(sk, cnt))
+		test_error("test_get_tcp_ao_counters()");
+
+	test_disable_repair(sk);
+	test_sock_state_free(img);
+	return sk;
+}
+
+static void *server_fn(void *arg)
+{
+	uint64_t before_good, after_good, after_bad;
+	struct tcp_ao_counters ao1, ao2;
+	struct tcp_sock_state img;
+	struct tcp_ao_repair ao_img;
+	sockaddr_af saddr;
+	ssize_t bytes;
+	int sk, lsk;
+
+	lsk = test_listen_socket(this_ip_addr, test_server_port, 1);
+
+	if (test_add_key(lsk, DEFAULT_TEST_PASSWORD, this_ip_dest, -1, 100, 100))
+		test_error("setsockopt(TCP_AO_ADD_KEY)");
+
+	synchronize_threads(); /* 1: MKT added => connect() */
+
+	if (test_wait_fd(lsk, TEST_TIMEOUT_SEC, 0))
+		test_error("test_wait_fd()");
+
+	sk = accept(lsk, NULL, NULL);
+	if (sk < 0)
+		test_error("accept()");
+
+	synchronize_threads(); /* 2: accepted => send data */
+	close(lsk);
+
+	bytes = test_server_run(sk, quota, TEST_TIMEOUT_SEC);
+	if (bytes != quota) {
+		if (bytes > 0)
+			test_fail("server served: %zd", bytes);
+		else
+			test_fail("server returned: %zd", bytes);
+		goto out;
+	}
+
+	before_good = netstat_get_one("TCPAOGood", NULL);
+
+	synchronize_threads(); /* 3: restore the connection on another port */
+
+	test_enable_repair(sk);
+	test_sock_checkpoint(sk, &img, &saddr);
+	test_ao_checkpoint(sk, &ao_img);
+	test_kill_sk(sk);
+#ifdef IPV6_TEST
+	saddr.sin6_port = htons(ntohs(saddr.sin6_port) + 1);
+#else
+	saddr.sin_port = htons(ntohs(saddr.sin_port) + 1);
+#endif
+	test_adjust_seqs(&img, &ao_img, true);
+	synchronize_threads(); /* 4: dump finished */
+	sk = test_sk_restore(&img, &ao_img, &saddr, this_ip_dest,
+			     client_new_port, &ao1);
+
+	synchronize_threads(); /* 5: verify counters during SEQ-number rollover */
+	bytes = test_server_run(sk, quota, TEST_TIMEOUT_SEC);
+	if (bytes != quota) {
+		if (bytes > 0)
+			test_fail("server served: %zd", bytes);
+		else
+			test_fail("server returned: %zd", bytes);
+	} else {
+		test_ok("server alive");
+	}
+
+	if (test_get_tcp_ao_counters(sk, &ao2))
+		test_error("test_get_tcp_ao_counters()");
+	after_good = netstat_get_one("TCPAOGood", NULL);
+
+	test_tcp_ao_counters_cmp(NULL, &ao1, &ao2, TEST_CNT_GOOD);
+
+	if (after_good <= before_good) {
+		test_fail("TCPAOGood counter did not increase: %zu <= %zu",
+			  after_good, before_good);
+	} else {
+		test_ok("TCPAOGood counter increased %zu => %zu",
+			before_good, after_good);
+	}
+	after_bad = netstat_get_one("TCPAOBad", NULL);
+	if (after_bad)
+		test_fail("TCPAOBad counter is non-zero: %zu", after_bad);
+	else
+		test_ok("TCPAOBad counter didn't increase");
+	test_enable_repair(sk);
+	test_ao_checkpoint(sk, &ao_img);
+	if (ao_img.snd_sne && ao_img.rcv_sne) {
+		test_ok("SEQ extension incremented: %u/%u",
+			ao_img.snd_sne, ao_img.rcv_sne);
+	} else {
+		test_fail("SEQ extension was not incremented: %u/%u",
+			  ao_img.snd_sne, ao_img.rcv_sne);
+	}
+
+	synchronize_threads(); /* 6: verified => closed */
+out:
+	close(sk);
+	return NULL;
+}
+
+static void *client_fn(void *arg)
+{
+	uint64_t before_good, after_good, after_bad;
+	struct tcp_ao_counters ao1, ao2;
+	struct tcp_sock_state img;
+	struct tcp_ao_repair ao_img;
+	sockaddr_af saddr;
+	int sk;
+
+	sk = socket(test_family, SOCK_STREAM, IPPROTO_TCP);
+	if (sk < 0)
+		test_error("socket()");
+
+	if (test_add_key(sk, DEFAULT_TEST_PASSWORD, this_ip_dest, -1, 100, 100))
+		test_error("setsockopt(TCP_AO_ADD_KEY)");
+
+	synchronize_threads(); /* 1: MKT added => connect() */
+	if (test_connect_socket(sk, this_ip_dest, test_server_port) <= 0)
+		test_error("failed to connect()");
+
+	synchronize_threads(); /* 2: accepted => send data */
+	if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC)) {
+		test_fail("pre-migrate verify failed");
+		return NULL;
+	}
+
+	before_good = netstat_get_one("TCPAOGood", NULL);
+
+	synchronize_threads(); /* 3: restore the connection on another port */
+	test_enable_repair(sk);
+	test_sock_checkpoint(sk, &img, &saddr);
+	test_ao_checkpoint(sk, &ao_img);
+	test_kill_sk(sk);
+#ifdef IPV6_TEST
+	client_new_port = ntohs(saddr.sin6_port) + 1;
+	saddr.sin6_port = htons(ntohs(saddr.sin6_port) + 1);
+#else
+	client_new_port = ntohs(saddr.sin_port) + 1;
+	saddr.sin_port = htons(ntohs(saddr.sin_port) + 1);
+#endif
+	test_adjust_seqs(&img, &ao_img, false);
+	synchronize_threads(); /* 4: dump finished */
+	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 */
+	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");
+
+	if (test_get_tcp_ao_counters(sk, &ao2))
+		test_error("test_get_tcp_ao_counters()");
+	after_good = netstat_get_one("TCPAOGood", NULL);
+
+	test_tcp_ao_counters_cmp(NULL, &ao1, &ao2, TEST_CNT_GOOD);
+
+	if (after_good <= before_good) {
+		test_fail("TCPAOGood counter did not increase: %zu <= %zu",
+			  after_good, before_good);
+	} else {
+		test_ok("TCPAOGood counter increased %zu => %zu",
+			before_good, after_good);
+	}
+	after_bad = netstat_get_one("TCPAOBad", NULL);
+	if (after_bad)
+		test_fail("TCPAOBad counter is non-zero: %zu", after_bad);
+	else
+		test_ok("TCPAOBad counter didn't increase");
+
+	synchronize_threads(); /* 6: verified => closed */
+	close(sk);
+
+	synchronize_threads(); /* don't race to exit: let server exit() */
+	return NULL;
+}
+
+int main(int argc, char *argv[])
+{
+	test_init(7, server_fn, client_fn);
+	return 0;
+}

-- 
2.43.0