* [PATCH net 1/7] selftests/net: Print TCP flags in more common format
2025-03-12 9:10 [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov via B4 Relay
@ 2025-03-12 9:10 ` Dmitry Safonov via B4 Relay
2025-03-12 9:10 ` [PATCH net 2/7] selftests/net: Provide tcp-ao counters comparison helper Dmitry Safonov via B4 Relay
` (6 subsequent siblings)
7 siblings, 0 replies; 15+ messages in thread
From: Dmitry Safonov via B4 Relay @ 2025-03-12 9:10 UTC (permalink / raw)
To: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan
Cc: netdev, linux-kselftest, linux-kernel, Dmitry Safonov
From: Dmitry Safonov <0x7f454c46@gmail.com>
Before:
># 13145[lib/ftrace-tcp.c:427] trace event filter tcp_ao_key_not_found [2001:db8:1::1:-1 => 2001:db8:254::1:7010, L3index 0, flags: !FS!R!P!., keyid: 100, rnext: 100, maclen: -1, sne: -1] = 1
After:
># 13487[lib/ftrace-tcp.c:427] trace event filter tcp_ao_key_not_found [2001:db8:1::1:-1 => 2001:db8:254::1:7010, L3index 0, flags: S, keyid: 100, rnext: 100, maclen: -1, sne: -1] = 1
For the history, I think the initial format was to emphasize the absence
of flags as well as their presence (!R meant no RST flag). But looking
again, it's just unreadable and hard to understand.
Make it the standard/expected one.
Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
---
tools/testing/selftests/net/tcp_ao/lib/ftrace-tcp.c | 7 ++-----
1 file changed, 2 insertions(+), 5 deletions(-)
diff --git a/tools/testing/selftests/net/tcp_ao/lib/ftrace-tcp.c b/tools/testing/selftests/net/tcp_ao/lib/ftrace-tcp.c
index 24380c68fec6b7613c363cdee68d13293e593aeb..27403f875054706718256e0007a4db77f82e60c8 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/ftrace-tcp.c
+++ b/tools/testing/selftests/net/tcp_ao/lib/ftrace-tcp.c
@@ -427,11 +427,8 @@ static void dump_trace_event(struct expected_trace_point *e)
test_print("trace event filter %s [%s:%d => %s:%d, L3index %d, flags: %s%s%s%s%s, keyid: %d, rnext: %d, maclen: %d, sne: %d] = %zu",
trace_event_names[e->type],
src, e->src_port, dst, e->dst_port, e->L3index,
- (e->fin > 0) ? "F" : (e->fin == 0) ? "!F" : "",
- (e->syn > 0) ? "S" : (e->syn == 0) ? "!S" : "",
- (e->rst > 0) ? "R" : (e->rst == 0) ? "!R" : "",
- (e->psh > 0) ? "P" : (e->psh == 0) ? "!P" : "",
- (e->ack > 0) ? "." : (e->ack == 0) ? "!." : "",
+ e->fin ? "F" : "", e->syn ? "S" : "", e->rst ? "R" : "",
+ e->psh ? "P" : "", e->ack ? "." : "",
e->keyid, e->rnext, e->maclen, e->sne, e->matched);
}
--
2.42.2
^ permalink raw reply [flat|nested] 15+ messages in thread* [PATCH net 2/7] selftests/net: Provide tcp-ao counters comparison helper
2025-03-12 9:10 [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov via B4 Relay
2025-03-12 9:10 ` [PATCH net 1/7] selftests/net: Print TCP flags in more common format Dmitry Safonov via B4 Relay
@ 2025-03-12 9:10 ` Dmitry Safonov via B4 Relay
2025-03-12 9:10 ` [PATCH net 3/7] selftests/net: Fetch and check TCP-MD5 counters Dmitry Safonov via B4 Relay
` (5 subsequent siblings)
7 siblings, 0 replies; 15+ messages in thread
From: Dmitry Safonov via B4 Relay @ 2025-03-12 9:10 UTC (permalink / raw)
To: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan
Cc: netdev, linux-kselftest, linux-kernel, Dmitry Safonov
From: Dmitry Safonov <0x7f454c46@gmail.com>
Rename __test_tcp_ao_counters_cmp() into test_assert_counters_ao() and
test_tcp_ao_key_counters_cmp() into test_assert_counters_key() as they
are asserts, rather than just compare functions.
Provide test_cmp_counters() helper, that's going to be used to compare
ao_info and netns counters as a stop condition for polling the sockets.
Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
---
tools/testing/selftests/net/tcp_ao/connect-deny.c | 4 +-
tools/testing/selftests/net/tcp_ao/connect.c | 2 +-
tools/testing/selftests/net/tcp_ao/icmps-discard.c | 4 +-
.../testing/selftests/net/tcp_ao/key-management.c | 8 +--
tools/testing/selftests/net/tcp_ao/lib/aolib.h | 33 ++++++---
tools/testing/selftests/net/tcp_ao/lib/sock.c | 83 ++++++++++++----------
tools/testing/selftests/net/tcp_ao/restore.c | 4 +-
tools/testing/selftests/net/tcp_ao/rst.c | 6 +-
tools/testing/selftests/net/tcp_ao/self-connect.c | 2 +-
tools/testing/selftests/net/tcp_ao/seq-ext.c | 4 +-
tools/testing/selftests/net/tcp_ao/unsigned-md5.c | 2 +-
11 files changed, 89 insertions(+), 63 deletions(-)
diff --git a/tools/testing/selftests/net/tcp_ao/connect-deny.c b/tools/testing/selftests/net/tcp_ao/connect-deny.c
index d418162d335f4c7b584fe81a7206f721d4efc20c..42223a1a7a822d84661c727ab1c8125fef534fa0 100644
--- a/tools/testing/selftests/net/tcp_ao/connect-deny.c
+++ b/tools/testing/selftests/net/tcp_ao/connect-deny.c
@@ -78,7 +78,7 @@ static void try_accept(const char *tst_name, unsigned int port, const char *pwd,
close(lsk);
if (pwd)
- test_tcp_ao_counters_cmp(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
+ test_assert_counters(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
if (!cnt_name)
goto out;
@@ -204,7 +204,7 @@ static void try_connect(const char *tst_name, unsigned int port,
if (pwd && ret > 0) {
if (test_get_tcp_ao_counters(sk, &ao_cnt2))
test_error("test_get_tcp_ao_counters()");
- test_tcp_ao_counters_cmp(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
+ test_assert_counters(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
}
out:
synchronize_threads(); /* close() */
diff --git a/tools/testing/selftests/net/tcp_ao/connect.c b/tools/testing/selftests/net/tcp_ao/connect.c
index f1d8d29e393f619d53ff0f370779ff5d227a7f5b..87647f38e7e6a6c0fde9a2789debe98e22a178dd 100644
--- a/tools/testing/selftests/net/tcp_ao/connect.c
+++ b/tools/testing/selftests/net/tcp_ao/connect.c
@@ -71,7 +71,7 @@ static void *client_fn(void *arg)
nr_packets, after_aogood, before_aogood);
return NULL;
}
- if (test_tcp_ao_counters_cmp("connect", &ao1, &ao2, TEST_CNT_GOOD))
+ if (test_assert_counters("connect", &ao1, &ao2, TEST_CNT_GOOD))
return NULL;
test_ok("connect TCPAOGood %" PRIu64 "/%" PRIu64 "/%" PRIu64 " => %" PRIu64 "/%" PRIu64 "/%" PRIu64 ", sent %zu",
diff --git a/tools/testing/selftests/net/tcp_ao/icmps-discard.c b/tools/testing/selftests/net/tcp_ao/icmps-discard.c
index a1614f0d8c44882e683adbdd6e010bb5a8bad9d3..e2958a6a0e28efa0c8a339b1a014f2fabf0dd7ce 100644
--- a/tools/testing/selftests/net/tcp_ao/icmps-discard.c
+++ b/tools/testing/selftests/net/tcp_ao/icmps-discard.c
@@ -91,9 +91,9 @@ static void serve_interfered(int sk)
return;
}
#ifdef TEST_ICMPS_ACCEPT
- test_tcp_ao_counters_cmp(NULL, &ao_cnt1, &ao_cnt2, TEST_CNT_GOOD);
+ test_assert_counters(NULL, &ao_cnt1, &ao_cnt2, TEST_CNT_GOOD);
#else
- test_tcp_ao_counters_cmp(NULL, &ao_cnt1, &ao_cnt2, TEST_CNT_GOOD | TEST_CNT_AO_DROPPED_ICMP);
+ test_assert_counters(NULL, &ao_cnt1, &ao_cnt2, TEST_CNT_GOOD | TEST_CNT_AO_DROPPED_ICMP);
#endif
if (icmp_ignored_a >= icmp_ignored_b) {
test_icmps_fail("%s counter didn't change: %" PRIu64 " >= %" PRIu64,
diff --git a/tools/testing/selftests/net/tcp_ao/key-management.c b/tools/testing/selftests/net/tcp_ao/key-management.c
index d4385b52c10b6bda6e047bbb09ae8f32f284042c..2c55c686285acedcdc5a836f7818e35be1ad9d91 100644
--- a/tools/testing/selftests/net/tcp_ao/key-management.c
+++ b/tools/testing/selftests/net/tcp_ao/key-management.c
@@ -633,7 +633,7 @@ static void verify_counters(const char *tst_name, bool is_listen_sk, bool server
{
unsigned int i;
- __test_tcp_ao_counters_cmp(tst_name, a, b, TEST_CNT_GOOD);
+ test_assert_counters_ao(tst_name, a, b, TEST_CNT_GOOD);
for (i = 0; i < collection.nr_keys; i++) {
struct test_key *key = &collection.keys[i];
@@ -652,9 +652,9 @@ static void verify_counters(const char *tst_name, bool is_listen_sk, bool server
rx_cnt_expected = key->used_on_server_tx;
}
- test_tcp_ao_key_counters_cmp(tst_name, a, b,
- rx_cnt_expected ? TEST_CNT_KEY_GOOD : 0,
- sndid, rcvid);
+ test_assert_counters_key(tst_name, a, b,
+ rx_cnt_expected ? TEST_CNT_KEY_GOOD : 0,
+ sndid, rcvid);
}
test_tcp_ao_counters_free(a);
test_tcp_ao_counters_free(b);
diff --git a/tools/testing/selftests/net/tcp_ao/lib/aolib.h b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
index 5db2f65cddc4e7f0186caf538e512416ddaf7286..d6557940ef1fa6bf52da9efea04156edd9b10b7e 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/aolib.h
+++ b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
@@ -528,6 +528,22 @@ extern int test_get_tcp_ao_counters(int sk, struct tcp_ao_counters *out);
#define TEST_CNT_NS_DROPPED_ICMP BIT(11)
typedef uint16_t test_cnt;
+#define _for_each_counter(f) \
+do { \
+ /* per-netns */ \
+ f(netns_ao_good, TEST_CNT_NS_GOOD); \
+ f(netns_ao_bad, TEST_CNT_NS_BAD); \
+ f(netns_ao_key_not_found, TEST_CNT_NS_KEY_NOT_FOUND); \
+ f(netns_ao_required, TEST_CNT_NS_AO_REQUIRED); \
+ f(netns_ao_dropped_icmp, TEST_CNT_NS_DROPPED_ICMP); \
+ /* per-socket */ \
+ f(ao_info_pkt_good, TEST_CNT_SOCK_GOOD); \
+ f(ao_info_pkt_bad, TEST_CNT_SOCK_BAD); \
+ f(ao_info_pkt_key_not_found, TEST_CNT_SOCK_KEY_NOT_FOUND); \
+ f(ao_info_pkt_ao_required, TEST_CNT_SOCK_AO_REQUIRED); \
+ f(ao_info_pkt_dropped_icmp, TEST_CNT_SOCK_DROPPED_ICMP); \
+} while (0)
+
#define TEST_CNT_AO_GOOD (TEST_CNT_SOCK_GOOD | TEST_CNT_NS_GOOD)
#define TEST_CNT_AO_BAD (TEST_CNT_SOCK_BAD | TEST_CNT_NS_BAD)
#define TEST_CNT_AO_KEY_NOT_FOUND (TEST_CNT_SOCK_KEY_NOT_FOUND | \
@@ -539,10 +555,10 @@ typedef uint16_t test_cnt;
#define TEST_CNT_GOOD (TEST_CNT_KEY_GOOD | TEST_CNT_AO_GOOD)
#define TEST_CNT_BAD (TEST_CNT_KEY_BAD | TEST_CNT_AO_BAD)
-extern int __test_tcp_ao_counters_cmp(const char *tst_name,
+extern int test_assert_counters_ao(const char *tst_name,
struct tcp_ao_counters *before, struct tcp_ao_counters *after,
test_cnt expected);
-extern int test_tcp_ao_key_counters_cmp(const char *tst_name,
+extern int test_assert_counters_key(const char *tst_name,
struct tcp_ao_counters *before, struct tcp_ao_counters *after,
test_cnt expected, int sndid, int rcvid);
extern void test_tcp_ao_counters_free(struct tcp_ao_counters *cnts);
@@ -552,18 +568,17 @@ extern void test_tcp_ao_counters_free(struct tcp_ao_counters *cnts);
* to test_get_tcp_ao_counters(). Check key counters manually if they
* may change.
*/
-static inline int test_tcp_ao_counters_cmp(const char *tst_name,
- struct tcp_ao_counters *before,
- struct tcp_ao_counters *after,
- test_cnt expected)
+static inline int test_assert_counters(const char *tst_name,
+ struct tcp_ao_counters *before,
+ struct tcp_ao_counters *after,
+ test_cnt expected)
{
int ret;
- ret = __test_tcp_ao_counters_cmp(tst_name, before, after, expected);
+ ret = test_assert_counters_ao(tst_name, before, after, expected);
if (ret)
goto out;
- ret = test_tcp_ao_key_counters_cmp(tst_name, before, after,
- expected, -1, -1);
+ ret = test_assert_counters_key(tst_name, before, after, expected, -1, -1);
out:
test_tcp_ao_counters_free(before);
test_tcp_ao_counters_free(after);
diff --git a/tools/testing/selftests/net/tcp_ao/lib/sock.c b/tools/testing/selftests/net/tcp_ao/lib/sock.c
index 0ffda966c677b36929ed939ddb42b25c6612e22c..2a2586b2ffa2bdf805562255e597ccd09597b33c 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/sock.c
+++ b/tools/testing/selftests/net/tcp_ao/lib/sock.c
@@ -403,55 +403,66 @@ int test_get_tcp_ao_counters(int sk, struct tcp_ao_counters *out)
return 0;
}
-int __test_tcp_ao_counters_cmp(const char *tst_name,
- struct tcp_ao_counters *before,
- struct tcp_ao_counters *after,
- test_cnt expected)
+int test_cmp_counters(struct tcp_ao_counters *before, struct tcp_ao_counters *after)
{
-#define __cmp_ao(cnt, expecting_inc) \
+#define __cmp(cnt, e_cnt) \
+do { \
+ if (before->cnt > after->cnt) \
+ return -1; \
+ if (before->cnt != after->cnt) \
+ ret |= e_cnt; \
+} while (0)
+
+ int ret = 0;
+ size_t i;
+
+ if (before->nr_keys != after->nr_keys)
+ return -1;
+
+ _for_each_counter(__cmp);
+
+ i = before->nr_keys;
+ while (i--) {
+ __cmp(key_cnts[i].pkt_good, TEST_CNT_KEY_GOOD);
+ __cmp(key_cnts[i].pkt_bad, TEST_CNT_KEY_BAD);
+ }
+#undef __cmp
+ return ret;
+}
+
+int test_assert_counters_ao(const char *tst_name,
+ struct tcp_ao_counters *before,
+ struct tcp_ao_counters *after,
+ test_cnt expected)
+{
+#define __cmp_ao(cnt, e_cnt) \
do { \
if (before->cnt > after->cnt) { \
test_fail("%s: Decreased counter " __stringify(cnt) " %" PRIu64 " > %" PRIu64, \
- tst_name ?: "", before->cnt, after->cnt); \
+ tst_name ?: "", before->cnt, after->cnt); \
return -1; \
} \
- if ((before->cnt != after->cnt) != (expecting_inc)) { \
+ if ((before->cnt != after->cnt) != !!(expected & e_cnt)) { \
test_fail("%s: Counter " __stringify(cnt) " was %sexpected to increase %" PRIu64 " => %" PRIu64, \
- tst_name ?: "", (expecting_inc) ? "" : "not ", \
+ tst_name ?: "", (expected & e_cnt) ? "" : "not ", \
before->cnt, after->cnt); \
return -1; \
} \
-} while(0)
+} while (0)
errno = 0;
- /* per-netns */
- __cmp_ao(netns_ao_good, !!(expected & TEST_CNT_NS_GOOD));
- __cmp_ao(netns_ao_bad, !!(expected & TEST_CNT_NS_BAD));
- __cmp_ao(netns_ao_key_not_found,
- !!(expected & TEST_CNT_NS_KEY_NOT_FOUND));
- __cmp_ao(netns_ao_required, !!(expected & TEST_CNT_NS_AO_REQUIRED));
- __cmp_ao(netns_ao_dropped_icmp,
- !!(expected & TEST_CNT_NS_DROPPED_ICMP));
- /* per-socket */
- __cmp_ao(ao_info_pkt_good, !!(expected & TEST_CNT_SOCK_GOOD));
- __cmp_ao(ao_info_pkt_bad, !!(expected & TEST_CNT_SOCK_BAD));
- __cmp_ao(ao_info_pkt_key_not_found,
- !!(expected & TEST_CNT_SOCK_KEY_NOT_FOUND));
- __cmp_ao(ao_info_pkt_ao_required, !!(expected & TEST_CNT_SOCK_AO_REQUIRED));
- __cmp_ao(ao_info_pkt_dropped_icmp,
- !!(expected & TEST_CNT_SOCK_DROPPED_ICMP));
+ _for_each_counter(__cmp_ao);
return 0;
#undef __cmp_ao
}
-int test_tcp_ao_key_counters_cmp(const char *tst_name,
- struct tcp_ao_counters *before,
- struct tcp_ao_counters *after,
- test_cnt expected,
- int sndid, int rcvid)
+int test_assert_counters_key(const char *tst_name,
+ struct tcp_ao_counters *before,
+ struct tcp_ao_counters *after,
+ test_cnt expected, int sndid, int rcvid)
{
size_t i;
-#define __cmp_ao(i, cnt, expecting_inc) \
+#define __cmp_ao(i, cnt, e_cnt) \
do { \
if (before->key_cnts[i].cnt > after->key_cnts[i].cnt) { \
test_fail("%s: Decreased counter " __stringify(cnt) " %" PRIu64 " > %" PRIu64 " for key %u:%u", \
@@ -461,16 +472,16 @@ do { \
before->key_cnts[i].rcvid); \
return -1; \
} \
- if ((before->key_cnts[i].cnt != after->key_cnts[i].cnt) != (expecting_inc)) { \
+ if ((before->key_cnts[i].cnt != after->key_cnts[i].cnt) != !!(expected & e_cnt)) { \
test_fail("%s: Counter " __stringify(cnt) " was %sexpected to increase %" PRIu64 " => %" PRIu64 " for key %u:%u", \
- tst_name ?: "", (expecting_inc) ? "" : "not ",\
+ tst_name ?: "", (expected & e_cnt) ? "" : "not ",\
before->key_cnts[i].cnt, \
after->key_cnts[i].cnt, \
before->key_cnts[i].sndid, \
before->key_cnts[i].rcvid); \
return -1; \
} \
-} while(0)
+} while (0)
if (before->nr_keys != after->nr_keys) {
test_fail("%s: Keys changed on the socket %zu != %zu",
@@ -485,8 +496,8 @@ do { \
continue;
if (rcvid >= 0 && before->key_cnts[i].rcvid != rcvid)
continue;
- __cmp_ao(i, pkt_good, !!(expected & TEST_CNT_KEY_GOOD));
- __cmp_ao(i, pkt_bad, !!(expected & TEST_CNT_KEY_BAD));
+ __cmp_ao(i, pkt_good, TEST_CNT_KEY_GOOD);
+ __cmp_ao(i, pkt_bad, TEST_CNT_KEY_BAD);
}
return 0;
#undef __cmp_ao
diff --git a/tools/testing/selftests/net/tcp_ao/restore.c b/tools/testing/selftests/net/tcp_ao/restore.c
index ecc6f1e3a41418204a541ec4859cb22bbcf36856..ad8505f77023bcfb0712108c8148f5c94def67b1 100644
--- a/tools/testing/selftests/net/tcp_ao/restore.c
+++ b/tools/testing/selftests/net/tcp_ao/restore.c
@@ -69,7 +69,7 @@ static void try_server_run(const char *tst_name, unsigned int port,
test_error("test_get_tcp_ao_counters()");
after_cnt = netstat_get_one(cnt_name, NULL);
- test_tcp_ao_counters_cmp(tst_name, &ao1, &ao2, cnt_expected);
+ test_assert_counters(tst_name, &ao1, &ao2, cnt_expected);
if (after_cnt <= before_cnt) {
test_fail("%s: %s counter did not increase: %" PRIu64 " <= %" PRIu64,
@@ -182,7 +182,7 @@ static void test_sk_restore(const char *tst_name, unsigned int server_port,
test_error("test_get_tcp_ao_counters()");
after_cnt = netstat_get_one(cnt_name, NULL);
- test_tcp_ao_counters_cmp(tst_name, &ao1, &ao2, cnt_expected);
+ test_assert_counters(tst_name, &ao1, &ao2, cnt_expected);
if (after_cnt <= before_cnt) {
test_fail("%s: %s counter did not increase: %" PRIu64 " <= %" PRIu64,
diff --git a/tools/testing/selftests/net/tcp_ao/rst.c b/tools/testing/selftests/net/tcp_ao/rst.c
index 6364facaa63ed79d17bdbfda57441c51d9039279..9cd257a995bb35fafad868faf0f325168cea5572 100644
--- a/tools/testing/selftests/net/tcp_ao/rst.c
+++ b/tools/testing/selftests/net/tcp_ao/rst.c
@@ -120,7 +120,7 @@ static void test_server_active_rst(unsigned int port)
synchronize_threads(); /* 5: closed active sk */
synchronize_threads(); /* 6: counters checks */
- if (test_tcp_ao_counters_cmp("active RST server", &cnt1, &cnt2, TEST_CNT_GOOD))
+ if (test_assert_counters("active RST server", &cnt1, &cnt2, TEST_CNT_GOOD))
test_fail("MKT counters (server) have not only good packets");
else
test_ok("MKT counters are good on server");
@@ -165,7 +165,7 @@ static void test_server_passive_rst(unsigned int port)
close(sk);
synchronize_threads(); /* 5: restore the socket, send more data */
- test_tcp_ao_counters_cmp("passive RST server", &ao1, &ao2, TEST_CNT_GOOD);
+ test_assert_counters("passive RST server", &ao1, &ao2, TEST_CNT_GOOD);
synchronize_threads(); /* 6: server exits */
}
@@ -431,7 +431,7 @@ static void test_client_passive_rst(unsigned int port)
synchronize_threads(); /* 6: server exits */
close(sk);
- test_tcp_ao_counters_cmp("client passive RST", &ao1, &ao2, TEST_CNT_GOOD);
+ test_assert_counters("client passive RST", &ao1, &ao2, TEST_CNT_GOOD);
}
static void *client_fn(void *arg)
diff --git a/tools/testing/selftests/net/tcp_ao/self-connect.c b/tools/testing/selftests/net/tcp_ao/self-connect.c
index 3ecd2b58de6a396dfa559fd10a6a00f64b4daf83..ebdf8e5d951f024efa4e41a9b67d6adf7cfa8ac7 100644
--- a/tools/testing/selftests/net/tcp_ao/self-connect.c
+++ b/tools/testing/selftests/net/tcp_ao/self-connect.c
@@ -93,7 +93,7 @@ static void tcp_self_connect(const char *tst, unsigned int port,
return;
}
- if (test_tcp_ao_counters_cmp(tst, &before_ao, &after_ao, TEST_CNT_GOOD)) {
+ if (test_assert_counters(tst, &before_ao, &after_ao, TEST_CNT_GOOD)) {
close(sk);
return;
}
diff --git a/tools/testing/selftests/net/tcp_ao/seq-ext.c b/tools/testing/selftests/net/tcp_ao/seq-ext.c
index 8901a6785dc8158227414cf90b461c5d2fe39689..ea8a84edbd10f9250e90a44c16dd0f971c6b385b 100644
--- a/tools/testing/selftests/net/tcp_ao/seq-ext.c
+++ b/tools/testing/selftests/net/tcp_ao/seq-ext.c
@@ -140,7 +140,7 @@ static void *server_fn(void *arg)
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);
+ test_assert_counters(NULL, &ao1, &ao2, TEST_CNT_GOOD);
if (after_good <= before_good) {
test_fail("TCPAOGood counter did not increase: %" PRIu64 " <= %" PRIu64,
@@ -226,7 +226,7 @@ static void *client_fn(void *arg)
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);
+ test_assert_counters(NULL, &ao1, &ao2, TEST_CNT_GOOD);
if (after_good <= before_good) {
test_fail("TCPAOGood counter did not increase: %" PRIu64 " <= %" PRIu64,
diff --git a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
index f779e5892bc162374fe6a64f5258b7227f19ba7a..f403ca25bbe6eb0b89f7467dc70b2a5a826838c4 100644
--- a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
+++ b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
@@ -108,7 +108,7 @@ static void try_accept(const char *tst_name, unsigned int port,
tst_name, cnt_name, before_cnt, after_cnt);
}
if (ao_addr)
- test_tcp_ao_counters_cmp(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
+ test_assert_counters(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
out:
synchronize_threads(); /* test_kill_sk() */
--
2.42.2
^ permalink raw reply [flat|nested] 15+ messages in thread* [PATCH net 3/7] selftests/net: Fetch and check TCP-MD5 counters
2025-03-12 9:10 [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov via B4 Relay
2025-03-12 9:10 ` [PATCH net 1/7] selftests/net: Print TCP flags in more common format Dmitry Safonov via B4 Relay
2025-03-12 9:10 ` [PATCH net 2/7] selftests/net: Provide tcp-ao counters comparison helper Dmitry Safonov via B4 Relay
@ 2025-03-12 9:10 ` Dmitry Safonov via B4 Relay
2025-03-12 9:10 ` [PATCH net 4/7] selftests/net: Add mixed select()+polling mode to TCP-AO tests Dmitry Safonov via B4 Relay
` (4 subsequent siblings)
7 siblings, 0 replies; 15+ messages in thread
From: Dmitry Safonov via B4 Relay @ 2025-03-12 9:10 UTC (permalink / raw)
To: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan
Cc: netdev, linux-kselftest, linux-kernel, Dmitry Safonov
From: Dmitry Safonov <0x7f454c46@gmail.com>
There are related TCP-MD5 <=> TCP and TCP-MD5 <=> TCP-AO tests
that can benefit from checking the related counters, not only from
validating operations timeouts.
It also prepares the code for introduction of mixed select()+poll mode,
see the follow-up patches.
Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
---
tools/testing/selftests/net/tcp_ao/connect-deny.c | 26 ++++----
tools/testing/selftests/net/tcp_ao/connect.c | 18 +++---
tools/testing/selftests/net/tcp_ao/icmps-discard.c | 14 ++---
.../testing/selftests/net/tcp_ao/key-management.c | 64 +++++++++----------
tools/testing/selftests/net/tcp_ao/lib/aolib.h | 63 ++++++++++++-------
tools/testing/selftests/net/tcp_ao/lib/sock.c | 72 ++++++++++++----------
tools/testing/selftests/net/tcp_ao/restore.c | 56 ++++++++---------
tools/testing/selftests/net/tcp_ao/rst.c | 34 +++++-----
tools/testing/selftests/net/tcp_ao/self-connect.c | 12 ++--
tools/testing/selftests/net/tcp_ao/seq-ext.c | 26 ++++----
tools/testing/selftests/net/tcp_ao/unsigned-md5.c | 23 ++++---
11 files changed, 219 insertions(+), 189 deletions(-)
diff --git a/tools/testing/selftests/net/tcp_ao/connect-deny.c b/tools/testing/selftests/net/tcp_ao/connect-deny.c
index 42223a1a7a822d84661c727ab1c8125fef534fa0..919e8d2b91345b453313f25b98cee27b80f1e984 100644
--- a/tools/testing/selftests/net/tcp_ao/connect-deny.c
+++ b/tools/testing/selftests/net/tcp_ao/connect-deny.c
@@ -34,7 +34,7 @@ static void try_accept(const char *tst_name, unsigned int port, const char *pwd,
const char *cnt_name, test_cnt cnt_expected,
fault_t inj)
{
- struct tcp_ao_counters ao_cnt1, ao_cnt2;
+ struct tcp_counters cnt1, cnt2;
uint64_t before_cnt = 0, after_cnt = 0; /* silence GCC */
int lsk, err, sk = 0;
time_t timeout;
@@ -46,8 +46,8 @@ static void try_accept(const char *tst_name, unsigned int port, const char *pwd,
if (cnt_name)
before_cnt = netstat_get_one(cnt_name, NULL);
- if (pwd && test_get_tcp_ao_counters(lsk, &ao_cnt1))
- test_error("test_get_tcp_ao_counters()");
+ if (pwd && test_get_tcp_counters(lsk, &cnt1))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* preparations done */
@@ -72,13 +72,13 @@ 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()");
+ if (pwd && test_get_tcp_counters(lsk, &cnt2))
+ test_error("test_get_tcp_counters()");
close(lsk);
if (pwd)
- test_assert_counters(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
+ test_assert_counters(tst_name, &cnt1, &cnt2, cnt_expected);
if (!cnt_name)
goto out;
@@ -163,7 +163,7 @@ static void try_connect(const char *tst_name, unsigned int port,
uint8_t sndid, uint8_t rcvid,
test_cnt cnt_expected, fault_t inj)
{
- struct tcp_ao_counters ao_cnt1, ao_cnt2;
+ struct tcp_counters cnt1, cnt2;
time_t timeout;
int sk, ret;
@@ -174,8 +174,8 @@ static void try_connect(const char *tst_name, unsigned int port,
if (pwd && test_add_key(sk, pwd, addr, prefix, sndid, rcvid))
test_error("setsockopt(TCP_AO_ADD_KEY)");
- if (pwd && test_get_tcp_ao_counters(sk, &ao_cnt1))
- test_error("test_get_tcp_ao_counters()");
+ if (pwd && test_get_tcp_counters(sk, &cnt1))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* preparations done */
@@ -202,9 +202,11 @@ static void try_connect(const char *tst_name, unsigned int port,
else
test_ok("%s: connected", tst_name);
if (pwd && ret > 0) {
- if (test_get_tcp_ao_counters(sk, &ao_cnt2))
- test_error("test_get_tcp_ao_counters()");
- test_assert_counters(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
+ if (test_get_tcp_counters(sk, &cnt2))
+ test_error("test_get_tcp_counters()");
+ test_assert_counters(tst_name, &cnt1, &cnt2, cnt_expected);
+ } else if (pwd) {
+ test_tcp_counters_free(&cnt1);
}
out:
synchronize_threads(); /* close() */
diff --git a/tools/testing/selftests/net/tcp_ao/connect.c b/tools/testing/selftests/net/tcp_ao/connect.c
index 87647f38e7e6a6c0fde9a2789debe98e22a178dd..45f2307483d3bab8c96df26c9dca28c898283c0c 100644
--- a/tools/testing/selftests/net/tcp_ao/connect.c
+++ b/tools/testing/selftests/net/tcp_ao/connect.c
@@ -35,7 +35,7 @@ static void *client_fn(void *arg)
uint64_t before_aogood, after_aogood;
const size_t nr_packets = 20;
struct netstat *ns_before, *ns_after;
- struct tcp_ao_counters ao1, ao2;
+ struct tcp_counters ao1, ao2;
if (sk < 0)
test_error("socket()");
@@ -50,8 +50,8 @@ static void *client_fn(void *arg)
ns_before = netstat_read();
before_aogood = netstat_get(ns_before, "TCPAOGood", NULL);
- if (test_get_tcp_ao_counters(sk, &ao1))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &ao1))
+ test_error("test_get_tcp_counters()");
if (test_client_verify(sk, 100, nr_packets, TEST_TIMEOUT_SEC)) {
test_fail("verify failed");
@@ -60,8 +60,8 @@ static void *client_fn(void *arg)
ns_after = netstat_read();
after_aogood = netstat_get(ns_after, "TCPAOGood", NULL);
- if (test_get_tcp_ao_counters(sk, &ao2))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &ao2))
+ test_error("test_get_tcp_counters()");
netstat_print_diff(ns_before, ns_after);
netstat_free(ns_before);
netstat_free(ns_after);
@@ -75,10 +75,10 @@ static void *client_fn(void *arg)
return NULL;
test_ok("connect TCPAOGood %" PRIu64 "/%" PRIu64 "/%" PRIu64 " => %" PRIu64 "/%" PRIu64 "/%" PRIu64 ", sent %zu",
- before_aogood, ao1.ao_info_pkt_good,
- ao1.key_cnts[0].pkt_good,
- after_aogood, ao2.ao_info_pkt_good,
- ao2.key_cnts[0].pkt_good,
+ before_aogood, ao1.ao.ao_info_pkt_good,
+ ao1.ao.key_cnts[0].pkt_good,
+ after_aogood, ao2.ao.ao_info_pkt_good,
+ ao2.ao.key_cnts[0].pkt_good,
nr_packets);
return NULL;
}
diff --git a/tools/testing/selftests/net/tcp_ao/icmps-discard.c b/tools/testing/selftests/net/tcp_ao/icmps-discard.c
index e2958a6a0e28efa0c8a339b1a014f2fabf0dd7ce..61a5e8e2d7e0d7bd1938c297e847384bf7e1d2c5 100644
--- a/tools/testing/selftests/net/tcp_ao/icmps-discard.c
+++ b/tools/testing/selftests/net/tcp_ao/icmps-discard.c
@@ -53,7 +53,7 @@ static void serve_interfered(int sk)
ssize_t test_quota = packet_size * packets_nr * 10;
uint64_t dest_unreach_a, dest_unreach_b;
uint64_t icmp_ignored_a, icmp_ignored_b;
- struct tcp_ao_counters ao_cnt1, ao_cnt2;
+ struct tcp_counters cnt1, cnt2;
bool counter_not_found;
struct netstat *ns_after, *ns_before;
ssize_t bytes;
@@ -61,16 +61,16 @@ static void serve_interfered(int sk)
ns_before = netstat_read();
dest_unreach_a = netstat_get(ns_before, dst_unreach, NULL);
icmp_ignored_a = netstat_get(ns_before, tcpao_icmps, NULL);
- if (test_get_tcp_ao_counters(sk, &ao_cnt1))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt1))
+ test_error("test_get_tcp_counters()");
bytes = test_server_run(sk, test_quota, 0);
ns_after = netstat_read();
netstat_print_diff(ns_before, ns_after);
dest_unreach_b = netstat_get(ns_after, dst_unreach, NULL);
icmp_ignored_b = netstat_get(ns_after, tcpao_icmps,
&counter_not_found);
- if (test_get_tcp_ao_counters(sk, &ao_cnt2))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt2))
+ test_error("test_get_tcp_counters()");
netstat_free(ns_before);
netstat_free(ns_after);
@@ -91,9 +91,9 @@ static void serve_interfered(int sk)
return;
}
#ifdef TEST_ICMPS_ACCEPT
- test_assert_counters(NULL, &ao_cnt1, &ao_cnt2, TEST_CNT_GOOD);
+ test_assert_counters(NULL, &cnt1, &cnt2, TEST_CNT_GOOD);
#else
- test_assert_counters(NULL, &ao_cnt1, &ao_cnt2, TEST_CNT_GOOD | TEST_CNT_AO_DROPPED_ICMP);
+ test_assert_counters(NULL, &cnt1, &cnt2, TEST_CNT_GOOD | TEST_CNT_AO_DROPPED_ICMP);
#endif
if (icmp_ignored_a >= icmp_ignored_b) {
test_icmps_fail("%s counter didn't change: %" PRIu64 " >= %" PRIu64,
diff --git a/tools/testing/selftests/net/tcp_ao/key-management.c b/tools/testing/selftests/net/tcp_ao/key-management.c
index 2c55c686285acedcdc5a836f7818e35be1ad9d91..c4087a15bdb3b68b51489a2d419bd911879c6c08 100644
--- a/tools/testing/selftests/net/tcp_ao/key-management.c
+++ b/tools/testing/selftests/net/tcp_ao/key-management.c
@@ -629,11 +629,11 @@ static int key_collection_socket(bool server, unsigned int port)
}
static void verify_counters(const char *tst_name, bool is_listen_sk, bool server,
- struct tcp_ao_counters *a, struct tcp_ao_counters *b)
+ struct tcp_counters *a, struct tcp_counters *b)
{
unsigned int i;
- test_assert_counters_ao(tst_name, a, b, TEST_CNT_GOOD);
+ test_assert_counters_sk(tst_name, a, b, TEST_CNT_GOOD);
for (i = 0; i < collection.nr_keys; i++) {
struct test_key *key = &collection.keys[i];
@@ -652,12 +652,12 @@ static void verify_counters(const char *tst_name, bool is_listen_sk, bool server
rx_cnt_expected = key->used_on_server_tx;
}
- test_assert_counters_key(tst_name, a, b,
+ test_assert_counters_key(tst_name, &a->ao, &b->ao,
rx_cnt_expected ? TEST_CNT_KEY_GOOD : 0,
sndid, rcvid);
}
- test_tcp_ao_counters_free(a);
- test_tcp_ao_counters_free(b);
+ test_tcp_counters_free(a);
+ test_tcp_counters_free(b);
test_ok("%s: passed counters checks", tst_name);
}
@@ -791,17 +791,17 @@ static void verify_keys(const char *tst_name, int sk,
}
static int start_server(const char *tst_name, unsigned int port, size_t quota,
- struct tcp_ao_counters *begin,
+ struct tcp_counters *begin,
unsigned int current_index, unsigned int rnext_index)
{
- struct tcp_ao_counters lsk_c1, lsk_c2;
+ struct tcp_counters lsk_c1, lsk_c2;
ssize_t bytes;
int sk, lsk;
synchronize_threads(); /* 1: key collection initialized */
lsk = key_collection_socket(true, port);
- if (test_get_tcp_ao_counters(lsk, &lsk_c1))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(lsk, &lsk_c1))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* 2: MKTs added => connect() */
if (test_wait_fd(lsk, TEST_TIMEOUT_SEC, 0))
test_error("test_wait_fd()");
@@ -809,12 +809,12 @@ static int start_server(const char *tst_name, unsigned int port, size_t quota,
sk = accept(lsk, NULL, NULL);
if (sk < 0)
test_error("accept()");
- if (test_get_tcp_ao_counters(sk, begin))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, begin))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* 3: accepted => send data */
- if (test_get_tcp_ao_counters(lsk, &lsk_c2))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(lsk, &lsk_c2))
+ test_error("test_get_tcp_counters()");
verify_keys(tst_name, lsk, true, true);
close(lsk);
@@ -830,12 +830,12 @@ static int start_server(const char *tst_name, unsigned int port, size_t quota,
}
static void end_server(const char *tst_name, int sk,
- struct tcp_ao_counters *begin)
+ struct tcp_counters *begin)
{
- struct tcp_ao_counters end;
+ struct tcp_counters end;
- if (test_get_tcp_ao_counters(sk, &end))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &end))
+ test_error("test_get_tcp_counters()");
verify_keys(tst_name, sk, false, true);
synchronize_threads(); /* 4: verified => closed */
@@ -848,7 +848,7 @@ static void end_server(const char *tst_name, int sk,
static void try_server_run(const char *tst_name, unsigned int port, size_t quota,
unsigned int current_index, unsigned int rnext_index)
{
- struct tcp_ao_counters tmp;
+ struct tcp_counters tmp;
int sk;
sk = start_server(tst_name, port, quota, &tmp,
@@ -860,7 +860,7 @@ static void server_rotations(const char *tst_name, unsigned int port,
size_t quota, unsigned int rotations,
unsigned int current_index, unsigned int rnext_index)
{
- struct tcp_ao_counters tmp;
+ struct tcp_counters tmp;
unsigned int i;
int sk;
@@ -886,7 +886,7 @@ static void server_rotations(const char *tst_name, unsigned int port,
static int run_client(const char *tst_name, unsigned int port,
unsigned int nr_keys, int current_index, int rnext_index,
- struct tcp_ao_counters *before,
+ struct tcp_counters *before,
const size_t msg_sz, const size_t msg_nr)
{
int sk;
@@ -904,8 +904,8 @@ static int run_client(const char *tst_name, unsigned int port,
if (test_set_key(sk, sndid, rcvid))
test_error("failed to set current/rnext keys");
}
- if (before && test_get_tcp_ao_counters(sk, before))
- test_error("test_get_tcp_ao_counters()");
+ if (before && test_get_tcp_counters(sk, before))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* 2: MKTs added => connect() */
if (test_connect_socket(sk, this_ip_dest, port++) <= 0)
@@ -922,7 +922,7 @@ static int run_client(const char *tst_name, unsigned int port,
test_fail("verify failed");
close(sk);
if (before)
- test_tcp_ao_counters_free(before);
+ test_tcp_counters_free(before);
return -1;
}
@@ -931,7 +931,7 @@ static int run_client(const char *tst_name, unsigned int port,
static int start_client(const char *tst_name, unsigned int port,
unsigned int nr_keys, int current_index, int rnext_index,
- struct tcp_ao_counters *before,
+ struct tcp_counters *before,
const size_t msg_sz, const size_t msg_nr)
{
if (init_default_key_collection(nr_keys, true))
@@ -943,9 +943,9 @@ static int start_client(const char *tst_name, unsigned int port,
static void end_client(const char *tst_name, int sk, unsigned int nr_keys,
int current_index, int rnext_index,
- struct tcp_ao_counters *start)
+ struct tcp_counters *start)
{
- struct tcp_ao_counters end;
+ struct tcp_counters end;
/* Some application may become dependent on this kernel choice */
if (current_index < 0)
@@ -955,8 +955,8 @@ static void end_client(const char *tst_name, int sk, unsigned int nr_keys,
verify_current_rnext(tst_name, sk,
collection.keys[current_index].client_keyid,
collection.keys[rnext_index].server_keyid);
- if (start && test_get_tcp_ao_counters(sk, &end))
- test_error("test_get_tcp_ao_counters()");
+ if (start && test_get_tcp_counters(sk, &end))
+ test_error("test_get_tcp_counters()");
verify_keys(tst_name, sk, false, false);
synchronize_threads(); /* 4: verify => closed */
close(sk);
@@ -1048,7 +1048,7 @@ static void check_current_back(const char *tst_name, unsigned int port,
unsigned int current_index, unsigned int rnext_index,
unsigned int rotate_to_index)
{
- struct tcp_ao_counters tmp;
+ struct tcp_counters tmp;
int sk;
sk = start_client(tst_name, port, nr_keys, current_index, rnext_index,
@@ -1081,7 +1081,7 @@ static void roll_over_keys(const char *tst_name, unsigned int port,
unsigned int nr_keys, unsigned int rotations,
unsigned int current_index, unsigned int rnext_index)
{
- struct tcp_ao_counters tmp;
+ struct tcp_counters tmp;
unsigned int i;
int sk;
@@ -1102,7 +1102,7 @@ static void roll_over_keys(const char *tst_name, unsigned int port,
if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC)) {
test_fail("verify failed");
close(sk);
- test_tcp_ao_counters_free(&tmp);
+ test_tcp_counters_free(&tmp);
return;
}
verify_current_rnext(tst_name, sk, -1,
@@ -1116,7 +1116,7 @@ static void roll_over_keys(const char *tst_name, unsigned int port,
static void try_client_run(const char *tst_name, unsigned int port,
unsigned int nr_keys, int current_index, int rnext_index)
{
- struct tcp_ao_counters tmp;
+ struct tcp_counters tmp;
int sk;
sk = start_client(tst_name, port, nr_keys, current_index, rnext_index,
diff --git a/tools/testing/selftests/net/tcp_ao/lib/aolib.h b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
index d6557940ef1fa6bf52da9efea04156edd9b10b7e..b88d15df76886b336cb5f5966fa1e4b2c8eb4a81 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/aolib.h
+++ b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
@@ -512,7 +512,15 @@ struct tcp_ao_counters {
size_t nr_keys;
struct tcp_ao_key_counters *key_cnts;
};
-extern int test_get_tcp_ao_counters(int sk, struct tcp_ao_counters *out);
+
+struct tcp_counters {
+ struct tcp_ao_counters ao;
+ uint64_t netns_md5_notfound;
+ uint64_t netns_md5_unexpected;
+ uint64_t netns_md5_failure;
+};
+
+extern int test_get_tcp_counters(int sk, struct tcp_counters *out);
#define TEST_CNT_KEY_GOOD BIT(0)
#define TEST_CNT_KEY_BAD BIT(1)
@@ -526,22 +534,29 @@ extern int test_get_tcp_ao_counters(int sk, struct tcp_ao_counters *out);
#define TEST_CNT_NS_KEY_NOT_FOUND BIT(9)
#define TEST_CNT_NS_AO_REQUIRED BIT(10)
#define TEST_CNT_NS_DROPPED_ICMP BIT(11)
+#define TEST_CNT_NS_MD5_NOT_FOUND BIT(12)
+#define TEST_CNT_NS_MD5_UNEXPECTED BIT(13)
+#define TEST_CNT_NS_MD5_FAILURE BIT(14)
typedef uint16_t test_cnt;
#define _for_each_counter(f) \
do { \
/* per-netns */ \
- f(netns_ao_good, TEST_CNT_NS_GOOD); \
- f(netns_ao_bad, TEST_CNT_NS_BAD); \
- f(netns_ao_key_not_found, TEST_CNT_NS_KEY_NOT_FOUND); \
- f(netns_ao_required, TEST_CNT_NS_AO_REQUIRED); \
- f(netns_ao_dropped_icmp, TEST_CNT_NS_DROPPED_ICMP); \
+ f(ao.netns_ao_good, TEST_CNT_NS_GOOD); \
+ f(ao.netns_ao_bad, TEST_CNT_NS_BAD); \
+ f(ao.netns_ao_key_not_found, TEST_CNT_NS_KEY_NOT_FOUND); \
+ f(ao.netns_ao_required, TEST_CNT_NS_AO_REQUIRED); \
+ f(ao.netns_ao_dropped_icmp, TEST_CNT_NS_DROPPED_ICMP); \
/* per-socket */ \
- f(ao_info_pkt_good, TEST_CNT_SOCK_GOOD); \
- f(ao_info_pkt_bad, TEST_CNT_SOCK_BAD); \
- f(ao_info_pkt_key_not_found, TEST_CNT_SOCK_KEY_NOT_FOUND); \
- f(ao_info_pkt_ao_required, TEST_CNT_SOCK_AO_REQUIRED); \
- f(ao_info_pkt_dropped_icmp, TEST_CNT_SOCK_DROPPED_ICMP); \
+ f(ao.ao_info_pkt_good, TEST_CNT_SOCK_GOOD); \
+ f(ao.ao_info_pkt_bad, TEST_CNT_SOCK_BAD); \
+ f(ao.ao_info_pkt_key_not_found, TEST_CNT_SOCK_KEY_NOT_FOUND); \
+ f(ao.ao_info_pkt_ao_required, TEST_CNT_SOCK_AO_REQUIRED); \
+ f(ao.ao_info_pkt_dropped_icmp, TEST_CNT_SOCK_DROPPED_ICMP); \
+ /* non-AO */ \
+ f(netns_md5_notfound, TEST_CNT_NS_MD5_NOT_FOUND); \
+ f(netns_md5_unexpected, TEST_CNT_NS_MD5_UNEXPECTED); \
+ f(netns_md5_failure, TEST_CNT_NS_MD5_FAILURE); \
} while (0)
#define TEST_CNT_AO_GOOD (TEST_CNT_SOCK_GOOD | TEST_CNT_NS_GOOD)
@@ -555,33 +570,37 @@ do { \
#define TEST_CNT_GOOD (TEST_CNT_KEY_GOOD | TEST_CNT_AO_GOOD)
#define TEST_CNT_BAD (TEST_CNT_KEY_BAD | TEST_CNT_AO_BAD)
-extern int test_assert_counters_ao(const char *tst_name,
- struct tcp_ao_counters *before, struct tcp_ao_counters *after,
+extern test_cnt test_cmp_counters(struct tcp_counters *before,
+ struct tcp_counters *after);
+extern int test_assert_counters_sk(const char *tst_name,
+ struct tcp_counters *before, struct tcp_counters *after,
test_cnt expected);
extern int test_assert_counters_key(const char *tst_name,
struct tcp_ao_counters *before, struct tcp_ao_counters *after,
test_cnt expected, int sndid, int rcvid);
-extern void test_tcp_ao_counters_free(struct tcp_ao_counters *cnts);
+extern void test_tcp_counters_free(struct tcp_counters *cnts);
+
/*
- * Frees buffers allocated in test_get_tcp_ao_counters().
+ * Frees buffers allocated in test_get_tcp_counters().
* The function doesn't expect new keys or keys removed between calls
- * to test_get_tcp_ao_counters(). Check key counters manually if they
+ * to test_get_tcp_counters(). Check key counters manually if they
* may change.
*/
static inline int test_assert_counters(const char *tst_name,
- struct tcp_ao_counters *before,
- struct tcp_ao_counters *after,
+ struct tcp_counters *before,
+ struct tcp_counters *after,
test_cnt expected)
{
int ret;
- ret = test_assert_counters_ao(tst_name, before, after, expected);
+ ret = test_assert_counters_sk(tst_name, before, after, expected);
if (ret)
goto out;
- ret = test_assert_counters_key(tst_name, before, after, expected, -1, -1);
+ ret = test_assert_counters_key(tst_name, &before->ao, &after->ao,
+ expected, -1, -1);
out:
- test_tcp_ao_counters_free(before);
- test_tcp_ao_counters_free(after);
+ test_tcp_counters_free(before);
+ test_tcp_counters_free(after);
return ret;
}
diff --git a/tools/testing/selftests/net/tcp_ao/lib/sock.c b/tools/testing/selftests/net/tcp_ao/lib/sock.c
index 2a2586b2ffa2bdf805562255e597ccd09597b33c..7ffde4dd794219cd43e056da3394deaea9696971 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/sock.c
+++ b/tools/testing/selftests/net/tcp_ao/lib/sock.c
@@ -333,12 +333,12 @@ do { \
return 0;
}
-int test_get_tcp_ao_counters(int sk, struct tcp_ao_counters *out)
+int test_get_tcp_counters(int sk, struct tcp_counters *out)
{
struct tcp_ao_getsockopt *key_dump;
socklen_t key_dump_sz = sizeof(*key_dump);
struct tcp_ao_info_opt info = {};
- bool c1, c2, c3, c4, c5;
+ bool c1, c2, c3, c4, c5, c6, c7, c8;
struct netstat *ns;
int err, nr_keys;
@@ -346,25 +346,30 @@ int test_get_tcp_ao_counters(int sk, struct tcp_ao_counters *out)
/* per-netns */
ns = netstat_read();
- out->netns_ao_good = netstat_get(ns, "TCPAOGood", &c1);
- out->netns_ao_bad = netstat_get(ns, "TCPAOBad", &c2);
- out->netns_ao_key_not_found = netstat_get(ns, "TCPAOKeyNotFound", &c3);
- out->netns_ao_required = netstat_get(ns, "TCPAORequired", &c4);
- out->netns_ao_dropped_icmp = netstat_get(ns, "TCPAODroppedIcmps", &c5);
+ out->ao.netns_ao_good = netstat_get(ns, "TCPAOGood", &c1);
+ out->ao.netns_ao_bad = netstat_get(ns, "TCPAOBad", &c2);
+ out->ao.netns_ao_key_not_found = netstat_get(ns, "TCPAOKeyNotFound", &c3);
+ out->ao.netns_ao_required = netstat_get(ns, "TCPAORequired", &c4);
+ out->ao.netns_ao_dropped_icmp = netstat_get(ns, "TCPAODroppedIcmps", &c5);
+ out->netns_md5_notfound = netstat_get(ns, "TCPMD5NotFound", &c6);
+ out->netns_md5_unexpected = netstat_get(ns, "TCPMD5Unexpected", &c7);
+ out->netns_md5_failure = netstat_get(ns, "TCPMD5Failure", &c8);
netstat_free(ns);
- if (c1 || c2 || c3 || c4 || c5)
+ if (c1 || c2 || c3 || c4 || c5 || c6 || c7 || c8)
return -EOPNOTSUPP;
err = test_get_ao_info(sk, &info);
+ if (err == -ENOENT)
+ return 0;
if (err)
return err;
/* per-socket */
- out->ao_info_pkt_good = info.pkt_good;
- out->ao_info_pkt_bad = info.pkt_bad;
- out->ao_info_pkt_key_not_found = info.pkt_key_not_found;
- out->ao_info_pkt_ao_required = info.pkt_ao_required;
- out->ao_info_pkt_dropped_icmp = info.pkt_dropped_icmp;
+ out->ao.ao_info_pkt_good = info.pkt_good;
+ out->ao.ao_info_pkt_bad = info.pkt_bad;
+ out->ao.ao_info_pkt_key_not_found = info.pkt_key_not_found;
+ out->ao.ao_info_pkt_ao_required = info.pkt_ao_required;
+ out->ao.ao_info_pkt_dropped_icmp = info.pkt_dropped_icmp;
/* per-key */
nr_keys = test_get_ao_keys_nr(sk);
@@ -372,7 +377,7 @@ int test_get_tcp_ao_counters(int sk, struct tcp_ao_counters *out)
return nr_keys;
if (nr_keys == 0)
test_error("test_get_ao_keys_nr() == 0");
- out->nr_keys = (size_t)nr_keys;
+ out->ao.nr_keys = (size_t)nr_keys;
key_dump = calloc(nr_keys, key_dump_sz);
if (!key_dump)
return -errno;
@@ -386,53 +391,54 @@ int test_get_tcp_ao_counters(int sk, struct tcp_ao_counters *out)
return -errno;
}
- out->key_cnts = calloc(nr_keys, sizeof(out->key_cnts[0]));
- if (!out->key_cnts) {
+ out->ao.key_cnts = calloc(nr_keys, sizeof(out->ao.key_cnts[0]));
+ if (!out->ao.key_cnts) {
free(key_dump);
return -errno;
}
while (nr_keys--) {
- out->key_cnts[nr_keys].sndid = key_dump[nr_keys].sndid;
- out->key_cnts[nr_keys].rcvid = key_dump[nr_keys].rcvid;
- out->key_cnts[nr_keys].pkt_good = key_dump[nr_keys].pkt_good;
- out->key_cnts[nr_keys].pkt_bad = key_dump[nr_keys].pkt_bad;
+ out->ao.key_cnts[nr_keys].sndid = key_dump[nr_keys].sndid;
+ out->ao.key_cnts[nr_keys].rcvid = key_dump[nr_keys].rcvid;
+ out->ao.key_cnts[nr_keys].pkt_good = key_dump[nr_keys].pkt_good;
+ out->ao.key_cnts[nr_keys].pkt_bad = key_dump[nr_keys].pkt_bad;
}
free(key_dump);
return 0;
}
-int test_cmp_counters(struct tcp_ao_counters *before, struct tcp_ao_counters *after)
+test_cnt test_cmp_counters(struct tcp_counters *before,
+ struct tcp_counters *after)
{
#define __cmp(cnt, e_cnt) \
do { \
if (before->cnt > after->cnt) \
- return -1; \
+ test_error("counter " __stringify(cnt) " decreased"); \
if (before->cnt != after->cnt) \
ret |= e_cnt; \
} while (0)
- int ret = 0;
+ test_cnt ret = 0;
size_t i;
- if (before->nr_keys != after->nr_keys)
- return -1;
+ if (before->ao.nr_keys != after->ao.nr_keys)
+ test_error("the number of keys has changed");
_for_each_counter(__cmp);
- i = before->nr_keys;
+ i = before->ao.nr_keys;
while (i--) {
- __cmp(key_cnts[i].pkt_good, TEST_CNT_KEY_GOOD);
- __cmp(key_cnts[i].pkt_bad, TEST_CNT_KEY_BAD);
+ __cmp(ao.key_cnts[i].pkt_good, TEST_CNT_KEY_GOOD);
+ __cmp(ao.key_cnts[i].pkt_bad, TEST_CNT_KEY_BAD);
}
#undef __cmp
return ret;
}
-int test_assert_counters_ao(const char *tst_name,
- struct tcp_ao_counters *before,
- struct tcp_ao_counters *after,
+int test_assert_counters_sk(const char *tst_name,
+ struct tcp_counters *before,
+ struct tcp_counters *after,
test_cnt expected)
{
#define __cmp_ao(cnt, e_cnt) \
@@ -503,9 +509,9 @@ do { \
#undef __cmp_ao
}
-void test_tcp_ao_counters_free(struct tcp_ao_counters *cnts)
+void test_tcp_counters_free(struct tcp_counters *cnts)
{
- free(cnts->key_cnts);
+ free(cnts->ao.key_cnts);
}
#define TEST_BUF_SIZE 4096
diff --git a/tools/testing/selftests/net/tcp_ao/restore.c b/tools/testing/selftests/net/tcp_ao/restore.c
index ad8505f77023bcfb0712108c8148f5c94def67b1..4b98f06725ef9e8be7742c951d8c160279bab0da 100644
--- a/tools/testing/selftests/net/tcp_ao/restore.c
+++ b/tools/testing/selftests/net/tcp_ao/restore.c
@@ -17,7 +17,7 @@ static void try_server_run(const char *tst_name, unsigned int port,
fault_t inj, test_cnt cnt_expected)
{
const char *cnt_name = "TCPAOGood";
- struct tcp_ao_counters ao1, ao2;
+ struct tcp_counters cnt1, cnt2;
uint64_t before_cnt, after_cnt;
int sk, lsk;
time_t timeout;
@@ -48,8 +48,8 @@ static void try_server_run(const char *tst_name, unsigned int port,
}
before_cnt = netstat_get_one(cnt_name, NULL);
- if (test_get_tcp_ao_counters(sk, &ao1))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt1))
+ test_error("test_get_tcp_counters()");
timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC;
bytes = test_server_run(sk, quota, timeout);
@@ -65,17 +65,17 @@ static void try_server_run(const char *tst_name, unsigned int port,
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()");
+ if (test_get_tcp_counters(sk, &cnt2))
+ test_error("test_get_tcp_counters()");
after_cnt = netstat_get_one(cnt_name, NULL);
- test_assert_counters(tst_name, &ao1, &ao2, cnt_expected);
+ test_assert_counters(tst_name, &cnt1, &cnt2, cnt_expected);
if (after_cnt <= before_cnt) {
- test_fail("%s: %s counter did not increase: %" PRIu64 " <= %" PRIu64,
- tst_name, cnt_name, after_cnt, before_cnt);
+ test_fail("%s(server): %s counter did not increase: %" PRIu64 " <= %" PRIu64,
+ tst_name, cnt_name, after_cnt, before_cnt);
} else {
- test_ok("%s: counter %s increased %" PRIu64 " => %" PRIu64,
+ test_ok("%s(server): counter %s increased %" PRIu64 " => %" PRIu64,
tst_name, cnt_name, before_cnt, after_cnt);
}
@@ -92,16 +92,16 @@ static void *server_fn(void *arg)
{
unsigned int port = test_server_port;
- try_server_run("TCP-AO migrate to another socket", port++,
+ try_server_run("TCP-AO migrate to another socket (server)", port++,
0, TEST_CNT_GOOD);
- try_server_run("TCP-AO with wrong send ISN", port++,
+ try_server_run("TCP-AO with wrong send ISN (server)", port++,
FAULT_TIMEOUT, TEST_CNT_BAD);
- try_server_run("TCP-AO with wrong receive ISN", port++,
+ try_server_run("TCP-AO with wrong receive ISN (server)", port++,
FAULT_TIMEOUT, TEST_CNT_BAD);
- try_server_run("TCP-AO with wrong send SEQ ext number", port++,
+ try_server_run("TCP-AO with wrong send SEQ ext number (server)", port++,
FAULT_TIMEOUT, TEST_CNT_BAD);
- try_server_run("TCP-AO with wrong receive SEQ ext number", port++,
- FAULT_TIMEOUT, TEST_CNT_NS_BAD | TEST_CNT_GOOD);
+ try_server_run("TCP-AO with wrong receive SEQ ext number (server)",
+ port++, FAULT_TIMEOUT, TEST_CNT_NS_BAD | TEST_CNT_GOOD);
synchronize_threads(); /* don't race to exit: client exits */
return NULL;
@@ -140,7 +140,7 @@ static void test_sk_restore(const char *tst_name, unsigned int server_port,
fault_t inj, test_cnt cnt_expected)
{
const char *cnt_name = "TCPAOGood";
- struct tcp_ao_counters ao1, ao2;
+ struct tcp_counters cnt1, cnt2;
uint64_t before_cnt, after_cnt;
time_t timeout;
int sk;
@@ -159,8 +159,8 @@ static void test_sk_restore(const char *tst_name, unsigned int server_port,
test_error("setsockopt(TCP_AO_ADD_KEY)");
test_ao_restore(sk, ao_img);
- if (test_get_tcp_ao_counters(sk, &ao1))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt1))
+ test_error("test_get_tcp_counters()");
test_disable_repair(sk);
test_sock_state_free(img);
@@ -178,11 +178,11 @@ static void test_sk_restore(const char *tst_name, unsigned int server_port,
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()");
+ if (test_get_tcp_counters(sk, &cnt2))
+ test_error("test_get_tcp_counters()");
after_cnt = netstat_get_one(cnt_name, NULL);
- test_assert_counters(tst_name, &ao1, &ao2, cnt_expected);
+ test_assert_counters(tst_name, &cnt1, &cnt2, cnt_expected);
if (after_cnt <= before_cnt) {
test_fail("%s: %s counter did not increase: %" PRIu64 " <= %" PRIu64,
@@ -203,7 +203,7 @@ static void *client_fn(void *arg)
sockaddr_af saddr;
test_get_sk_checkpoint(port, &saddr, &tcp_img, &ao_img);
- test_sk_restore("TCP-AO migrate to another socket", port++,
+ test_sk_restore("TCP-AO migrate to another socket (client)", port++,
&saddr, &tcp_img, &ao_img, 0, TEST_CNT_GOOD);
test_get_sk_checkpoint(port, &saddr, &tcp_img, &ao_img);
@@ -212,7 +212,7 @@ static void *client_fn(void *arg)
-1, port, 0, -1, -1, -1, -1, -1, 100, 100, -1);
trace_ao_event_expect(TCP_AO_MISMATCH, this_ip_dest, this_ip_addr,
port, -1, 0, -1, -1, -1, -1, -1, 100, 100, -1);
- test_sk_restore("TCP-AO with wrong send ISN", port++,
+ test_sk_restore("TCP-AO with wrong send ISN (client)", port++,
&saddr, &tcp_img, &ao_img, FAULT_TIMEOUT, TEST_CNT_BAD);
test_get_sk_checkpoint(port, &saddr, &tcp_img, &ao_img);
@@ -221,7 +221,7 @@ static void *client_fn(void *arg)
-1, port, 0, -1, -1, -1, -1, -1, 100, 100, -1);
trace_ao_event_expect(TCP_AO_MISMATCH, this_ip_dest, this_ip_addr,
port, -1, 0, -1, -1, -1, -1, -1, 100, 100, -1);
- test_sk_restore("TCP-AO with wrong receive ISN", port++,
+ test_sk_restore("TCP-AO with wrong receive ISN (client)", port++,
&saddr, &tcp_img, &ao_img, FAULT_TIMEOUT, TEST_CNT_BAD);
test_get_sk_checkpoint(port, &saddr, &tcp_img, &ao_img);
@@ -229,8 +229,8 @@ static void *client_fn(void *arg)
trace_ao_event_expect(TCP_AO_MISMATCH, this_ip_addr, this_ip_dest,
-1, port, 0, -1, -1, -1, -1, -1, 100, 100, -1);
/* not expecting server => client mismatches as only snd sne is broken */
- test_sk_restore("TCP-AO with wrong send SEQ ext number", port++,
- &saddr, &tcp_img, &ao_img, FAULT_TIMEOUT,
+ test_sk_restore("TCP-AO with wrong send SEQ ext number (client)",
+ port++, &saddr, &tcp_img, &ao_img, FAULT_TIMEOUT,
TEST_CNT_NS_BAD | TEST_CNT_GOOD);
test_get_sk_checkpoint(port, &saddr, &tcp_img, &ao_img);
@@ -238,8 +238,8 @@ static void *client_fn(void *arg)
/* not expecting client => server mismatches as only rcv sne is broken */
trace_ao_event_expect(TCP_AO_MISMATCH, this_ip_dest, this_ip_addr,
port, -1, 0, -1, -1, -1, -1, -1, 100, 100, -1);
- test_sk_restore("TCP-AO with wrong receive SEQ ext number", port++,
- &saddr, &tcp_img, &ao_img, FAULT_TIMEOUT,
+ test_sk_restore("TCP-AO with wrong receive SEQ ext number (client)",
+ port++, &saddr, &tcp_img, &ao_img, FAULT_TIMEOUT,
TEST_CNT_NS_GOOD | TEST_CNT_BAD);
return NULL;
diff --git a/tools/testing/selftests/net/tcp_ao/rst.c b/tools/testing/selftests/net/tcp_ao/rst.c
index 9cd257a995bb35fafad868faf0f325168cea5572..f752df511ba2e646384cf02ccae3c2f807ebaed1 100644
--- a/tools/testing/selftests/net/tcp_ao/rst.c
+++ b/tools/testing/selftests/net/tcp_ao/rst.c
@@ -84,15 +84,15 @@ static void close_forced(int sk)
static void test_server_active_rst(unsigned int port)
{
- struct tcp_ao_counters cnt1, cnt2;
+ struct tcp_counters cnt1, cnt2;
ssize_t bytes;
int sk, lsk;
lsk = test_listen_socket(this_ip_addr, port, backlog);
if (test_add_key(lsk, DEFAULT_TEST_PASSWORD, this_ip_dest, -1, 100, 100))
test_error("setsockopt(TCP_AO_ADD_KEY)");
- if (test_get_tcp_ao_counters(lsk, &cnt1))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(lsk, &cnt1))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* 1: MKT added */
if (test_wait_fd(lsk, TEST_TIMEOUT_SEC, 0))
@@ -103,8 +103,8 @@ static void test_server_active_rst(unsigned int port)
test_error("accept()");
synchronize_threads(); /* 2: connection accept()ed, another queued */
- if (test_get_tcp_ao_counters(lsk, &cnt2))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(lsk, &cnt2))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* 3: close listen socket */
close(lsk);
@@ -128,7 +128,7 @@ static void test_server_active_rst(unsigned int port)
static void test_server_passive_rst(unsigned int port)
{
- struct tcp_ao_counters ao1, ao2;
+ struct tcp_counters cnt1, cnt2;
int sk, lsk;
ssize_t bytes;
@@ -147,8 +147,8 @@ static void test_server_passive_rst(unsigned int port)
synchronize_threads(); /* 2: accepted => send data */
close(lsk);
- if (test_get_tcp_ao_counters(sk, &ao1))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt1))
+ test_error("test_get_tcp_counters()");
bytes = test_server_run(sk, quota, TEST_TIMEOUT_SEC);
if (bytes != quota) {
@@ -160,12 +160,12 @@ static void test_server_passive_rst(unsigned int port)
synchronize_threads(); /* 3: checkpoint the client */
synchronize_threads(); /* 4: close the server, creating twsk */
- if (test_get_tcp_ao_counters(sk, &ao2))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt2))
+ test_error("test_get_tcp_counters()");
close(sk);
synchronize_threads(); /* 5: restore the socket, send more data */
- test_assert_counters("passive RST server", &ao1, &ao2, TEST_CNT_GOOD);
+ test_assert_counters("passive RST server", &cnt1, &cnt2, TEST_CNT_GOOD);
synchronize_threads(); /* 6: server exits */
}
@@ -323,7 +323,7 @@ static void test_client_active_rst(unsigned int port)
static void test_client_passive_rst(unsigned int port)
{
- struct tcp_ao_counters ao1, ao2;
+ struct tcp_counters cnt1, cnt2;
struct tcp_ao_repair ao_img;
struct tcp_sock_state img;
sockaddr_af saddr;
@@ -397,8 +397,8 @@ static void test_client_passive_rst(unsigned int port)
test_error("setsockopt(TCP_AO_ADD_KEY)");
test_ao_restore(sk, &ao_img);
- if (test_get_tcp_ao_counters(sk, &ao1))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt1))
+ test_error("test_get_tcp_counters()");
test_disable_repair(sk);
test_sock_state_free(&img);
@@ -426,12 +426,12 @@ static void test_client_passive_rst(unsigned int port)
else
test_fail("client sock is yet connected post-seq-adjust");
- if (test_get_tcp_ao_counters(sk, &ao2))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt2))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* 6: server exits */
close(sk);
- test_assert_counters("client passive RST", &ao1, &ao2, TEST_CNT_GOOD);
+ test_assert_counters("client passive RST", &cnt1, &cnt2, TEST_CNT_GOOD);
}
static void *client_fn(void *arg)
diff --git a/tools/testing/selftests/net/tcp_ao/self-connect.c b/tools/testing/selftests/net/tcp_ao/self-connect.c
index ebdf8e5d951f024efa4e41a9b67d6adf7cfa8ac7..93732732c60c03ccdec093de4964cf9f8caa7b0d 100644
--- a/tools/testing/selftests/net/tcp_ao/self-connect.c
+++ b/tools/testing/selftests/net/tcp_ao/self-connect.c
@@ -30,7 +30,7 @@ static void setup_lo_intf(const char *lo_intf)
static void tcp_self_connect(const char *tst, unsigned int port,
bool different_keyids, bool check_restore)
{
- struct tcp_ao_counters before_ao, after_ao;
+ struct tcp_counters before, after;
uint64_t before_aogood, after_aogood;
struct netstat *ns_before, *ns_after;
const size_t nr_packets = 20;
@@ -60,8 +60,8 @@ static void tcp_self_connect(const char *tst, unsigned int port,
ns_before = netstat_read();
before_aogood = netstat_get(ns_before, "TCPAOGood", NULL);
- if (test_get_tcp_ao_counters(sk, &before_ao))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &before))
+ test_error("test_get_tcp_counters()");
if (__test_connect_socket(sk, "lo", (struct sockaddr *)&addr,
sizeof(addr), TEST_TIMEOUT_SEC) < 0) {
@@ -78,8 +78,8 @@ static void tcp_self_connect(const char *tst, unsigned int port,
ns_after = netstat_read();
after_aogood = netstat_get(ns_after, "TCPAOGood", NULL);
- if (test_get_tcp_ao_counters(sk, &after_ao))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &after))
+ test_error("test_get_tcp_counters()");
if (!check_restore) {
/* to debug: netstat_print_diff(ns_before, ns_after); */
netstat_free(ns_before);
@@ -93,7 +93,7 @@ static void tcp_self_connect(const char *tst, unsigned int port,
return;
}
- if (test_assert_counters(tst, &before_ao, &after_ao, TEST_CNT_GOOD)) {
+ if (test_assert_counters(tst, &before, &after, TEST_CNT_GOOD)) {
close(sk);
return;
}
diff --git a/tools/testing/selftests/net/tcp_ao/seq-ext.c b/tools/testing/selftests/net/tcp_ao/seq-ext.c
index ea8a84edbd10f9250e90a44c16dd0f971c6b385b..ec2c93b6dc8e721e25b5fc342ef204ecdcab3642 100644
--- a/tools/testing/selftests/net/tcp_ao/seq-ext.c
+++ b/tools/testing/selftests/net/tcp_ao/seq-ext.c
@@ -40,7 +40,7 @@ static void test_adjust_seqs(struct tcp_sock_state *img,
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)
+ struct tcp_counters *cnt)
{
int sk;
@@ -54,8 +54,8 @@ static int test_sk_restore(struct tcp_sock_state *img,
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()");
+ if (test_get_tcp_counters(sk, cnt))
+ test_error("test_get_tcp_counters()");
test_disable_repair(sk);
test_sock_state_free(img);
@@ -65,7 +65,7 @@ static int test_sk_restore(struct tcp_sock_state *img,
static void *server_fn(void *arg)
{
uint64_t before_good, after_good, after_bad;
- struct tcp_ao_counters ao1, ao2;
+ struct tcp_counters cnt1, cnt2;
struct tcp_sock_state img;
struct tcp_ao_repair ao_img;
sockaddr_af saddr;
@@ -114,7 +114,7 @@ static void *server_fn(void *arg)
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);
+ client_new_port, &cnt1);
trace_ao_event_sne_expect(TCP_AO_SND_SNE_UPDATE, this_ip_addr,
this_ip_dest, test_server_port + 1, client_new_port, 1);
@@ -136,11 +136,11 @@ static void *server_fn(void *arg)
}
synchronize_threads(); /* 6: verify counters after SEQ-number rollover */
- if (test_get_tcp_ao_counters(sk, &ao2))
- test_error("test_get_tcp_ao_counters()");
+ if (test_get_tcp_counters(sk, &cnt2))
+ test_error("test_get_tcp_counters()");
after_good = netstat_get_one("TCPAOGood", NULL);
- test_assert_counters(NULL, &ao1, &ao2, TEST_CNT_GOOD);
+ test_assert_counters(NULL, &cnt1, &cnt2, TEST_CNT_GOOD);
if (after_good <= before_good) {
test_fail("TCPAOGood counter did not increase: %" PRIu64 " <= %" PRIu64,
@@ -173,7 +173,7 @@ static void *server_fn(void *arg)
static void *client_fn(void *arg)
{
uint64_t before_good, after_good, after_bad;
- struct tcp_ao_counters ao1, ao2;
+ struct tcp_counters cnt1, cnt2;
struct tcp_sock_state img;
struct tcp_ao_repair ao_img;
sockaddr_af saddr;
@@ -213,7 +213,7 @@ static void *client_fn(void *arg)
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);
+ test_server_port + 1, &cnt1);
synchronize_threads(); /* 5: verify the connection during SEQ-number rollover */
if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC))
@@ -222,11 +222,11 @@ static void *client_fn(void *arg)
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()");
+ if (test_get_tcp_counters(sk, &cnt2))
+ test_error("test_get_tcp_counters()");
after_good = netstat_get_one("TCPAOGood", NULL);
- test_assert_counters(NULL, &ao1, &ao2, TEST_CNT_GOOD);
+ test_assert_counters(NULL, &cnt1, &cnt2, TEST_CNT_GOOD);
if (after_good <= before_good) {
test_fail("TCPAOGood counter did not increase: %" PRIu64 " <= %" PRIu64,
diff --git a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
index f403ca25bbe6eb0b89f7467dc70b2a5a826838c4..602dc04e3087be428106b0c03667c44a1aa0e53c 100644
--- a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
+++ b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
@@ -41,7 +41,7 @@ static void try_accept(const char *tst_name, unsigned int port,
const char *cnt_name, test_cnt cnt_expected,
int needs_tcp_md5, fault_t inj)
{
- struct tcp_ao_counters ao_cnt1, ao_cnt2;
+ struct tcp_counters cnt1, cnt2;
uint64_t before_cnt = 0, after_cnt = 0; /* silence GCC */
int lsk, err, sk = 0;
time_t timeout;
@@ -63,8 +63,8 @@ static void try_accept(const char *tst_name, unsigned int port,
if (cnt_name)
before_cnt = netstat_get_one(cnt_name, NULL);
- if (ao_addr && test_get_tcp_ao_counters(lsk, &ao_cnt1))
- test_error("test_get_tcp_ao_counters()");
+ if (ao_addr && test_get_tcp_counters(lsk, &cnt1))
+ test_error("test_get_tcp_counters()");
synchronize_threads(); /* preparations done */
@@ -89,8 +89,8 @@ static void try_accept(const char *tst_name, unsigned int port,
}
}
- if (ao_addr && test_get_tcp_ao_counters(lsk, &ao_cnt2))
- test_error("test_get_tcp_ao_counters()");
+ if (ao_addr && test_get_tcp_counters(lsk, &cnt2))
+ test_error("test_get_tcp_counters()");
close(lsk);
if (!cnt_name) {
@@ -108,7 +108,7 @@ static void try_accept(const char *tst_name, unsigned int port,
tst_name, cnt_name, before_cnt, after_cnt);
}
if (ao_addr)
- test_assert_counters(tst_name, &ao_cnt1, &ao_cnt2, cnt_expected);
+ test_assert_counters(tst_name, &cnt1, &cnt2, cnt_expected);
out:
synchronize_threads(); /* test_kill_sk() */
@@ -158,7 +158,7 @@ static void *server_fn(void *arg)
TEST_CNT_GOOD, 0, 0);
try_accept("AO server (INADDR_ANY): MD5 client", port++, NULL, 0,
&addr_any, 0, 0, 100, 100, 0, "TCPMD5Unexpected",
- 0, 1, FAULT_TIMEOUT);
+ TEST_CNT_NS_MD5_UNEXPECTED, 1, FAULT_TIMEOUT);
try_accept("AO server (INADDR_ANY): no sign client", port++, NULL, 0,
&addr_any, 0, 0, 100, 100, 0, "TCPAORequired",
TEST_CNT_AO_REQUIRED, 0, FAULT_TIMEOUT);
@@ -204,10 +204,12 @@ static void *server_fn(void *arg)
100, 100, 0, NULL, 0, 1, 0);
try_accept("AO+MD5 server: MD5 client (misconfig, matching AO)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
- 100, 100, 0, "TCPMD5Unexpected", 0, 1, FAULT_TIMEOUT);
+ 100, 100, 0, "TCPMD5Unexpected",
+ TEST_CNT_NS_MD5_UNEXPECTED, 1, FAULT_TIMEOUT);
try_accept("AO+MD5 server: MD5 client (misconfig, non-matching)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
- 100, 100, 0, "TCPMD5Unexpected", 0, 1, FAULT_TIMEOUT);
+ 100, 100, 0, "TCPMD5Unexpected",
+ TEST_CNT_NS_MD5_UNEXPECTED, 1, FAULT_TIMEOUT);
try_accept("AO+MD5 server: no sign client (unmatched)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "CurrEstab", 0, 1, 0);
@@ -217,7 +219,8 @@ static void *server_fn(void *arg)
TEST_CNT_AO_REQUIRED, 1, FAULT_TIMEOUT);
try_accept("AO+MD5 server: no sign client (misconfig, matching MD5)",
port++, &this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
- 100, 100, 0, "TCPMD5NotFound", 0, 1, FAULT_TIMEOUT);
+ 100, 100, 0, "TCPMD5NotFound",
+ TEST_CNT_NS_MD5_NOT_FOUND, 1, FAULT_TIMEOUT);
try_accept("AO+MD5 server: client with both [TCP-MD5] and TCP-AO keys",
port++, &this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
--
2.42.2
^ permalink raw reply [flat|nested] 15+ messages in thread* [PATCH net 4/7] selftests/net: Add mixed select()+polling mode to TCP-AO tests
2025-03-12 9:10 [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov via B4 Relay
` (2 preceding siblings ...)
2025-03-12 9:10 ` [PATCH net 3/7] selftests/net: Fetch and check TCP-MD5 counters Dmitry Safonov via B4 Relay
@ 2025-03-12 9:10 ` Dmitry Safonov via B4 Relay
2025-03-18 14:47 ` Paolo Abeni
2025-03-12 9:10 ` [PATCH net 5/7] selftests/net: Print the testing side in unsigned-md5 Dmitry Safonov via B4 Relay
` (3 subsequent siblings)
7 siblings, 1 reply; 15+ messages in thread
From: Dmitry Safonov via B4 Relay @ 2025-03-12 9:10 UTC (permalink / raw)
To: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan
Cc: netdev, linux-kselftest, linux-kernel, Dmitry Safonov
From: Dmitry Safonov <0x7f454c46@gmail.com>
Currently, tcp_ao tests have two timeouts: TEST_RETRANSMIT_SEC and
TEST_TIMEOUT_SEC [by default 1 and 5 seconds]. The first one,
TEST_RETRANSMIT_SEC is used for operations that are expected to succeed
in order for a test to pass. It is usually not consumed and exists only
to avoid indefinite test run if the operation didn't complete.
The second one, TEST_RETRANSMIT_SEC exists for the tests that checking
operations, that are expected to fail/timeout. It is shorter as it is
fully consumed, with an expectation that if operation didn't succeed
during that period, it will timeout. And the related test that expects
the timeout is passing. The actual operation failure is then
cross-verified by other means like counters checks.
The issue with TEST_RETRANSMIT_SEC timeout is that 1 second is the exact
initial TCP timeout. So, in case the initial segment gets lost (quite
unlikely on local veth interface between two net namespaces, yet happens
in slow VMs), the retransmission never happens and as a result, the test
is not actually testing the functionality. Which in the end fails
counters checks.
As I want tcp_ao selftests to be fast and finishing in a reasonable
amount of time on manual run, I didn't consider increasing
TEST_RETRANSMIT_SEC.
Rather, initially, BPF_SOCK_OPS_TIMEOUT_INIT looked promising as a lever
to make the initial TCP timeout shorter. But as it's not a socket bpf
attached thing, but sock_ops (attaches to cgroups), the selftests would
have to use libbpf, which I wanted to avoid if not absolutely required.
Instead, use a mixed select() and counters polling mode with the longer
TEST_TIMEOUT_SEC timeout to detect running-away failed tests. It
actually not only allows losing segments and succeeding after
the previous TEST_RETRANSMIT_SEC timeout was consumed, but makes
the tests expecting timeout/failure pass faster.
The only test case taking longer (TEST_TIMEOUT_SEC) now is connect-deny
"wrong snd id", which checks for no key on SYN-ACK for which there is no
counter in the kernel (see tcp_make_synack()). Yet it can be speed up
by poking skpair from the trace event (see trace_tcp_ao_synack_no_key).
Reported-by: Jakub Kicinski <kuba@kernel.org>
Closes: https://lore.kernel.org/netdev/20241205070656.6ef344d7@kernel.org/
Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
---
tools/testing/selftests/net/tcp_ao/connect-deny.c | 32 +++--
tools/testing/selftests/net/tcp_ao/lib/aolib.h | 37 ++++-
tools/testing/selftests/net/tcp_ao/lib/sock.c | 162 +++++++++++++++++++---
tools/testing/selftests/net/tcp_ao/restore.c | 17 ++-
tools/testing/selftests/net/tcp_ao/unsigned-md5.c | 51 ++++---
5 files changed, 232 insertions(+), 67 deletions(-)
diff --git a/tools/testing/selftests/net/tcp_ao/connect-deny.c b/tools/testing/selftests/net/tcp_ao/connect-deny.c
index 919e8d2b91345b453313f25b98cee27b80f1e984..93b61e9a36f1fc77a5fb4d05a5326e0177dd81be 100644
--- a/tools/testing/selftests/net/tcp_ao/connect-deny.c
+++ b/tools/testing/selftests/net/tcp_ao/connect-deny.c
@@ -4,6 +4,7 @@
#include "aolib.h"
#define fault(type) (inj == FAULT_ ## type)
+static volatile int sk_pair;
static inline int test_add_key_maclen(int sk, const char *key, uint8_t maclen,
union tcp_addr in_addr, uint8_t prefix,
@@ -36,8 +37,8 @@ static void try_accept(const char *tst_name, unsigned int port, const char *pwd,
{
struct tcp_counters cnt1, cnt2;
uint64_t before_cnt = 0, after_cnt = 0; /* silence GCC */
+ test_cnt poll_cnt = (cnt_expected == TEST_CNT_GOOD) ? 0 : cnt_expected;
int lsk, err, sk = 0;
- time_t timeout;
lsk = test_listen_socket(this_ip_addr, port, 1);
@@ -51,16 +52,19 @@ static void try_accept(const char *tst_name, unsigned int port, const char *pwd,
synchronize_threads(); /* preparations done */
- timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC;
- err = test_wait_fd(lsk, timeout, 0);
+ err = test_skpair_wait_poll(lsk, 0, poll_cnt, &sk_pair);
if (err == -ETIMEDOUT) {
+ sk_pair = err;
if (!fault(TIMEOUT))
- test_fail("timed out for accept()");
+ test_fail("%s: timed out for accept()", tst_name);
+ } else if (err == -EKEYREJECTED) {
+ if (!fault(KEYREJECT))
+ test_fail("%s: key was rejected", tst_name);
} else if (err < 0) {
- test_error("test_wait_fd()");
+ test_error("test_skpair_wait_poll()");
} else {
if (fault(TIMEOUT))
- test_fail("ready to accept");
+ test_fail("%s: ready to accept", tst_name);
sk = accept(lsk, NULL, NULL);
if (sk < 0) {
@@ -109,7 +113,7 @@ static void *server_fn(void *arg)
try_accept("Non-AO server + AO client", port++, NULL,
this_ip_dest, -1, 100, 100, 0,
- "TCPAOKeyNotFound", 0, FAULT_TIMEOUT);
+ "TCPAOKeyNotFound", TEST_CNT_NS_KEY_NOT_FOUND, FAULT_TIMEOUT);
try_accept("AO server + Non-AO client", port++, DEFAULT_TEST_PASSWORD,
this_ip_dest, -1, 100, 100, 0,
@@ -135,8 +139,9 @@ static void *server_fn(void *arg)
wrong_addr, -1, 100, 100, 0,
"TCPAOKeyNotFound", TEST_CNT_AO_KEY_NOT_FOUND, FAULT_TIMEOUT);
+ /* Key rejected by the other side, failing short through skpair */
try_accept("Client: Wrong addr", port++, NULL,
- this_ip_dest, -1, 100, 100, 0, NULL, 0, FAULT_TIMEOUT);
+ this_ip_dest, -1, 100, 100, 0, NULL, 0, FAULT_KEYREJECT);
try_accept("rcv id != snd id", port++, DEFAULT_TEST_PASSWORD,
this_ip_dest, -1, 200, 100, 0,
@@ -164,7 +169,6 @@ static void try_connect(const char *tst_name, unsigned int port,
test_cnt cnt_expected, fault_t inj)
{
struct tcp_counters cnt1, cnt2;
- time_t timeout;
int sk, ret;
sk = socket(test_family, SOCK_STREAM, IPPROTO_TCP);
@@ -179,11 +183,10 @@ static void try_connect(const char *tst_name, unsigned int port,
synchronize_threads(); /* preparations done */
- timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC;
- ret = _test_connect_socket(sk, this_ip_dest, port, timeout);
-
+ ret = test_skpair_connect_poll(sk, this_ip_dest, port, cnt_expected, &sk_pair);
synchronize_threads(); /* before counter checks */
if (ret < 0) {
+ sk_pair = ret;
if (fault(KEYREJECT) && ret == -EKEYREJECTED) {
test_ok("%s: connect() was prevented", tst_name);
} else if (ret == -ETIMEDOUT && fault(TIMEOUT)) {
@@ -243,6 +246,11 @@ static void *client_fn(void *arg)
try_connect("Wrong rcv id", port++, DEFAULT_TEST_PASSWORD,
this_ip_dest, -1, 100, 100, 0, FAULT_TIMEOUT);
+ /*
+ * XXX: The test doesn't increase any counters, see tcp_make_synack().
+ * Potentially, it can be speed up by setting sk_pair = -ETIMEDOUT
+ * but the price would be increased complexity of the tracer thread.
+ */
trace_ao_event_sk_expect(TCP_AO_SYNACK_NO_KEY, this_ip_dest, addr_any,
port, 0, 100, 100);
try_connect("Wrong snd id", port++, DEFAULT_TEST_PASSWORD,
diff --git a/tools/testing/selftests/net/tcp_ao/lib/aolib.h b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
index b88d15df76886b336cb5f5966fa1e4b2c8eb4a81..3c8c5ac2a878a7a4f5949055bdc2a9f05f3d1de5 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/aolib.h
+++ b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
@@ -331,6 +331,7 @@ static inline int test_listen_socket(const union tcp_addr taddr,
* If set to 0 - kernel will try to retransmit SYN number of times, set in
* /proc/sys/net/ipv4/tcp_syn_retries
* By default set to 1 to make tests pass faster on non-busy machine.
+ * [in process of removal, don't use in new tests]
*/
#ifndef TEST_RETRANSMIT_SEC
#define TEST_RETRANSMIT_SEC 1
@@ -483,8 +484,6 @@ static inline int test_set_ao_flags(int sk, bool ao_required, bool accept_icmps)
}
extern ssize_t test_server_run(int sk, ssize_t quota, time_t timeout_sec);
-extern ssize_t test_client_loop(int sk, char *buf, size_t buf_sz,
- const size_t msg_len, time_t timeout_sec);
extern int test_client_verify(int sk, const size_t msg_len, const size_t nr,
time_t timeout_sec);
@@ -580,6 +579,40 @@ extern int test_assert_counters_key(const char *tst_name,
test_cnt expected, int sndid, int rcvid);
extern void test_tcp_counters_free(struct tcp_counters *cnts);
+/*
+ * Polling for netns and socket counters during select()/connect() and also
+ * client/server messaging. Instead of constant timeout on underlying select(),
+ * check the counters and return early. This allows to pass the tests where
+ * timeout is expected without waiting for that fixing timeout (tests speed-up).
+ * Previously shorter timeouts were used for tests expecting to time out,
+ * but that leaded to sporadic false positives on counter checks failures,
+ * as one second timeouts aren't enough for TCP retransmit.
+ *
+ * Two sides of the socketpair (client/server) should synchronize failures
+ * using a shared variable *err, so that they can detect the other side's
+ * failure.
+ */
+extern int test_skpair_wait_poll(int sk, bool write, test_cnt cond,
+ volatile int *err);
+extern int _test_skpair_connect_poll(int sk, const char *device,
+ void *addr, size_t addr_sz,
+ test_cnt cond, volatile int *err);
+static inline int test_skpair_connect_poll(int sk, const union tcp_addr taddr,
+ unsigned int port,
+ test_cnt cond, volatile int *err)
+{
+ sockaddr_af addr;
+
+ tcp_addr_to_sockaddr_in(&addr, &taddr, htons(port));
+ return _test_skpair_connect_poll(sk, veth_name,
+ (void *)&addr, sizeof(addr), cond, err);
+}
+
+extern int test_skpair_client(int sk, const size_t msg_len, const size_t nr,
+ test_cnt cond, volatile int *err);
+extern int test_skpair_server(int sk, ssize_t quota,
+ test_cnt cond, volatile int *err);
+
/*
* Frees buffers allocated in test_get_tcp_counters().
* The function doesn't expect new keys or keys removed between calls
diff --git a/tools/testing/selftests/net/tcp_ao/lib/sock.c b/tools/testing/selftests/net/tcp_ao/lib/sock.c
index 7ffde4dd794219cd43e056da3394deaea9696971..1befaf0aa98f9de83fa466cd68a24f467c541a2d 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/sock.c
+++ b/tools/testing/selftests/net/tcp_ao/lib/sock.c
@@ -34,10 +34,8 @@ int __test_listen_socket(int backlog, void *addr, size_t addr_sz)
return sk;
}
-int test_wait_fd(int sk, time_t sec, bool write)
+static int __test_wait_fd(int sk, struct timeval *tv, bool write)
{
- struct timeval tv = { .tv_sec = sec };
- struct timeval *ptv = NULL;
fd_set fds, efds;
int ret;
socklen_t slen = sizeof(ret);
@@ -47,14 +45,11 @@ int test_wait_fd(int sk, time_t sec, bool write)
FD_ZERO(&efds);
FD_SET(sk, &efds);
- if (sec)
- ptv = &tv;
-
errno = 0;
if (write)
- ret = select(sk + 1, NULL, &fds, &efds, ptv);
+ ret = select(sk + 1, NULL, &fds, &efds, tv);
else
- ret = select(sk + 1, &fds, NULL, &efds, ptv);
+ ret = select(sk + 1, &fds, NULL, &efds, tv);
if (ret < 0)
return -errno;
if (ret == 0) {
@@ -69,6 +64,52 @@ int test_wait_fd(int sk, time_t sec, bool write)
return 0;
}
+int test_wait_fd(int sk, time_t sec, bool write)
+{
+ struct timeval tv = { .tv_sec = sec, };
+
+ return __test_wait_fd(sk, sec ? &tv : NULL, write);
+}
+
+static bool __skpair_poll_should_stop(int sk, struct tcp_counters *c,
+ test_cnt condition)
+{
+ struct tcp_counters c2;
+ test_cnt diff;
+
+ if (test_get_tcp_counters(sk, &c2))
+ test_error("test_get_tcp_counters()");
+
+ diff = test_cmp_counters(c, &c2);
+ test_tcp_counters_free(&c2);
+ return (diff & condition) == condition;
+}
+
+/* How often wake up and check netns counters & paired (*err) */
+#define POLL_USEC 150
+static int __test_skpair_poll(int sk, bool write, uint64_t timeout,
+ struct tcp_counters *c, test_cnt cond,
+ volatile int *err)
+{
+ uint64_t t;
+
+ for (t = 0; t <= timeout * 1000000; t += POLL_USEC) {
+ struct timeval tv = { .tv_usec = POLL_USEC, };
+ int ret;
+
+ ret = __test_wait_fd(sk, &tv, write);
+ if (ret != -ETIMEDOUT)
+ return ret;
+ if (c && cond && __skpair_poll_should_stop(sk, c, cond))
+ break;
+ if (err && *err)
+ return *err;
+ }
+ if (err)
+ *err = -ETIMEDOUT;
+ return -ETIMEDOUT;
+}
+
int __test_connect_socket(int sk, const char *device,
void *addr, size_t addr_sz, time_t timeout)
{
@@ -113,6 +154,43 @@ int __test_connect_socket(int sk, const char *device,
return err;
}
+int test_skpair_wait_poll(int sk, bool write,
+ test_cnt cond, volatile int *err)
+{
+ struct tcp_counters c;
+ int ret;
+
+ *err = 0;
+ if (test_get_tcp_counters(sk, &c))
+ test_error("test_get_tcp_counters()");
+ synchronize_threads(); /* 1: init skpair & read nscounters */
+
+ ret = __test_skpair_poll(sk, write, TEST_TIMEOUT_SEC, &c, cond, err);
+ test_tcp_counters_free(&c);
+ return ret;
+}
+
+int _test_skpair_connect_poll(int sk, const char *device,
+ void *addr, size_t addr_sz,
+ test_cnt condition, volatile int *err)
+{
+ struct tcp_counters c;
+ int ret;
+
+ *err = 0;
+ if (test_get_tcp_counters(sk, &c))
+ test_error("test_get_tcp_counters()");
+ synchronize_threads(); /* 1: init skpair & read nscounters */
+ ret = __test_connect_socket(sk, device, addr, addr_sz, -1);
+ if (ret < 0) {
+ test_tcp_counters_free(&c);
+ return (*err = ret);
+ }
+ ret = __test_skpair_poll(sk, 1, TEST_TIMEOUT_SEC, &c, condition, err);
+ test_tcp_counters_free(&c);
+ return ret;
+}
+
int __test_set_md5(int sk, void *addr, size_t addr_sz, uint8_t prefix,
int vrf, const char *password)
{
@@ -515,7 +593,9 @@ void test_tcp_counters_free(struct tcp_counters *cnts)
}
#define TEST_BUF_SIZE 4096
-ssize_t test_server_run(int sk, ssize_t quota, time_t timeout_sec)
+static ssize_t _test_server_run(int sk, ssize_t quota, struct tcp_counters *c,
+ test_cnt cond, volatile int *err,
+ time_t timeout_sec)
{
ssize_t total = 0;
@@ -524,7 +604,7 @@ ssize_t test_server_run(int sk, ssize_t quota, time_t timeout_sec)
ssize_t bytes, sent;
int ret;
- ret = test_wait_fd(sk, timeout_sec, 0);
+ ret = __test_skpair_poll(sk, 0, timeout_sec, c, cond, err);
if (ret)
return ret;
@@ -535,7 +615,7 @@ ssize_t test_server_run(int sk, ssize_t quota, time_t timeout_sec)
if (bytes == 0)
break;
- ret = test_wait_fd(sk, timeout_sec, 1);
+ ret = __test_skpair_poll(sk, 1, timeout_sec, c, cond, err);
if (ret)
return ret;
@@ -550,13 +630,41 @@ ssize_t test_server_run(int sk, ssize_t quota, time_t timeout_sec)
return total;
}
-ssize_t test_client_loop(int sk, char *buf, size_t buf_sz,
- const size_t msg_len, time_t timeout_sec)
+ssize_t test_server_run(int sk, ssize_t quota, time_t timeout_sec)
+{
+ return _test_server_run(sk, quota, NULL, 0, NULL,
+ timeout_sec ?: TEST_TIMEOUT_SEC);
+}
+
+int test_skpair_server(int sk, ssize_t quota, test_cnt cond, volatile int *err)
+{
+ struct tcp_counters c;
+ ssize_t ret;
+
+ *err = 0;
+ if (test_get_tcp_counters(sk, &c))
+ test_error("test_get_tcp_counters()");
+ synchronize_threads(); /* 1: init skpair & read nscounters */
+
+ ret = _test_server_run(sk, quota, &c, cond, err, TEST_TIMEOUT_SEC);
+ test_tcp_counters_free(&c);
+ return ret;
+}
+
+static ssize_t test_client_loop(int sk, size_t buf_sz, const size_t msg_len,
+ struct tcp_counters *c, test_cnt cond,
+ volatile int *err, time_t timeout_sec)
{
char msg[msg_len];
int nodelay = 1;
+ char *buf;
size_t i;
+ buf = alloca(buf_sz);
+ if (!buf)
+ return -ENOMEM;
+ randomize_buffer(buf, buf_sz);
+
if (setsockopt(sk, IPPROTO_TCP, TCP_NODELAY, &nodelay, sizeof(nodelay)))
test_error("setsockopt(TCP_NODELAY)");
@@ -564,7 +672,7 @@ ssize_t test_client_loop(int sk, char *buf, size_t buf_sz,
size_t sent, bytes = min(msg_len, buf_sz - i);
int ret;
- ret = test_wait_fd(sk, timeout_sec, 1);
+ ret = __test_skpair_poll(sk, 1, timeout_sec, c, cond, err);
if (ret)
return ret;
@@ -578,7 +686,7 @@ ssize_t test_client_loop(int sk, char *buf, size_t buf_sz,
do {
ssize_t got;
- ret = test_wait_fd(sk, timeout_sec, 0);
+ ret = __test_skpair_poll(sk, 0, timeout_sec, c, cond, err);
if (ret)
return ret;
@@ -601,11 +709,29 @@ int test_client_verify(int sk, const size_t msg_len, const size_t nr,
time_t timeout_sec)
{
size_t buf_sz = msg_len * nr;
- char *buf = alloca(buf_sz);
ssize_t ret;
- randomize_buffer(buf, buf_sz);
- ret = test_client_loop(sk, buf, buf_sz, msg_len, timeout_sec);
+ ret = test_client_loop(sk, buf_sz, msg_len, NULL, 0, NULL, timeout_sec);
+ if (ret < 0)
+ return (int)ret;
+ return ret != buf_sz ? -1 : 0;
+}
+
+int test_skpair_client(int sk, const size_t msg_len, const size_t nr,
+ test_cnt cond, volatile int *err)
+{
+ struct tcp_counters c;
+ size_t buf_sz = msg_len * nr;
+ ssize_t ret;
+
+ *err = 0;
+ if (test_get_tcp_counters(sk, &c))
+ test_error("test_get_tcp_counters()");
+ synchronize_threads(); /* 1: init skpair & read nscounters */
+
+ ret = test_client_loop(sk, buf_sz, msg_len, &c, cond, err,
+ TEST_TIMEOUT_SEC);
+ test_tcp_counters_free(&c);
if (ret < 0)
return (int)ret;
return ret != buf_sz ? -1 : 0;
diff --git a/tools/testing/selftests/net/tcp_ao/restore.c b/tools/testing/selftests/net/tcp_ao/restore.c
index 4b98f06725ef9e8be7742c951d8c160279bab0da..1c7264a5498a693b246c11414cdc20aa73c55117 100644
--- a/tools/testing/selftests/net/tcp_ao/restore.c
+++ b/tools/testing/selftests/net/tcp_ao/restore.c
@@ -16,11 +16,11 @@ const size_t quota = nr_packets * msg_len;
static void try_server_run(const char *tst_name, unsigned int port,
fault_t inj, test_cnt cnt_expected)
{
+ test_cnt poll_cnt = (cnt_expected == TEST_CNT_GOOD) ? 0 : cnt_expected;
const char *cnt_name = "TCPAOGood";
struct tcp_counters cnt1, cnt2;
uint64_t before_cnt, after_cnt;
- int sk, lsk;
- time_t timeout;
+ int sk, lsk, dummy;
ssize_t bytes;
if (fault(TIMEOUT))
@@ -51,8 +51,7 @@ static void try_server_run(const char *tst_name, unsigned int port,
if (test_get_tcp_counters(sk, &cnt1))
test_error("test_get_tcp_counters()");
- timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC;
- bytes = test_server_run(sk, quota, timeout);
+ bytes = test_skpair_server(sk, quota, poll_cnt, &dummy);
if (fault(TIMEOUT)) {
if (bytes > 0)
test_fail("%s: server served: %zd", tst_name, bytes);
@@ -139,11 +138,11 @@ static void test_sk_restore(const char *tst_name, unsigned int server_port,
struct tcp_ao_repair *ao_img,
fault_t inj, test_cnt cnt_expected)
{
+ test_cnt poll_cnt = (cnt_expected == TEST_CNT_GOOD) ? 0 : cnt_expected;
const char *cnt_name = "TCPAOGood";
struct tcp_counters cnt1, cnt2;
uint64_t before_cnt, after_cnt;
- time_t timeout;
- int sk;
+ int sk, dummy;
if (fault(TIMEOUT))
cnt_name = "TCPAOBad";
@@ -165,18 +164,18 @@ static void test_sk_restore(const char *tst_name, unsigned int server_port,
test_disable_repair(sk);
test_sock_state_free(img);
- timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC;
- if (test_client_verify(sk, msg_len, nr_packets, timeout)) {
+ if (test_skpair_client(sk, msg_len, nr_packets, poll_cnt, &dummy)) {
if (fault(TIMEOUT))
test_ok("%s: post-migrate connection is broken", tst_name);
else
test_fail("%s: post-migrate connection is working", tst_name);
} else {
if (fault(TIMEOUT))
- test_fail("%s: post-migrate connection still working", tst_name);
+ test_fail("%s: post-migrate connection is working", tst_name);
else
test_ok("%s: post-migrate connection is alive", tst_name);
}
+
synchronize_threads(); /* 3: counters checks */
if (test_get_tcp_counters(sk, &cnt2))
test_error("test_get_tcp_counters()");
diff --git a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
index 602dc04e3087be428106b0c03667c44a1aa0e53c..4cde2c9f7f49c9ae5a7e1bf73745ef77c7b4706a 100644
--- a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
+++ b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
@@ -6,6 +6,7 @@
#define fault(type) (inj == FAULT_ ## type)
static const char *md5_password = "Some evil genius, enemy to mankind, must have been the first contriver.";
static const char *ao_password = DEFAULT_TEST_PASSWORD;
+static volatile int sk_pair;
static union tcp_addr client2;
static union tcp_addr client3;
@@ -43,8 +44,8 @@ static void try_accept(const char *tst_name, unsigned int port,
{
struct tcp_counters cnt1, cnt2;
uint64_t before_cnt = 0, after_cnt = 0; /* silence GCC */
- int lsk, err, sk = 0;
- time_t timeout;
+ test_cnt poll_cnt = (cnt_expected == TEST_CNT_GOOD) ? 0 : cnt_expected;
+ int lsk, err, sk = -1;
if (needs_tcp_md5 && should_skip_test(tst_name, KCONFIG_TCP_MD5))
return;
@@ -68,17 +69,20 @@ static void try_accept(const char *tst_name, unsigned int port,
synchronize_threads(); /* preparations done */
- timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC;
- err = test_wait_fd(lsk, timeout, 0);
+ err = test_skpair_wait_poll(lsk, 0, poll_cnt, &sk_pair);
synchronize_threads(); /* connect()/accept() timeouts */
if (err == -ETIMEDOUT) {
+ sk_pair = err;
if (!fault(TIMEOUT))
- test_fail("timed out for accept()");
+ test_fail("%s: timed out for accept()", tst_name);
+ } else if (err == -EKEYREJECTED) {
+ if (!fault(KEYREJECT))
+ test_fail("%s: key was rejected", tst_name);
} else if (err < 0) {
- test_error("test_wait_fd()");
+ test_error("test_skpair_wait_poll()");
} else {
if (fault(TIMEOUT))
- test_fail("ready to accept");
+ test_fail("%s: ready to accept", tst_name);
sk = accept(lsk, NULL, NULL);
if (sk < 0) {
@@ -112,7 +116,7 @@ static void try_accept(const char *tst_name, unsigned int port,
out:
synchronize_threads(); /* test_kill_sk() */
- if (sk > 0)
+ if (sk >= 0)
test_kill_sk(sk);
}
@@ -172,19 +176,19 @@ static void *server_fn(void *arg)
try_accept("MD5 server (INADDR_ANY): AO client", port++, &addr_any, 0,
NULL, 0, 0, 0, 0, 0, "TCPAOKeyNotFound",
- 0, 1, FAULT_TIMEOUT);
+ TEST_CNT_NS_KEY_NOT_FOUND, 1, FAULT_TIMEOUT);
try_accept("MD5 server (INADDR_ANY): MD5 client", port++, &addr_any, 0,
NULL, 0, 0, 0, 0, 0, NULL, 0, 1, 0);
try_accept("MD5 server (INADDR_ANY): no sign client", port++, &addr_any,
0, NULL, 0, 0, 0, 0, 0, "TCPMD5NotFound",
- 0, 1, FAULT_TIMEOUT);
+ TEST_CNT_NS_MD5_NOT_FOUND, 1, FAULT_TIMEOUT);
try_accept("no sign server: AO client", port++, NULL, 0,
NULL, 0, 0, 0, 0, 0, "TCPAOKeyNotFound",
- TEST_CNT_AO_KEY_NOT_FOUND, 0, FAULT_TIMEOUT);
+ TEST_CNT_NS_KEY_NOT_FOUND, 0, FAULT_TIMEOUT);
try_accept("no sign server: MD5 client", port++, NULL, 0,
NULL, 0, 0, 0, 0, 0, "TCPMD5Unexpected",
- 0, 1, FAULT_TIMEOUT);
+ TEST_CNT_NS_MD5_UNEXPECTED, 1, FAULT_TIMEOUT);
try_accept("no sign server: no sign client", port++, NULL, 0,
NULL, 0, 0, 0, 0, 0, "CurrEstab", 0, 0, 0);
@@ -222,12 +226,13 @@ static void *server_fn(void *arg)
100, 100, 0, "TCPMD5NotFound",
TEST_CNT_NS_MD5_NOT_FOUND, 1, FAULT_TIMEOUT);
+ /* Key rejected by the other side, failing short through skpair */
try_accept("AO+MD5 server: client with both [TCP-MD5] and TCP-AO keys",
port++, &this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
- 100, 100, 0, NULL, 0, 1, FAULT_TIMEOUT);
+ 100, 100, 0, NULL, 0, 1, FAULT_KEYREJECT);
try_accept("AO+MD5 server: client with both TCP-MD5 and [TCP-AO] keys",
port++, &this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
- 100, 100, 0, NULL, 0, 1, FAULT_TIMEOUT);
+ 100, 100, 0, NULL, 0, 1, FAULT_KEYREJECT);
server_add_fail_tests(&port);
@@ -262,7 +267,6 @@ static void try_connect(const char *tst_name, unsigned int port,
uint8_t sndid, uint8_t rcvid, uint8_t vrf,
fault_t inj, int needs_tcp_md5, union tcp_addr *bind_addr)
{
- time_t timeout;
int sk, ret;
if (needs_tcp_md5 && should_skip_test(tst_name, KCONFIG_TCP_MD5))
@@ -284,11 +288,10 @@ static void try_connect(const char *tst_name, unsigned int port,
synchronize_threads(); /* preparations done */
- timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC;
- ret = _test_connect_socket(sk, this_ip_dest, port, timeout);
-
+ ret = test_skpair_connect_poll(sk, this_ip_dest, port, 0, &sk_pair);
synchronize_threads(); /* connect()/accept() timeouts */
if (ret < 0) {
+ sk_pair = ret;
if (fault(KEYREJECT) && ret == -EKEYREJECTED)
test_ok("%s: connect() was prevented", tst_name);
else if (ret == -ETIMEDOUT && fault(TIMEOUT))
@@ -308,8 +311,7 @@ static void try_connect(const char *tst_name, unsigned int port,
out:
synchronize_threads(); /* test_kill_sk() */
- /* _test_connect_socket() cleans up on failure */
- if (ret > 0)
+ if (ret > 0) /* test_skpair_connect_poll() cleans up on failure */
test_kill_sk(sk);
}
@@ -440,7 +442,6 @@ static void try_to_add(const char *tst_name, unsigned int port,
int ao_vrf, uint8_t sndid, uint8_t rcvid,
int needs_tcp_md5, fault_t inj)
{
- time_t timeout;
int sk, ret;
if (needs_tcp_md5 && should_skip_test(tst_name, KCONFIG_TCP_MD5))
@@ -453,11 +454,10 @@ static void try_to_add(const char *tst_name, unsigned int port,
synchronize_threads(); /* preparations done */
- timeout = fault(TIMEOUT) ? TEST_RETRANSMIT_SEC : TEST_TIMEOUT_SEC;
- ret = _test_connect_socket(sk, this_ip_dest, port, timeout);
+ ret = test_skpair_connect_poll(sk, this_ip_dest, port, 0, &sk_pair);
synchronize_threads(); /* connect()/accept() timeouts */
- if (ret <= 0) {
+ if (ret < 0) {
test_error("%s: connect() returned %d", tst_name, ret);
goto out;
}
@@ -493,8 +493,7 @@ static void try_to_add(const char *tst_name, unsigned int port,
out:
synchronize_threads(); /* test_kill_sk() */
- /* _test_connect_socket() cleans up on failure */
- if (ret > 0)
+ if (ret > 0) /* test_skpair_connect_poll() cleans up on failure */
test_kill_sk(sk);
}
--
2.42.2
^ permalink raw reply [flat|nested] 15+ messages in thread* Re: [PATCH net 4/7] selftests/net: Add mixed select()+polling mode to TCP-AO tests
2025-03-12 9:10 ` [PATCH net 4/7] selftests/net: Add mixed select()+polling mode to TCP-AO tests Dmitry Safonov via B4 Relay
@ 2025-03-18 14:47 ` Paolo Abeni
2025-03-18 23:04 ` Dmitry Safonov
0 siblings, 1 reply; 15+ messages in thread
From: Paolo Abeni @ 2025-03-18 14:47 UTC (permalink / raw)
To: Dmitry Safonov
Cc: netdev, linux-kselftest, linux-kernel, David S. Miller,
Eric Dumazet, Jakub Kicinski, Simon Horman, Shuah Khan
On 3/12/25 10:10 AM, Dmitry Safonov wrote:
> Currently, tcp_ao tests have two timeouts: TEST_RETRANSMIT_SEC and
> TEST_TIMEOUT_SEC [by default 1 and 5 seconds]. The first one,
> TEST_RETRANSMIT_SEC is used for operations that are expected to succeed
> in order for a test to pass. It is usually not consumed and exists only
> to avoid indefinite test run if the operation didn't complete.
> The second one, TEST_RETRANSMIT_SEC exists for the tests that checking
> operations, that are expected to fail/timeout. It is shorter as it is
> fully consumed, with an expectation that if operation didn't succeed
> during that period, it will timeout. And the related test that expects
> the timeout is passing. The actual operation failure is then
> cross-verified by other means like counters checks.
>
> The issue with TEST_RETRANSMIT_SEC timeout is that 1 second is the exact
> initial TCP timeout. So, in case the initial segment gets lost (quite
> unlikely on local veth interface between two net namespaces, yet happens
> in slow VMs), the retransmission never happens and as a result, the test
> is not actually testing the functionality. Which in the end fails
> counters checks.
>
> As I want tcp_ao selftests to be fast and finishing in a reasonable
> amount of time on manual run, I didn't consider increasing
> TEST_RETRANSMIT_SEC.
>
> Rather, initially, BPF_SOCK_OPS_TIMEOUT_INIT looked promising as a lever
> to make the initial TCP timeout shorter. But as it's not a socket bpf
> attached thing, but sock_ops (attaches to cgroups), the selftests would
> have to use libbpf, which I wanted to avoid if not absolutely required.
>
> Instead, use a mixed select() and counters polling mode with the longer
> TEST_TIMEOUT_SEC timeout to detect running-away failed tests. It
> actually not only allows losing segments and succeeding after
> the previous TEST_RETRANSMIT_SEC timeout was consumed, but makes
> the tests expecting timeout/failure pass faster.
>
> The only test case taking longer (TEST_TIMEOUT_SEC) now is connect-deny
> "wrong snd id", which checks for no key on SYN-ACK for which there is no
> counter in the kernel (see tcp_make_synack()). Yet it can be speed up
> by poking skpair from the trace event (see trace_tcp_ao_synack_no_key).
>
> Reported-by: Jakub Kicinski <kuba@kernel.org>
> Closes: https://lore.kernel.org/netdev/20241205070656.6ef344d7@kernel.org/
> Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
Could you please provide a suitable Fixes tag here?
Also given a good slices of the patches here are refactor, I think the
whole series could land on net-next - so that we avoid putting a bit of
stuff in the last 6.14-net PR - WDYT?
Thanks,
Paolo
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH net 4/7] selftests/net: Add mixed select()+polling mode to TCP-AO tests
2025-03-18 14:47 ` Paolo Abeni
@ 2025-03-18 23:04 ` Dmitry Safonov
0 siblings, 0 replies; 15+ messages in thread
From: Dmitry Safonov @ 2025-03-18 23:04 UTC (permalink / raw)
To: Paolo Abeni
Cc: netdev, linux-kselftest, linux-kernel, David S. Miller,
Eric Dumazet, Jakub Kicinski, Simon Horman, Shuah Khan
Hi Paolo,
On Tue, 18 Mar 2025 at 14:47, Paolo Abeni <pabeni@redhat.com> wrote:
>
> On 3/12/25 10:10 AM, Dmitry Safonov wrote:
[..]
> > Reported-by: Jakub Kicinski <kuba@kernel.org>
> > Closes: https://lore.kernel.org/netdev/20241205070656.6ef344d7@kernel.org/
> > Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
>
> Could you please provide a suitable Fixes tag here?
>
> Also given a good slices of the patches here are refactor, I think the
> whole series could land on net-next - so that we avoid putting a bit of
> stuff in the last 6.14-net PR - WDYT?
Fair enough, let me resend it for net-next and add a Fixes tag.
Thanks for taking a look,
Dmitry
^ permalink raw reply [flat|nested] 15+ messages in thread
* [PATCH net 5/7] selftests/net: Print the testing side in unsigned-md5
2025-03-12 9:10 [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov via B4 Relay
` (3 preceding siblings ...)
2025-03-12 9:10 ` [PATCH net 4/7] selftests/net: Add mixed select()+polling mode to TCP-AO tests Dmitry Safonov via B4 Relay
@ 2025-03-12 9:10 ` Dmitry Safonov via B4 Relay
2025-03-12 9:11 ` [PATCH net 6/7] selftests/net: Delete timeout from test_connect_socket() Dmitry Safonov via B4 Relay
` (2 subsequent siblings)
7 siblings, 0 replies; 15+ messages in thread
From: Dmitry Safonov via B4 Relay @ 2025-03-12 9:10 UTC (permalink / raw)
To: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan
Cc: netdev, linux-kselftest, linux-kernel, Dmitry Safonov
From: Dmitry Safonov <0x7f454c46@gmail.com>
As both client and server print the same test name on failure or pass,
add "[server]" so that it's more obvious from a log which side printed
"ok" or "not ok".
Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
---
tools/testing/selftests/net/tcp_ao/unsigned-md5.c | 44 +++++++++++------------
1 file changed, 22 insertions(+), 22 deletions(-)
diff --git a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
index 4cde2c9f7f49c9ae5a7e1bf73745ef77c7b4706a..a1467b64390adcb5904c02fcfc378a37e989a8bd 100644
--- a/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
+++ b/tools/testing/selftests/net/tcp_ao/unsigned-md5.c
@@ -157,80 +157,80 @@ static void *server_fn(void *arg)
server_add_routes();
- try_accept("AO server (INADDR_ANY): AO client", port++, NULL, 0,
+ try_accept("[server] AO server (INADDR_ANY): AO client", port++, NULL, 0,
&addr_any, 0, 0, 100, 100, 0, "TCPAOGood",
TEST_CNT_GOOD, 0, 0);
- try_accept("AO server (INADDR_ANY): MD5 client", port++, NULL, 0,
+ try_accept("[server] AO server (INADDR_ANY): MD5 client", port++, NULL, 0,
&addr_any, 0, 0, 100, 100, 0, "TCPMD5Unexpected",
TEST_CNT_NS_MD5_UNEXPECTED, 1, FAULT_TIMEOUT);
- try_accept("AO server (INADDR_ANY): no sign client", port++, NULL, 0,
+ try_accept("[server] AO server (INADDR_ANY): no sign client", port++, NULL, 0,
&addr_any, 0, 0, 100, 100, 0, "TCPAORequired",
TEST_CNT_AO_REQUIRED, 0, FAULT_TIMEOUT);
- try_accept("AO server (AO_REQUIRED): AO client", port++, NULL, 0,
+ try_accept("[server] AO server (AO_REQUIRED): AO client", port++, NULL, 0,
&this_ip_dest, TEST_PREFIX, true,
100, 100, 0, "TCPAOGood", TEST_CNT_GOOD, 0, 0);
- try_accept("AO server (AO_REQUIRED): unsigned client", port++, NULL, 0,
+ try_accept("[server] AO server (AO_REQUIRED): unsigned client", port++, NULL, 0,
&this_ip_dest, TEST_PREFIX, true,
100, 100, 0, "TCPAORequired",
TEST_CNT_AO_REQUIRED, 0, FAULT_TIMEOUT);
- try_accept("MD5 server (INADDR_ANY): AO client", port++, &addr_any, 0,
+ try_accept("[server] MD5 server (INADDR_ANY): AO client", port++, &addr_any, 0,
NULL, 0, 0, 0, 0, 0, "TCPAOKeyNotFound",
TEST_CNT_NS_KEY_NOT_FOUND, 1, FAULT_TIMEOUT);
- try_accept("MD5 server (INADDR_ANY): MD5 client", port++, &addr_any, 0,
+ try_accept("[server] MD5 server (INADDR_ANY): MD5 client", port++, &addr_any, 0,
NULL, 0, 0, 0, 0, 0, NULL, 0, 1, 0);
- try_accept("MD5 server (INADDR_ANY): no sign client", port++, &addr_any,
+ try_accept("[server] MD5 server (INADDR_ANY): no sign client", port++, &addr_any,
0, NULL, 0, 0, 0, 0, 0, "TCPMD5NotFound",
TEST_CNT_NS_MD5_NOT_FOUND, 1, FAULT_TIMEOUT);
- try_accept("no sign server: AO client", port++, NULL, 0,
+ try_accept("[server] no sign server: AO client", port++, NULL, 0,
NULL, 0, 0, 0, 0, 0, "TCPAOKeyNotFound",
TEST_CNT_NS_KEY_NOT_FOUND, 0, FAULT_TIMEOUT);
- try_accept("no sign server: MD5 client", port++, NULL, 0,
+ try_accept("[server] no sign server: MD5 client", port++, NULL, 0,
NULL, 0, 0, 0, 0, 0, "TCPMD5Unexpected",
TEST_CNT_NS_MD5_UNEXPECTED, 1, FAULT_TIMEOUT);
- try_accept("no sign server: no sign client", port++, NULL, 0,
+ try_accept("[server] no sign server: no sign client", port++, NULL, 0,
NULL, 0, 0, 0, 0, 0, "CurrEstab", 0, 0, 0);
- try_accept("AO+MD5 server: AO client (matching)", port++,
+ try_accept("[server] AO+MD5 server: AO client (matching)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "TCPAOGood", TEST_CNT_GOOD, 1, 0);
- try_accept("AO+MD5 server: AO client (misconfig, matching MD5)", port++,
+ try_accept("[server] AO+MD5 server: AO client (misconfig, matching MD5)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "TCPAOKeyNotFound", TEST_CNT_AO_KEY_NOT_FOUND,
1, FAULT_TIMEOUT);
- try_accept("AO+MD5 server: AO client (misconfig, non-matching)", port++,
+ try_accept("[server] AO+MD5 server: AO client (misconfig, non-matching)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "TCPAOKeyNotFound", TEST_CNT_AO_KEY_NOT_FOUND,
1, FAULT_TIMEOUT);
- try_accept("AO+MD5 server: MD5 client (matching)", port++,
+ try_accept("[server] AO+MD5 server: MD5 client (matching)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, NULL, 0, 1, 0);
- try_accept("AO+MD5 server: MD5 client (misconfig, matching AO)", port++,
+ try_accept("[server] AO+MD5 server: MD5 client (misconfig, matching AO)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "TCPMD5Unexpected",
TEST_CNT_NS_MD5_UNEXPECTED, 1, FAULT_TIMEOUT);
- try_accept("AO+MD5 server: MD5 client (misconfig, non-matching)", port++,
+ try_accept("[server] AO+MD5 server: MD5 client (misconfig, non-matching)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "TCPMD5Unexpected",
TEST_CNT_NS_MD5_UNEXPECTED, 1, FAULT_TIMEOUT);
- try_accept("AO+MD5 server: no sign client (unmatched)", port++,
+ try_accept("[server] AO+MD5 server: no sign client (unmatched)", port++,
&this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "CurrEstab", 0, 1, 0);
- try_accept("AO+MD5 server: no sign client (misconfig, matching AO)",
+ try_accept("[server] AO+MD5 server: no sign client (misconfig, matching AO)",
port++, &this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "TCPAORequired",
TEST_CNT_AO_REQUIRED, 1, FAULT_TIMEOUT);
- try_accept("AO+MD5 server: no sign client (misconfig, matching MD5)",
+ try_accept("[server] AO+MD5 server: no sign client (misconfig, matching MD5)",
port++, &this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, "TCPMD5NotFound",
TEST_CNT_NS_MD5_NOT_FOUND, 1, FAULT_TIMEOUT);
/* Key rejected by the other side, failing short through skpair */
- try_accept("AO+MD5 server: client with both [TCP-MD5] and TCP-AO keys",
+ try_accept("[server] AO+MD5 server: client with both [TCP-MD5] and TCP-AO keys",
port++, &this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, NULL, 0, 1, FAULT_KEYREJECT);
- try_accept("AO+MD5 server: client with both TCP-MD5 and [TCP-AO] keys",
+ try_accept("[server] AO+MD5 server: client with both TCP-MD5 and [TCP-AO] keys",
port++, &this_ip_dest, TEST_PREFIX, &client2, TEST_PREFIX, 0,
100, 100, 0, NULL, 0, 1, FAULT_KEYREJECT);
--
2.42.2
^ permalink raw reply [flat|nested] 15+ messages in thread* [PATCH net 6/7] selftests/net: Delete timeout from test_connect_socket()
2025-03-12 9:10 [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov via B4 Relay
` (4 preceding siblings ...)
2025-03-12 9:10 ` [PATCH net 5/7] selftests/net: Print the testing side in unsigned-md5 Dmitry Safonov via B4 Relay
@ 2025-03-12 9:11 ` Dmitry Safonov via B4 Relay
2025-03-12 9:11 ` [PATCH net 7/7] selftests/net: Drop timeout argument from test_client_verify() Dmitry Safonov via B4 Relay
2025-03-19 3:20 ` [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov
7 siblings, 0 replies; 15+ messages in thread
From: Dmitry Safonov via B4 Relay @ 2025-03-12 9:11 UTC (permalink / raw)
To: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan
Cc: netdev, linux-kselftest, linux-kernel, Dmitry Safonov
From: Dmitry Safonov <0x7f454c46@gmail.com>
Unused: it's always either the default timeout or asynchronous
connect().
Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
---
tools/testing/selftests/net/tcp_ao/lib/aolib.h | 8 ++++----
tools/testing/selftests/net/tcp_ao/lib/sock.c | 19 ++++++-------------
tools/testing/selftests/net/tcp_ao/rst.c | 5 ++---
tools/testing/selftests/net/tcp_ao/self-connect.c | 2 +-
4 files changed, 13 insertions(+), 21 deletions(-)
diff --git a/tools/testing/selftests/net/tcp_ao/lib/aolib.h b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
index 3c8c5ac2a878a7a4f5949055bdc2a9f05f3d1de5..e4d3d5016b619ef30362bc03802a2455b6aee99c 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/aolib.h
+++ b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
@@ -289,7 +289,7 @@ extern int link_set_up(const char *intf);
extern const unsigned int test_server_port;
extern int test_wait_fd(int sk, time_t sec, bool write);
extern int __test_connect_socket(int sk, const char *device,
- void *addr, size_t addr_sz, time_t timeout);
+ void *addr, size_t addr_sz, bool async);
extern int __test_listen_socket(int backlog, void *addr, size_t addr_sz);
static inline int test_listen_socket(const union tcp_addr taddr,
@@ -338,19 +338,19 @@ static inline int test_listen_socket(const union tcp_addr taddr,
#endif
static inline int _test_connect_socket(int sk, const union tcp_addr taddr,
- unsigned int port, time_t timeout)
+ unsigned int port, bool async)
{
sockaddr_af addr;
tcp_addr_to_sockaddr_in(&addr, &taddr, htons(port));
return __test_connect_socket(sk, veth_name,
- (void *)&addr, sizeof(addr), timeout);
+ (void *)&addr, sizeof(addr), async);
}
static inline int test_connect_socket(int sk, const union tcp_addr taddr,
unsigned int port)
{
- return _test_connect_socket(sk, taddr, port, TEST_TIMEOUT_SEC);
+ return _test_connect_socket(sk, taddr, port, false);
}
extern int __test_set_md5(int sk, void *addr, size_t addr_sz,
diff --git a/tools/testing/selftests/net/tcp_ao/lib/sock.c b/tools/testing/selftests/net/tcp_ao/lib/sock.c
index 1befaf0aa98f9de83fa466cd68a24f467c541a2d..7863ebc36f3c6ced463bd3c8d2c082d5ad152e01 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/sock.c
+++ b/tools/testing/selftests/net/tcp_ao/lib/sock.c
@@ -111,7 +111,7 @@ static int __test_skpair_poll(int sk, bool write, uint64_t timeout,
}
int __test_connect_socket(int sk, const char *device,
- void *addr, size_t addr_sz, time_t timeout)
+ void *addr, size_t addr_sz, bool async)
{
long flags;
int err;
@@ -123,15 +123,6 @@ int __test_connect_socket(int sk, const char *device,
test_error("setsockopt(SO_BINDTODEVICE, %s)", device);
}
- if (!timeout) {
- err = connect(sk, addr, addr_sz);
- if (err) {
- err = -errno;
- goto out;
- }
- return 0;
- }
-
flags = fcntl(sk, F_GETFL);
if ((flags < 0) || (fcntl(sk, F_SETFL, flags | O_NONBLOCK) < 0))
test_error("fcntl()");
@@ -141,9 +132,9 @@ int __test_connect_socket(int sk, const char *device,
err = -errno;
goto out;
}
- if (timeout < 0)
+ if (async)
return sk;
- err = test_wait_fd(sk, timeout, 1);
+ err = test_wait_fd(sk, TEST_TIMEOUT_SEC, 1);
if (err)
goto out;
}
@@ -181,12 +172,14 @@ int _test_skpair_connect_poll(int sk, const char *device,
if (test_get_tcp_counters(sk, &c))
test_error("test_get_tcp_counters()");
synchronize_threads(); /* 1: init skpair & read nscounters */
- ret = __test_connect_socket(sk, device, addr, addr_sz, -1);
+ ret = __test_connect_socket(sk, device, addr, addr_sz, true);
if (ret < 0) {
test_tcp_counters_free(&c);
return (*err = ret);
}
ret = __test_skpair_poll(sk, 1, TEST_TIMEOUT_SEC, &c, condition, err);
+ if (ret < 0)
+ close(sk);
test_tcp_counters_free(&c);
return ret;
}
diff --git a/tools/testing/selftests/net/tcp_ao/rst.c b/tools/testing/selftests/net/tcp_ao/rst.c
index f752df511ba2e646384cf02ccae3c2f807ebaed1..e51be747e2b4f5e0a02f413afd011168a8554302 100644
--- a/tools/testing/selftests/net/tcp_ao/rst.c
+++ b/tools/testing/selftests/net/tcp_ao/rst.c
@@ -271,8 +271,7 @@ static void test_client_active_rst(unsigned int port)
synchronize_threads(); /* 1: MKT added */
for (i = 0; i < last; i++) {
- err = _test_connect_socket(sk[i], this_ip_dest, port,
- (i == 0) ? TEST_TIMEOUT_SEC : -1);
+ err = _test_connect_socket(sk[i], this_ip_dest, port, i != 0);
if (err < 0)
test_error("failed to connect()");
}
@@ -283,7 +282,7 @@ static void test_client_active_rst(unsigned int port)
test_error("test_wait_fds(): %d", err);
/* async connect() with third sk to get into request_sock_queue */
- err = _test_connect_socket(sk[last], this_ip_dest, port, -1);
+ err = _test_connect_socket(sk[last], this_ip_dest, port, 1);
if (err < 0)
test_error("failed to connect()");
diff --git a/tools/testing/selftests/net/tcp_ao/self-connect.c b/tools/testing/selftests/net/tcp_ao/self-connect.c
index 93732732c60c03ccdec093de4964cf9f8caa7b0d..10927cfa23840d31a67e4de34fa2211f9b54cf4b 100644
--- a/tools/testing/selftests/net/tcp_ao/self-connect.c
+++ b/tools/testing/selftests/net/tcp_ao/self-connect.c
@@ -64,7 +64,7 @@ static void tcp_self_connect(const char *tst, unsigned int port,
test_error("test_get_tcp_counters()");
if (__test_connect_socket(sk, "lo", (struct sockaddr *)&addr,
- sizeof(addr), TEST_TIMEOUT_SEC) < 0) {
+ sizeof(addr), 0) < 0) {
ns_after = netstat_read();
netstat_print_diff(ns_before, ns_after);
test_error("failed to connect()");
--
2.42.2
^ permalink raw reply [flat|nested] 15+ messages in thread* [PATCH net 7/7] selftests/net: Drop timeout argument from test_client_verify()
2025-03-12 9:10 [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov via B4 Relay
` (5 preceding siblings ...)
2025-03-12 9:11 ` [PATCH net 6/7] selftests/net: Delete timeout from test_connect_socket() Dmitry Safonov via B4 Relay
@ 2025-03-12 9:11 ` Dmitry Safonov via B4 Relay
2025-03-19 3:20 ` [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov
7 siblings, 0 replies; 15+ messages in thread
From: Dmitry Safonov via B4 Relay @ 2025-03-12 9:11 UTC (permalink / raw)
To: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan
Cc: netdev, linux-kselftest, linux-kernel, Dmitry Safonov
From: Dmitry Safonov <0x7f454c46@gmail.com>
It's always TEST_TIMEOUT_SEC, with an unjustified exception in rst test,
that is more paranoia-long timeout rather than based on requirements.
Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
---
tools/testing/selftests/net/tcp_ao/connect.c | 2 +-
tools/testing/selftests/net/tcp_ao/icmps-discard.c | 3 +--
tools/testing/selftests/net/tcp_ao/key-management.c | 8 ++++----
tools/testing/selftests/net/tcp_ao/lib/aolib.h | 3 +--
tools/testing/selftests/net/tcp_ao/lib/sock.c | 15 +++++++--------
tools/testing/selftests/net/tcp_ao/restore.c | 2 +-
tools/testing/selftests/net/tcp_ao/rst.c | 6 +++---
tools/testing/selftests/net/tcp_ao/self-connect.c | 4 ++--
tools/testing/selftests/net/tcp_ao/seq-ext.c | 4 ++--
9 files changed, 22 insertions(+), 25 deletions(-)
diff --git a/tools/testing/selftests/net/tcp_ao/connect.c b/tools/testing/selftests/net/tcp_ao/connect.c
index 45f2307483d3bab8c96df26c9dca28c898283c0c..340f00e979eaf6a134a61fa679b37bfcdb273323 100644
--- a/tools/testing/selftests/net/tcp_ao/connect.c
+++ b/tools/testing/selftests/net/tcp_ao/connect.c
@@ -53,7 +53,7 @@ static void *client_fn(void *arg)
if (test_get_tcp_counters(sk, &ao1))
test_error("test_get_tcp_counters()");
- if (test_client_verify(sk, 100, nr_packets, TEST_TIMEOUT_SEC)) {
+ if (test_client_verify(sk, 100, nr_packets)) {
test_fail("verify failed");
return NULL;
}
diff --git a/tools/testing/selftests/net/tcp_ao/icmps-discard.c b/tools/testing/selftests/net/tcp_ao/icmps-discard.c
index 61a5e8e2d7e0d7bd1938c297e847384bf7e1d2c5..85c1a1e958c68fd4f2a609f7ae48664149541ff7 100644
--- a/tools/testing/selftests/net/tcp_ao/icmps-discard.c
+++ b/tools/testing/selftests/net/tcp_ao/icmps-discard.c
@@ -395,7 +395,6 @@ static void icmp_interfere(const size_t nr, uint32_t rcv_nxt, void *src, void *d
static void send_interfered(int sk)
{
- const unsigned int timeout = TEST_TIMEOUT_SEC;
struct sockaddr_in6 src, dst;
socklen_t addr_sz;
@@ -409,7 +408,7 @@ static void send_interfered(int sk)
while (1) {
uint32_t rcv_nxt;
- if (test_client_verify(sk, packet_size, packets_nr, timeout)) {
+ if (test_client_verify(sk, packet_size, packets_nr)) {
test_fail("client: connection is broken");
return;
}
diff --git a/tools/testing/selftests/net/tcp_ao/key-management.c b/tools/testing/selftests/net/tcp_ao/key-management.c
index c4087a15bdb3b68b51489a2d419bd911879c6c08..69d9a7a05d5c1a29741dd45afacad18f933bd6f6 100644
--- a/tools/testing/selftests/net/tcp_ao/key-management.c
+++ b/tools/testing/selftests/net/tcp_ao/key-management.c
@@ -918,7 +918,7 @@ static int run_client(const char *tst_name, unsigned int port,
collection.keys[rnext_index].used_on_server_tx = 1;
synchronize_threads(); /* 3: accepted => send data */
- if (test_client_verify(sk, msg_sz, msg_nr, TEST_TIMEOUT_SEC)) {
+ if (test_client_verify(sk, msg_sz, msg_nr)) {
test_fail("verify failed");
close(sk);
if (before)
@@ -1016,7 +1016,7 @@ static void try_unmatched_keys(int sk, int *rnext_index, unsigned int port)
trace_ao_event_expect(TCP_AO_RNEXT_REQUEST, this_ip_addr, this_ip_dest,
-1, port, 0, -1, -1, -1, -1, -1,
-1, key->server_keyid, -1);
- if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC))
+ if (test_client_verify(sk, msg_len, nr_packets))
test_fail("verify failed");
*rnext_index = i;
}
@@ -1061,7 +1061,7 @@ static void check_current_back(const char *tst_name, unsigned int port,
port, -1, 0, -1, -1, -1, -1, -1,
collection.keys[rotate_to_index].client_keyid,
collection.keys[current_index].client_keyid, -1);
- if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC))
+ if (test_client_verify(sk, msg_len, nr_packets))
test_fail("verify failed");
/* There is a race here: between setting the current_key with
* setsockopt(TCP_AO_INFO) and starting to send some data - there
@@ -1099,7 +1099,7 @@ static void roll_over_keys(const char *tst_name, unsigned int port,
collection.keys[i].server_keyid, -1);
if (test_set_key(sk, -1, collection.keys[i].server_keyid))
test_error("Can't change the Rnext key");
- if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC)) {
+ if (test_client_verify(sk, msg_len, nr_packets)) {
test_fail("verify failed");
close(sk);
test_tcp_counters_free(&tmp);
diff --git a/tools/testing/selftests/net/tcp_ao/lib/aolib.h b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
index e4d3d5016b619ef30362bc03802a2455b6aee99c..ebb2899c12fe0abfe8b8292fdf6fa69b800eb8e0 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/aolib.h
+++ b/tools/testing/selftests/net/tcp_ao/lib/aolib.h
@@ -484,8 +484,7 @@ static inline int test_set_ao_flags(int sk, bool ao_required, bool accept_icmps)
}
extern ssize_t test_server_run(int sk, ssize_t quota, time_t timeout_sec);
-extern int test_client_verify(int sk, const size_t msg_len, const size_t nr,
- time_t timeout_sec);
+extern int test_client_verify(int sk, const size_t msg_len, const size_t nr);
struct tcp_ao_key_counters {
uint8_t sndid;
diff --git a/tools/testing/selftests/net/tcp_ao/lib/sock.c b/tools/testing/selftests/net/tcp_ao/lib/sock.c
index 7863ebc36f3c6ced463bd3c8d2c082d5ad152e01..ef8e9031d47a3e91786290e558b5d203cbd126f7 100644
--- a/tools/testing/selftests/net/tcp_ao/lib/sock.c
+++ b/tools/testing/selftests/net/tcp_ao/lib/sock.c
@@ -646,7 +646,7 @@ int test_skpair_server(int sk, ssize_t quota, test_cnt cond, volatile int *err)
static ssize_t test_client_loop(int sk, size_t buf_sz, const size_t msg_len,
struct tcp_counters *c, test_cnt cond,
- volatile int *err, time_t timeout_sec)
+ volatile int *err)
{
char msg[msg_len];
int nodelay = 1;
@@ -665,7 +665,7 @@ static ssize_t test_client_loop(int sk, size_t buf_sz, const size_t msg_len,
size_t sent, bytes = min(msg_len, buf_sz - i);
int ret;
- ret = __test_skpair_poll(sk, 1, timeout_sec, c, cond, err);
+ ret = __test_skpair_poll(sk, 1, TEST_TIMEOUT_SEC, c, cond, err);
if (ret)
return ret;
@@ -679,7 +679,8 @@ static ssize_t test_client_loop(int sk, size_t buf_sz, const size_t msg_len,
do {
ssize_t got;
- ret = __test_skpair_poll(sk, 0, timeout_sec, c, cond, err);
+ ret = __test_skpair_poll(sk, 0, TEST_TIMEOUT_SEC,
+ c, cond, err);
if (ret)
return ret;
@@ -698,13 +699,12 @@ static ssize_t test_client_loop(int sk, size_t buf_sz, const size_t msg_len,
return i;
}
-int test_client_verify(int sk, const size_t msg_len, const size_t nr,
- time_t timeout_sec)
+int test_client_verify(int sk, const size_t msg_len, const size_t nr)
{
size_t buf_sz = msg_len * nr;
ssize_t ret;
- ret = test_client_loop(sk, buf_sz, msg_len, NULL, 0, NULL, timeout_sec);
+ ret = test_client_loop(sk, buf_sz, msg_len, NULL, 0, NULL);
if (ret < 0)
return (int)ret;
return ret != buf_sz ? -1 : 0;
@@ -722,8 +722,7 @@ int test_skpair_client(int sk, const size_t msg_len, const size_t nr,
test_error("test_get_tcp_counters()");
synchronize_threads(); /* 1: init skpair & read nscounters */
- ret = test_client_loop(sk, buf_sz, msg_len, &c, cond, err,
- TEST_TIMEOUT_SEC);
+ ret = test_client_loop(sk, buf_sz, msg_len, &c, cond, err);
test_tcp_counters_free(&c);
if (ret < 0)
return (int)ret;
diff --git a/tools/testing/selftests/net/tcp_ao/restore.c b/tools/testing/selftests/net/tcp_ao/restore.c
index 1c7264a5498a693b246c11414cdc20aa73c55117..9a059b6c45231d118a6f5324afba511b7d339ece 100644
--- a/tools/testing/selftests/net/tcp_ao/restore.c
+++ b/tools/testing/selftests/net/tcp_ao/restore.c
@@ -124,7 +124,7 @@ static void test_get_sk_checkpoint(unsigned int server_port, sockaddr_af *saddr,
test_error("failed to connect()");
synchronize_threads(); /* 2: accepted => send data */
- if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC))
+ if (test_client_verify(sk, msg_len, nr_packets))
test_fail("pre-migrate verify failed");
test_enable_repair(sk);
diff --git a/tools/testing/selftests/net/tcp_ao/rst.c b/tools/testing/selftests/net/tcp_ao/rst.c
index e51be747e2b4f5e0a02f413afd011168a8554302..883cddf377cff899f79220439bc58fe66f5da4d8 100644
--- a/tools/testing/selftests/net/tcp_ao/rst.c
+++ b/tools/testing/selftests/net/tcp_ao/rst.c
@@ -287,7 +287,7 @@ static void test_client_active_rst(unsigned int port)
test_error("failed to connect()");
synchronize_threads(); /* 3: close listen socket */
- if (test_client_verify(sk[0], packet_sz, quota / packet_sz, TEST_TIMEOUT_SEC))
+ if (test_client_verify(sk[0], packet_sz, quota / packet_sz))
test_fail("Failed to send data on connected socket");
else
test_ok("Verified established tcp connection");
@@ -340,7 +340,7 @@ static void test_client_passive_rst(unsigned int port)
test_error("failed to connect()");
synchronize_threads(); /* 2: accepted => send data */
- if (test_client_verify(sk, packet_sz, quota / packet_sz, TEST_TIMEOUT_SEC))
+ if (test_client_verify(sk, packet_sz, quota / packet_sz))
test_fail("Failed to send data on connected socket");
else
test_ok("Verified established tcp connection");
@@ -416,7 +416,7 @@ static void test_client_passive_rst(unsigned int port)
* IP 10.0.254.1.7011 > 10.0.1.1.59772: Flags [R], seq 3215596252, win 0,
* options [tcp-ao keyid 100 rnextkeyid 100 mac 0x0bcfbbf497bce844312304b2], length 0
*/
- err = test_client_verify(sk, packet_sz, quota / packet_sz, 2 * TEST_TIMEOUT_SEC);
+ err = test_client_verify(sk, packet_sz, quota / packet_sz);
/* Make sure that the connection was reset, not timeouted */
if (err && err == -ECONNRESET)
test_ok("client sock was passively reset post-seq-adjust");
diff --git a/tools/testing/selftests/net/tcp_ao/self-connect.c b/tools/testing/selftests/net/tcp_ao/self-connect.c
index 10927cfa23840d31a67e4de34fa2211f9b54cf4b..73b2f2276f3f5410aaa74bede7f366f81761bd6e 100644
--- a/tools/testing/selftests/net/tcp_ao/self-connect.c
+++ b/tools/testing/selftests/net/tcp_ao/self-connect.c
@@ -70,7 +70,7 @@ static void tcp_self_connect(const char *tst, unsigned int port,
test_error("failed to connect()");
}
- if (test_client_verify(sk, 100, nr_packets, TEST_TIMEOUT_SEC)) {
+ if (test_client_verify(sk, 100, nr_packets)) {
test_fail("%s: tcp connection verify failed", tst);
close(sk);
return;
@@ -136,7 +136,7 @@ static void tcp_self_connect(const char *tst, unsigned int port,
test_ao_restore(sk, &ao_img);
test_disable_repair(sk);
test_sock_state_free(&img);
- if (test_client_verify(sk, 100, nr_packets, TEST_TIMEOUT_SEC)) {
+ if (test_client_verify(sk, 100, nr_packets)) {
test_fail("%s: tcp connection verify failed", tst);
close(sk);
return;
diff --git a/tools/testing/selftests/net/tcp_ao/seq-ext.c b/tools/testing/selftests/net/tcp_ao/seq-ext.c
index ec2c93b6dc8e721e25b5fc342ef204ecdcab3642..f00245263b20d95942f49b2fcb7c90604b43f39f 100644
--- a/tools/testing/selftests/net/tcp_ao/seq-ext.c
+++ b/tools/testing/selftests/net/tcp_ao/seq-ext.c
@@ -191,7 +191,7 @@ static void *client_fn(void *arg)
test_error("failed to connect()");
synchronize_threads(); /* 2: accepted => send data */
- if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC)) {
+ if (test_client_verify(sk, msg_len, nr_packets)) {
test_fail("pre-migrate verify failed");
return NULL;
}
@@ -216,7 +216,7 @@ static void *client_fn(void *arg)
test_server_port + 1, &cnt1);
synchronize_threads(); /* 5: verify the connection during SEQ-number rollover */
- if (test_client_verify(sk, msg_len, nr_packets, TEST_TIMEOUT_SEC))
+ if (test_client_verify(sk, msg_len, nr_packets))
test_fail("post-migrate verify failed");
else
test_ok("post-migrate connection alive");
--
2.42.2
^ permalink raw reply [flat|nested] 15+ messages in thread* Re: [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests
2025-03-12 9:10 [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov via B4 Relay
` (6 preceding siblings ...)
2025-03-12 9:11 ` [PATCH net 7/7] selftests/net: Drop timeout argument from test_client_verify() Dmitry Safonov via B4 Relay
@ 2025-03-19 3:20 ` Dmitry Safonov
2025-03-19 3:26 ` Dmitry Safonov
7 siblings, 1 reply; 15+ messages in thread
From: Dmitry Safonov @ 2025-03-19 3:20 UTC (permalink / raw)
To: 0x7f454c46
Cc: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan, netdev, linux-kselftest, linux-kernel
On Wed, 12 Mar 2025 at 09:11, Dmitry Safonov via B4 Relay
<devnull+0x7f454c46.gmail.com@kernel.org> wrote:
>
> Should fix flaky tcp-ao/connect-deny-ipv6 test.
> Begging pardon for the delay since the report and for sending it this
> late in the release cycle.
>
> Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
I think patchwork can't detect v2 to be the same series, so doing it
manually here:
pw-bot: changes-requested
Just in case, v2 link:
https://lore.kernel.org/netdev/20250319-tcp-ao-selftests-polling-v2-0-da48040153d1@gmail.com/
Thanks,
Dmitry
^ permalink raw reply [flat|nested] 15+ messages in thread* Re: [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests
2025-03-19 3:20 ` [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests Dmitry Safonov
@ 2025-03-19 3:26 ` Dmitry Safonov
2025-03-19 13:18 ` Konstantin Ryabitsev
0 siblings, 1 reply; 15+ messages in thread
From: Dmitry Safonov @ 2025-03-19 3:26 UTC (permalink / raw)
To: 0x7f454c46
Cc: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan, netdev, linux-kselftest, linux-kernel
On Wed, 19 Mar 2025 at 03:20, Dmitry Safonov <0x7f454c46@gmail.com> wrote:
>
> On Wed, 12 Mar 2025 at 09:11, Dmitry Safonov via B4 Relay
> <devnull+0x7f454c46.gmail.com@kernel.org> wrote:
> >
> > Should fix flaky tcp-ao/connect-deny-ipv6 test.
> > Begging pardon for the delay since the report and for sending it this
> > late in the release cycle.
> >
> > Signed-off-by: Dmitry Safonov <0x7f454c46@gmail.com>
>
> I think patchwork can't detect v2 to be the same series, so doing it
> manually here:
> pw-bot: changes-requested
* potentially because of the -net vs -net-next subject.
However, as I sent the patches with b4 relay, it seems patchwork is
unable to authorise me as the author of the series. Ugh :(
> Just in case, v2 link:
> https://lore.kernel.org/netdev/20250319-tcp-ao-selftests-polling-v2-0-da48040153d1@gmail.com/
--
Dmitry
^ permalink raw reply [flat|nested] 15+ messages in thread* Re: [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests
2025-03-19 3:26 ` Dmitry Safonov
@ 2025-03-19 13:18 ` Konstantin Ryabitsev
2025-03-20 13:51 ` Konstantin Ryabitsev
0 siblings, 1 reply; 15+ messages in thread
From: Konstantin Ryabitsev @ 2025-03-19 13:18 UTC (permalink / raw)
To: Dmitry Safonov, helpdesk
Cc: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan, netdev, linux-kselftest, linux-kernel
On Wed, Mar 19, 2025 at 03:26:27AM +0000, Dmitry Safonov wrote:
> However, as I sent the patches with b4 relay, it seems patchwork is
> unable to authorise me as the author of the series. Ugh :(
Oops. We should definitely properly handle this.
Sending a note to self via the helpdesk.
-K
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests
2025-03-19 13:18 ` Konstantin Ryabitsev
@ 2025-03-20 13:51 ` Konstantin Ryabitsev
2025-03-21 2:57 ` Dmitry Safonov
0 siblings, 1 reply; 15+ messages in thread
From: Konstantin Ryabitsev @ 2025-03-20 13:51 UTC (permalink / raw)
To: Dmitry Safonov, helpdesk
Cc: David S. Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Simon Horman, Shuah Khan, netdev, linux-kselftest, linux-kernel
On Wed, Mar 19, 2025 at 09:18:37AM -0400, Konstantin Ryabitsev wrote:
> On Wed, Mar 19, 2025 at 03:26:27AM +0000, Dmitry Safonov wrote:
> > However, as I sent the patches with b4 relay, it seems patchwork is
> > unable to authorise me as the author of the series. Ugh :(
>
> Oops. We should definitely properly handle this.
> Sending a note to self via the helpdesk.
This should be handled correctly now, e.g. see this series sent via the relay
yesterday:
https://patchwork.kernel.org/project/netdevbpf/patch/20250319-rfc2203-seqnum-cache-v2-2-2c98b859f2dd@janestreet.com/
vs.
https://lore.kernel.org/netdev/20250319-rfc2203-seqnum-cache-v2-2-2c98b859f2dd@janestreet.com/
-K
^ permalink raw reply [flat|nested] 15+ messages in thread
* Re: [PATCH net 0/7] selftests/net: Mixed select()+polling mode for TCP-AO tests
2025-03-20 13:51 ` Konstantin Ryabitsev
@ 2025-03-21 2:57 ` Dmitry Safonov
0 siblings, 0 replies; 15+ messages in thread
From: Dmitry Safonov @ 2025-03-21 2:57 UTC (permalink / raw)
To: Konstantin Ryabitsev
Cc: helpdesk, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Simon Horman, Shuah Khan, netdev, linux-kselftest,
linux-kernel
On Thu, 20 Mar 2025 at 13:51, Konstantin Ryabitsev
<konstantin@linuxfoundation.org> wrote:
>
> On Wed, Mar 19, 2025 at 09:18:37AM -0400, Konstantin Ryabitsev wrote:
> > On Wed, Mar 19, 2025 at 03:26:27AM +0000, Dmitry Safonov wrote:
> > > However, as I sent the patches with b4 relay, it seems patchwork is
> > > unable to authorise me as the author of the series. Ugh :(
> >
> > Oops. We should definitely properly handle this.
> > Sending a note to self via the helpdesk.
>
> This should be handled correctly now, e.g. see this series sent via the relay
> yesterday:
>
> https://patchwork.kernel.org/project/netdevbpf/patch/20250319-rfc2203-seqnum-cache-v2-2-2c98b859f2dd@janestreet.com/
> vs.
> https://lore.kernel.org/netdev/20250319-rfc2203-seqnum-cache-v2-2-2c98b859f2dd@janestreet.com/
Thank you, Konstantin!
I appreciate your fixes and b4 development.
Thanks,
Dmitry
^ permalink raw reply [flat|nested] 15+ messages in thread