Skip to contentSkip to conversation

selftests: tls: avoid flakiness in data_steal

PATCH Applied Jakub Kicinski 3 participants 3 messages Jan 6 – Jan 10

Subsystems (4)
lore
Share & tools
lore

Jakub Kicinski

author maintainer
#0 lore raw

PATCH 1/1 selftests: tls: avoid flakiness in data_steal

We see the following failure a few times a week:

# RUN global.data_steal ...
# tls.c:3280:data_steal:Expected recv(cfd, buf2, sizeof(buf2), MSG_DONTWAIT) (10000) == -1 (-1) # data_steal: Test failed
# FAIL global.data_steal
not ok 8 global.data_steal

The 10000 bytes read suggests that the child process did a recv() of half of the data using the TLS ULP and we're now getting the remaining half. The intent of the test is to get the child to enter _TCP_ recvmsg handler, so it needs to enter the syscall before parent installed the TLS recvmsg with setsockopt(SOL_TLS).

Instead of the 10msec sleep send 1 byte of data and wait for the child to consume it.

---
CC: [email protected]
CC: [email protected]
CC: [email protected]

---
 tools/testing/selftests/net/tls.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)
tools/testing/selftests/net/tls.c +12 -4
index a4d16a460fbe..9e2ccea13d70 100644
@@ -3260,17 +3260,25 @@ TEST(data_steal) {
 	ASSERT_EQ(setsockopt(cfd, IPPROTO_TCP, TCP_ULP, "tls", sizeof("tls")), 0);
 
 	/* Spawn a child and get it into the read wait path of the underlying
-	 * TCP socket.
+	 * TCP socket (before kernel .recvmsg is replaced with the TLS one).
 	 */
 	pid = fork();
 	ASSERT_GE(pid, 0);
 	if (!pid) {
-		EXPECT_EQ(recv(cfd, buf, sizeof(buf) / 2, MSG_WAITALL),
-			  sizeof(buf) / 2);
+		EXPECT_EQ(recv(cfd, buf, sizeof(buf) / 2 + 1, MSG_WAITALL),
+			  sizeof(buf) / 2 + 1);
 		exit(!__test_passed(_metadata));
 	}
 
-	usleep(10000);
+	/* Send a sync byte and poll until it's consumed to ensure
+	 * the child is in recv() before we proceed to install TLS.
+	 */
+	ASSERT_EQ(send(fd, buf, 1, 0), 1);
+	do {
+		usleep(500);
+	} while (recv(cfd, buf, 1, MSG_PEEK | MSG_DONTWAIT) == 1);
+	EXPECT_EQ(errno, EAGAIN);
+
 	ASSERT_EQ(setsockopt(fd, SOL_TLS, TLS_TX, &tls, tls.len), 0);
 	ASSERT_EQ(setsockopt(cfd, SOL_TLS, TLS_RX, &tls, tls.len), 0);
 
Signature · 1 line
2.52.0
maintainer Reviewed-by on v1 1/1
#1 lore raw
Quoted 17 lines · Jakub Kicinski

We see the following failure a few times a week:

# RUN global.data_steal ...
# tls.c:3280:data_steal:Expected recv(cfd, buf2, sizeof(buf2), MSG_DONTWAIT) (10000) == -1 (-1) # data_steal: Test failed
# FAIL global.data_steal
not ok 8 global.data_steal

The 10000 bytes read suggests that the child process did a recv() of half of the data using the TLS ULP and we're now getting the remaining half. The intent of the test is to get the child to enter _TCP_ recvmsg handler, so it needs to enter the syscall before parent installed the TLS recvmsg with setsockopt(SOL_TLS).

Instead of the 10msec sleep send 1 byte of data and wait for the child to consume it.

Quoted 8 lines

---
CC: [email protected]
CC: [email protected]
CC: [email protected]

---
 tools/testing/selftests/net/tls.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)
Signature · 1 line
Sabrina
#2 lore raw
In reply to Jakub Kicinski
This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <[email protected]>:

Quoted 9 lines · On Tue, 6 Jan 2026 12:02:05 -0800 you wrote:

We see the following failure a few times a week:

# RUN global.data_steal ...
# tls.c:3280:data_steal:Expected recv(cfd, buf2, sizeof(buf2), MSG_DONTWAIT) (10000) == -1 (-1) # data_steal: Test failed
# FAIL global.data_steal
not ok 8 global.data_steal

[...]

Here is the summary with links:
- [net-next] selftests: tls: avoid flakiness in data_steal

    https://git.kernel.org/netdev/net-next/c/96ea4fa60c45

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html