On 11/7/25 08:05, Kuniyuki Iwashima wrote:
From: Sunday Adelodun <[email protected]>
Date: Tue, 4 Nov 2025 11:42:47 +0100
On 11/4/25 01:30, Kuniyuki Iwashima wrote:
On Mon, Nov 3, 2025 at 4:08 PM Sunday Adelodun
<[email protected]> wrote:
On 11/2/25 08:32, Kuniyuki Iwashima wrote:
On Sat, Nov 1, 2025 at 10:23 AM Sunday Adelodun
<[email protected]> wrote:
Add selftests to verify and document Linux’s intended behaviour for
UNIX domain sockets (SOCK_STREAM and SOCK_DGRAM) when a peer closes.
The tests verify that:
1. SOCK_STREAM returns EOF when the peer closes normally.
2. SOCK_STREAM returns ECONNRESET if the peer closes with unread data.
3. SOCK_SEQPACKET returns EOF when the peer closes normally.
4. SOCK_SEQPACKET returns ECONNRESET if the peer closes with unread data.
5. SOCK_DGRAM does not return ECONNRESET when the peer closes.
This follows up on review feedback suggesting a selftest to clarify
Linux’s semantics.
Suggested-by: Kuniyuki Iwashima <[email protected]>
Signed-off-by: Sunday Adelodun <[email protected]>
---
tools/testing/selftests/net/af_unix/Makefile | 1 +
.../selftests/net/af_unix/unix_connreset.c | 179 ++++++++++++++++++
2 files changed, 180 insertions(+)
create mode 100644 tools/testing/selftests/net/af_unix/unix_connreset.c
diff --git a/tools/testing/selftests/net/af_unix/Makefile
b/tools/testing/selftests/net/af_unix/Makefile
index de805cbbdf69..5826a8372451 100644
--- a/tools/testing/selftests/net/af_unix/Makefile
+++ b/tools/testing/selftests/net/af_unix/Makefile
@@ -7,6 +7,7 @@ TEST_GEN_PROGS := \
scm_pidfd \
scm_rights \
unix_connect \
+ unix_connreset \
patchwork caught this test is not added to .gitignore.
https://patchwork.kernel.org/project/netdevbpf/patch/[email protected]/
Could you add it to this file ?
tools/testing/selftests/net/.gitignore
Oh, thank you for this. will add it
# end of TEST_GEN_PROGS
include ../../lib.mk
diff --git a/tools/testing/selftests/net/af_unix/unix_connreset.c
b/tools/testing/selftests/net/af_unix/unix_connreset.c
new file mode 100644
index 000000000000..6f43435d96e2
--- /dev/null
+++ b/tools/testing/selftests/net/af_unix/unix_connreset.c
@@ -0,0 +1,179 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Selftest for AF_UNIX socket close and ECONNRESET behaviour.
+ *
+ * This test verifies:
+ * 1. SOCK_STREAM returns EOF when the peer closes normally.
+ * 2. SOCK_STREAM returns ECONNRESET if peer closes with unread data.
+ * 3. SOCK_SEQPACKET returns EOF when the peer closes normally.
+ * 4. SOCK_SEQPACKET returns ECONNRESET if the peer closes with unread data.
+ * 5. SOCK_DGRAM does not return ECONNRESET when the peer closes.
+ *
+ * These tests document the intended Linux behaviour.
+ *
+ */
+
+#define _GNU_SOURCE
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <errno.h>
+#include <sys/socket.h>
+#include <sys/un.h>
+#include "../../kselftest_harness.h"
+
+#define SOCK_PATH "/tmp/af_unix_connreset.sock"
+
+static void remove_socket_file(void)
+{
+ unlink(SOCK_PATH);
+}
+
+FIXTURE(unix_sock)
+{
+ int server;
+ int client;
+ int child;
+};
+
+FIXTURE_VARIANT(unix_sock)
+{
+ int socket_type;
+ const char *name;
+};
+
+/* Define variants: stream and datagram */
nit: outdated, maybe simply remove ?
oh..skipped me.
will do so.
+FIXTURE_VARIANT_ADD(unix_sock, stream) {
+ .socket_type = SOCK_STREAM,
+ .name = "SOCK_STREAM",
+};
+
+FIXTURE_VARIANT_ADD(unix_sock, dgram) {
+ .socket_type = SOCK_DGRAM,
+ .name = "SOCK_DGRAM",
+};
+
+FIXTURE_VARIANT_ADD(unix_sock, seqpacket) {
+ .socket_type = SOCK_SEQPACKET,
+ .name = "SOCK_SEQPACKET",
+};
+
+FIXTURE_SETUP(unix_sock)
+{
+ struct sockaddr_un addr = {};
+ int err;
+
+ addr.sun_family = AF_UNIX;
+ strcpy(addr.sun_path, SOCK_PATH);
+ remove_socket_file();
+
+ self->server = socket(AF_UNIX, variant->socket_type, 0);
+ ASSERT_LT(-1, self->server);
+
+ err = bind(self->server, (struct sockaddr *)&addr, sizeof(addr));
+ ASSERT_EQ(0, err);
+
+ if (variant->socket_type == SOCK_STREAM ||
+ variant->socket_type == SOCK_SEQPACKET) {
patchwork caught mis-alignment here and other places.
I'm using this for emacs, and other editors will have a similar config.
(setq-default c-default-style "linux")
You can check if lines are aligned properly by
$ git show --format=email | ./scripts/checkpatch.pl
+ err = listen(self->server, 1);
+ ASSERT_EQ(0, err);
+
+ self->client = socket(AF_UNIX, variant->socket_type, 0);
Could you add SOCK_NONBLOCK here too ?
This is noted
+ ASSERT_LT(-1, self->client);
+
+ err = connect(self->client, (struct sockaddr *)&addr,
sizeof(addr));
+ ASSERT_EQ(0, err);
+
+ self->child = accept(self->server, NULL, NULL);
+ ASSERT_LT(-1, self->child);
+ } else {
+ /* Datagram: bind and connect only */
+ self->client = socket(AF_UNIX, SOCK_DGRAM | SOCK_NONBLOCK, 0);
+ ASSERT_LT(-1, self->client);
+
+ err = connect(self->client, (struct sockaddr *)&addr,
sizeof(addr));
+ ASSERT_EQ(0, err);
+ }
+}
+
+FIXTURE_TEARDOWN(unix_sock)
+{
+ if (variant->socket_type == SOCK_STREAM ||
+ variant->socket_type == SOCK_SEQPACKET)
+ close(self->child);
+
+ close(self->client);
+ close(self->server);
+ remove_socket_file();
+}
+
+/* Test 1: peer closes normally */
+TEST_F(unix_sock, eof)
+{
+ char buf[16] = {};
+ ssize_t n;
+
+ /* Peer closes normally */
+ if (variant->socket_type == SOCK_STREAM ||
+ variant->socket_type == SOCK_SEQPACKET)
+ close(self->child);
+ else
+ close(self->server);
+
+ n = recv(self->client, buf, sizeof(buf), 0);
+ TH_LOG("%s: recv=%zd errno=%d (%s)", variant->name, n, errno,
strerror(errno));
errno is undefined if not set, and same for strerror(errno).
Also, if ASSERT_XXX() below fails, the same information
(test case, errno) is logged. So, TH_LOG() seems unnecessary.
Maybe try modifying the condition below and see how the
assertion is logged.
Oh..thank you. Didn't it through that way.
I understand.
I will remove the TH_LOG()'s
+ if (variant->socket_type == SOCK_STREAM ||
+ variant->socket_type == SOCK_SEQPACKET) {
+ ASSERT_EQ(0, n);
+ } else {
+ ASSERT_EQ(-1, n);
+ ASSERT_EQ(EAGAIN, errno);
+ }
+}
+
+/* Test 2: peer closes with unread data */
+TEST_F(unix_sock, reset_unread)
+{
+ char buf[16] = {};
+ ssize_t n;
+
+ /* Send data that will remain unread by client */
+ send(self->client, "hello", 5, 0);
+ close(self->child);
+
+ n = recv(self->client, buf, sizeof(buf), 0);
+ TH_LOG("%s: recv=%zd errno=%d (%s)", variant->name, n, errno,
strerror(errno));
+ if (variant->socket_type == SOCK_STREAM ||
+ variant->socket_type == SOCK_SEQPACKET) {
+ ASSERT_EQ(-1, n);
+ ASSERT_EQ(ECONNRESET, errno);
+ } else {
+ ASSERT_EQ(-1, n);
+ ASSERT_EQ(EAGAIN, errno);
+ }
+}
+
+/* Test 3: SOCK_DGRAM peer close */
Now Test 2 and Test 3 look identical ;)
seems so, but the only difference is:
close(self->child); is used in Test 2, while
close(self->server); is used in Test 3.
Maybe I should find a way to collapse Tests 2 and 3 (if statement might
work)
I am just afraid the tests to run will reduce to 6 from 9 and we will have 6
cases passed as against 7 as before.
What do you think?
The name of Test 2 is a bit confusing, which is not true
for SOCK_DGRAM. So, I'd use "if" to change which fd
to close() depending on the socket type.
Also, close(self->server) does nothing for SOCK_STREAM
and SOCK_SEQPACKET after accept(). Rather, that close()
should have the same effect if self->child is not accept()ed.
(In this case, Skip() for SOCK_DGRAM makes sense)
I think covering that scenario would be nicer.
If interested, you can check the test coverage with this patch.
https://lore.kernel.org/linux-kselftest/[email protected]/
Thanks!
Thank you!
kindly check these if any conforms to what it should be:
TEST_F(unix_sock, reset_unread_behavior)
{
char buf[16] = {};
ssize_t n;
/* Send data that will remain unread by client */
send(self->client, "hello", 5, 0);
if (variant->socket_type == SOCK_DGRAM) {
close(self->server);
}
else {
if (!self->child)
SKIP(return);
close(self->child);
}
n = recv(self->client, buf, sizeof(buf), 0);
ASSERT_EQ(-1, n);
if (variant->socket_type == SOCK_STREAM ||
variant->socket_type == SOCK_SEQPACKET)
do { ASSERT_EQ(ECONNRESET, errno); } while (0);
else
ASSERT_EQ(EAGAIN, errno);
}
OR
TEST_F(unix_sock, reset_unread_behavior)
{
char buf[16] = {};
ssize_t n;
/* Send data that will remain unread by client */
send(self->client, "hello", 5, 0);
if (variant->socket_type == SOCK_DGRAM) {
close(self->server);
}
else {
if (self->child)
close(self->child);
else
close(self->server);
}
n = recv(self->client, buf, sizeof(buf), 0);
ASSERT_EQ(-1, n);
if (variant->socket_type == SOCK_STREAM ||
variant->socket_type == SOCK_SEQPACKET)
do { ASSERT_EQ(ECONNRESET, errno); } while (0);
else
ASSERT_EQ(EAGAIN, errno);
}
OR
is there a better way to handle this?
Sorry for late!
What I had in mind is to move accept() in FIXTURE_SETUP() to
Test 1 & 2 (then, only listen() is conditional in FIXTURE_SETUP())
and rewrite Test 3 to cover the last ECONNRESET case caused by
close()ing un-accept()ed sockets:
TEST_F(unix_sock, reset_closed_embryo)
{
if (variant->socket_type == SOCK_DGRAM)
SKIP(return, "This test only applies to SOCK_STREAM and
SOCK_SEQPACKET");
close(self->server);
n = recv(self->client, ...);
ASSERT_EQ(-1, n);
ASSERT_EQ(ECONNRESET, errno);
}
I ran the KCOV_OUTPUT command using the first *TEST_F above* as the Test
2 and got the output below:
*$ KCOV_OUTPUT=kcov KCOV_SLOTS=8192
./tools/testing/selftests/net/af_unix/unix_connreset*
You should be able to see line-by-line coverage by decoding
files under kcov with addr2line or vock/report.py.
Thanks!
So sorry for the late reply.
I have noted all and I will send v4 shortly
Thanks once again
