Add TCP Listen and Accept by MarkusLassila · Pull Request #196 · nrfconnect/ncs-serial-modem

MarkusLassila · 2026-02-27T12:28:46Z

Add AT#XLISTEN and AT#XACCEPT for TCP server functionality.

Jira: SM-240

Copilot

Pull request overview

Adds TCP server support to the Serial Modem AT socket layer by documenting and implementing AT#XLISTEN / AT#XACCEPT, plus extending AT#XSOCKET? output to include remote peer info for accepted sockets.

Changes:

Extend AT#XSOCKET? read response to optionally include accepted peer address/port.
Add AT#XLISTEN (listen) and AT#XACCEPT (accept) AT commands in implementation and docs.
Update unit tests and test harness wiring (AT wrapper routing + kernel alloc stubs) to cover listen/accept behavior.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
doc/app/at_socket.rst	Documents `#XLISTEN` / `#XACCEPT` and extends `#XSOCKET?` response format with peer info.
app/src/sm_at_socket.c	Implements listen/accept, tracks bound port + listen state, and stores peer address for accepted sockets.
app/tests/at_socket/src/nrf_modem_at_wrapper.c	Routes `AT#XLISTEN` / `AT#XACCEPT` to custom-command wrappers in tests.
app/tests/at_socket/src/test_at_socket.c	Adds tests for listen/accept and updates socket-count assumptions.
app/tests/at_socket/CMakeLists.txt	Links new kernel stubs into the at_socket test target.
app/tests/stubs/kernel_stubs.c	Provides `k_malloc`/`k_free` stubs for the unit-test environment.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-27T12:35:16Z

app/src/sm_at_socket.c

+{
+	int ret;
+
+	if (sock->type != NRF_SOCK_STREAM || sock->local_port == 0 ||


do_listen() rejects listening unless sock->local_port != 0, but #XBIND explicitly documents the port range as 0–65535. If a socket is bound with port 0 (ephemeral), it is still bound but will be treated as “not bound” here. Consider tracking a separate “is_bound” flag (or querying getsockname) instead of using local_port == 0 as a proxy.

Suggested change

if (sock->type != NRF_SOCK_STREAM || sock->local_port == 0 ||

if (sock->type != NRF_SOCK_STREAM ||

This looks valid, right? 0 is valid port number.
Add a unit test for this.

It depends on how the modem deals with bind to 0. But in any case either this or the port range in bind is wrong.

Actually, the way this works is that the 0 assigns an ephemeral port, which we cannot currently be aware as we lack the getsockname(). So we are right to deny this here as without being bound to specific port we would not know which port the listen went to, which makes it useless.

Added a note in the bind about 0 being a special value.

app/src/sm_at_socket.c

app/tests/at_socket/src/test_at_socket.c

app/src/sm_at_socket.c

trantanen

Going through doc and parts of code.

app/tests/at_socket/src/test_at_socket.c

trantanen · 2026-03-02T11:20:12Z

app/src/sm_at_socket.c

+{
+	int ret;
+
+	if (sock->type != NRF_SOCK_STREAM || sock->local_port == 0 ||


This looks valid, right? 0 is valid port number.
Add a unit test for this.

doc/app/at_socket.rst

trantanen · 2026-03-02T11:40:02Z

doc/app/at_socket.rst

+
+   The listening socket is set to non-blocking mode when ``#XLISTEN`` command is executed.
+   If there are no incoming connections when the ``#XACCEPT`` command is executed, it will return an error.
+   The ``#XAPOLL`` command can be used to receive asynchronous notifications (``POLLIN``) for incoming connections on the listening socket.


Is POLLIN something we've defined as our proprietary solution?
Should this be mentioned in XAPOLL description of POLLIN too?

No, the poll() returns POLLIN if there is a new connection.

doc/app/at_socket.rst

app/src/sm_at_socket.c

SeppoTakalo · 2026-03-04T18:35:45Z

app/src/sm_at_socket.c

+	new_sock->fd = ret;
+	new_sock->family = remote.sa_family;
+	new_sock->type = NRF_SOCK_STREAM;
+	new_sock->role = AT_SOCKET_ROLE_SERVER;


Is the new socket anymore a server?

Where do we use this role information?

Nowhere. This the desired TLS role on the socket. Modem takes this parameter but doesn't really support TLS server. We should have removed this altogether before 1.0 but we didn't so it will hang in there for nothing...

But yes, this should probably be client socket here.

The I recommend that we use this "role" to mark sockets that are listening.

So we can drop the sock->local_port and sock->listen values as well.

It is not relevant to know in a boolean if you already called nrf_listen() or not, because nrf_accept() will fail if you did not.

Given that we have no means of getting the ephemeral port from the modem, I think that it still makes sense to track whether we have bound to non-ephemeral port.

It is better to give an error than allow the user to do listen() to a port that they have no hope of getting.

Actually, the role makes 0 sense as it is. No need to waste uint16_t for it when a bit field is enough.

Edit: That is, if we can just ignore the server parameter given in AT#XSOCKET and not return the same value when AT#XSOCKET? is done.

@trantanen: Does that constitute as breaking AT-interface?

-> Discussed offline, it does, so we keep it and return it in AT#XSOCKET?. Parameter does not have any effect internally.

Also, AT#XLISTEN? currently returns the listen sockets and the ports that they are bound. Keep or remove?

-> Discussed offline: Keep.

Add AT#XLISTEN and AT#XACCEPT for TCP server functionality. Signed-off-by: Markus Lassila <markus.lassila@nordicsemi.no>

Basic unit tests for AT#XLISTEN and AT#XACCEPT. Signed-off-by: Markus Lassila <markus.lassila@nordicsemi.no>

MarkusLassila requested review from SeppoTakalo, Copilot, juhaylinen and trantanen February 27, 2026 12:28

Copilot started reviewing on behalf of MarkusLassila February 27, 2026 12:29 View session

MarkusLassila changed the title ~~Tcp listen accept~~ Add TCP Listen and Accept Feb 27, 2026

Copilot AI reviewed Feb 27, 2026

View reviewed changes

trantanen reviewed Mar 2, 2026

View reviewed changes

SeppoTakalo reviewed Mar 4, 2026

View reviewed changes

MarkusLassila force-pushed the tcp-listen-accept branch 2 times, most recently from 207ba63 to ac3bea8 Compare March 10, 2026 09:16

MarkusLassila added 2 commits March 10, 2026 16:19

app: Add AT#XLISTEN and AT#XACCEPT

c698367

Add AT#XLISTEN and AT#XACCEPT for TCP server functionality. Signed-off-by: Markus Lassila <markus.lassila@nordicsemi.no>

tests: Add unit tests for AT#XLISTEN and AT#XACCEPT

67ca2cf

Basic unit tests for AT#XLISTEN and AT#XACCEPT. Signed-off-by: Markus Lassila <markus.lassila@nordicsemi.no>

MarkusLassila force-pushed the tcp-listen-accept branch from ac3bea8 to 67ca2cf Compare March 10, 2026 14:22

MarkusLassila requested review from SeppoTakalo, divipillai and trantanen March 10, 2026 14:31

	if (sock->type != NRF_SOCK_STREAM \|\| sock->local_port == 0 \|\|
	if (sock->type != NRF_SOCK_STREAM \|\|

Conversation

MarkusLassila commented Feb 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

trantanen left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MarkusLassila Mar 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MarkusLassila Mar 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

MarkusLassila commented Feb 27, 2026 •

edited

Loading

MarkusLassila Mar 10, 2026 •

edited

Loading

MarkusLassila Mar 10, 2026 •

edited

Loading