Openthread border router fix socket close #99495

Cristib05 · 2025-11-17T10:12:11Z

Main goal of this PR is to handle a WI-Fi connect/disconnect or link lost and link re-established scenario.
OpenThread's platform code has been enhanced to handle those events.
Border Router application code has been updated to de-initialize platform modules and remove its multicast routes.
TREL code has been updated to accommodate user enable/disable interaction from CLI.

rlubos · 2025-11-17T11:11:59Z

modules/openthread/platform/trel.c

+		if (!net_mgmt(NET_REQUEST_WIFI_IFACE_STATUS, ail_iface_ptr,
+				  &status, sizeof(status))) {
+			if (status.state == WIFI_STATE_COMPLETED) {
+				(void)trel_plat_init(ot_instance_ptr, ail_iface_ptr);
+			}
+		}


Doesn't this make a hardcoded assumption that the backbone link is Wi-Fi? Please correct me if I'm wrong, but I think this module should also work with other interface types like Ethernet?

A generic way to check if the interface is ready to transmit data would be to check net_if_is_up() - for Wi-Fi interfaces, this should only return true if the interface is connected to a Wi-Fi network.

OTBR is planned to support both interfaces, Wi-FI or Ethernet, you are right. It would be a better idea to stick with your approach.

jukkar · 2025-11-18T06:48:37Z

@Cristib05 you need to rebase to latest main, the network APIs were name spaced by #99169 and there is now conflict with this PR.

modules/openthread/platform/infra_if.c

rlubos · 2025-11-18T09:31:39Z

subsys/net/l2/openthread/openthread_border_router.c

+			entry = net_route_mcast_add(ail_iface_ptr, &addr, 16);
+			if (entry != NULL) {
+				mcast_addr = net_if_ipv6_maddr_add(ail_iface_ptr,
+								(const struct in6_addr *)&addr);


The typecast here should also be updated.

rlubos · 2025-11-18T09:31:50Z

subsys/net/l2/openthread/openthread_border_router.c

-				net_if_ipv6_maddr_join(ail_iface_ptr, mcast_addr);
+				net_if_ipv6_maddr_leave(ail_iface_ptr, mcast_addr);
+				net_if_ipv6_maddr_rm(ail_iface_ptr,
+						(const struct in6_addr *)&addr);


jukkar · 2025-11-18T12:26:46Z

modules/openthread/platform/mdns_socket.c

 	net_ipv6_addr_create(&mreq_v6.ipv6mr_multiaddr, 0xff02, 0, 0, 0, 0, 0, 0, 0x00fb);
-	VerifyOrExit(zsock_setsockopt(mdns_sock_v6, NET_IPPROTO_IPV6, ZSOCK_IPV6_ADD_MEMBERSHIP,
-				      &mreq_v6, sizeof(mreq_v6)) == 0,
+	mcast_join_ret = zsock_setsockopt(mdns_sock_v6, IPPROTO_IPV6, IPV6_ADD_MEMBERSHIP,


Also these needs to be changed
IPPROTO_IPV6 -> NET_IPPROTO_IPV6
IPV6_ADD_MEMBERSHIP -> ZSOCK_IPV6_ADD_MEMBERSHIP

jukkar · 2025-11-18T12:27:07Z

modules/openthread/platform/mdns_socket.c

-	VerifyOrExit(zsock_setsockopt(mdns_sock_v4, NET_IPPROTO_IP, ZSOCK_IP_ADD_MEMBERSHIP,
-				      &mreq_v4, sizeof(mreq_v4)) == 0,
+	mreq_v4.imr_multiaddr.s_addr = htonl(0xE00000FB);
+	mcast_join_ret = zsock_setsockopt(mdns_sock_v4, IPPROTO_IP, IP_ADD_MEMBERSHIP,


similar issue here

jukkar · 2025-11-18T12:29:03Z

I will investigate why the CI did not fail because some symbols were not converted, probably openthread was not part of tests/net/all tests.

jukkar · 2025-11-18T12:29:57Z

modules/openthread/platform/mdns_socket.c

-	mreq_v4.imr_multiaddr.s_addr = net_htonl(0xE00000FB);
-	VerifyOrExit(zsock_setsockopt(mdns_sock_v4, NET_IPPROTO_IP, ZSOCK_IP_ADD_MEMBERSHIP,
-				      &mreq_v4, sizeof(mreq_v4)) == 0,
+	mreq_v4.imr_multiaddr.s_addr = htonl(0xE00000FB);


also htonl() -> net_htonl()

Cristib05 · 2025-11-18T12:32:54Z

@jukkar , I will discard #99567 and do all the work in this PR since I'm touching most of the files in here.

Cristib05 · 2025-11-18T13:27:24Z

I will investigate why the CI did not fail because some symbols were not converted, probably openthread was not part of tests/net/all tests.

Rebased and added changes, but having tests enabled would definitely ease the process. Hopefully I haven't missed much this time

jukkar · 2025-11-18T13:50:06Z

If needed, you can try things by setting CONFIG_NET_NAMESPACE_COMPAT_MODE=n and compiling a non-POSIX enabled application. It should output errors for all the used old network APIs.

OpenThread interface is initialized beform Wi-Fi interface. `otPlatTrelEnable` is called by OpenThread stack when it's interface is being initialized. Given this scenario, socket operation, like `bind`, will fail. There is also no mean to get a valid pointer to backbone interface. This is why, `trel_plat_init` was declared and called when backbone interface reported connectivity. This commit handles the `ot trel disable/ot trel enable` scenario that can be initialized from CLI. `otPlatTrelEnable` will be called, but `trel_plat_init` will not be called anymore, leaving trel socket without any options set, and `net_socket_service_register` is not called anymore, to handle incoming traffic. Now, when `otPlatTrelEnable` is called, it will verify if Wi-Fi interface is connected and will call `trel_plat_init` if needed. Signed-off-by: Cristian Bulacu <cristian.bulacu@nxp.com>

In this commit, `net_socket_service_register` is called when a platform module that has sockets is deinitialized, and it's socket/sockets are closed. This commit also handles a case when a module tries to join a multicast group and a subscription, from another module, is already present. Also, covered network namespace changes done in zephyrproject-rtos#99169 Signed-off-by: Cristian Bulacu <cristian.bulacu@nxp.com>

This commit deinitializes platform modules when external network interface is brought down. It also delets the multicast routes that are added for OpenThread interface. Signed-off-by: Cristian Bulacu <cristian.bulacu@nxp.com>

sonarqubecloud · 2025-11-20T16:43:07Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

Cristib05 · 2025-11-20T17:36:19Z

Rebased to include #99593

pdgendt · 2025-11-21T08:24:34Z

modules/openthread/platform/dhcp6_pd.c

 	struct net_sockaddr_in6 dest_sock_addr = {.sin6_family = NET_AF_INET6,
-					      .sin6_port = net_htons(DHCPV6_SERVER_PORT),
-					      .sin6_addr = NET_IN6ADDR_ANY_INIT,
-					      .sin6_scope_id = 0};
+						  .sin6_port = net_htons(DHCPV6_SERVER_PORT),
+						  .sin6_addr = NET_IN6ADDR_ANY_INIT,
+						  .sin6_scope_id = 0};


Nitpicking here, if you'd add a trailing comma to the last item, it would be formatted like:

struct net_sockaddr_in6 dest_sock_addr = { .sin6_family = NET_AF_INET6, .sin6_port = net_htons(DHCPV6_SERVER_PORT), .sin6_addr = NET_IN6ADDR_ANY_INIT, .sin6_scope_id = 0, };

Which is a best nicer IMO.

pdgendt · 2025-11-21T08:26:07Z

modules/openthread/platform/trel.c

+
+

Nit, single empty line is sufficient

Suggested change

zephyrbot added area: Networking area: OpenThread labels Nov 17, 2025

zephyrbot requested review from canisLupus1313, edmont, jukkar, kkasperczyk-no, maciejbaczmanski, mariuszpos, pdgendt, rlubos and ssharks November 17, 2025 10:13

zephyrbot assigned rlubos Nov 17, 2025

rlubos reviewed Nov 17, 2025

View reviewed changes

Cristib05 force-pushed the openthread_border_router_fix_socket_close branch from 2523147 to 7311d26 Compare November 17, 2025 11:29

Cristib05 requested a review from rlubos November 17, 2025 11:34

Cristib05 force-pushed the openthread_border_router_fix_socket_close branch from 7311d26 to 6de7a35 Compare November 17, 2025 11:55

rlubos previously approved these changes Nov 17, 2025

View reviewed changes

jukkar previously approved these changes Nov 17, 2025

View reviewed changes

Cristib05 dismissed stale reviews from jukkar and rlubos via 66ff636 November 18, 2025 07:30

Cristib05 force-pushed the openthread_border_router_fix_socket_close branch from 6de7a35 to 66ff636 Compare November 18, 2025 07:30

pdgendt reviewed Nov 18, 2025

View reviewed changes

modules/openthread/platform/infra_if.c Outdated Show resolved Hide resolved

Cristib05 force-pushed the openthread_border_router_fix_socket_close branch from 66ff636 to aaa4f94 Compare November 18, 2025 07:57

Cristib05 requested review from jukkar, pdgendt and rlubos November 18, 2025 09:11

rlubos reviewed Nov 18, 2025

View reviewed changes

Cristib05 force-pushed the openthread_border_router_fix_socket_close branch from aaa4f94 to 452c5be Compare November 18, 2025 09:48

rlubos previously approved these changes Nov 18, 2025

View reviewed changes

pdgendt previously approved these changes Nov 18, 2025

View reviewed changes

jukkar requested changes Nov 18, 2025

View reviewed changes

jukkar reviewed Nov 18, 2025

View reviewed changes

Cristib05 dismissed stale reviews from pdgendt and rlubos via 082b945 November 18, 2025 13:26

Cristib05 force-pushed the openthread_border_router_fix_socket_close branch from 452c5be to 082b945 Compare November 18, 2025 13:26

Cristib05 force-pushed the openthread_border_router_fix_socket_close branch from 082b945 to 4744add Compare November 18, 2025 14:24

Cristib05 added 3 commits November 20, 2025 18:39

net: l2: openthread: border_router: Enhance modules deinit

4c2b107

This commit deinitializes platform modules when external network interface is brought down. It also delets the multicast routes that are added for OpenThread interface. Signed-off-by: Cristian Bulacu <cristian.bulacu@nxp.com>

Cristib05 force-pushed the openthread_border_router_fix_socket_close branch from 4744add to 4c2b107 Compare November 20, 2025 16:39

Cristib05 requested review from jukkar, pdgendt and rlubos November 21, 2025 07:09

jukkar approved these changes Nov 21, 2025

View reviewed changes

pdgendt approved these changes Nov 21, 2025

View reviewed changes

rlubos approved these changes Nov 21, 2025

View reviewed changes

nashif merged commit b9542ab into zephyrproject-rtos:main Nov 21, 2025
29 checks passed

Cristib05 deleted the openthread_border_router_fix_socket_close branch November 26, 2025 11:15

Openthread border router fix socket close #99495

Openthread border router fix socket close #99495

Uh oh!

Conversation

Cristib05 commented Nov 17, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jukkar commented Nov 18, 2025

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!

jukkar commented Nov 18, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Cristib05 commented Nov 18, 2025

Uh oh!

Cristib05 commented Nov 18, 2025

Uh oh!

jukkar commented Nov 18, 2025

Uh oh!

sonarqubecloud bot commented Nov 20, 2025

Quality Gate passed

Uh oh!

Cristib05 commented Nov 20, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants