net: coap_client: Stop all socket activities when cancelling requests

Calling coap_client_cancel_requests() clears the internal request
context only for active requests (i. e. not replied yet). However,
if there are any pending request context monitoring ACK duplicates,
those would still make the corresponding client socket being monitored
by poll(). In result, when application closes the socket, the polling
thread will throw POLLNVAL error for the socket.

Fix this, by resetting all request contexts unconditionally. The request
callback will only be called for the active requests.

Signed-off-by: Robert Lubos <[email protected]>

Author: rlubos

include/zephyr/net/coap_client.h

@@ -157,6 +157,8 @@ int coap_client_req(struct coap_client *client, int sock, const struct sockaddr
  *
  * This is intended for canceling long-running requests (e.g. GETs with the OBSERVE option set)
  * which has gone stale for some reason.
+ * The function should also be called before the corresponding client socket is closed,
+ * to prevent the socket from being monitored by the internal polling thread.
  *
  * @param client Client instance.
  */

subsys/net/lib/coap/coap_client.c

@@ -1013,14 +1013,12 @@ static void cancel_requests_with(struct coap_client *client, int error)
 			 * request was cancelled anyway.
 			 */
 			report_callback_error(&client->requests[i], error);
-			release_internal_request(&client->requests[i]);
 		}
-		/* If our socket has failed, clear all requests, even completed ones,
-		 * so that our handle_poll() does not poll() anymore for this socket.
+
+		/* Clear all requests, even completed ones, so that our
+		 * handle_poll() does not poll() anymore for this socket.
 		 */
-		if (error == -EIO) {
-			reset_internal_request(&client->requests[i]);
-		}
+		reset_internal_request(&client->requests[i]);
 	}
 	k_mutex_unlock(&client->lock);