Achieve non-concurrent switching between ESB/BLE

RE: Achieve non-concurrent switching between ESB/BLE

john.liu — Thu, 23 Jul 2026 13:36:21 GMT

Hi Priyanka:
Thanks for your help!
CONFIG_BT_UNINIT_MPSL_ON_DISABLE=y is already enabled.

When I start up, it defaults to ESB mode.

static K_SEM_DEFINE(bt_init_sem, 0, 1);

static void bt_ready_cb(int err)
{
    if (err) {
        printk("bt_enable failed in callback: %d\n", err);
    }
    /* 通知 main bt_enable 已完成 */
    k_sem_give(&bt_init_sem);
}

int main(void)
{
	int err;
    err = bt_enable(bt_ready_cb);
    if (err) {
        printk("bt_enable failed: %d\n", err);
        return 0;
    }
    /* 阻塞等待初始化完成 */
    k_sem_take(&bt_init_sem, K_FOREVER);
    printk("bt_enable complete\n");

    err = bt_disable();
    if (err) {
        printk("bt_disable failed: %d\n", err);
        return 0;
    }
	k_sleep(K_MSEC(500));
	printk("k_sleep done\n");
	err = clocks_start();
	printk("clocks_start ret: %d\n", err);
	if (err) {
		 printk("clocks_start failed: %d\n", err);
		return 0;
	}
	..........

It will get stuck at clocks_start()...

static int clocks_start(void)
{
	int err;
	int res;
	struct onoff_manager *clk_mgr;
	struct onoff_client clk_cli;

	clk_mgr = z_nrf_clock_control_get_onoff(CLOCK_CONTROL_NRF_SUBSYS_HF);
	if (!clk_mgr) {
		printk("Unable to get the Clock manager\n");
		return -ENXIO;
	}

	sys_notify_init_spinwait(&clk_cli.notify);

	err = onoff_request(clk_mgr, &clk_cli);
	if (err < 0) {
		printk("Clock request failed: %d", err);
		return err;
	}

	do {
		err = sys_notify_fetch_result(&clk_cli.notify, &res);
		if (!err && res) {
			printk("Clock could not be started: %d\n", res);
			return res;
		}
	} while (err);

#if NRF54L_ERRATA_20_PRESENT
	if (nrf54l_errata_20()) {
		nrf_power_task_trigger(NRF_POWER, NRF_POWER_TASK_CONSTLAT);
	}
#endif /* NRF54L_ERRATA_20_PRESENT */

#if defined(NRF54LM20A_ENGA_XXAA)
	/* MLTPAN-39 */
	nrf_clock_task_trigger(NRF_CLOCK, NRF_CLOCK_TASK_PLLSTART);
#endif

	printk("HF clock started\n");
	return 0;
}

RE: Achieve non-concurrent switching between ESB/BLE

Priyanka — Thu, 23 Jul 2026 12:37:30 GMT

Hi,

The crash might not be caused by k_msgq_put(). That is most likely where the failure becomes visible.

The MPSL assertion shows that Bluetooth/MPSL and ESB are conflicting over shared resources such as the radio, clock, or interrupts.

Although you wish to run ESB and Bluetooth non-concurrently, enabling CONFIG_BT=y initializes components that manage these resources. Therefore, the transition between ESB and Bluetooth must fully release one protocol before starting the other. i.e. use only one protocol at a time.

Please check the following:

Configure MPSL and ESB to use dynamic interrupts.
Fully stop ESB before starting Bluetooth.
Call bt_disable() before returning to ESB.
Enable CONFIG_BT_UNINIT_MPSL_ON_DISABLE=y so disabling Bluetooth also releases MPSL’s radio-related resources.
Avoid manually controlling clocks or peripherals while MPSL owns them.

[quote user="john.liu"]But after canceling CONFIG_BT=y, there's no problem.[/quote]

when you do this, the Bluetooth controller and its MPSL-related resource management are no longer included or initialized. ESB then has exclusive ownership of the radio, clocks, and interrupts, so the conflict disappears.

In simple words, without Bluetooth enabled, only ESB is trying to drive the radio;

with Bluetooth enabled, two different systems believe they control it.

-Priyanka

RE: Achieve non-concurrent switching between ESB/BLE

john.liu — Thu, 23 Jul 2026 08:29:43 GMT

Hi:
The exact location of the collapse.

                    struct main_msg msg;
                    msg.type = MAIN_MSG_MOTION;
                    msg.cnt = 1;
                    int err = k_msgq_put(&main_msgq, &msg, K_NO_WAIT);

But after canceling CONFIG_BT=y, there's no problem.