how to disable interrupts before SD gets active

[Peter_E]

Hi there,

I am using the uart example with SD130 and added neopixel code from GIT to drive the WS2812B LEDs using bit banging. I want to update the LED every ms. The bitbanging sequence used in neopixel_show(&m_strip); must not be interrupted. To achieve this I am using this code

ble_radio_notification_init(NRF_RADIO_PRIORITY_NORMAL ,
       NRF_RADIO_NOTIFICATION_DISTANCE_800US,
       led_radio_callback_handler);

void led_radio_callback_handler(bool radio_active)                    
{
			if (radio_active == false)
			{
				softdevice_active = false;                                                                                 
			} else {                                                                       
				softdevice_active = true;
			}					             
}

static void ms_timeout_handler(void * p_context)																// 1ms Timer
{
    // uint32_t err_code;
    UNUSED_PARAMETER(p_context);
	
		// nrf_drv_gpiote_out_toggle(BSP_LED_2);
		// ms_timer_ticks++;
	  if (softdevice_active == false)
		{
			// CRITICAL_REGION_ENTER(); 
			// sd_nvic_critical_region_enter(&p_is_nested_critical_region);
			__disable_irq();
			neopixel_show(&m_strip);
			__enable_irq(); 
			// CRITICAL_REGION_EXIT();
			// sd_nvic_critical_region_exit(p_is_nested_critical_region);
                } else {
			nrf_drv_gpiote_out_toggle(BSP_LED_2);
		}
}

The above code works but I read in a different thread that it is not recommened to use __disable_irq() together with a SD. The the other two methods that are outcommented do not work proberly. I still get interrupts just before the notfication:

The question is:

Can I use __disable_irq() in this special case or what would be your recommendation?

Thanks

Peter

[Peter_E]

Thanks for the answer, I will have to find some kind of workaround.

[Petter Myhre]

It is not recommended to disable all interrupts while you are in a connection, this may break the connection, and may result in SoftDevice asserts. See this and this.

And there is no way to guarantee that you have the CPU every ms, however you can use the timeslot API to guarantee that you have the CPU (and everything else) for a certain timeslot.

According to this it seems that you can disable interrupts safely for 10us, but this isn't something I have tested.