nrfx_saadc_abort() seems to be broken

RE: nrfx_saadc_abort() seems to be broken

David Meiklejohn — Wed, 29 Jan 2020 10:01:29 GMT

Got it! Thanks!

RE: nrfx_saadc_abort() seems to be broken

Simon — Wed, 29 Jan 2020 09:48:34 GMT

The problem is not related to the SAADC driver, but the example itself. The function nrf_drv_saadc_abort() is called from an interrupt context, which causes a deadlock, since nrf_drv_saadc_abort() is waiting for an interrupt itself.

If you move nrf_drv_saadc_abort to main(), the problem goes away.

Best regards,

Simon

RE: nrfx_saadc_abort() seems to be broken

David Meiklejohn — Mon, 27 Jan 2020 22:12:10 GMT

Thanks Simon.

I hesitate to modify the SDK code (messy when it comes to version control and tracking dependencies), but I could certainly make my own version of nrfx_saadc_abort() based on this and use that.

Thanks again.

RE: nrfx_saadc_abort() seems to be broken

Simon — Mon, 27 Jan 2020 15:52:15 GMT

This seems like a bug, since nrfx_saadc_irq_handler()-->if (nrf_saadc_event_check(NRF_SAADC_EVENT_STOPPED))-->m_cb.adc_state = NRF_SAADC_STATE_IDLE is never triggered

I have reported the issue and I'm currently waiting for an answer. However, if nrfx_saadc_abort() is modified accordingly it works fine:

void nrfx_saadc_abort(void)
{
    .
    .
    .
        {
            // Wait for ADC being stopped.
            bool result;
            //NRFX_WAIT_FOR((m_cb.adc_state == NRF_SAADC_STATE_IDLE), HW_TIMEOUT, 0, result);
            //NRFX_ASSERT(result);
            NRFX_WAIT_FOR(nrf_saadc_event_check(NRF_SAADC_EVENT_STOPPED), HW_TIMEOUT, 0, result);
            NRFX_ASSERT(result);
        }
    .
    .
}

Best regards,

Simon