Calling `spi.init()` twice causes a panic with error code 090

[microbit-carlos]

Panic 090 is the DEVICE_HARDWARE_CONFIGURATION_ERROR.

Likely hitting this CODAL panic: https://github.com/lancaster-university/codal-nrf52/blob/ac01d74dc13236beddde78a5dab23c9e41b42c23/source/NRF52SPI.cpp#L55

But technically MicroPython is deleting the CODAL SPI instance before recreating it, which you'd guess it would free up the peripheral: https://github.com/microbit-foundation/micropython-microbit-v2/blob/f3db1e6215cd747e8a09f80cb7c35cf9d8f9cad5/src/codal_app/microbithal.cpp#L245-L259

This might be an issue in the CODAL side, in which case we can move this issue there.

[microbit-carlos]

Right, I don't think there is a destructor in NRF52SPI.cpp and I don't see anywhere else where free_alloc_peri() might be called for the SPI, so it looks like this is something that could be added to CODAL: https://github.com/lancaster-university/codal-microbit-v2/issues/219

Alternatively, if CODAL provided a way to change the frequency, mode and pins that could be used instead.

@dpgeorge do you have any preference?

[dpgeorge]

Alternatively, if CODAL provided a way to change the frequency, mode and pins that could be used instead

CODAL already provides a way to change the frequency and mode. So all that's needed is a way to change the pins. That would be better than destructing and recreating the NRF52SPI object (although I still think it would be good to fix CODAL so that it does call free_alloc_peri() in the destructor).

[microbit-carlos]

The CODAL changes required to reassign pins spawn several repos and files, so to reduce risk for the v2.1.0 final release we'll do this for v2.2.

[microbit-carlos]

CODAL has been updated and this can be implemented using the latest tag (0.2.48 at the time of writing).

[dpgeorge]

CODAL was updated to 0.2.48 and SPI fixed in 3af0abd392fd2e325c5e824a2ffb508088ac849a

[microbit-carlos]

Thanks Damien!