nrfx_ppi_channel_disable() sometimes returns NRFX_ERROR_INVALID_STATE

RE: nrfx_ppi_channel_disable() sometimes returns NRFX_ERROR_INVALID_STATE

Einar Thorsrud — Fri, 31 May 2019 08:51:05 GMT

Hi,

Are you sure that the libuarte is initialized when this happens? Perhaps there is a problem with the logic in your application so that it is not called? Even if you have basic handling of this (which I don't see any trace of in your code snippets), another explanation could be that since the call stack shows that everyting here happen in the SoftDevice interrupt handling, there could be that you started to initialize libuarte, but the BLE event came before it finished and interrupted the initialization procedure. Then the uninitialization happens before it is actually initialized, which in turn leads to this error.

RE: nrfx_ppi_channel_disable() sometimes returns NRFX_ERROR_INVALID_STATE

gordie — Thu, 30 May 2019 05:41:01 GMT

Hi Einar, from what I've seen so far, it seems to happen when I start the bluetooth connection process on the peripheral device and then abruptly stop it by turning the central off. This then causes the code on the peripheral to hit the case statement: BLE_GAP_EVT_DISCONNECTED where nrf_libuarte_async_uninit(&libuarte); gets called.

I have attached a screencap of the call stack, which is the same everytime (I've been able to reproduce it many times):

RE: nrfx_ppi_channel_disable() sometimes returns NRFX_ERROR_INVALID_STATE

Einar Thorsrud — Wed, 29 May 2019 13:38:56 GMT

Hi,

This would happen if nrfx_ppi_channel_free() has been called before nrfx_ppi_channel_disable(), but that should not happen. Can you add some logging in nrfx_ppi.c and nrf_libuarte_async.c to get a good understanding of the program flow up to when this error occurs?