nrf52 write buffer

RE: nrf52 write buffer

Stephane D'Alu — Mon, 11 Jul 2016 13:00:31 GMT

Thanks for your help

RE: nrf52 write buffer

Hung Bui — Mon, 11 Jul 2016 12:46:58 GMT

Hi Stephane,

My understanding is that the possible issue with clearing the interrupt event register is more critical than writing to execute a task or to configure a register as the interrupt can be triggered multiple time if it's not cleared properly before existing the IRQ handler. I don't see potential issue when there is a delay when executing writing to task or register.

When writing to a task, usually you have to make sure the task has been executed with a while loop to read the register (_STARTED register).

RE: nrf52 write buffer

Stephane D'Alu — Mon, 11 Jul 2016 12:00:06 GMT

Thanks.

I have a side question, if write-buffer could be a problem when clearing events. Should we also be aware of a possible issue when writing to tasks or other registers?

RE: nrf52 write buffer

Hung Bui — Mon, 11 Jul 2016 11:45:15 GMT

Hi Stephane,

It's true that in some of our driver we still haven't implemented that. I guess it was due to we ported the driver partly from the code from nRF51 and haven't updated it to meet with the requirement for nRF52. We haven't seen consequence of doing that but we should update them.

You can find in some driver we actually did that, for example in nrf_clock_event_clear() function.

I have reported it to R&D and we will try to update the code.

I would suggest you to add the code to do a dummy read every time your clear an interrupt register.