Potential MPSL (err -35) issue with DM subsystem requires power cycle

[cal.rubbo]

This is a hard to reproduce issue, so I'm wondering if somebody from support could tell me what this failure might imply:

<dbg> nrf_dm: mpsl_nonpreemptible_thread: MPSL timeslot request failed (err -35)

• When running the stock "nrf_dm" I sometimes have seen my device get stuck with this error message, forever unable to range successfully. 
• Pressing the reset button doesn't fix this state
• Power cycling does fix this state

[Einar Thorsrud]

Hi,

-35 is NRF_EAGAIN in this case. Can you say more about in which cases you see this? Also, are you testing on a DK or a custom board? (I am asking to understand what the reset button does, as I woudl expect a pin reset to give the same result as a power-cycle).

[cal.rubbo]

I'm finding it hard to reproduce, but it seemed to happen a few times over the course of a day when I was flashing other boards in my test setup. Right now, I'm using nRF5340 dev kits for everything. 

For context, I'm doing one-direction ranging and I have one "reflector" and three "initiators". It seemed like when I'd flash the "reflector" sometimes I'd suddenly notice this MPSL err (-35) and all rangings fail on one of the "initiators" (while the other "initiators" continued to range).

After a power cycle, the broken "initiator" would come back to life and start ranging successfully again. I could be wrong, but I'm positive I tried the "reboot" button on the dev kit and that didn't fix it. Only a power cycle.

I know this is vague but I was hoping support could have some insight on what would cause the "initiator" to suddenly stop ranging successfully. 

[Einar Thorsrud]

Hi,

Just so I understand this, is it so that you get the -NRF_EAGAIN error on one of the initiators, and no issues (or warnings in the log) with the reflector or other two initiators? And you recover by power-cycling the problematic initiator?

Can you double-check that just resetting will not recover but a power-cycle will? And if a powr-cycle is needed, does this look differently seen from the reflector? (The only thing I can think of that could make a difference is that power-cycling typically takes a longer time than just a pin reset).

[cal.rubbo]

Einar Thorsrud said:

Just so I understand this, is it so that you get the -NRF_EAGAIN error on one of the initiators, and no issues (or warnings in the log) with the reflector or other two initiators? And you recover by power-cycling the problematic initiator?

That's correct.

Einar Thorsrud said:

Can you double-check that just resetting will not recover but a power-cycle will? And if a powr-cycle is needed, does this look differently seen from the reflector? (The only thing I can think of that could make a difference is that power-cycling typically takes a longer time than just a pin reset).

It'll probably be a few days before I have that same setup again, so if I notice it happen again I'll be sure to try out the reset button.

Thanks!