EVENTS_DEVMATCH never seen active, and invalid ACCESS CODE possibly passes the BASE0/PREFIX0 filter?

[Oleksandr K.]

Hello,

Using own BLE stack here (not SoftDevice) and having troubles with LL certification test cases with particular problem:

• the tester establishes connection and then starts sending LL_DATA packets with incorrect (modified) access code for connection events.
• out of 20 such packets at most 4 may be accepted (for any reason?) by the IUT, but in our case this number is as large as 8.
  
  

We are puzzled by the possibility to get a false-positive EVENT_END events at all and packets that pass the CRC check.

The code relies on EVENTS_END and CRC_STATUS in handling of all incoming packets.

Another discovery, which I cannot find explanation is that the EVENTS_DEVMATCH is always 0 in the RADIO IRQ, even when BLE connections are surely running and use own access code.

EVENTS_DEVMISS is "1" all the time too!?

How can it be that the MATCH/MISS events are not right somehow?

       

My understanding is that single bit errors are possible, so invalid access code is sometimes could be accepted, but 40% failure rate is to high still.

Regards,

Oleksandr

[Håkon Alseth]

Hi,

 

My understanding is that single bit errors are possible, so invalid access code is sometimes could be accepted, but 40% failure rate is to high still.

Are you implementing the errata workarounds for #102 #106 #107?

https://infocenter.nordicsemi.com/topic/errata_nRF52832_Rev3/ERR/nRF52832/Rev3/latest/anomaly_832_182.html?cp=4_2_1_0_1_47

 

For revision 1, there's a specific workaround for handling the above mentioned erratas, while for rev 2 and newer, you can do as described in erratum #182 to apply the fixes.

 

Kind regards,

Håkon

[Oleksandr K.]

Hi Håkon,

We have added the 102/106/107 + 182 errata afterwards (yesterday) and new results tell that a couple of troubled test cases pass.

The errata fend of reception errors (false negatives), but apparently also could help with invalid packet reception (false positives). Is this a common knowledge?

EVENT_DEVMATCH/MISSED are still not seen correct. Should they be avoided altogether in development?

Thanks,
Oleksandr

[Håkon Alseth]

Hi,

 

Oleksandr K. said:

The errata fend of reception errors (false negatives), but apparently also could help with invalid packet reception (false positives). Is this a common knowledge?

EVENT_DEVMATCH/MISSED are still not seen correct. Should they be avoided altogether in development?

Given that these erratas are handled, we do not currently have any known issues wrt. these events.

It is hard to say why this happens on your end, as I do not have detailed overview of what you're doing on a firmware level.

Do you clear your events?

 

Kind regards,

Håkon