[Replicant] Exposure Notification not working

[Nick]

Hi doak,

Thanks very much for your response, your questions are things that I 
hadn't generally thought of!

> Did you check if your device actually sends BLE beacons?

I did not! I checked that it received beacons, and it does, so I 
naively assumed therefore that all would be perfect.

> At the beginning of the german tracing app I configured an i9305 (S3 LTE) with Lineage, OpenGApps and the
> app (this not free SW) and *it seemed* to work but did **not**!
> Reason was that BLE beacons could be received, but not sended. I checked that with 'nRF Connect' which is
> able to trace bluetooth packets and shows additional information also for BLE beacons. Furthermore it is
> able to list device capabilities. In my case 'Periphal mode' was not availabble.
> At the end I have tried dozen different (questionable) ROMs till I found one which in fact supports pheriphal
> mode and thus was able to send BLE beacons.

I found an apk for the proprietary 'nRF Connect' app 
(no.nordicsemi.android.mcp) online and installed it. It does indeed 
list 'Peripheral mode' as not supported on this device (basic 
Bluetooth Low Energy is, though). I only have one other android 
device, which has very old software on, and I couldn't get nRF 
Connect to install, so I couldn't test whether my phone was indeed 
successfully sending BLE beacons, but presumably not. There is more 
information about this "peripheral mode" being essential to the 
sending aspect of exposure notifications at [0].

There is still some utility in running the app in "receive only" 
mode, as I would still be alerted if someone I was in contact with 
sufficiently closely tested positive and could have been infectious 
at the time. But obviously it's nicer to be able to return the 
favour.

> I am still not sure if the signal strength is as strong as it should be. My gut feeling is that it is too
> weak which would make the estimated risk wrong and thus useless.

I have no idea how one could test such a thing. It does seem to be 
collecting lots of IDs when I've been out, more than I would expect 
if it was weaker than it should be. But I have no way of knowing 
either.

> Furthermore I at least know that the i9305 is able to send them if used kernel/firmware supports it. Dunni
> about i9300.

That is interesting indeed. I wonder what would be required of 
replicant to support it. Are you able to compare whether the 
firmware in your working i9305 is the same as the one that you would 
have used with replicant, presumably that listed in 
i9305-fwinfo/bluetooth.txt from 
https://code.fossencdi.org/firmwares_nonfree.git ? If it is the 
same, then at least we know that it's likely to be a kernel, rather 
than firmware, issue. And if not, then potentially finding some 
newer firmware could be enough to fix it.

Thanks again,

Nick

0: https://github.com/corona-warn-app/cwa-app-android/issues/688