Our project uses a Raspberry Pi compute module 3 connected via UART to 2 nRF51822. I'll go through what is working and than describe the issues I'm having
What's going on without issues:
btattach -B /dev/ttyAMA0 -S 115200 -P h4
btmgmt --index 0
Static address successfully set
LE Device Address: FF:22:33:44:55:FF
systemctl status bluetooth
The BD address is always 00:00:00:00:00 and this seems to have implications on tools like bluetoothctland others.
I'm basically trying every tool every command line I can find on Google (e.g. this link) to set the address and there's always errors.
BD Address: 00:00:00:00:00:00
Manufacturer: internal use (65535). Device address: 00:00:00:00:00:00 (XEROX CORPORATION)
bdaddr -i hci0 FF:22:33:44:55:FF
No default controller available
Controller 0 not available
As per "Raju" on the comments section of the nRF5x support within the Zephyr Project RTOS blog post it seems that it seems that there's some issue on mapping the address on Zephyr, but I'm not sure if that is my issue or something unrelated.
I've tried to find on the source code of Zephyr anything related to the bd address, how to set it, or how to change it, and I couldn't find anything. If needed I can recompile it with extras flags, or include some params, extra protocol. But I'm really running out of options here.
Any help will be deeply appreciated.
Hi, thanks for pointing it out the direction of the code. But the issue here is the public address. It's always zero on the linux host which causes issues. We do have a address space from IEEE. How can…
Please address your questions to the Zephyr support team. They can be reached for technical questions, through the mailing lists or IRC channels for prompt assistance.
Thanks, Nordic employees help follow up the requests in Zephyr support portals, see Carles' answer in this thread.
Valid observation. We have our own OUI acquired from IEEE that we should use at the final product. But for now, just something that works would be great.
The code that set random address from FICR on the nRF5x devices in not included in the hci_uart example, only on host + controller samples. If you want to use random addresses, you can patch the sample code with this code that ge and set the random values from FICR. You can also set a public address the same way, if you have acquired one from IEEE.
Hi, thanks for pointing it out the direction of the code. But the issue here is the public address. It's always zero on the linux host which causes issues. We do have a address space from IEEE. How can I set it using the hci_uart?
If you are not getting any adapter when calling bluetoothctl show or bluetoothctl list that means bluetoothd was not able to initialize your interface.
The lack of a public address is expected here, but bluez should still set a random or static address when starting (depending on your config). Try manually running bluetoothd with debug (-d) and see if you get any error.
Had a similar issue on a different linux host and my problem was because bluez was failing to generate a random address (bt_crypto_new was returning an error). Investigating further, my issue was because my kernel was missin CRYPTO_USER, so I just enabled the following configs:
After rebuilding my kernel with the additional options, bluez was able to generate a random address and successfully initialize the interface:
root@hikey:~# bluetoothctl show
[NEW] Controller C5:62:65:D2:5C:3E hikey #1 [default]
This readme has a link to a patch that helped me with the BD Address issue: github.com/.../README.md. First I could not completely run diretcly the patch, but then I made the changes manually. There were some folder names wrong also I think that's because of versions of zephyr. hope that helps.