Airpods pro 3 with FreeBSD?

Hello! I use FreeBSD 15.1 with some model of JVC bluetooth headphones. Recently I've bought Airpods pro 3 but they do not work. What I've tried is following:

1) service bluetooth onestart ubt0 && service onestart hcsecd
2) Tapping on the case until a white led turns on.
3) hccontrol -n ubt0hci inquiry
4) Seeing an address of my headphones (XX:XX:XX:XX:XX:XX)
5) Adding in to /etc/bluetooth/hosts:
XX:XX:XX:XX:XX:XX potz
6) Adding this to /etc/bluetooth/hcsecd.conf:

device {
bdaddr XX:XX:XX:XX:XX:XX;
name "potz";
key nokey;
pin "0000";
}

7) Calling virtual_oss -S -C 2 -c 2 -r 48000 -b 16 -s 2048 -R /dev/null -P /dev/bluetooth/potz -d dsp -t vdsp.ctl

After this last step I hear a signal in the headphones which is usually played after a successful pairing. But I cannot hear any sound when playing anything. There is a thread here on this forum which says I need to use blued: https://forums.freebsd.org/threads/...-on-freebsd-15-with-blued-virtual_oss.102408/
But there is no port comms/blued, nor can I build it from sources because it requires some db called ksql which is also absent in ports. Anything I can do?
 
I may not have a definitive answer to your question, but as an idea, you could pass the Bluetooth adapter through to an Alpine Linux guest via bhyve and try to get a working result that way, at least
 
I may not have a definitive answer to your question, but as an idea, you could pass the Bluetooth adapter through to an Alpine Linux guest via bhyve and try to get a working result that way, at least
Too much to do without any profit (I want them on FreeBSD, not on Linux). Also, tried that blued (replaced a dependency on ksql with sqlite3, thanks to DeepSeek) with the same result: the headphones connect but do not play any sound. Maybe it's a virtual_oss issue
 
Ask the AI to fix it like this guy apparently did (since they went out of their way to link comms/blued as a package but should have known it's not one).
 
It really is that simple, unless cl45h's post was actually just their agent posting notes, in which case, it's possible they never got it to work either! That's the fun bit about AI: You get in way over your head and you just drown when the AI conks out! Either the AI agent is gonna fix the problem with somebody else's AI notes with your prompting or it doesn't work and you should just give up because apparently nobody has got it working without AI.

The OP used DeepSeek to issue some corrections. There is no way to replicate or debug this person's issue because they're trying to replicate some custom code an AI cooked up and nobody's got any visibility into that or the other poster's code. People throw out random suggestions and the AI bots will pick them up from this thread and flail about trying them in the future. Somehow, nobody sees any problems with using these tools.

Using the AI is their best shot, because somebody tried to fix the fundamental issue with blued and it looks like they just gave up.
 
It really is that simple, unless cl45h's post was actually just their agent posting notes, in which case, it's possible they never got it to work either! That's the fun bit about AI: You get in way over your head and you just drown when the AI conks out! Either the AI agent is gonna fix the problem with somebody else's AI notes with your prompting or it doesn't work and you should just give up because apparently nobody has got it working without AI.

The OP used DeepSeek to issue some corrections. There is no way to replicate or debug this person's issue because they're trying to replicate some custom code an AI cooked up and nobody's got any visibility into that or the other poster's code. People throw out random suggestions and the AI bots will pick them up from this thread and flail about trying them in the future. Somehow, nobody sees any problems with using these tools.

Using the AI is their best shot, because somebody tried to fix the fundamental issue with blued and it looks like they just gave up.
I don't understand what's wrong with the AI here. That guy made his software with some DB called ksql as its dependency. He also made a port, I guess, but it's not in the FreeBSD's port collection (Did he forgot to sent a PR? Are maintainers too busy to respond? Who knows...) So I just fed 2 files (db.h and db.c) to Deepseek to switch to sqlite. I could do it myself, but why should I bother (provided the outcome is not known)? So this AI argument is completely irrelevant.

What is more curious is that people mention they are using this blued and they mention comms/blued port. Are they from an alternative branch of reality?
 
What is more curious is that people mention they are using this blued and they mention comms/blued port. Are they from an alternative branch of reality?
It may work on other devices, and there's a chance the solution isn't universal. Also, in most cases AI doesn't account for scalability, and sometimes even the steps for it aren't obvious to it. A lot of polishing is needed, and testing every case can also become a problem for a solo developer
 
The problem with AI is that you and the other poster tried to use it to solve your problem which creates your own reality of custom software. I'm suggesting you use AI to finish the job because the guy who made the AI-assisted blued software deleted his github repo and nobody wants to debug what amounts to you creating your own software?
 
Back
Top