Description of problem: For weeks bluetooth audio devices are broken on Intel wifi cards like my AX200. What happens: At some point, the sound will stop playing and after ~10-20 seconds, it will reconnect. If there are other bluetooth device connected like mice, they will also be interrupted for that time. This happens only with bluetooth headset/speakers. A mouse by itself does not cause it. This can happen several times an hour, once an hour or sometimes it can take a day until it happens. The issue was also reported on several other online forums, like this arch thread, which is the most active on this issue: https://bbs.archlinux.org/viewtopic.php?id=263040&p=3 Users there suggest that it might be due to linux-firmware update and/or a combination of that and bluez. The last working linux-firmware version is supposed to be 20201120. This problem persists for several weeks with different versions and kernels but here are the current non-working versions: Version-Release number of selected component (if applicable): linux-firmware: 20210315-119.fc33 bluez: 5.56-1 kernel: 5.11.11-200.fc33.x86_64 How reproducible: All it needs to take is connecting a bluetooth headset or speaker and wait until it hangs and disconnects. As said this can take a while, maybe due to a race condition. Steps to Reproduce: 1. Connect bluetooth headset/speaker 2. Play 3. Wait Actual results: At random times, all bluetooth devices stop working for maybe 20 seconds and then reconnect. Expected results: No random hang and reconnect. Additional info: I posted more diagnostics here: https://ask.fedoraproject.org/t/bluetooth-devices-disconnect-and-reconnect-right-after-several-times-a-day/12400 But I think most of this is useless. The only diagnostics that seems to be consistent is this: Feb 17 12:24:04 pc bluetoothd[1849]: src/profile.c:ext_io_disconnected() Unable to get io data for Headset Voice gateway: getp> I know that this is maybe a problem in linux-firmware or bluez but it happens in Fedora and there is no access to the former project to actually report and track this issue.
Can confirm, the problem persists on F34 beta. Disconnection happen due to timeouts, see here: https://bugzilla.redhat.com/show_bug.cgi?id=1929434 Reverting the firmware as a workaround seems to fix it for me, at least I didn't encounter any new disconnects since. I also opened a thread in the Intel wireless support forum, but Linux support is pretty basic. I really hope their devs take the time to analyze it. But I guess that's up to us endusers. Any recommendations how to debug the firmware?
Manually replacing Intel firmware files from [0] fixed it for me. Lets hope this makes it into the Fedora releases soon. [0]: https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/log/