Comment 18 for bug 1750146

Revision history for this message
In , hurikhan77+bko (hurikhan77+bko-linux-kernel-bugs) wrote :

Yeah, it seems to get worse with every 5.4 LTS update. Worked almost flawless about some months ago (despite taking something from 30-60s to successfully connect). But I don't see a correlation to any commits. But now, it's almost impossible to connect successfully, it usually disconnects after a 1-2 minutes (coincidentally just when entering a game).

I usually also see messages like this:

During use of the controller (probably around disconnect time):
[ 5304.480502] Bluetooth: Unexpected continuation frame (len 52)

When plugging the BT dongle:
[ 5241.926752] Bluetooth: hci0: unexpected event for opcode 0x0000

While it's in a loop of constantly connecting and disconnecting:
[ 1120.736530] Bluetooth: hci0: hardware error 0x58
[ 1120.923550] debugfs: File 'le_min_key_size' in directory 'hci0' already present!
[ 1120.923555] debugfs: File 'le_max_key_size' in directory 'hci0' already present!
[ 1120.923557] debugfs: File 'force_bredr_smp' in directory 'hci0' already present!

Other BT devices seem to just work fine tho I don't use a lot of them and only occasionally. It seems that the Xbox One S firmware does something strange in the BT protocol. It also needs either ERTM disabled to connect. Alternatively, one can apply this patch to be able to use ERTM:

https://github.com/kakra/linux/commit/c8b24d83f227a7fecfa9420d6756074e8f9b542c

I'm also feeling there's some internal state getting messed up within the controller on each BT pairing: Results are not really reproducible. I managed to get it to work flawlessly one time (stable connection despite 20-30s connect handshake time), then purged my Bluetooth config from /var/lib, re-paired the controller and now it's a mess - still with the same kernel version.