Error on Linux while init'ing HCI about ble HOT 4 CLOSED

As far as I can tell, something in here is failing: https://github.com/currantlabs/ble/blob/master/linux/hci/hci.go#L187-L198

from ble.

I've discovered that after the socket is bound, there is sometimes data to be read on the socket. This is throwing everything off because the code doesn't expect anything yet. When it does start sending commands, it gets confused because the response isn't what it was expecting.

After the socket is bound (https://github.com/currantlabs/ble/blob/master/linux/hci/socket/socket.go#L107) if I add some code to sleep for 50ms and then read from the socket to clear it out, then it works flawlessly. I can submit a PR with this, but I'm still a little unsure if it's just an issue with my particular hardware or what. It may be that after an "hci down" and "hci up", something gets written to the socket when the bluetooth hardware comes back up, which may explain the time delay issue.

Any thoughts?

from ble.

Let me try it out on some of our hardware here; we haven't seen that issue, but doesn't seem like this would hurt aside from the (very slight) delay.

from ble.

Yeah, it may be something very specific to my hardware. I'm testing this with ubuntu running in VirtualBox on macOS, with a USB bluetooth dongle (you can't get a VM to attach to a Mac's builtin bluetooth). It may be something with the dongle I'm using. I've previously had good success with running bluetooth stuff in linux with this setup though.

from ble.