DF-0683 / run.log
=== DF-0683 live-trigger check (6.5-DEVELOPMENT #0) === $ ifconfig | grep -i bt # (empty) $ ls /dev/bt* /dev/bluetooth* ls: No such file or directory $ sysctl net.bluetooth.hci # (no hci_unit registered) === Conclusion === No Bluetooth hardware in the VM -> no hci_unit -> hci_disable is never called. Race cannot be exercised here. Source-level analysis (see VERDICT.md) shows the race is real: hci_disable tears down hci_links / rx queues / eventqlen WITHOUT hci_devlock, while hci_intr (netisr-driven) holds hci_devlock around its concurrent reads of the same fields.