Bluetooth: hci_sync: Don't wait peer's reply when powering off

Currently, when we initiate disconnection, we will wait for the peer's
reply unless when we are suspending, where we fire and forget the
disconnect request.

A similar case is when adapter is powering off. However, we still wait
for the peer's reply in this case. Therefore, if the peer is
unresponsive, the command will time out and the power off sequence
will fail, causing "bluetooth powered on by itself" to users.

This patch makes the host doesn't wait for the peer's reply when the
disconnection reason is powering off.
Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@google.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>

net/bluetooth/hci_sync.c

@@ -5108,10 +5108,12 @@ static int hci_disconnect_sync(struct hci_dev *hdev, struct hci_conn *conn,
 	cp.handle = cpu_to_le16(conn->handle);
 	cp.reason = reason;
 
-	/* Wait for HCI_EV_DISCONN_COMPLETE not HCI_EV_CMD_STATUS when not
-	 * suspending.
+	/* Wait for HCI_EV_DISCONN_COMPLETE, not HCI_EV_CMD_STATUS, when the
+	 * reason is anything but HCI_ERROR_REMOTE_POWER_OFF. This reason is
+	 * used when suspending or powering off, where we don't want to wait
+	 * for the peer's response.
 	 */
-	if (!hdev->suspended)
+	if (reason != HCI_ERROR_REMOTE_POWER_OFF)
 		return __hci_cmd_sync_status_sk(hdev, HCI_OP_DISCONNECT,
 						sizeof(cp), &cp,
 						HCI_EV_DISCONN_COMPLETE,