As I'm sure anyone who has ever written an autonomous OpMode knows, if your thread does not exit within 1000ms of a stop being requested, the watchdog growls, and the app force-crashes itself in order to terminate your rogue thread. (This is necessary because it is not possible to kill -9 a single thread).

However, I have found a serious bug in the implementation of the watchdog: it appears that the watchdog of the active OpMode is abandoned when a "Restart Robot" command is issued. This causes the RC device to perform a restart robot as if everything were normal, but the rogue thread continues to run in the background.

Here's a sample OpMode:

@TeleOp
public class RogueLoop extends LinearOpMode
{
    @Override
    public void runOpMode()
    {
        waitForStart();

        while (true) //muahaha Thread.interrupt() ain't gonna do you any good here!
        {
            long iT = System.currentTimeMillis();

            while (System.currentTimeMillis() - iT < 500)
            {
                Thread.yield();
            }

            System.out.println("Sysout from rogue loop");
        }
    }
}

Steps to reproduce:

1. Deploy the SDK to the RC phone with the above OpMode added to the TeamCode module
2. Ensure that Android Studio is printing logcat to the bottom pane
3. Ensure the DS and RC phones are connected as normal
4. Init and run the "RogueLoop" OpMode
5. Observe that the OpMode prints a message to logcat every 500ms
6. While the OpMode is still running, press "Restart Robot".
7. Wait for the "Restart Robot" to complete
8. Observe that the rogue OpMode continues to print messages to the logcat every 500ms, despite the DS and RC showing every indication that only the "StopRobot" OpMode is being run