Okay, but here's a potential bug.
Ah yes, I am seeing that now for the first time here, and I have a pretty good idea of what to look for. Definitely caused by Hijack, I think, and only recently being noticed due to Hijack injecting extra lines at the serial port.
I think it is somehow tossing data into the output FIFO without re-enabling the serial port TX interrupt in the cases where that interrupt is not already enabled.
Hitting ENTER is a great clue, because that would trigger a serial RX interrupt, and the processing thereof would also re-enable TX.
So I know where to look for it.
