I've gotten this before on RHL9, with a USB keyboard. It
doesn't seem
to be happening recently though.
It could be kernel, X, or the desktop env, pretty hard to track down
especially when it's not reproduceable.
The older uhci USB code would repeat events - I saw that with USB mice
too. That seems to be fixed. The other case I know about is toshiba laptops
but we *should* have a workaround in place that stops this occuring.