James Wetterau writes:
> > hypothesis would be to kill your X server the same way that init would 
> > kill it in going single user and see if your console is latched.
> ...
> 
> Say this turns out to be correct.  Is there an easy remedy for this
> symptom then?

If you kill X (with a HUP or whatever else), the server should be
calling a handler routine to clean up the console state before
exiting. Given that it is not doing this, what you should do is get
out the sources for the X server you are running and figure out what
is going wrong. The first step is to look at the xsrc sources and see
if you can figure out what they are *supposed* to be doing on a clean
exit and why they aren't doing this in the signal handler. If it
appears that the signal handler is there, it might help to try running
gdb on the X server while it is running from another host on your
network -- I'd ssh in from Ruth's laptop, attach to the X server, and
see if you can figure out what's going wrong.

In any case, ultimately you'll need to develop a code patch and submit 
it back to the XFree folks -- luckily, this being open source
software, they are likely to take it.

BTW, arguably, the console driver itself should detect that the thing
that opened it in non-text mode has exited and clean up. Figuring that 
out requires some kernel code hunting, and is even harder.

Let me know in private mail if you need help in the course of tracking 
down the issue.

Perry