X Windows appears to lock up

[Thomas Molina]

I've got this problem where the X Windows system appears to malfunction 
after some indeterminite time under Fedora Core 2 test 1.  Unless I am 
doing something wrong it probably needs putting in bugzilla, but my 
observations to this point are rather amorphous so I will describe it and 
solicit suggestions on how to proceed for further data gathering I can put 
into bugzilla.

I have the screensaver set to blank after five minutes, cycle after five 
minutes, and lock the screen after zero minutes.  This gives me a random 
screensaver which cycles between random patterns every five minutes and 
locks the screen as soon as it kicks in, which is excatly what I want.  If 
I watch, it cycles, blanks, and locks as expected.  However, after some 
indeterminite time greater than one hour and less than six hours X goes 
away and refuses to come back.  I say greater than one hour and less than 
six hours because I have watched it for an hour at a time and saw it act 
normally, but if I leave it operating overnight X is locked up when I get 
up in the morning.  

When I say locked up I mean I see a blank screen with no graphics, no 
mouse cursor, and no ability to kill X from that screen.  Ctrl+Alt+Bkspc 
does not kill the X Server, and Ctrl+Alt+Fx doesn't switch to other 
virtual consoles.  However, everything else is unaffected.  The mailserver 
is functioning normally, nfs shares are served out, and I can telnet in 
with no problem.  I just can't kill X, nor switch to another console to 
kill it.  I can telnet in and kill -9 it or, if I have previously switched 
to another console, I can kill -9 it there.  

There does not appear to be anything relevant in /var/log/messages, nor in 
XFree86.0.log.  It has happened with all the Fedora Core 2 test kernels, 
as well as with the stock Linus 2.6 kernels. It does not happen with 
Fedora Core 1 under any conditions.  

So there you have it.  I have a description of a problem, but not a lot of 
hard data or debugging material.  It seems I may have missed something, 
but I am not sure what.  It also seems to be an X problem, but that is not 
conclusive.  Otherwise it wouold have happened in Core 1 or behaviour 
would have changed with the switch to x11-org.

Can someone suggest something I have missed or some way to gather more 
data for bugzilla?