XOrg freeze that affects a lot of people

[Charles Goodwin]

On Sun, 2005-03-20 at 15:13 -0500, Michel Dänzer wrote:
> On Sun, 2005-03-20 at 14:53 -0500, Anton Markov wrote:
> > Once again, it is weird that the keyboard can not be used to kill the 
> > Xorg server; often not even Alt-SysRq works. Would a driver problem lock 
> > the Xorg server so badly that it doesn't let the kernel see the 
> > keyboard? Or is it the CPU usage that prevents keyboard events from 
> > being read?
> 
> If Alt-SysRq doesn't work, the machine is probably either spinning in
> kernel mode or dead at the hardware level.

I consider myself fairly technologically literate (programmer, used
Linux/X for 3 years as my desktop, computers for 20+).  I did not know
that Alt-SysRq existed.  What do you expect a normal user (the average
user of a desktop) to do?

> > Yes, the ATI drivers have been suffering from this problem to a varing 
> > degree since at least Xorg 6.8 (that's when I started using Xorg).
> 
> It's very unlikely the same issue in both drivers though. And as has
> been pointed out several times already in this thread, issues that are
> specific to drivers that don't ship with X.org are off topic here.

Again the issue I raise is being misconstrued / confused / diverted.
The issue is not a specific bug with a specific driver.  The issue is
that XOrg locks up when something unexpected occurs with the driver.  If
it's locked up, it's still running.  If it's still running, it hasn't
crashed.  If it's still running, it could surely be recovered.  Right?

Or am I way off base?  (Not implausible.)

If I'm not, then please don't misunderstand me.  I'm not condoning buggy
graphics drivers.  Any such bugs should obviously be fixed.  However, to
use a bad car metaphor, just because the engine dies doesn't mean the
wheels should fall off.

- Charlie

Charles Goodwin <charlie at vexi.org>