Re: Weird display problem in recent current

From: Kevin Oberman <oberman_at_es.net>
Date: Fri, 15 Apr 2005 14:39:14 -0700
> Date: Fri, 15 Apr 2005 10:00:32 -0700 (PDT)
> From: Doug White <dwhite_at_gumbysoft.com>
> 
> On Tue, 12 Apr 2005, Kevin Oberman wrote:
> 
> > Since I updated my current kernel yesterday, I have been seeing a weird
> > problem that appears to be in X. I did not have the problem with the
> > build I did on Friday, Apr. 8.
> >
> > I run gkrellm, a GTK based system monitor. I also run Gnome 2.10 and
> > X.org. All of these ports are up to date. System is a P4M IBM T30.
> >
> > After my new kernel was installed yesterday, gkrellm would work until
> > the screen blanks (screensaver set to blank). When the screen is
> > re-activated, all of the applications refresh except that one. I just
> > have an empty grey are where gkrellm should be.
> 
> This is indicative of gkrellm not monitoring the X socket and getting the
> refresh message.  This means its stuck doing something.

That's exactly what I had assumed, although I was less sure when I sent
the message than I am now. 

> > top shows that gkrellm is continually in RUN state, although it is using
> > very little CPU.
> >
> > The second problem is that Gnome will no longer shut down. (This may be
> > an artifact of th problems gkrellm is having.) I have to
> > CTRL-ALT-BS. (Ugh!)
> >
> > I'd look at gkrellm except that I only updated the kernel on Monday, not
> > world, let alone the gkrellm port. I thought some user-land/kernel
> > issue may have cropped up, so I just finished rebuilding both the kernel
> > and world. No change in behavior.
> >
> > Any ideas what might have changed between Friday morning and today to
> > cause this? I'm not even sure where to start looking.
> 
> Being that gkrellm grubs around in the kernel, its possible some data
> structre it was looking at changed and now its looping off into space.
> I'd ktrace it and see what its doing when it goes awry -- that might help
> isolate the affected module. If that isn't useful then remove modules
> until you find the broken one.

This is probably in the right direction, it won't be fast. The time it
takes to lock up can be hours. That's why I am suspicious that it is
some sort of resource exhaustion rather than simply pointing out to
nowhere. ktrace is the obvious tool, but, having never used it before, I
did not think of it.

Thanks so much for getting back to me on this. My -current system is
down for a while as I am building a new system on it and need to get the
5.4-RC2 disk burned.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman_at_es.net			Phone: +1 510 486-8634
Received on Fri Apr 15 2005 - 19:39:16 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:32 UTC