[TriLUG] FC1, nvidia, switch to console causes fatal system lock

[Timothy A. Chagnon]

On Sun, 2004-02-22 at 18:22, crimsun at fungus.sh.nu wrote:
> On Sun, Feb 22, 2004 at 11:43:43AM -0500, Timothy A. Chagnon wrote:
> > Here's the problem.  After having my computer running a couple days or
> > more, if I try to switch to a console, logout, or shutdown the computer
> > the screen goes black and stays black.
> 
> Check and make sure you're not using framebuffer for anything in the
> kernel, eg. grep FB /path/to/kernel/config should return:
> # CONFIG_FB is not set
> (in 2.6.3+ it will return # CONFIG_BLK_DEV_OFFBOARD is not set as well).
> If you have any entries besides those two, you will need to reconfigure
> your kernel by disabling those entries, and then you will need to
> recompile your kernel and the Nvidia kernel module.

Thanks for the idea.

Well, I took out CONFIG_FB as well as a bunch of other stuff, but after
a couple days of uptime I still had the problem.  Looks like it's off to
submit a bug report.

What I'm curious about is why it never did this with RH9.  Do you think
it could have anything to do with the fact that I upgraded to FC1
instead of doing a fresh install?  I notice I don't get graphical
startup/shutdown which is a holdover from my RH9.  Or could it be
something with the ntpl pthreads kernels that are on FC1?  Also, do you
think this is a kernel panic or what?  I don't get any information when
it happens.  Shouldn't a kernel panic at least write to
/var/log/messages?

> Rivafb is a known no-go with the binary-only Nvidia drivers. The two
> enjoy stepping on each other's toes. Vesafb is more problematic because
> it "works" for some people some of the time. The last I looked at it I
> was seeing register stomping, but that was several revisions ago.

I'm assuming these things were sub sections of CONFIG_FB, I didn't see
anything about them after I took out framebuffer support in make
xconfig.

-Tim
-- 
Timothy A. Chagnon <tchagnon at nc.rr.com>