Strange crash on booting X with nvidia drivers

[Arthur]

I've recently had a problem with the nvidia drivers, that causes X to crash on bootup. Just after it has initialised agp, and it switchs to my desired resolution. All I get it a screen full of ascii garbage, mostly the letter 'P' though.

When this happens I'm totally locked out, all I can do is ctrl-alt-del, and boot into single user mode, so I can get gdm to use a different driver.

I'm running redhat 7.3, XFree86 4.2.0, kernel 2.4.18-3. I've tried driver releases 3123 and 2960, both with the same problem. I've tried disabling the AGP stuff (option NvAGP 0). I have a SIS 650 chipset on a Gigabyte 8SIML mobo, celeron 1.7Gig cpu. The graphics card is an Asus 6800 32Mb (geforce 1 DDR).

Yes, my kernel version is different to the rpms you have precompiled, so I rebuilt the source rpm.

The 'nv' driver that comes with XFree86 works. As does the latest windows nvidia drivers.

As I mentioned earlier this has only happened recently, I have had the nvidia drivers working previously, which strikes me as strange. Possibly some redhat update not compatible. Could be down to me not running kernel 2.4.18-10...

Any ideas??

Arthur

[hapena]

This crash happens every time X starts? Could you provide XF86err.log?
I have had these kind of troubles in an very stage after switching to 2880 drivers and it was somewhat buggy in the xfconfig file.

[Arthur]

Yes this happens everytime I boot X with the nvidia drivers. With any other driver ('nv' or 'vesa') it works fine, albeit dog slow.

The log file is a bit big for this forum, so I've put it on my website:-

XFree86.0.log (http://fubaby.com/XFree86.0.log)

I couldn't spot anything blindingly obvious in it.
Note this one, is with the agp disabled.

Arthur

[bwkaz]

If you load "glx" in your config file, does that help?

[Arthur]

I've tried enabling and disabling the 'glx' module - made no difference.

Tried 16 & 24 colour depths.

Tried going back to driver vers 2880. Made no difference.

One interesting thing is, when it crashes, according to top, X is hogging like 99% of the CPU time. Which makes me wonder if it busy waiting for something. Something blocking it.

So I'm still at a loss.

Arthur

[Arthur]

Been digging around at this problem and I'm getting some hints it could be an overlapping mtrr problem.

my /proc/mtrr looks like this:-
reg00: base=0x00000000 ( 0MB), size= 128MB: write-back, count=1
reg01: base=0xe8000000 (3712MB), size= 32MB: write-combining, count=1
reg02: base=0xe0000000 (3584MB), size= 32MB: write-combining, count=1

The last two obviously overlap. I've had a couple of warnings about this in /var/log/messages when I start/crash X. And also when I try to use the rivatv drivers to get my video-in working.

My motherboard has its own Savage4 graphics card on board. Which I'm wondering, may be causing some kind of mtrr clash...

Any thoughts?

Arthur.

[Arthur]

Turned out to be some dodgy memory, causing all kinds of problems. Has been basically causing all kinds of other instabilties too.

Lots of NULL pointer errors in the kernel...

Learn me for buying cheap no name memory.