FC4 system freeze after nvidia driver install

[sr20dem0n]

I recently got a new machine up and running but have hit some snags with the nvidia drivers. As far as I can tell there are no indications of a problem in the logs, the system simply freezes solid anywhere between 5 minutes and 1 day after the boot. It runs fine with the default linux driver, it's simply after upgrading to the nvidia driver that it has problems. Nothing "graphically challenging" is being done on this computer, so I don't need the added performance, but the default linux driver has resolution issues with the DVI output and clarity issues with the VGA output, neither of which are a problem with the nvidia driver (at least until it freezes), so I'd like to get this stability issue ironed out.

System specs:
Dual dual core 3.2GHz Xeon (yes dual, dual core, 4 procs in total)
8GB system memory
500GB boot drive & 3TB raid array
GeForce FX5200 PCI (regular PCI, not express)
Fedora Core 4 x86_64
2.6.17-1.2142_FC4smp kernel
nvidia driver being used:
NVIDIA-Linux-x86_64-1.0-8776-pkg2

Attached is the nvidia-bug-report, it was generated with X running, however the nvidia driver was not in use, I don't know if that will affect the contents. Unfortunately I can't generate the bug report after the problem occurs because the system is frozen solid, I can't even ssh into it.

Also attached is /var/log/dmesg from the most recent boot (unfortunately not using the nvidia driver, I don't have the luxury of rebooting this computer at will ) and /var/log/messages trimmed to the boot right after the lockup (using the nvidia driver)

I checked the stability sticky and am not experiencing any of the symptoms described, however I haven't tried a couple of the fixes that are presented. I am not what you would call a linux expert...I have no problem using linux, fixing most issues, or even setting up a system from scratch like the previously mentioned one, but I have no experience when it comes to recompiling kernels or fixing stability issues such as this one.

If there's anything else that you need I can provide it, just let me know. Any help would be greatly appreciated, thanks

[netllama]

I have a few questions:
0) Does this problem persist with the 1.0-9629 or 1.0-9742 drivers?
1) Does this problem persist with a more recent kernel (2.6.18.x)?
2) Have you verified that you're using the latest BIOS?
3) Does booting with the irq=poll and/or noapic kernel parameters have any impact?

Thanks,
Lonni

[sr20dem0n]

0 - I didn't realize there were newer drivers than 1.0-8776, for some reason it shows that is the newest one for EM64T unless you click on the "Linux, FreeBSD, and Solaris Drivers" link, which I hadn't done - I'll try it out on Monday once once some of these models finish running and I'm back in the office.

1 - All of the kernels that I've installed have been through the YUM manager, where the newest is 2.6.17-1.2142 - I'm sure I could get it going manually with a newer kernel from kernel.org, but we can't handle the down time in case something goes wrong and needs fixing, not at the end of the year at least. I'm willing to try it, but only as a last resort.

2 - It is not running the latest BIOS, I didn't think that would be the problem since it doesn't have a problem with the video card, just the nvidia drivers. If you think it might I'll update it, but again it will have to wait for a few days because it's busy running models right now.

3 - I haven't tried them, but I will and I'll let you know


It's hard experimenting with this since stability is such a grave concern. If it died right on the boot that would be one thing, but sometimes it takes up to a day before it locks up, sometimes right in the middle of a week-long model run (as it did today).

Thanks for the help thus far

[piotrq__]

Guys, if this is such a mission-critical system, then why don't you try the "nv" driver, which is shipped with Xorg ? It should give you stability and performance good enough for normal use (unless you fire-up Quake when nobody's around ;-) ). This way you'll be able to use that system and wait with the "experiments" 'till you have time and a chance to do that.

Just my 2 cents.

[sr20dem0n]

Originally Posted by piotrq__

Guys, if this is such a mission-critical system, then why don't you try the "nv" driver, which is shipped with Xorg ? It should give you stability and performance good enough for normal use (unless you fire-up Quake when nobody's around ;-) ). This way you'll be able to use that system and wait with the "experiments" 'till you have time and a chance to do that.

Just my 2 cents.

That's what I'm doing in the meantime, sorry if I didn't make that clear in the first post. It has been running the nv driver for the last week or so, but I moved it to a new monitor that ran on DVI which is when the resolution issues popped up, so I decided to give the nvidia driver another try after I had updated everything (though not enough, apparently ). It locked up again this afternoon so I reverted back to the nv driver and started searching for the cause of the problem.