kernel oops with 2.6.16 and 1.0.8762

logan · 06-23-06, 04:00 PM

Quote:

NVRM: VM: nv_kern_close: 0xf7f97a80, 257 page(s), count = 1, flags = 0x00020002, 0xf1636000, 0xe9ef6000
NVRM: VM: nv_kern_close: 0xf7f97940, 1 page(s), count = 1, flags = 0x00000001, 0x37d63000, 0xdba0b840
<snip>
NVRM: VM: nv_kern_close: 0xc9d6f800, 1 page(s), count = 1, flags = 0x00000001, 0x1c1ec000, 0xdba0bda0
NVRM: VM: nv_kern_close: 0xc9d6f680, 259 page(s), count = 1, flags = 0x00020002, 0xf0616000, 0xcac57800
slab error in kmem_cache_destroy(): cache `nv_pte_t': Can't free all objects
[<c0141922>] kmem_cache_destroy+0x79/0xf4
[<f917a4a1>] nvidia_exit_module+0x1bd/0x1cf [nvidia]
[<c013839f>] unmap_region+0xa4/0xeb
[<c01290c1>] sys_delete_module+0x164/0x18c
[<c01389ac>] do_munmap+0x196/0x1b2
[<c0102541>] syscall_call+0x7/0xb

This has happened twice (with 2.6.16+1.0.8762) and the oops showed up when trying to unload the nvidia module.

The first timed seemed to happen out of the blue on a stable system (60+ day uptime) and no problems with any games (jk2, jk3, mainly ut2004 - played daily). I assume the reload was attempted after I saw the NVRM messages when investigating horribly slow games, but don't really remember... q3a was basicly stuck on the menus with ~0fps and although ut2004 was more or less ok, it wasn't 100% as I experienced extremely low fps in areas that were ok before and after. The second happened shortly before or during my attempted run of the Prey demo (cedega). This time q3a (menus) was ok, but jk3 was not.

I've recently switched to debian's nvidia-glx/kernel-source packages, but I don't believe there's any relation. Nothing on the forum or google, maybe it's just me... Anyone else?

netllama · 06-23-06, 04:15 PM

I have a few questions:
0) You stated that this Oops was with 2.6.16, yet your bug reports shows that you're running 2.6.17. Can you clarify which kernel version triggered the Oops?
1) Is there any specific series of events that reliably triggers this Oops, or is it seemingly random?
2) Have you tried unsetting NvAGP, to see if that has any impact?
3) It looks like you have an Asus K8V SE Deluxe motherboard. If so, can you also try updating to the latest BIOS, as the one you're using is 3 versions behind the latest?

Thanks,
Lonni

logan · 06-23-06, 04:53 PM

Quote:

Originally Posted by netllama

I have a few questions:
0) You stated that this Oops was with 2.6.16, yet your bug reports shows that you're running 2.6.17. Can you clarify which kernel version triggered the Oops?

I was running 2.6.16 at the time of both, but rebooted to 2.6.17 yesterday evening. Didn't think about creating the bug report prior to this and I'm at work now.

Quote:

Originally Posted by netllama

1) Is there any specific series of events that reliably triggers this Oops, or is it seemingly random?

I thought it could have been caused by abnormal termination of GL games. ut2004 frequently hangs on exit (possibly due to TLS), so I used to always CTRL+C the process. If that were actually the problem, it should not have happened the second time because Prey would have been the first game killed off and it was the first game to display awful performance. Nothing else comes to mind...

Quote:

Originally Posted by netllama

2) Have you tried unsetting NvAGP, to see if that has any impact?

Should that make a difference since the default is to try AGPGART first? I'd have to boot with agp=off to use anything other than AGPGART.

Quote:

Originally Posted by netllama

3) It looks like you have an Asus K8V SE Deluxe motherboard. If so, can you also try updating to the latest BIOS, as the one you're using is 3 versions behind the latest?

All of the changes in newer versions amount to:

Quote:

Solve keyboard may be no response when cmos checksum bad
Support new CPUs.

... but I can try it anyway.

netllama · 06-23-06, 04:59 PM

Its unlikely that the BIOS changelog is the full extent of the changes. Typically there are hundreds of bugs fixed, and only the highlights are posted publicly.

It sounds like you haven't hit this (yet) with 2.6.17. Please let me know if and when it reproduces under 2.6.17 (plus with a newer BIOS).

Thanks,
Lonni

logan · 06-23-06, 05:06 PM

If and when it happens again, I'll create the log right after unloading the module and post my kernel config.

Thanks for the suggestions.

-- edit

Second time around with Prey was no sweat, runs great with cedega.

K8V SE Deluxe BIOS upgraded to 1007.

06-23-06, 04:15 PM	#2
netllama NVIDIA Corporation Join Date: Dec 2004 Posts: 8,763	Re: kernel oops with 2.6.16 and 1.0.8762 I have a few questions: 0) You stated that this Oops was with 2.6.16, yet your bug reports shows that you're running 2.6.17. Can you clarify which kernel version triggered the Oops? 1) Is there any specific series of events that reliably triggers this Oops, or is it seemingly random? 2) Have you tried unsetting NvAGP, to see if that has any impact? 3) It looks like you have an Asus K8V SE Deluxe motherboard. If so, can you also try updating to the latest BIOS, as the one you're using is 3 versions behind the latest? Thanks, Lonni

06-23-06, 04:59 PM	#4
netllama NVIDIA Corporation Join Date: Dec 2004 Posts: 8,763	Re: kernel oops with 2.6.16 and 1.0.8762 Its unlikely that the BIOS changelog is the full extent of the changes. Typically there are hundreds of bugs fixed, and only the highlights are posted publicly. It sounds like you haven't hit this (yet) with 2.6.17. Please let me know if and when it reproduces under 2.6.17 (plus with a newer BIOS). Thanks, Lonni

06-23-06, 05:06 PM	#5
logan Registered User Join Date: Aug 2002 Posts: 492	Re: kernel oops with 2.6.16 and 1.0.8762 If and when it happens again, I'll create the log right after unloading the module and post my kernel config. Thanks for the suggestions. -- edit Second time around with Prey was no sweat, runs great with cedega. K8V SE Deluxe BIOS upgraded to 1007. Last edited by logan; 06-23-06 at 09:55 PM.