Bug Report: Hard crash (freeze) with nVidia GTX 460

[artem]

If someone in this thread really wants to be useful (instead of occasionally whining) and really wants to identify/help solving this problem, then try to debug this issue using a serial console: http://www.av8n.com/computer/htm/kernel-lockup.htm

I have attached this document just in case.

[drhirsch]

Originally Posted by elvanor

*all* 460 GPUs; what I mean is that is not related to a particular, faulty, card.

No definetly not. I have a gtx 460 (Gigabyte 1 GB OC), and I am using it a lot playing wow under wine, in linux. I also use kde with a lot of desktop effects activated. I have yet to see a freeze in X.

I am using a Gentoo (so it's not Gentoo related) and a recent self built 2.6.37, arch is an amd64.

It may be a hardware incompatibility or even a other defective part, which would explain why the symptom is still there after a card replacement. I really recommend running the usual stress tests to make sure it isn't a hardware error.

[TwinFinity]

Try adding "pcie_aspm=off" to the boot config.

This command sorted out problems for a few people at the end of last year but isn't referred to outside of the thread.

[N3o]

Same happens to me, moving the mouse around or switching between windows and suddenly all hangs and freezes. I have to shut down the system manually. This thing happens at least once a day.

I've even reinstalled the system and this problem is still present.

Video card: Nvidia 9600 M GS @ Dell Vostro 1720
OS: Mandriva Linux 2010.2 Free - 64 bit
video driver: dkms-nvidia-current-260.19.29-2mdv2010.2
kernel: kernel-desktop-2.6.36.2-2mnb-1-1mnb2

About the bug-report log, I will try to get one...

[mondragor]

@artem: Thanks for the input. Actually, that really helps me as I am not experienced in these sort of things, and getting advice on how to systematically search for errors is highly appreciated.
However, it will take some time before I can really sit down and approach this, as this will be a rather lengthy procedure for a layman like me (it's ~10 years since I last compiled a kernel ) and I am rather involved at work these days.

@TwinFinity: Will try this when I get home

[artem]

mondragor,

I'm not really forcing anyone to follow those procedures (they are quite complicated, tedious and time consuming), but since no one can give a kernel backtrace, NVIDIA guys certainly won't help you. So, either you help yourself or no one will.

And don't thank me as yet, googling for those instructions took just a few minutes for me, it's now your turn.

[elvanor]

Adding "pcie_aspm=off" to the boot config does not help for me. I added it and still get random freezes.

[mondragor]

Originally Posted by elvanor

Adding "pcie_aspm=off" to the boot config does not help for me. I added it and still get random freezes.

Same for me, sadly. Actually, I capitulate. I just don't have the time and nerves to continue searching for an error that seemingly only a handful of people in the whole world have. A colleague gave me his 260GTX for testing which after 5 days of running non-stop seems to be completely stable. As the 460 was a free replacement for my 9800 anyway, I will exchange cards with him and we are both happy ...

[Civilus]

I've triggered something that looks similar on my GTX 476.0 (Zotac AMP) after recent upgrade (was Intel C2Q Q6600, 4GB Ram, ASUS P5Q Pro Turbo, now - Core i7 2600k, 8GB Ram (2x4), ASUS P8P67 Deluxe):
Sometimes I see garbage on screen (I can connect via SSH, but can't restart X), but sometimes it just hangs up (no ping or something else). Temporary workaround is to disable Composite in KDE

Sorry for almost non informative log, but it usually hangs up, it was the only time I've could ssh to PC.

I've checked drivers: 260.19.26, 260.19.29, 260.19.32, 270.18 (log with 270.18).
Kernel: 2.6.37, 2.6.38-rc1-linux-next (log with 2.6.38).

Last (and the most suspicions) thing in dmesg:

Jan 26 00:01:53 civil-home kernel: NVRM: Xid (0001:00): 31, Ch 00000001, engmask 00000101, intr 10000000
Jan 26 00:01:57 civil-home kernel: NVRM: Xid (0001:00): 32, Channel ID 00000001 intr 00804000
Jan 26 00:01:57 civil-home kernel: NVRM: Xid (0001:00): 55, CMDre 00000000 00000000 00000000 00000001 00000001
Jan 26 00:01:57 civil-home kernel: NVRM: Xid (0001:00): 31, Ch 00000001, engmask 00000101, intr 10000000
Jan 26 00:01:57 civil-home kernel: NVRM: Xid (0001:00): 32, Channel ID 00000001 intr 00800000
Jan 26 00:01:57 civil-home kernel: NVRM: Xid (0001:00): 32, Channel ID 00000001 intr 00800000
Jan 26 00:01:57 civil-home kernel: NVRM: Xid (0001:00): 32, Channel ID 00000001 intr 00800000
Jan 26 00:01:57 civil-home kernel: NVRM: Xid (0001:00): 32, Channel ID 00000001 intr 00800000

Maybe it's different bugreport, but...

p.s. system is perfectly stable in Windows.

[Civilus]

It's soft/hard lockup + X-crash (black screen after crash).

Will memory dump help? I'm trying to get it from kdump...

[elvanor]

Disabling Composite in KDE seems to help; I have not experienced a crash in 5 days.

I am still waiting a bit more before declaring the system totally stable though...

[artem]

Civilus

It'll probably help, but it'd be nice to see OOPS/panic message as well.