X Freezes at random with NVIDIA driver

MrAster · 2014-04-18 21:33:47

Hi,

I've recently (re)installed Archlinux on my desktop, which is running two GTX 770s in SLI.

At seemingly random moments, the whole system will lock up. Mouse won't move, can't switch to a TTY, nothing. If I press the power button, the HDD light flickers, but no visual output other than that. And the machine doesn't shut down.
Obviously, I want to give my machine a fighting chance, and I go and grab some coffee thinking that all I need to do is wait. But 5 minutes later, nothing has changed.
So after a cold reset, I try again, and everything works as usual, until it happens again.

So, I tried to find a reason for this.
It happened once when I was compiling clang from source with 32 threads (all my cores were at 100%).
It happened once when I was watching a TV show in VLC.
It happened once when I switched tabs in xfce4-settings-manager.

Also, Xorg uses between 7-15% of my CPU at all times. Which seems weird to me.

I kind of assumed it was related to resource usage, but I have nothing to back that.

So.. yeah. I'm at a loss, I have no idea what is causing this, and it's pretty annoying to have to reboot every few hours.

Any help is appreciated, and I'll try my best to help you to help me.

Thanks.

blackout23 · 2014-04-18 21:44:55

Look at your kernel message journal or connect to your PC with another device via SSH and follow the kernel messages live with journalctl -kf and reproduce the freeze.

MrAster · 2014-04-19 08:44:22

My machine has been running for twelve hours without an issue, I'm not sure if that means I should wait longer or not, and I'll be absent for the weekend.
I guess I'll have to pick this back up when I return.

MrAster · 2014-04-22 00:07:05

I'm back, and my machine just froze again.

Apr 22 01:52:36 enterprise kernel: NVRM: GPU at 0000:01:00: GPU-db8532d6-d93f-7534-ca97-3431c6e99f9a
Apr 22 01:52:36 enterprise kernel: NVRM: Xid (0000:01:00): 31, Ch 00000001, engmask 00000101, intr 10000000
Apr 22 01:52:36 enterprise kernel: NVRM: Xid (0000:01:00): 13, 0001 00000000 0000902d 000005e0 00140539 00000000
Apr 22 01:52:37 enterprise kernel: NVRM: GPU at 0000:02:00: GPU-31eb024a-cd2a-1fc8-048d-c3137e466247
Apr 22 01:52:37 enterprise kernel: NVRM: Xid (0000:02:00): 61, 12a4(1f48) 00000000 00000000

So, I guess it is an nvidia issue.
What can I do about this?

Baryon · 2014-04-22 12:28:59

I'm possibly having this problem as well. Do you have KMS enabled? I have disabled it with nomodeset, but the problem still seems to be there, so wondering if it's necessary to add i915.modeset=0 and nouveau.modeset=0 as well.

MrAster · 2014-04-22 12:36:58

No, I don't believe I have KMS enabled. Previously, I had vga=0x34D in my kernel line, but removing it didn't seem to fix anything.

Baryon · 2014-04-22 15:50:56

I believe it is enabled by default.
I am just updating to version 337 of the nvidia driver to see if that fixes it.

Last edited by Baryon (2014-04-22 15:51:33)

Beelzebud · 2014-04-22 20:12:14

I've also been having X freeze on me using KDE and the newest 337 driver. I have 2 GTX760's in SLI mode.

MrAster · 2014-04-22 20:59:17

I added nomodeset to my kernel params and it didn't work, just got a freeze again. I'm also on the newest 337.12 driver.
Looks like this might be an issue with SLI.
*sigh* This is frustrating, because Nouveau doesn't even want to try to detect my second GPU, and then my third screen is useless.

Also, I might mention that I previously had "CoolBits" set to "12" in my xorg.conf, but I removed that and it still freezes.

Taylor001 · 2014-04-26 00:46:13

Same here. Random freezes, no root cause found, nothing strange on logs.
When it does happen I can only hard-reset, sometimes I have to wait like two minutes before logging for a second time, otherwise it freezes right after login.

linux 3.14.1-1
nvidia 337.12-1
xorg-server 1.15.1-1

*UPDATE*
After a freeze, the PC automatically reboots after a minute or so.

Last edited by Taylor001 (2014-04-26 01:03:18)

MrAster · 2014-04-26 00:53:14

Are you using SLI? Which GPUs?

Taylor001 · 2014-04-26 01:04:25

No SLI, I've a single GTX770. Read the updated post, do you have a similar issue?

MrAster · 2014-04-26 01:13:38

No, my PC doesn't reboot after a minute or so. But this is Arch, everybody has a different config.
I also have a (two) GTX 770 in my machine. Maybe it's a bug with drivers for the GTX7XX series?
Ludum Dare is going on right now, but when it's done, I'll post a bug report with nvidia (provided they cared enough to set that up).

Taylor001 · 2014-04-26 01:27:28

Good luck with your Ludum Dare.
In the meantime, I disabled nvidia PowerMizer and freezes seem gone for now. I'll post an update to confirm stability.

MrAster · 2014-04-26 01:29:04

Thanks!
PowerMizer, huh? I didn't have any freezes for two days, and then I rebooted and had three in a 60-minute period.

Taylor001 · 2014-04-26 01:43:32

Yeah, me too.
I had freezes on a regular basis in the last 3 hours, but nothing after disabling PowerMizer. Maybe I'm wrong, we'll see.

MrAster · 2014-04-26 01:49:19

I have no idea how to disable PowerMizer other than set it to maximum performance, so I've done that.
If you're right, I love you.

Taylor001 · 2014-04-26 01:57:56

Haha! However, I disabled it that way too, just to try.

MrAster · 2014-04-26 14:58:41

Look like this is the right solution, albeit very expensive power-wise, and turning my computer into a room heater.
I hope this is something nvidia will fix in an upcoming driver release.

Taylor001 · 2014-04-26 19:33:55

I confirm the stability.
The PowerMizer stuff was a very well-known issue years ago, I thought it was fixed nowadays. Regression?

Last edited by Taylor001 (2014-04-26 19:34:22)

Beelzebud · 2014-04-26 20:46:28

Hmm I really hate the idea of having to run the cards hot to remain stable. I hope they fix this issue quickly.

MrAster · 2014-04-26 20:55:45

No idea. It's bothersome though.

Beelzebud · 2014-04-27 04:11:24

Ok tonight I updated my other machine, and am not getting the crashes on it at all.

Both machines are now running totally current Arch installs. Both machines are Intel based (one a Q6600, and one an i5 3570k), with the Q6600 having a Geforce GTX 650Ti Boost, and the i5 having 2x GTX760 4gb in SLI mode.

The machine with the single GPU is perfectly stable. The one with SLI randomly crashes.

Baryon · 2014-05-04 12:07:41

I am still getting random freezes even with PowerMizer set to Prefer Maximum Performance. It's very occasional which is why I delayed a while before posting. Of course there is every possibility that the problem in my case is a different problem. I have a GTX 750 Ti, no SLI.

Taylor001 wrote:

Haha! However, I disabled it that way too, just to try.

So what is the correct way to disable it?

MrAster · 2014-05-04 12:15:50

Looks like the best way is:

Option         "RegistryDwords" "PowerMizerEnable=0x1; PowerMizerDefaultAC=0x1"

In the "device" section of your xorg.conf

Arch Linux

#1 2014-04-18 21:33:47

X Freezes at random with NVIDIA driver

#2 2014-04-18 21:44:55

Re: X Freezes at random with NVIDIA driver

#3 2014-04-19 08:44:22

Re: X Freezes at random with NVIDIA driver

#4 2014-04-22 00:07:05

Re: X Freezes at random with NVIDIA driver

#5 2014-04-22 12:28:59

Re: X Freezes at random with NVIDIA driver

#6 2014-04-22 12:36:58

Re: X Freezes at random with NVIDIA driver

#7 2014-04-22 15:50:56

Re: X Freezes at random with NVIDIA driver

#8 2014-04-22 20:12:14

Re: X Freezes at random with NVIDIA driver

#9 2014-04-22 20:59:17

Re: X Freezes at random with NVIDIA driver

#10 2014-04-26 00:46:13

Re: X Freezes at random with NVIDIA driver

#11 2014-04-26 00:53:14

Re: X Freezes at random with NVIDIA driver

#12 2014-04-26 01:04:25

Re: X Freezes at random with NVIDIA driver

#13 2014-04-26 01:13:38

Re: X Freezes at random with NVIDIA driver

#14 2014-04-26 01:27:28

Re: X Freezes at random with NVIDIA driver

#15 2014-04-26 01:29:04

Re: X Freezes at random with NVIDIA driver

#16 2014-04-26 01:43:32

Re: X Freezes at random with NVIDIA driver

#17 2014-04-26 01:49:19

Re: X Freezes at random with NVIDIA driver

#18 2014-04-26 01:57:56

Re: X Freezes at random with NVIDIA driver

#19 2014-04-26 14:58:41

Re: X Freezes at random with NVIDIA driver

#20 2014-04-26 19:33:55

Re: X Freezes at random with NVIDIA driver

#21 2014-04-26 20:46:28

Re: X Freezes at random with NVIDIA driver

#22 2014-04-26 20:55:45

Re: X Freezes at random with NVIDIA driver

#23 2014-04-27 04:11:24

Re: X Freezes at random with NVIDIA driver

#24 2014-05-04 12:07:41

Re: X Freezes at random with NVIDIA driver

#25 2014-05-04 12:15:50

Re: X Freezes at random with NVIDIA driver

Board footer