kernel: NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt co

[uOpt]

dmesg says: kernel: NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context"

I get that every 30 minutes or so of watching a movie, usually but not always when messing with tabs in google chrome at the same time. The X11 server will freeze for 10 seconds or so and then continue normally.

Linux 2.6.38.8, NVidia 285.05.09 but I also had it with 280.x drivers. Dual-card 7900GTX and 240 GT, all activity on the 7900.

Is this a known issue? Any workaround? Worth filing a bug report about?

[uOpt]

This is really annoying.

Am I really the only one seeing it?

[Licaon]

Flash version? hardware acceleration active for Flash ?

[uOpt]

Flash isn't involved. It it caused by Xine with a Xv backend.

While I get it most of the time when fiddling with Chrome tabs at the same time, this is required (just had one without touching anything) and in any case this Chrome doesn't have Flash installed in the first place.

I would entertain a theory that this is an interaction with sound. I had these hangs when playing music in Xine. That displays a light animation in a Xv video and caused these hangs and it went away by hiding the video window.

[uOpt]

Upgraded to kernel 3.1.5 and still get these. Extremely annoying.

Still can't think I'm the only one?

[mlauss]

I have that too, on a Q4000M. It happens sporadically actually: just 30 minutes ago it started to throw
these errors out of the blue, then reported the device fell off the bus, all after running 48hrs without a hitch.
Another system with a 8800GT but otherwise completely identical userland (KDE 4.7.4) runs like a champ.

[uOpt]

Not funny anymore. Lost an X11 server from this today. After a couple of them in one movie I had tearing in both Xv and GL backend drivers that even switching to and back from a text console wouldn't fix. A restart later I got it back.

I am also a little concerned think that since this apparently stops my machine cold dead for a couple seconds what happens to outstanding harddrive requests?

[uOpt]

I previously suspected that this might be related to sound, namely the driver for my particular card and the NVidia driver maybe incorrectly being blamed. I now run with an entirely different card (USB instead of onboard) and still get this.

[tzp]

I started to get this error message a few days ago, out of the blue. No driver or kernel update was done for long time before. Since then it happens always when starting up the X server (using startx from command line).

I updated my kernel to the latest stable 3.2.*, (and still used the 285. NVIDIA driver) but it didn't help. I saw some noisy Nvidia splash page, then my machine freezes. (probably X server running at 100% CPU).
I had to fall back to the Xorg vesa driver that works fine.

Could this be the sign of a dying GPU ? (it's an old 8500 GT)

[uOpt]

I would entertain the thought of a broken graphics card.

But I got to tell you, if I have to replace this one I'd be not amused and if the Xorg drivers for ATI are up to some better standards now that would be it. I just moved this card from my gaming machine to my workstation because the 7800 GTX in there had a dual-link DVI port that tried to do dual-link but unsuccessfully so, and there is a whole stack of dead NVidia cards right next to my chair. In any case this card worked fine in the gaming machine.

It could also be some issue in this dual-card setup in an Asus board (who know what they did with the BIOS again) but the second card is just sitting there with a background image, sounds not too likely.

[tzp]

Well, there could be some (critical or less critical) HW problem too, but at also the SW seems to be unable to handle it.

I compiled a 3.2.3 kernel and installed the 290.10. The same story as before:
startx, then slow start, can't get beyond the initial blank screen and a mouse cursor, the CPU loaded so that the machine is unresponsive. Somehow still managed to stop the X with Ctrl-C and this is what I saw in the log:

[  2450.319] (WW) NVIDIA(0): WAIT (2, 6, 0x8000, 0x0000c6dc, 0x0000c94c)
[  2459.678] (WW) NVIDIA(0): WAIT (0, 6, 0x8000, 0x0000f1e0, 0x0000f1e0)
[  2462.679] (WW) NVIDIA(0): WAIT (0, 6, 0x8000, 0x0000f3bc, 0x0000f3bc)
[  2465.681] (WW) NVIDIA(0): WAIT (0, 6, 0x8000, 0x0000f4c4, 0x0000f4c4)
[  2468.683] (WW) NVIDIA(0): WAIT (0, 6, 0x8000, 0x0000f5cc, 0x0000f5cc)
[  2471.685] (WW) NVIDIA(0): WAIT (0, 7, 0x8000, 0x0000f624, 0x0000f624)
[  2474.686] (WW) NVIDIA(0): WAIT (0, 7, 0x8000, 0x0000fb64, 0x0000fb64)
[  2477.687] (WW) NVIDIA(0): WAIT (0, 7, 0x8000, 0x0000045c, 0x0000045c)
[  2480.688] (WW) NVIDIA(0): WAIT (2, 7, 0x8000, 0x00000518, 0x00000984)
[  2484.103] [mi] EQ overflowing. The server is probably stuck in an infinite loop.
[  2484.103] 
Backtrace:
[  2484.116] 0: /usr/bin/X11/X (xorg_backtrace+0x37) [0x80a7ec7]
[  2484.116] 1: /usr/bin/X11/X (mieqEnqueue+0x1d1) [0x80a21b1]
[  2484.116] 2: /usr/bin/X11/X (xf86PostMotionEventM+0xb0) [0x80c8fc0]
[  2484.116] 3: /usr/bin/X11/X (xf86PostMotionEventP+0x6f) [0x80c910f]
[  2484.116] 4: /usr/lib/xorg/modules/input/evdev_drv.so (0xb44f3000+0x2cfe) [0xb44f5cfe]
[  2484.116] 5: /usr/lib/xorg/modules/input/evdev_drv.so (0xb44f3000+0x3e1d) [0xb44f6e1d]
[  2484.116] 6: /usr/bin/X11/X (0x8048000+0x6dee1) [0x80b5ee1]
[  2484.116] 7: /usr/bin/X11/X (0x8048000+0x1281c2) [0x81701c2]
[  2484.116] 8: (vdso) (__kernel_sigreturn+0x0) [0xb779d400]
[  2484.116] 9: /usr/lib/xorg/modules/drivers/nvidia_drv.so (0xb4ae2000+0x5f646) [0xb4b41646]

See also the Xorg log in the attached bug report.
Could someone from Nvidia look at this ?

[uOpt]

Getting more interesting. I so hope I'm not looking at general memory corruption here.

Mainboard exchange probably tomorrow, I also hope it's the mainboard and not the NVidia drivers.

Feb 25 23:12:14 hostname kernel: NVRM: os_cheue tepe oyedteCUwiei tmco nerp otx
Feb 25 23:22:41 hostname kernel: NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context
Feb 25 23:24:38 hostname kernel: NVRM: os_schedule: Attempted to yield th CP hl naoi ritrutcnet<4>NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt contet
Feb 25 23:24:40 hostname kernel: NVRM: os_schedule: Attempted to yield the CPU while in atoico nerp otx
Feb 25 23:25:18 hostname kernel: NVRM: os_schedule: Attempted to yield the CPU while in atomi o nerp otx
Feb 26 00:01:15 hostname kernel: NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context