Bug 508246 – x11-drivers/nvidia-drivers-337.12 hangs with GeForce GT 440

• Status: RESOLVED UPSTREAM
• Product: Gentoo Linux

[Description Johan Ymerson 2014-04-20 15:41:45 UTC]

After upgrade to x11-drivers/nvidia-drivers-337.12, the computer hangs when X is starting on my system with a GeForce GT 440 board.
Additinally, the startup hangs for 30 seconds when the nvidia kernel module is loaded.

Reproducible: Always

Steps to Reproduce:
1. emerge =x11-drivers/nvidia-drivers-337.12
2. reboot
3.
Actual Results:  
System hangs for 30 seconds when loading nvidia kernel module, then continues bootup sequence until X starts, then it locks up completely.


Snip from messages log when nvidia module loads/hangs:
Apr 20 16:48:44 lilo kernel: [    3.197158] input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input21
Apr 20 16:48:44 lilo kernel: [    3.197327] input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input19
Apr 20 16:48:44 lilo kernel: [    3.197494] input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input18
Apr 20 16:48:44 lilo kernel: [    3.197661] input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:01.0/0000:01:00.1/sound/card1/input17
Apr 20 16:48:44 lilo kernel: [    3.198199] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=io+mem,decodes=none:owns=none
Apr 20 16:48:44 lilo kernel: [   34.038308] vboxdrv: fAsync=0 offMin=0x1e3 offMax=0xf05c
Apr 20 16:48:44 lilo kernel: [   34.038413] vboxdrv: TSC mode is 'synchronous', kernel timer mode is 'normal'.
Apr 20 16:48:44 lilo kernel: [   34.068229] device-mapper: ioctl: 4.26.0-ioctl (2013-08-15) initialised: [e-mail]


End of Xorg.0.log:
[   146.018] (II) NVIDIA dlloader X Driver  337.12  Fri Apr  4 14:13:45 PDT 2014
[   146.018] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[   146.018] (++) using VT number 7
[   146.022] (II) Loading sub module "fb"
[   146.022] (II) LoadModule: "fb"
[   146.022] (II) Loading /usr/lib64/xorg/modules/libfb.so
[   146.023] (II) Module fb: vendor="X.Org Foundation"
[   146.023]    compiled for 1.15.0, module version = 1.0.0
[   146.023]    ABI class: X.Org ANSI C Emulation, version 0.4
[   146.023] (WW) Unresolved symbol: fbGetGCPrivateKey
[   146.023] (II) Loading sub module "wfb"
[   146.023] (II) LoadModule: "wfb"
[   146.023] (II) Loading /usr/lib64/xorg/modules/libwfb.so
[   146.024] (II) Module wfb: vendor="X.Org Foundation"
[   146.024]    compiled for 1.15.0, module version = 1.0.0
[   146.024]    ABI class: X.Org ANSI C Emulation, version 0.4
[   146.024] (II) Loading sub module "ramdac"
[   146.024] (II) LoadModule: "ramdac"
[   146.024] (II) Module "ramdac" already built-in



A healty version (334.21) would have continued with:
[    18.539] (**) NVIDIA(0): Depth 24, (--) framebuffer bpp 32
[    18.539] (==) NVIDIA(0): RGB weight 888
[    18.539] (==) NVIDIA(0): Default visual is TrueColor
[    18.539] (==) NVIDIA(0): Using gamma correction (1.0, 1.0, 1.0)
[    18.540] (**) NVIDIA(0): Enabling 2D acceleration
[    19.198] (II) NVIDIA(0): Display (DELL U2412M (DFP-0)) does not support NVIDIA 3D
[    19.198] (II) NVIDIA(0):     Vision stereo.
[    19.198] (II) NVIDIA(GPU-0): Found DRM driver nvidia-drm (20130102)
[    19.199] (II) NVIDIA(0): NVIDIA GPU GeForce GT 440 (GF108) at PCI:1:0:0 (GPU-0)
[    19.199] (--) NVIDIA(0): Memory: 1048576 kBytes
[    19.199] (--) NVIDIA(0): VideoBIOS: 70.08.4d.00.00
[    19.199] (II) NVIDIA(0): Detected PCI Express Link width: 16X



Kernel version is 3.12.17, but 3.10.36 tested as well with the same result.

[Comment 1 Jeroen Roovers (RETIRED) 2014-04-21 00:15:09 UTC]

Run nvidia-bug-report.sh as root, read the output and send the results to Nvidia.

[Comment 2 Jeroen Roovers (RETIRED) 2014-04-21 00:23:50 UTC]

(In reply to Johan Ymerson from comment #0)
> After upgrade to x11-drivers/nvidia-drivers-337.12, the computer hangs when
> X is starting on my system with a GeForce GT 440 board.
> Additinally, the startup hangs for 30 seconds when the nvidia kernel module
> is loaded.

This is really suspicious. The only stuff that normally has such ridiculously long time-outs is networking. Maybe /first/ check your .nvidia-settings-rc and other configuration files for spurious hostname usage and alike?

[Comment 3 Johan Ymerson 2014-04-21 08:58:03 UTC]

No hostnames or similar in .nvidia-settings-rc.
The 30-second hang occurs when the nvidia kernel module is loaded, that is pretty early in the boot process. No network available then.

Some additional info:
If I remove xdm from the default runlevel, ie. X never starts, the system hangs when rebooted/shut down. It is the nvidia.ko module itself that hangs, not the X server.

I'll file an upstream bug report.