Nvidia-kernel-dkms crashes whole system.

[Taras Zakharchenko]

Hello, Vincent.
You mean mailing list? OK

>Edit /etc/bumblebee/bumblebee.conf and change "KernelDriver=nvidia" to
>"KernelDriver=nvidia-current", and then restart bumblebeed ("service
>bumblebeed restart").

Congrats, amazing! Whole system crashed again, as I described earlier. It
is like emergency hardware shutdown.
Last kernel log entries before it:
[47270.992541] bbswitch: disabling discrete graphics
[47271.100620] pci 0000:01:00.0: power state changed by ACPI to D3cold
[47278.961307] bbswitch: enabling discrete graphics
[47279.412284] pci 0000:01:00.0: power state changed by ACPI to D0
[47279.623965] nvidia: module license 'NVIDIA' taints kernel.
[47279.623969] Disabling lock debugging due to kernel taint
[47279.634763] vgaarb: device changed decodes:
PCI:0000:01:00.0,olddecodes=io+mem,decodes=none:owns=none
[47279.634852] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  304.108  Wed
Jul 31 19:35:17 PDT 2013
No error mesages or something like it.

Can the problem be connected with ACPI. I got some suspicious messages
during boot:
[   15.409041] ACPI: If an ACPI driver is available for this device, you
should use it instead of the native driver
[   15.436610] ACPI Warning: 0x000000000000f040-0x000000000000f05f SystemIO
conflicts with Region \SMB0 1 (20130328/utaddress-251)
[   15.436615] ACPI Warning: 0x000000000000f040-0x000000000000f05f SystemIO
conflicts with Region \_SB_.PCI0.SBUS.SMBI 2 (20130328/utaddress-251)

Taras Zakharchenko


2013/10/27 Vincent Cheng <vincentc1208 at gmail.com>

> Hi Taras,
>
> First off, please don't forget to cc the list when replying.
>
> On Sat, Oct 26, 2013 at 3:20 AM, Taras Zakharchenko
> <taras at zakharchenko.in.ua> wrote:
> >>Do you have bbswitch-dkms, bumblebee, and primus installed from the
> >>repositories?
> > Yes, from repositories. I have tried to alter settings, compile source to
> > figure out where problem is. But it seems it is not in bumblebee. For
> now I
> > have clean install from repos.
> >
> >>Correct, neither nouveau nor nvidia should be loaded at boottime.
> >>Indeed, Debian's bumblebee package ships a file
> >>(/etc/modprobe.d/bumblebee.conf) blacklisting both nouveau and nvidia,
> >>to prevent either module from being loaded before bbswitch. The
> >>bbswitch kernel module is basically in charge of loading and unloading
> >>nouveau/nvidia when needed, for power management purposes.
> >
> > Yes, I found this lines. This explains why drivers did not load in
> boottime.
> >
> >>Don't manually load the nvidia module yourself. optirun/primusrun will
> >>take care of doing so for you.
> >
> > In fact it can not do this.
> > I constantly have error:
> >
> > From primus:
> > primus: fatal: Bumblebee daemon reported: error: Could not load GPU
> driver
> >
> > From optirun:
> > [ 1097.292155] [ERROR]Cannot access secondary GPU - error: Could not load
> > GPU driver
> > [ 1097.292196] [ERROR]Aborting because fallback start is disabled.
> >
> > Syslog:
> > Module nvidia could not be loaded (timeout?)
> > Could not load GPU driver
>
> Edit /etc/bumblebee/bumblebee.conf and change "KernelDriver=nvidia" to
> "KernelDriver=nvidia-current", and then restart bumblebeed ("service
> bumblebeed restart").
>
> Regards,
> Vincent
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-nvidia-devel/attachments/20131027/ce4592c0/attachment.html>