Re: [gentoo-user] Nvidia kernel module API version mismatch

[J. Roeleveld]

On March 15, 2018 1:14:58 PM UTC, Aleksander Okonski  
wrote:
>Hey Joost,
>
>Thank you! This was my problem, rebuilt the kernel and then everything
>worked smoothly.
>
>Aleks
>
>On Thu, Mar 15, 2018 at 12:53 PM, J. Roeleveld 
>wrote:
>
>> On Thursday, March 15, 2018 12:09:23 PM CET Aleksander Okonski wrote:
>> > Hey,
>> >
>> > I have run into a strange problem with my nvidia drivers and
>gentoo. I am
>> > currently running kernel 4.14.14 and I upgraded my
>> > x11-drivers/nvidia-drivers to 390.42 from 390.25. Once the new
>drivers
>> were
>> > installed I rebooted my laptop. Once rebooted I was unable to start
>the
>> > xorg server using startx and was greeted with errors. The
>> > /var/logs/xorg.0.log said that the problem was with the kernel
>module.
>> > Looking at dmesg I see that I am getting the error:
>> >
>> > NVRM: API mismatch: the client has the version 390.42, but this
>kernel
>> > module has version 390.25. please make sure that this kernel module
>and
>> all
>> > NVIDIA driver components have the same version.
>> >
>> > I therefor thought that maybe I forgot to "emerge @module-rebuild"
>so I
>> ran
>> > that and rebooted to make sure that everything was ok. However I am
>still
>> > getting the same problem.
>> >
>> > Looking at /usr/src/linux it is pointing to the correct kernel of
>4.14.14
>> > and running "modinfo nvidia" it shows that "filename:
>> > /lib/modules/4/14/14-gentoo/video/nvidia.ko" and that version is
>390.42.
>> > The other nvidia modules (nvidia-drm, nvidia-modeset, and
>nvidia-uvm)
>> show
>> > that they are built for the correct kernel and that they are also
>version
>> > 390.42.
>> >
>> > I have tried to debug this issue however all the recourse that I
>was able
>> > to find mostly said to reboot or perform rmmod nvidia then modprobe
>> nvidia.
>> > However when I try to rmmod nvidia it is dependent on
>nvidia-modeset and
>> > nvidia-drm. Trying to rmmod nvidia-drm will not work as I get the
>error
>> > "rmmod error: module nvidia_drm in in use".
>> >
>> > I therefor cannot understand why and where the error is coming from
>as it
>> > seems like the kernel has the correct nvidia modules loaded. Any
>ideas on
>> > how to fix it or any other suggestions?
>> >
>> > Thanks,
>> > Aleks
>>
>> Do you use an initramfs? And if yes, does it maybe contain the nvidia
>> module?
>>
>> --
>> Joost
>>
>>
>>

Aleksander,

You might want to change your initramfs config to not include the nvidia module 
to avoid this in future.
I don't have it in my initramfs and don't experience any issues as it ends up 
being loaded correctly after the initramfs.

--
Joost

Ps. Please do not top-post
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-user] Nvidia kernel module API version mismatch

[Aleksander Okonski]

Hey Joost,

Thank you! This was my problem, rebuilt the kernel and then everything
worked smoothly.

Aleks

On Thu, Mar 15, 2018 at 12:53 PM, J. Roeleveld  wrote:

> On Thursday, March 15, 2018 12:09:23 PM CET Aleksander Okonski wrote:
> > Hey,
> >
> > I have run into a strange problem with my nvidia drivers and gentoo. I am
> > currently running kernel 4.14.14 and I upgraded my
> > x11-drivers/nvidia-drivers to 390.42 from 390.25. Once the new drivers
> were
> > installed I rebooted my laptop. Once rebooted I was unable to start the
> > xorg server using startx and was greeted with errors. The
> > /var/logs/xorg.0.log said that the problem was with the kernel module.
> > Looking at dmesg I see that I am getting the error:
> >
> > NVRM: API mismatch: the client has the version 390.42, but this kernel
> > module has version 390.25. please make sure that this kernel module and
> all
> > NVIDIA driver components have the same version.
> >
> > I therefor thought that maybe I forgot to "emerge @module-rebuild" so I
> ran
> > that and rebooted to make sure that everything was ok. However I am still
> > getting the same problem.
> >
> > Looking at /usr/src/linux it is pointing to the correct kernel of 4.14.14
> > and running "modinfo nvidia" it shows that "filename:
> > /lib/modules/4/14/14-gentoo/video/nvidia.ko" and that version is 390.42.
> > The other nvidia modules (nvidia-drm, nvidia-modeset, and nvidia-uvm)
> show
> > that they are built for the correct kernel and that they are also version
> > 390.42.
> >
> > I have tried to debug this issue however all the recourse that I was able
> > to find mostly said to reboot or perform rmmod nvidia then modprobe
> nvidia.
> > However when I try to rmmod nvidia it is dependent on nvidia-modeset and
> > nvidia-drm. Trying to rmmod nvidia-drm will not work as I get the error
> > "rmmod error: module nvidia_drm in in use".
> >
> > I therefor cannot understand why and where the error is coming from as it
> > seems like the kernel has the correct nvidia modules loaded. Any ideas on
> > how to fix it or any other suggestions?
> >
> > Thanks,
> > Aleks
>
> Do you use an initramfs? And if yes, does it maybe contain the nvidia
> module?
>
> --
> Joost
>
>
>

Re: [gentoo-user] Nvidia kernel module API version mismatch

[Raffaele Belardi]

Aleksander Okonski wrote:
> Hey,
> 
> I have run into a strange problem with my nvidia drivers and gentoo. I am 
> currently
> running kernel 4.14.14 and I upgraded my x11-drivers/nvidia-drivers to 390.42 
> from 390.25.
> Once the new drivers were installed I rebooted my laptop. Once rebooted I was 
> unable to
> start the xorg server using startx and was greeted with errors. The 
> /var/logs/xorg.0.log
> said that the problem was with the kernel module. Looking at dmesg I see that 
> I am getting
> the error:
> 
> NVRM: API mismatch: the client has the version 390.42, but this kernel module 
> has version
> 390.25. please make sure that this kernel module and all NVIDIA driver 
> components have the
> same version. 
> 

I had a similar problem for months, finally I looked into /lib or /usr/lib (I 
think it was
an nvidia/ subdirectory of these) and found that some symlinks where pointing 
to old
versions of the libraries provided by nvidia. I adjusted the symlinks to point 
to the new
versions and the message disappeared.

Sorry I cannot be more precise but it was some time ago and right now I'm on an
nvidia-less system.

Note that in my case it was just an annoying log message, the server did start 
successfully.

raffaele

Re: [gentoo-user] Nvidia kernel module API version mismatch

[J. Roeleveld]

On Thursday, March 15, 2018 12:09:23 PM CET Aleksander Okonski wrote:
> Hey,
> 
> I have run into a strange problem with my nvidia drivers and gentoo. I am
> currently running kernel 4.14.14 and I upgraded my
> x11-drivers/nvidia-drivers to 390.42 from 390.25. Once the new drivers were
> installed I rebooted my laptop. Once rebooted I was unable to start the
> xorg server using startx and was greeted with errors. The
> /var/logs/xorg.0.log said that the problem was with the kernel module.
> Looking at dmesg I see that I am getting the error:
> 
> NVRM: API mismatch: the client has the version 390.42, but this kernel
> module has version 390.25. please make sure that this kernel module and all
> NVIDIA driver components have the same version.
> 
> I therefor thought that maybe I forgot to "emerge @module-rebuild" so I ran
> that and rebooted to make sure that everything was ok. However I am still
> getting the same problem.
> 
> Looking at /usr/src/linux it is pointing to the correct kernel of 4.14.14
> and running "modinfo nvidia" it shows that "filename:
> /lib/modules/4/14/14-gentoo/video/nvidia.ko" and that version is 390.42.
> The other nvidia modules (nvidia-drm, nvidia-modeset, and nvidia-uvm) show
> that they are built for the correct kernel and that they are also version
> 390.42.
> 
> I have tried to debug this issue however all the recourse that I was able
> to find mostly said to reboot or perform rmmod nvidia then modprobe nvidia.
> However when I try to rmmod nvidia it is dependent on nvidia-modeset and
> nvidia-drm. Trying to rmmod nvidia-drm will not work as I get the error
> "rmmod error: module nvidia_drm in in use".
> 
> I therefor cannot understand why and where the error is coming from as it
> seems like the kernel has the correct nvidia modules loaded. Any ideas on
> how to fix it or any other suggestions?
> 
> Thanks,
> Aleks

Do you use an initramfs? And if yes, does it maybe contain the nvidia module?

--
Joost

Re: [gentoo-user] Nvidia kernel module API version mismatch

[Alexander Kapshuk]

On Thu, Mar 15, 2018 at 1:09 PM, Aleksander Okonski
 wrote:
> Hey,
>
> I have run into a strange problem with my nvidia drivers and gentoo. I am
> currently running kernel 4.14.14 and I upgraded my
> x11-drivers/nvidia-drivers to 390.42 from 390.25. Once the new drivers were
> installed I rebooted my laptop. Once rebooted I was unable to start the xorg
> server using startx and was greeted with errors. The /var/logs/xorg.0.log
> said that the problem was with the kernel module. Looking at dmesg I see
> that I am getting the error:
>
> NVRM: API mismatch: the client has the version 390.42, but this kernel
> module has version 390.25. please make sure that this kernel module and all
> NVIDIA driver components have the same version.
>
> I therefor thought that maybe I forgot to "emerge @module-rebuild" so I ran
> that and rebooted to make sure that everything was ok. However I am still
> getting the same problem.
>
> Looking at /usr/src/linux it is pointing to the correct kernel of 4.14.14
> and running "modinfo nvidia" it shows that "filename:
> /lib/modules/4/14/14-gentoo/video/nvidia.ko" and that version is 390.42.
> The other nvidia modules (nvidia-drm, nvidia-modeset, and nvidia-uvm) show
> that they are built for the correct kernel and that they are also version
> 390.42.
>
> I have tried to debug this issue however all the recourse that I was able to
> find mostly said to reboot or perform rmmod nvidia then modprobe nvidia.
> However when I try to rmmod nvidia it is dependent on nvidia-modeset and
> nvidia-drm. Trying to rmmod nvidia-drm will not work as I get the error
> "rmmod error: module nvidia_drm in in use".
>
> I therefor cannot understand why and where the error is coming from as it
> seems like the kernel has the correct nvidia modules loaded. Any ideas on
> how to fix it or any other suggestions?
>
> Thanks,
> Aleks

Just to clarify, does:
modinfo nvidia
display version 390.42 or 390.25?