FS#59962 - [nvidia-lts] Doesn't match actual linux-lts version

Attached to Project: Arch Linux
Opened by Gustav Brauer (Gustav_2017) - Thursday, 06 September 2018, 21:24 GMT
Last edited by Andreas Radke (AndyRTR) - Monday, 10 September 2018, 16:59 GMT
Task Type Bug Report
Category Packages: Extra
Status Closed
Assigned To Andreas Radke (AndyRTR)
Sven-Hendrik Haase (Svenstaro)
Architecture All
Severity Medium
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 3
Private No

Details

Description:
After update linux-lts from 4.14.67-1 -> 4.14.68-1 the nvidia module could not be loaded. Seems that the vermagic info doesn't match.


Additional info:
* package version(s)
# pacman -Q linux-lts
linux-lts 4.14.68-1
# pacman -Q nvidia-lts
nvidia-lts 1:396.54-2
# pacman -Si nvidia-lts
Repositorium : extra
Name : nvidia-lts
Version : 1:396.54-2

* config and/or log files etc.
# modprobe nvidia
insmod /lib/modules/4.14.68-1-lts/extramodules/nvidia.ko.gz
could not be inserted (similar)
# modinfo nvidia
filename: /lib/modules/4.14.68-1-lts/extramodules/nvidia.ko.gz
alias: char-major-195-*
version: 396.54
supported: external
license: NVIDIA
srcversion: CDDF5F6229BAD6360A195B1
alias: pci:v000010DEd00000E00sv*sd*bc04sc80i00*
alias: pci:v000010DEd*sv*sd*bc03sc02i00*
alias: pci:v000010DEd*sv*sd*bc03sc00i00*
depends: ipmi_msghandler
retpoline: Y
name: nvidia
vermagic: 4.14.67-1-lts SMP mod_unload modversions

Steps to reproduce:
Upgrade zu latest linux-lts package, install or use nvidia-lts package.
This task depends upon

Closed by  Andreas Radke (AndyRTR)
Monday, 10 September 2018, 16:59 GMT
Reason for closing:  Fixed
Additional comments about closing:  all modules have been rebuilt for kernel 4.14.69
Comment by Doug Newgard (Scimmia) - Thursday, 06 September 2018, 22:59 GMT
# modprobe nvidia
insmod /lib/modules/4.14.68-1-lts/extramodules/nvidia.ko.gz
could not be inserted (similar)

This makes no sense.
Comment by f (bakgwailo) - Friday, 07 September 2018, 00:38 GMT
I am also seeing this, results in an SDDM segfault for me.
Comment by Gustav Brauer (Gustav_2017) - Friday, 07 September 2018, 06:38 GMT
@Scimmia: Sorry, i havn't had the exact error message yesterday. Here it is (nvidia-drm automatically loads other depended nvidia_* modules):
# modprobe -v nvidia-drm
insmod /lib/modules/4.14.68-1-lts/extramodules/nvidia.ko.gz
modprobe: ERROR: could not insert 'nvidia_drm': Invalid argument

The related dmesg output from above modprobe is:
nvidia: disagrees about version of symbol boot_cpu_data
nvidia: Unknown symbol boot_cpu_data (err -22)

I could force loading - as a workaround - the current (old) modules with:
# modprobe -v --force-vermagic --force-modversion nvidia-drm
Then further start of XOrg works again.

Users who use the nvidia-dkms have no problems after current linux-lts package update, only those who use the nvidia-lts package. It should be rebuild to match current LTS-Kernel version.h, that's the reason I think.

//Edit: I bet the other nvidia-*-lts packages are affected too, esp. extra/nvidia-340xx-lts and extra/nvidia-390xx-lts. It's the same maintainer, so i guess he/she will know what to do ;-)

Comment by V.S. (vence) - Friday, 07 September 2018, 07:51 GMT
With linux-lts 4.14.68-1:

<...>
kernel: nvidia: disagrees about version of symbol boot_cpu_data
kernel: nvidia: Unknown symbol boot_cpu_data (err -22)
<...>
sddm[492]: Failed to read display number from pipe
sddm[492]: Display server failed to start. Exiting
<...>
systemd[1]: sddm.service: Main process exited, code=dumped, status=6/ABRT
systemd[1]: sddm.service: Failed with result 'core-dump'.
Comment by Hussam Al-Tayeb (hussam) - Friday, 07 September 2018, 10:34 GMT
I recompiled the nvidia-lts kernel module package and Xorg starts.

Loading...