Bug#928497: nvidia-persistenced: Error in nvidia-persistenced source (postinst)

[Andreas Beckmann]

Control: tag -1 moreinfo

On 2019-05-06 08:01, Marcelo "Elppans" Klumpp wrote:
> There is an ambiguous configuration error in the file "debian/nvidia-persistenced.postinst":

> It means that you should create a "home" folder in /var/run/nvpd/, but at the end of the code it says DO NOT CREATE the folder.
> 
> If you want to create a folder, how can you NOT CREATE the folder you are creating?

/var/run is a volatile path, anything needed there has to be created at runtime, not during installation.
And the home directory of nvpd is clearly *not* needed.

> Because of this, when installing the package, the following error occurs and the service does not start:
> 
> Alert: The home directory /var/run/nvpd/ you specified can not be accessed: No such file or directory
> 
> Alert: The home directory /var/run/nvpd/ you specified can not be accessed: No such file or directory

This is just noisy.

> Adding system user 'nvpd' (UID 118) ...
> Adding new group 'nvpd' (GID 124) ...
> Adding new user 'nvpd' (UID 118) with group 'nvpd' ...
> Not creating personal directory '/var/run/nvpd/'.
> Created symlink /etc/systemd/system/multi-user.target.wants/nvidia-persistenced.service → /lib/systemd/system/nvidia-persistenced.service.

> Job for nvidia-persistenced.service failed because the control process exited with error code.
> See "systemctl status nvidia-persistenced.service" and "journalctl -xe" for details.

What is the actual error encountered while starting the service?

I can reproduce this (on a systemv system) with a (deliberately)
mismatching kernel module loaded:

Preparing to unpack .../nvidia-persistenced_418.56-1_amd64.deb ...
Unpacking nvidia-persistenced (418.56-1) over (390.87-1~deb9u1) ...
Setting up nvidia-persistenced (418.56-1) ...
Warning: The home dir /var/run/nvpd/ you specified can't be accessed: No such file or directory
Adding system user `nvpd' (UID 133) ...
Adding new group `nvpd' (GID 144) ...
Adding new user `nvpd' (UID 133) with group `nvpd' ...
Not creating home directory `/var/run/nvpd/'.
Stopping NVIDIA Persistence Daemon
Starting NVIDIA Persistence Daemon
nvidia-persistenced failed to initialize. Check syslog for more details.
invoke-rc.d: initscript nvidia-persistenced, action "restart" failed.
dpkg: error processing package nvidia-persistenced (--configure):
 installed nvidia-persistenced package post-installation script subprocess returned error exit status 1

in syslog I have:

nvidia-persistenced: Started (7837)
nvidia-persistenced: Failed to query NVIDIA devices. Please ensure that the NVIDIA device files (/dev/nvidia*) exist, and that user 133 has read and write permissions for those files.
nvidia-persistenced: Shutdown (7837)
kernel: [12558015.162992] NVRM: API mismatch: the client has the version 410.104, but
kernel: [12558015.162992] NVRM: this kernel module has the version 396.54.  Please
kernel: [12558015.162992] NVRM: make sure that this kernel module and all NVIDIA driver
kernel: [12558015.162992] NVRM: components have the same version.

With the correct kernel module loaded, nvidia-persistenced runs fine.


Andreas