Cuda 5.0 on Debian, first use

Accelerated Computing CUDA CUDA Setup and Installation
#1

Hello, I’ve installed CUDA 5.0 on a new Debian 6.0. It works great, however I have a small issue, every single time I boot the OS, I have to run at least one cuda program as root so that my normal user can actually use it.
Any help on how to fix this?

Thanks you very much!

#2

Are the /dev nodes created at boot time? From the Getting Started Guide at http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-linux/index.html#install-cuda-software, there is a small script to load the NVIDIA driver and create the required dev nodes:

#!/bin/bash

/sbin/modprobe nvidia

if [ "$?" -eq 0 ]; then
  # Count the number of NVIDIA controllers found.
  NVDEVS=`lspci | grep -i NVIDIA`
  N3D=`echo "$NVDEVS" | grep "3D controller" | wc -l`
  NVGA=`echo "$NVDEVS" | grep "VGA compatible controller" | wc -l`

  N=`expr $N3D + $NVGA - 1`
  for i in `seq 0 $N`; do
    mknod -m 666 /dev/nvidia$i c 195 $i
  done

  mknod -m 666 /dev/nvidiactl c 195 255

else
  exit 1
fi
#3

I have that script on boot time, however it always fails, here’s the output I could find o dmesg:

[26167.688569] nvidia 0000:01:00.0: PCI INT A disabled
[26171.334627] nvidia 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
[26171.334634] nvidia 0000:01:00.0: setting latency timer to 64
[26171.334638] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=none,decodes=none:owns=io+mem
[26171.334730] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  304.54  Sat Sep 29 00:05:49 PDT 2012
[26190.712684] NVRM: Your system is not currently configured to drive a VGA console
[26190.712686] NVRM: on the primary VGA device. The NVIDIA Linux graphics driver
[26190.712688] NVRM: requires the use of a text-mode VGA console. Use of other console
[26190.712690] NVRM: drivers including, but not limited to, vesafb, may result in
[26190.712691] NVRM: corruption and stability problems, and is not supported.

Seeing it now, I wonder if it can be something about the framebuffer?
I have this on the grub.conf:

menuentry 'Debian GNU/Linux, with Linux 2.6.32-5-amd64' --class debian --class gnu-linux --class gnu --class os {
        set gfxpayload=1024x768
        insmod part_msdos
        insmod ext2
        set root='(hd3,msdos1)'
        search --no-floppy --fs-uuid --set cb634d3e-fd3b-4862-8ee6-1816cdbd2f1f
        echo    'Loading Linux 2.6.32-5-amd64 ...'
        linux   /vmlinuz-2.6.32-5-amd64 root=/dev/mapper/hellish-root ro  quiet
        echo    'Loading initial ramdisk ...'
        initrd  /initrd.img-2.6.32-5-amd64
}
#4

Ok, I’ve finally fixed it.
It appears that it wasn’t about the framebuffer in the end, but I simply could not put the script directly on /etc/rc.local, so instead, I create a /etc/rc.cuda with the script to load it, and /etc/rc.local calls that script.

Thanks for the help Alban.