Bug#1016595: cuda: CUDA Error : system not yet initialized

[Antonio]

Package: nvidia-cuda-dev
Version: 11.4.3-4
Severity: important
File: cuda
X-Debbugs-Cc: antonio.bertoni at gmail.com

Dear Maintainer,

After the last system update on 07/30/2022, the CUDA driver has stopped working. These are
some error messages i got after updating:

- CUDA Error : system not yet initialized
- Unable to enum CUDA GPUs: unknown error
- No OpenCL platforms found
- No avaiable GPUs for mining. Please check your drivers and/or hardware.

Same problem using Debian bullseye or bookworm.
this is the nvidia-smi output:
Wed Aug  3 21:52:35 2022       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.129.06   Driver Version: 470.129.06   CUDA Version: 11.4     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  NVIDIA GeForce ...  On   | 00000000:01:00.0  On |                  N/A |
| 79%   34C    P8     9W /  97W |    152MiB /  8119MiB |     12%      Default |
|                               |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|    0   N/A  N/A      1102      G   /usr/lib/xorg/Xorg                 88MiB |
|    0   N/A  N/A      1266      G   /usr/bin/gnome-shell               28MiB |
|    0   N/A  N/A      6646      G   ...181797667042908025,131072       32MiB |
+-----------------------------------------------------------------------------+

Before the system upgrade everything worked fine.

Thanks
Antonio

-- System Information:
Debian Release: bookworm/sid
  APT prefers testing
  APT policy: (500, 'testing')
Architecture: amd64 (x86_64)

Kernel: Linux 5.18.0-3-amd64 (SMP w/2 CPU threads; PREEMPT)
Kernel taint flags: TAINT_PROPRIETARY_MODULE, TAINT_OOT_MODULE, TAINT_UNSIGNED_MODULE
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8), LANGUAGE=en_GB:en
Shell: /bin/sh linked to /usr/bin/dash
Init: systemd (via /run/systemd/system)
LSM: AppArmor: enabled

Versions of packages nvidia-cuda-dev:amd64 depends on:
ii  libaccinj64-11.4                             11.4.120~11.4.3-4
ii  libcublas11                                  11.6.5.2~11.4.3-4
ii  libcublaslt11                                11.6.5.2~11.4.3-4
ii  libcuda1 [libcuda.so.1]                      470.129.06-6
ii  libcudart11.0                                11.4.148~11.4.3-4
ii  libcufft10                                   11.1.1+~10.5.2.100~11.4.3-4
ii  libcufftw10                                  11.1.1+~10.5.2.100~11.4.3-4
ii  libcuinj64-11.4                              11.4.120~11.4.3-4
ii  libcupti-dev                                 11.4.120~11.4.3-4
ii  libcurand10                                  11.1.1+~10.2.5.120~11.4.3-4
ii  libcusolver11                                11.2.0.120~11.4.3-4
ii  libcusolvermg11                              11.2.0.120~11.4.3-4
ii  libcusparse11                                11.6.0.120~11.4.3-4
ii  libnppc11                                    11.4.0.110~11.4.3-4
ii  libnppial11                                  11.4.0.110~11.4.3-4
ii  libnppicc11                                  11.4.0.110~11.4.3-4
ii  libnppidei11                                 11.4.0.110~11.4.3-4
ii  libnppif11                                   11.4.0.110~11.4.3-4
ii  libnppig11                                   11.4.0.110~11.4.3-4
ii  libnppim11                                   11.4.0.110~11.4.3-4
ii  libnppist11                                  11.4.0.110~11.4.3-4
ii  libnppisu11                                  11.4.0.110~11.4.3-4
ii  libnppitc11                                  11.4.0.110~11.4.3-4
ii  libnpps11                                    11.4.0.110~11.4.3-4
ii  libnvblas11                                  11.6.5.2~11.4.3-4
ii  libnvidia-legacy-390xx-cuda1 [libcuda.so.1]  390.151-2
ii  libnvidia-ml-dev                             11.4.120~11.4.3-4
ii  libnvjpeg11                                  11.5.2.120~11.4.3-4
ii  libnvrtc-builtins11.4                        11.4.152~11.4.3-4
ii  libnvrtc11.2                                 11.4.152~11.4.3-4
ii  libnvtoolsext1                               11.4.120~11.4.3-4
ii  libnvvm4                                     11.4.152~11.4.3-4
ii  libthrust-dev                                1.16.0-1

Versions of packages nvidia-cuda-dev:amd64 recommends:
ii  libgl-dev        1.4.0-1
ii  libgl1-mesa-dev  22.0.5-1
ii  libnvcuvid1      470.129.06-6
ii  libvdpau-dev     1.5-1

nvidia-cuda-dev:amd64 suggests no packages.

-- no debconf information