Re: Random system freezes with Linux 6.12 + Xenomai 3.3.1 on Intel Ultra 5 125H

On 29.07.26 08:05, 刘杨 wrote:
> Hi.
>     I have already performed the following tests:
> 1. Linux 6.18.20 + Xenomai 3.3.1: the system hangs during boot, but SSH login is still possible.
> 2. Linux 6.12.90 + Xenomai 3.3.1: the system completely freezes (hard lock).

Self-ported? Latest 6.12 is only v6.12.85-cip22-dovetail1

> 3. Linux 6.12.55 + GRUB configuration with xenomai.state=disabled: the system freezes.
> 4. Linux 6.12.55 without Dovetail and Cobalt: the system runs normally.
> 5.Linux 6.12.55 + xenomai 3.3.1  Using the Xe driver instead of i915   the system freezes.
> 6.Linux 6.12.55 + xenomai 3.3.1 and nvidia_uvn.ko with uvm_disable_hmm=1 uvm_perf_prefetch_enable=0    the system freezes.

That's quite a lot of variations. Pick the simplest one, with least
dependencies on special things, reproduce the issue and compare to a
working setup, I would say. Such hard freezes left and right are
unusual, and there must be one more or less special variable that
triggers them. E.g., if you just use our x86 6.12 image from
xenomai-images, does that run the basic tests on your hardware?

> 
> CONFIG_XENO_OPT_WATCHDOG is enbale and set to 30.
> 
> On Ubuntu 24.04, I monitored the CPU; the CPU utilisation is around 30%. A Gemini 360L camera is connected to the USB controller. I ran nvidia-smi and obtained the following output (please provide the output if you need further analysis).
> There is also an f81601a CAN interface connected to the machine, which uses CAN FD to connect six motors.

Does adding/removing this part of your workload make a difference? I
guess not when you describe those boot freeze with 6.18 above, but still
a variation from a generic system setup.

> +-----------------------------------------------------------------------------------------+
> | NVIDIA-SMI 580.173.02             Driver Version: 580.173.02     CUDA Version: 13.0     |
> +-----------------------------------------+------------------------+----------------------+
> | 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 RTX 4060 ...    Off |   00000000:01:00.0 Off |                  N/A |
> | N/A   37C    P0             27W /  115W |     141MiB /   8188MiB |      1%      Default |
> |                                         |                        |                  N/A |
> +-----------------------------------------+------------------------+----------------------+
> 
> +-----------------------------------------------------------------------------------------+
> | Processes:                                                                              |
> |  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
> |        ID   ID                                                               Usage      |
> |=========================================================================================|
> |    0   N/A  N/A            1892      C   /usr/NX/bin/nxnode.bin                  131MiB |
> +-----------------------------------------------------------------------------------------+
> 

Not loading the nvidia driver does not make a difference for the
freezes? Or is it mandatory to even start some reasonable load that is
needed to trigger the freezes?

Jan

-- 
Siemens AG, Foundational Technologies
Linux Expert Center