NovaCustom: Difference between revisions

Cafkafk (talk | contribs)
Added basic page with a solution to the spd5118 problem. The goal is mainly to save my current progress and keep adding additional notes.
 
Cafkafk (talk | contribs)
Added details about the nvidia i2c timeout bus blacklisting
 
Line 15: Line 15:
The current latest release of the Dasharo (coreboot+UEFI) firmware [https://docs.dasharo.com/variants/novacustom_v540tnx/releases/ supporting V540TNx is v0.9.1], although a v1.0 is expected in Q4 2025.
The current latest release of the Dasharo (coreboot+UEFI) firmware [https://docs.dasharo.com/variants/novacustom_v540tnx/releases/ supporting V540TNx is v0.9.1], although a v1.0 is expected in Q4 2025.


====== SMBus/I2C Issues ======
====== spd5118 Issues ======
The [https://docs.kernel.org/hwmon/spd5118.html spd5118 driver] responsible for RAM temperature sensors is known to be problematic. E.g. lm_sensors will report it as failing. Likewise, there are reports of [https://github.com/Dasharo/dasharo-issues/issues/1125 48GB SODIMMS overheating].  
The [https://docs.kernel.org/hwmon/spd5118.html spd5118 driver] responsible for RAM temperature sensors is known to be problematic. E.g. lm_sensors will report it as failing. Likewise, there are reports of [https://github.com/Dasharo/dasharo-issues/issues/1125 48GB SODIMMS overheating].  


Line 30: Line 30:
];
];
</syntaxhighlight>
</syntaxhighlight>
===== Nvidia i2c bus issues =====
Because of the other I2C bus instability issues, an issue has been observed, specially with RTX 4070 dGPUs where the kernel reports the dGPU "failling off the bus" with lines such as <code>kernel: nvidia-modeset: ERROR: GPU:0: Error while waiting for GPU progress:</code>
One theory for this failure is issues with the Dasharo i2c/smbus firmware code. A potential soluton is to disable the nvidia gpu's i2c timeout trigger.<syntaxhighlight lang="nixos">
boot.blacklistedKernelModules = [ "i2c_nvidia_gpu" ];
</syntaxhighlight>Notice that this can disable thermal reading or fan control, but has been observed as stabilizing dGPU based systems.


=== V560 Series ===
=== V560 Series ===