Jonas Malaco

I agree with @aleksamagicka. What about having the driver try to detect the issue and, in that case, make the suggestion to the user?

Related: #448 I'm not sure if the response there is sill applicable. > There is currently no path to fixing this as liquidctl only supports hardware mode. And we might...

The limitation seems to be of how the driver implements its hardware modes (=modes in which no program is required to remain running after the initial setup/configuration). In software mode...

LGTM, thanks!

Like Aleksa mentioned, we need CRC-8 (CRC-8/SMBUS) and CRC-16/USB, so there are no implementations from the standard library we can use. While [`crcmod`](https://pypi.org/project/crcmod/) hasn't seen changes since 2010, it still...

Packaging statuses according to Repology: - `crcmod`: https://repology.org/project/python:crcmod/versions - `crc`: https://repology.org/project/python:crc/versions

Can you add the output immediately preceding (i.e. current and previous two liquidctl executions) an error from `watch -e -n1 liquidctl status -n 0 --debug`?

> Does it fail if you try with `--direct-access` multiple times in a row? It might be wise to manually unload the kernel driver instead of trying `--direct-access` in succession....

Have you noticed any unreasonable values being reported now? We don't check the specific (supposed) error flag that the kernel driver checks, so we might just be reporting garbage. On...

> Edit: It seems like the RPM spikes have actually stopped as soon as I terminated the watch command. Maybe they were caused by multiple applications trying to access the...