[Bug] linux-cachyos iwlwifi crashes about nyx HOT 7 CLOSED

@ptr1337 I have updated my flake recently and uninstalled sched_ext. Now I don't get any crashes.

I am not sure what exactly caused it (older kernel or schedext). But from my side it seems fixed and can be closed.

Should this issues be kept open, in order to track if the issue is related to sched_ext? (I mean it is probaply just a random kernel bug)

If you don't use any scx-sched scheduler, then this is completly unrelated.

from nyx.

Cc @ptr1337

from nyx.

Sometimes iwlwifi crashes and I have no wireless network anymore. Seems to happen more often when using nixos rebuild or playing around with sched_ext.

So, it seems when there is a higher load it does crash?
We are actually not changing anything on the driver part, besides on ATH11k for the steam deck.

Is this reproduceable on other kernels, like Zen?

from nyx.

I wasn't able to reproduce on latest kernel, I will test with zen.

I tried running sysbench cpu for high cpu load, to test if it would cause the same problem, but I wasn't able to cause iwifi crashing.

from nyx.

@ptr1337 I was trying with zen/xanmod kernels and not iwlwifi issues yet. I am only getting lag spikes from time to time with zen, but nothing related to iwlwifi.

It seems that short high load spikes are the issue, since running benchmarks pushing cpu and gpu to 100% cause no wifi loss. Starting firefox or high load compilations sometimes crashes iwlwifi. It doesn't seem to appear at random with no app starting. Higher load in general however causes the issue with iwlwifi to appear more frequent.

Note this is pure assumption since I didn't found a way to reproduce a 100%.

from nyx.

I wasn't able to reproduce on latest kernel, I will test with zen.

My guess is that this was fixed in upstream. We dont change anything on the drivers.

from nyx.

@ptr1337 I have updated my flake recently and uninstalled sched_ext. Now I don't get any crashes.

I am not sure what exactly caused it (older kernel or schedext). But from my side it seems fixed and can be closed.

Should this issues be kept open, in order to track if the issue is related to sched_ext? (I mean it is probaply just a random kernel bug)

from nyx.