Network (or hardware) related issues with ASUS TUF A15 FA506QM laptop (EndeavourOS)

[rare violet]

Hi! I've been having this issue since roughly April 17th (or the week of). I'm running EndeavourOS with Asus-Linux's custom linux-g14 kernel but I've tried linux, linux-lts as well as versions 6.13.8, 6.13.7 and 6.14.4 and experience the same. Network drivers are r1869 and mt7921e, DE is Plasma and I use the standard network plasmoid to see, connect, disconnect from networks since i am not too familiar with any other method; hope it's fine if I refer to that.

When I boot, for a couple minutes, all is fine. I automatically connect to the Wi-Fi, and I can use Ethernet just fine. Then, Ethernet will suddenly stop working, showing me as connected to Wi-Fi only; then Wi-Fi will silently stop working too. If I try to disconnect from the Wi-Fi at any point, it will not work; the plasmoid will show that I am disconnecting from the Wi-Fi but will stay stuck there, until the Wi-Fi stops working and then I am no longer shown any networks to connect to in the plasmoid.

If I run inxi -Nnz during this, the ethernet driver shows as down the whole time even from boot, and the wifi driver is up until it's down when the wifi stops working or I try to disconnect. Weirder still, my CPU usage goes up after I try to disconnect (temp and fans too). A few times, I noticed a ton of odd 'napi' and 'ksoftirc' processes starting and taking up CPU after it happens (somehow not all the time). I've also experienced straight up kernel panics from this occasionally.

Here's where it gets weird. I tried running a live USB to troubleshoot twice, once with EndeavourOS, another time with Fedora, and had the EXACT SAME ISSUE both times. This made me think it could be hardware, especially since I'd recently replaced a fan and thermal paste on my heatsink; maybe something had damaged the network cards. However, I also tried a Windows To Go USB on a suggestion and it had zero issue. Hence why I tried different kernel versions, since both distros i tried and my system were on 13.8.

Attached are logs from journalctl -u NetworkManager -b 0; first one is just me booting, plugging in an rj45 and using the machine normally until the ethernet crashes. Second one goes until the Wi-Fi stops working too and I try to disconnect it.

I'm a bit at a loss as to what to try here. I'm really hoping it's not hardware and fixable but I don't know how to make sure. Any help?

Will add some command outputs in a sec

dmesg: https://pastebin.com/EUQhze9i
inxi -Nnz: https://pastebin.com/xv8avM3E
ps -ef, which interestingly shows a number of "napi" and "ksoftirqd" processes that are taking up a lot of my CPU cycles when I try to disconnect from the Wi-Fi: https://pastebin.com/rA2cismv

forgot to add something important: when the laptop tries to shut down, network manager's stop job hangs for a while, and then it fails to fully shut down. screen turns off and it stays powered until I force shutdown; at least on any boot where the aforementioned issue happens

[small orbit]

@rare violet which one? https://wiki.archlinux.org/title/Laptop/ASUS#The_Ultimate_Force_(TUF)

[rare violet]

uhh, none of them, it's fa506qm; but I see where you're getting at. though can it really be r8169 that's causing the issue if it affects everything else too?

can't try immediately but will when i can

[small orbit]

?

[solemn wigeon]

i copied/pasted some log analysis i did and it it attached as a text

its kind of a lot but the main but i was focussingon was

avril 28 17:54:50 gb-eos2 NetworkManager[739]: <info> [1745855690.2329] device (enp2s0): conflict detected for IP address 192.168.1.79 with hostB4:8C:9D:31:5E:F3

[rare violet]

i'm only half understanding, but are you saying this is in part an issue on my router's dhcp server? it seems unlikely, this issue happens even when i'm using public wifi and affects none of my other devices

[solemn wigeon]

so that's the ethernet dev complaining about a dhcp conflict with the wlan0 device

not necessarily the router, but more of a network manager issue

you may have issues with wifi stil tbh as meditek devices have pretty bad drivers from the looks of it

[rare violet]

heard of that ya

[solemn wigeon]

but the reason ethernet seems to be messing up is dhcp conflicts, probably the result of the wifi constantly connecting/disconnecting

those seem to result in duplicate leases getting created so the eth device just doesn't get an ip address because the router's already assigned one previously to wlan0

id expect fixing a lease/setting enps0's mac on its dhcp request to make that device more stable and ideally you don't see issues with both

[rare violet]

okay, i'll need to try that then; won't fixing a lease cause an issue when connecting to public wifi as mentioned before though?

i hardly remember my classes on dhcp but it seems logical you'd need to have it fixed on both sides

[solemn wigeon]

nah, the lease will only be fixed on your own router. basically everything will be handled via dhcp but the wifi dev will ideally just stick with the same address when it makes a renewal request

and by not setting the mac address of wlan0 in the request, it doesn't affect network manager profiles for when mac address randomization is done. e.g. for something like public wifi

fixing it on the router's side is exactly how you want it to be done, otherwise you can't manage it/your router won't register it

and THAT would definitely lead to dhcp conflicts, because if someone defines a static ip address, the router has no way to know this is the case. so it may actually hand out a lease that was statically defined by the client

in those cases, when defining a static ip on the client, you want to define the ip to be outside the router's dhcp pool

[rare violet]

so, quick update: r8168 does not work better @small orbit

@solemn wigeon question: how do i remove unused leases?