Computer automatically wakes up from sleep after enabling wake on lan

[kayvee]

I am trying to set up Wake on Lan on my machine... I followed instructions from https://help.ubuntu.com/community/WakeOnLan and https://necromuralist.github.io/post...g-wake-on-lan/ for this. I first made relevant changes to my BIOS. I am using an MSI B550 Unify motherboard with the latest version of BIOS. In my BIOS I have set the "Wake Up Event" to "BIOS" (other option is "OS") and enabled "Resume by PCI devices" (my ethernet is built-in).

After enabling wake on lan using ethtool on my machine, I checked if it is receiving magic packets using tcpdump as described here: https://necromuralist.github.io/post...g-wake-on-lan/

$ sudo tcpdump -i enp42s0 '(udp and port 7) or (udp and port 9)'
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on enp42s0, link-type EN10MB (Ethernet), capture size 262144 bytes
15:19:24.316234 IP recon.58451 > 255.255.255.255.discard: UDP, length 102

At this point, I put my machine to sleep using the KDE power menu. It goes to sleep for a few seconds and automatically restarts on its own without any input. I don't even have to touch the mouse or keyboard... it just wakes up on its own. The motherboard has a WiFi adapter as well. I have disabled Wireless in KDE but not from BIOS. I doubt that is what's causing it though as the ethtool tells me that interface (wlo1) is disabled. Any ideas where the signal to wake the computer is coming from?

[kayvee]

So, I followed instructions from here https://wiki.archlinux.org/title/Wake-on-LAN and used nmcli to get the correct settings in. I am now able to wake my machine remotely from sleep by trying to ssh to it. Long story short, I had to run the following two commands:

## this command to wake on lan using magic packet
nmcli connection modify "Wired connection 1" 802-3-ethernet.wake-on-lan magic
## or this command to wake on ssh:
nmcli connection modify "Wired connection 1" 802-3-ethernet.wake-on-lan unicast
nmcli connection modify "Wired connection 1" 802-3-ethernet.auto-negotiate yes

But my computer still wakes up on its own from sleep without having to do anything. After some searching, I found multiple posts referring to something like this: https://forums.linuxmint.com/viewtopic.php?p=1995055
My machine has a bunch of devices that are in the enabled state:

$ cat /proc/acpi/wakeup
Device  S-state   Status   Sysfs node
GP12      S4    *enabled   pci:0000:00:07.1
GP13      S4    *enabled   pci:0000:00:08.1
XHC0      S4    *enabled   pci:0000:2d:00.3
GP30      S4    *disabled
GP31      S4    *disabled
GPP0      S4    *enabled   pci:0000:00:01.1
GPP8      S4    *enabled   pci:0000:00:03.1
GPP1      S4    *enabled   pci:0000:00:01.2
PTXH      S4    *enabled   pci:0000:05:00.0
PT20      S4    *disabled
PT24      S4    *disabled
PT26      S4    *disabled
PT27      S4    *disabled
PT28      S4    *enabled   pci:0000:06:08.0
PT29      S4    *enabled   pci:0000:06:09.0

And these device codes mean the following:

$ cat /proc/acpi/wakeup | grep -P '\d.:\d+\.\d+' | sed 's/[ ]*S4.*pci:0000:/\t/' | while read -r a b; do echo -ne "$a\t"; lspci | grep "$b" ; done                                   
GP12    00:07.1 PCI bridge: Advanced Micro Devices, Inc. [AMD] Starship/Matisse Internal PCIe GPP Bridge 0 to bus[E:B]
GP13    00:08.1 PCI bridge: Advanced Micro Devices, Inc. [AMD] Starship/Matisse Internal PCIe GPP Bridge 0 to bus[E:B]
XHC0    2d:00.3 USB controller: Advanced Micro Devices, Inc. [AMD] Matisse USB 3.0 Host Controller
GPP0    00:01.1 PCI bridge: Advanced Micro Devices, Inc. [AMD] Starship/Matisse GPP Bridge
GPP8    00:03.1 PCI bridge: Advanced Micro Devices, Inc. [AMD] Starship/Matisse GPP Bridge
GPP1    00:01.2 PCI bridge: Advanced Micro Devices, Inc. [AMD] Starship/Matisse GPP Bridge
PTXH    05:00.0 USB controller: Advanced Micro Devices, Inc. [AMD] Device 43ee
PT28    06:08.0 PCI bridge: Advanced Micro Devices, Inc. [AMD] Device 43ea
PT29    06:09.0 PCI bridge: Advanced Micro Devices, Inc. [AMD] Device 43ea

I started to disable them one after another finally getting to a state where I had disabled every single one of them. Still, the machine just wakes up from sleep on its own. So I am still not sure what is going on...

[jlittle]

Forgive my curiosity, but does it wake with the ethernet cable disconnected?

[kayvee]

Forgive my curiosity, but does it wake with the ethernet cable disconnected?

I did not try that. I started on this path with the goal to put the computer to sleep when I am not using it but still be able to access via ssh should I need. If I pull the ethernet cable I won't be able to use wake-on-lan which will defeat the purpose. Nevertheless I am going to give it a try...

EDIT: Yes, pulling the ethernet cable keeps the computer asleep (at least for the brief period I was monitoring). Unfortunately, that's not a solution that works for me. I will need to keep it connected.

[kayvee]

jlittle suggestion to pull the ethernet cable got me thinking...
I had wake-on-lan set up such that the computer wakes up when I ssh to it. If I tweak the setting just so that the machine wakes up only on receiving a magic packet, the computer no longer automatically wakes up from sleep without any external trigger. So... progress!

Now I need to figure out how to (1) make the unicast setting that allows wake-on-ssh work with sleep state, or (2) send magic packets from an external network. I prefer the first one over second because then I won't have to worry about having a wakeonlan application installed on my client.

[GreyGeek]

Thanks for posting the question AND the solution, Kayvee!
You'll never know how many people you will help with that post but it will be many more than just yourself!

[jlittle]

I suspect you'll have to learn a bit of Wireshark. You've run tcpdump with the system asleep, but that won't in principle see the activity that woke the computer. Ideally, you'd run a dump using another computer on the same subnet (I'm unsure if that's the right term) and look closely in the second before the computer wakes. I suspect your router is doing something. If you don't have another computer to look at stuff, a tcpdump with the computer idling might reveal the culprit.

tcpdump output bamboozled me when I looked at it for something, years ago now. Wireshark organizes stuff to let you see what's going on, IIRC.

[kayvee]

Thank you GreyGeek ... I hope it is helpful to others (or even future me :-D)
jlittle -- I am not sure that I will able to do much even if I figure out what it is that is currently causing activity on my LAN at all times. Say, I find out it is the router or the home printer or another device at home. Unfortunately, shutting them down is not an option. I am sure the curiosity to find out what it is will get me eventually :-D I will update the thread with my findings.

After searching for a bit I could not come up with a way to keep the unicast setting. So I changed it to magic packet and now I have to rely on sending a magic packet to wake the computer up. It's easy enough to do it from within the network but to be able to do it over the internet I had to add appropriate settings in my WiFi router to route UDP ports 7 and 9 to my desktop. I have dynamic IP but have a DDNS set up so now I can send magic packets to wake up my computer as follows:

wakeonlan -i mysite.example.com a0:7b:c0:3d:9e:0f

Apparently there are cellphone apps that can do this too. So my plan is to just use the cellphone app to wake the machine up when needed.--

[jlittle]

That's great, thank you for sharing the solution.