[Solved] System entering wrong suspend mem sleep mode

heyimnyl · 2024-04-01 06:01:25

Hello,

I have this problem of sometimes the automatic sleep does work correctly, sometimes it tries to enter a deeper mode(?) as in it wants to shut off or rather it does for a short time power cycle the psu and then instantly comes back.
I noticed this because fan speeds of a connected usb fan controller are spinning faster in the morning than when I left for the night.

What I'm trying to say is: It's nice that it wants to automatically sleep but its sometimes entering the wrong sleep mode, the one it should be in is suspend to ram and usually thats fine.

https://pastebin.com/raw/QhPKwAuc
^ journalctl entries from 1 cycle, this just repeats every time the system wants to sleep over and over until now.

https://pastebin.com/raw/pYLSUCJM
^ this is from me pressing sleep in start menue

It's almost like it's mistaking hibernate for sleep but hibernate is the same and I don't hibernate and never set it up to go hibernate as far as I remember.
Just to try this and see what happens I pressed hibernate but it just goes back to login screen after showing tty for a bit. I don't hibernate so not sure if relevant but here is the log from that:
https://pastebin.com/raw/JpZ1dJA4

If I go and press sleep in system menue now or in the login screen it also tries to enter deeper sleep mode which is wrong. Usually I fix this by just rebooting but it's rather annoying and I would like to know why or what causes this.

Thank you!

Last edited by heyimnyl (2024-04-04 16:42:57)

seth · 2024-04-01 16:11:20

https://pastebin.com/raw/QhPKwAuc simply shows the system entering S3 and then returning ~8s later, there's nothing about "deeper" or hiberanting.
If the problem is that the system automatically wakes up from S3, see https://wiki.archlinux.org/title/Power_ … om_suspend and also check cat /proc/acpi/wakeup

There's some noise around

Apr 01 04:09:46 nyArch kernel: usb 5-1.2: new full-speed USB device number 35 using xhci_hcd
Apr 01 04:09:46 nyArch kernel: usb 5-1.2: device descriptor read/64, error -32
Apr 01 04:09:46 nyArch kernel: usb 5-1.2: device descriptor read/64, error -32
Apr 01 04:09:47 nyArch kernel: usb 5-1.2: new full-speed USB device number 36 using xhci_hcd
Apr 01 04:09:47 nyArch kernel: usb 5-1.2: device descriptor read/64, error -32
Apr 01 04:09:47 nyArch kernel: usb 5-1.2: device descriptor read/64, error -32
Apr 01 04:09:47 nyArch kernel: usb 5-1-port2: attempt power cycle

so if you have sub in the wakeup triggers and depending on what that device is, it might be the cause.

heyimnyl · 2024-04-03 07:35:27

Sorry for potentially mixing up technical terms about sleep states

seth wrote:

https://pastebin.com/raw/QhPKwAuc simply shows the system entering S3 and then returning ~8s later, there's nothing about "deeper" or hiberanting.

Yes I think that's the problem. I didn't wake it up and normally a log looks like this: https://pastebin.com/raw/zJTrLhEC
Which is not S3 if I'm reading this correct and thats totally fine, I like not S3 - it works

I tried to look in Suspend and Hibernate Wiki 7.4 but my devices are ??
cat /sys/bus/i2c/devices/i2c-
i2c-0/ i2c-1/ i2c-2/ i2c-3/ i2c-4/ i2c-5/ i2c-6/ i2c-7/ i2c-8/

So I'm a bit stuck on the names.

The only thing enabled in acpi/wakeup is this
XHC0 S4 *enabled pci:0000:10:00.3

the rest is all disabled.
until now I have not had the problem again of system entering wrong mode so I could not check what would have been enabled in the case if it's in wake-up loop again.

Also I added
options pcie_port_pm=off
options pcie_pme=nomsi

to my boot conf after not being able to have the system go to idle when I set everything up and came across a thread here on the forums, that basically solved it but sometimes it still tries to to do this exact thing of entering and just waking up.
I think it's a different idle mode.. What I can see from the good journal is s2idle and its all fine, I can just power it off when I go away for longer but it's rather weird and I don't know how to fix these other occurrences when it doesn't go to s2idle.

Last edited by heyimnyl (2024-04-03 07:43:36)

seth · 2024-04-03 08:22:13

Apr 03 09:12:30 nyArch kernel: PM: suspend entry (deep)
Apr 03 09:12:52 nyArch kernel: PM: suspend entry (s2idle)

cat /sys/power/mem_sleep

You can control the behavior w/ eg.

echo s2idle | sudo tee /sys/power/mem_sleep

https://wiki.archlinux.org/title/Power_ … end_method

heyimnyl · 2024-04-04 16:42:16

Thank you!
I set s2idle as kernel parameter after noticing it defaults back to deep.
would be nice if this fixes the wrong sleep occurrences (it probably will), marking as solved.

Arch Linux

#1 2024-04-01 06:01:25

[Solved] System entering wrong suspend mem sleep mode

#2 2024-04-01 16:11:20

Re: [Solved] System entering wrong suspend mem sleep mode

#3 2024-04-03 07:35:27

Re: [Solved] System entering wrong suspend mem sleep mode

#4 2024-04-03 08:22:13

Re: [Solved] System entering wrong suspend mem sleep mode

#5 2024-04-04 16:42:16

Re: [Solved] System entering wrong suspend mem sleep mode

Board footer