Sometimes, resuming from hibernate puts it in a frozen state

porridgewithraisins · 2023-11-11 07:12:29

I followed the instructions in the wiki and have the following:

$ cat /boot/loader/entries/arch.conf
title Arch Linux
linux vmlinuz-linux-zen
initrd amd-ucode.img
initrd initramfs-linux-zen.img
options root=LABEL=Arch rw nowatchdog nmi_watchdog=0 mitigations=off resume=LABEL=Swap

initramfs

HOOKS=(base udev autodetect modconf kms block filesystems fsck resume)

Resume works 90% of the time. Sometimes, even though it boots properly, and puts me in the desktop, in the same state as before, and the time in the taskbar is updated correctly, before I can interact with it, it freezes completely. The logs from the time I hibernated, to the time I had to force shutdown, point to the issue being `amdgpu` related.

$ journalctl -b -1 --since="Nov 10 23:39:54" | curl -F'file=@-' http://0x0.st
http://0x0.st/Ht7l.txt

Nov 10 23:39:54 was when I hibernated yesterday.

porridgewithraisins · 2023-11-11 07:25:31

UPDATE: I have managed to freeze it in a reproducible way.

I hadn't realized until now, but when I `systemctl hibernate`, the following happens:
1. The screen goes black, while the keyboard backlight, etc, remain lit up.
2. The screen comes _back_ on, in a frozen state.
3. Everything goes off together.

I usually have laptop lid close => Turn off display.

So, if I close my laptop lid before steps 2 and 3 complete, then when I resume, it lands in step 2, and freezes there.

For now, the temporary solution is to wait for steps 2 and 3 to complete before I close the lid. But how do I fix this properly?

Last edited by porridgewithraisins (2023-11-11 07:26:33)

seth · 2023-11-11 12:50:06

Please try the non-zen and LTS kernels and post the entire journal for the boot, the error is probably beforehand.
A generic test would be to add "amdgpu.runpm=0 amdgpu.bapm=0 amdgpu.aspm=0 pcie_aspm=off" to the kernel parameters, but it's not even clear whether the system supports ASPM.
Also try the behavior from a multi-user.target login only (no GUI, not even GDM or so) or, maybe even better, a basic GUI context (just X11 + xterm, maybe twm) in case you're running into a race condition when the DE wants to respond to the output loss (internal display)

skyw4ll · 2023-11-12 16:40:17

sudo systemctl stop nvidia-suspend.service
sudo systemctl stop nvidia-hibernate.service
sudo systemctl stop nvidia-resume.service

sudo systemctl disable nvidia-suspend.service
sudo systemctl disable nvidia-hibernate.service
sudo systemctl disable nvidia-resume.service

sudo mv /lib/systemd/system-sleep/nvidia ~/nvidia.bak

and reboot.

In case you nuke it, the contents of nvidia.bak are:

#!/bin/sh

case "$1" in
    post)
        /usr/bin/nvidia-sleep.sh "resume"
        ;;
esac

It worked for me

related: https://forums.developer.nvidia.com/t/f … 0/187150/2

seth · 2023-11-12 16:55:51

The OP doesn't have an nvidia GPU - nor are there traces of the nvidia S3 hooks in the, limited, journal segment they posted.

Arch Linux

#1 2023-11-11 07:12:29

Sometimes, resuming from hibernate puts it in a frozen state

#2 2023-11-11 07:25:31

Re: Sometimes, resuming from hibernate puts it in a frozen state

#3 2023-11-11 12:50:06

Re: Sometimes, resuming from hibernate puts it in a frozen state

#4 2023-11-12 16:40:17

Re: Sometimes, resuming from hibernate puts it in a frozen state

#5 2023-11-12 16:55:51

Re: Sometimes, resuming from hibernate puts it in a frozen state

Board footer