This is an issue that’s been plauging me for awhile now (I used to use Nobara and this problem persisted there as well). I use Bazzite on a desktop with an AMD GPU. I realized that, after I walked away from the computer for awhile, I’d hear the GPU running at full speed but couldn’t awaken the computer or get any signal from it (my bluetooth mose wouldn’t connect).
I realized today that the issue was something with suspending to RAM. Is there any suggestion for what the problem is and how to fix it? Or if not fix it, at least prevent the computer from doing this?
More than happy to provide specs/logs if helpful. I am relatively new to Linux so pardon my ignorance if the solution to this is obvious.
I noticed this too after Fedora updated to 41. Computer, or screen rather, wouldnt wake up from sleep mode. For me it was solved when I turned off bluetooth.
Also, I tweaked the power settings so it wouldn’t go into sleep, and so far it seems to have fixed the problem. But I’m curious if hakinger’s point about Bluetooth might be involved.
I too am having a similar problem, except in my case if suspend in either ‘Game’ SteamOS environment or the KDE desktop. It seems to go into sleep mode, allowing me to press a key on the keyboard awake it. However, 24-30 seconds later it reboots. I even loaded ChimeraOS (ARCH Kernel) onto another NVME as well Mint OS onto a USB port with the same results so it appears it is the Linux kernel that is the culprit. I looked for settings like I have used in older PCs BIOSes for ACPI settings, but it appears the GigaByte A20I-AX mini ITX motherboard (AM5) BISO (with Ryzen 8600G APU) has none or at least nothing with a similar naming convention as in other BIOSes I have worked with.
I have also tried to run a python tool called ‘amd_s2idle.py’ created by AMD to run on Linux to see if the kernel has the S2idle set up, which appears it does not by Indicating “System isn’t configured for S2idle in firmware setup.” I also “forced” it to do test iterations of suspend/wake and it will only do as many cycles in that 30 second window (?) which is two.
I will set up a new post on this, but it would be great if this could be figured out since it appears this might be widespread AMD problem.