Hey everyone,
I’ve got a laptop with a Lunar Lake CPU (Intel Core Ultra 200V, Arc Graphics 130V/140V) running Fedora, and I’ve been fighting a really annoying issue: every time I resume from suspend, the system graphic freezes and becomes unusable. So I need to perform manual reboot.
Here’s my setup:
- CPU/GPU: Intel Core Ultra 200V (Lunar Lake),
lspcishows:00:02.0 VGA compatible controller: Intel Corporation Core Ultra 200V Series Processors Arc Graphics 130V/140V GPU (rev 04) - Driver:
xe - Kernel: 7.1.8, so this is not some old-kernel issue that’s already been fixed
- Suspend mode is s2idle
I went digging and it looks like this isn’t just me. there’s a decent amount of chatter about Lunar Lake having trouble with resume from deep s2idle (PC10), possibly a firmware/C-state issue rather than a pure driver bug:
- A Level1Techs thread with basically the exact same symptoms on a ThinkPad X1 Carbon Gen 13, also on Fedora: GPU stuck at low frequency after resume, freeze, watchdog reboot. https://forum.level1techs.com/t/lunar-lake-thinkpad-x1-carbon-gen-13-xe-driver-freeze-on-s2idle-resume-fedora-43/246487
- An Arch forum thread pointing at deep s2idle/PC10 resume specifically, suspecting a BIOS/firmware bug rather than the kernel: https://bbs.archlinux.org/viewtopic.php?id=314185
- The big “Lunar Lake Laptop Issues Collection” thread on Arch, where several people mention stuttering/freezing tied to C-states, worked around with
intel_idle.max_cstate, though nobody there considers it a real fix: https://bbs.archlinux.org/viewtopic.php?id=300155
One thing I found interesting: someone in one of those threads tested this by disabling specific BIOS options one at a time, and narrowed it down to CPU Power Management and intel SpeedStep. Turning those two off (and leaving everything else alone) got rid of the freezing entirely, but battery drain during suspend went up noticeably (roughly 1.5-2%/h instead of 0.3-0.5%/h). So it really seems like this is about CPU C-state transitions on resume, not the xe driver being broken on its own.
So, a few questions for anyone who’s run into this:
- Is anyone else seeing this on Fedora specifically with Lunar Lake / Arc 130V-140V?
- Has anyone found a workaround that doesn’t tank battery life as much, a specific
intel_idle.max_cstatevalue, a narrower BIOS tweak, anything like that? - Is this already being tracked somewhere upstream (kernel bugzilla, drm/xe gitlab) that’s worth piling onto instead of starting fresh here?
Happy to grab more logs if it helps track this down.