Kernel freezes every few days due to the i915 driver (GPU hang)

Sprinter05 · 2025-09-25 22:53:32

I've been having random system freezes since I switched to Arch, and when I mean freezes I mean actual kernel freezes (SysRq does not work). Kernel says:

i915 [drm] ERROR GT0: rcs0 reset request timed out: {request: 00000001, RESET_CTL: 00000001}

Followed up by:

[drm] GPU HANG: ecode 9:1:85dffffa

.
Freezes happen in both X11 and Wayland and its completely random and without a pattern, it can happen while AFK, while writing code, while watching a video, while playing a game, etc.
I have managed to reduce the amount of freezes by installing the "vulkan-intel" package and adding:

intel_idle.max_cstate=1 i915.enable_dc=0

as specified in https://wiki.archlinux.org/title/Intel_ … Intel_CPUs, even though my CPU is not of any of the mentioned generations (I am running an Intel 10th gen i7-10750H specifically); but it has not removed the freezes entirely.
Using X11 I can disable the iGPU by messing with the configuration and it seems to work but Wayland does not allow me to select a single GPU so the problem persists.
Also my BIOS (HP laptop) does not allow for disable the iGPU at the hardware level so I'm stuck with this.

I have kinda tried everything at this point and I'm out of ideas and tired of having kernel freezes 1-2 times a week, specially because I don't know how bad could it be for my laptop and installation to have hard resets this often (at least I keep some backups).
Any help? I really have ran out of ideas and I have already scrutinized the web for information on this issue.
Thanks.

seth · 2025-09-26 19:39:14

You could randomly try to throw "i915.enable_psr=0 pcie_aspm=off" at it (do you have context to the GPU HANG message?) and just as as sanity check: you do have https://wiki.archlinux.org/title/Microcode and https://archlinux.org/packages/core/any … are-intel/ ?
Do you load https://wiki.archlinux.org/title/Intel_ … re_loading (and what if you don't)?

Edit, forgot:

but Wayland does not allow me to select a single GPU

There's no "wayland", which compositor do you use? Couple of them actually do allow you to control the GPU (there's just no standard way, "wayland" is a scant IO protocol)

Last edited by seth (2025-09-26 19:51:49)

Sprinter05 · 2025-09-29 11:57:48

What do you mean context to the GPU hang? The logs dont give me more info to what is shown there and I stated it can happen in any situation.
I have installed both the intel microcode and intel firmware packages.
I do not have GuC/HuC because the power management part is for Gen12 onwards (for my Gen 10 I only have media offloading functionalities which I found not relevant to my situation) but I could try it anyways
I will try the psr and aspm parameters you provided too see if they fix anything
Respecting wayland I am currently running KDE Plasma 6

seth · 2025-09-29 14:21:01

https://wiki.archlinux.org/title/KDE#Me … -specific)

I meant whether there're userspace or kernel messages around the hang, documenting what might cause it.

Sprinter05 · 2025-09-30 10:17:31

I think I tried that KWIN_DRM_DEVICES environment variable and it caused severe lag but I'll try it again.
The messages where the GPU hangs are from the i915 driver which is a kernel driver, but it crashes on a userspace program (e.g Xorg or Kwin)

seth · 2025-09-30 12:20:04

The messages where the GPU hangs are from the i915 driver which is a kernel driver, but it crashes on a userspace program (e.g Xorg or Kwin)

Yeah, I'm remotely aware of the design…

I mean freezes I mean actual kernel freezes (SysRq does not work)

Still: any kind of errors in kernel or userspace (system journal, previous xorg log) might provide hints as to *why* you're ending up w/ the botched reset.
Fwwi there's pending bug in iwlwifi that apparently corrupts the kernel memory and frequently leads to i915 faceplants - this is esp. the case in 6.16 and the later LTS kernels, since when do you face this?

Sprinter05 · 2025-10-03 18:29:52

I have been facing this error since I installed arch 2 months ago (it was 6.16.0 at the time, and got my first freeze after upgrading to 6.16.3). I rolled back to 6.16.1 and the problem persisted so i switched to Zen same issue so I switched to LTS same issue.
In respects to journal/dmesg the only thing before the gpu hang and kernel crash its just discord messages trying to access remote resources but they're just normal messages and before that there's really nothing relevant, most things are from several minutes prior.
Also, would you mind sending a link to the upstream bug on iwlwifi?

seth · 2025-10-03 19:41:35

https://bbs.archlinux.org/viewtopic.php?id=307628
https://gitlab.archlinux.org/archlinux/ … issues/155
https://bugzilla.kernel.org/show_bug.cgi?id=220558
https://lkml.org/lkml/2025/7/23/1544

It is of yet not entirely clear how much these various bugs are interconnected and what leads up to the i915 corruption, but 6.15.9 is likely not and 6.13.x not likely to be affected.

https://wiki.archlinux.org/title/Kdump will allow you to inspect the kernel crash better.

Sprinter05 · 2025-10-06 16:24:11

Okay thanks for the information, if I get the crash again I'll try to get a dump

Arch Linux

#1 2025-09-25 22:53:32

Kernel freezes every few days due to the i915 driver (GPU hang)

#2 2025-09-26 19:39:14

Re: Kernel freezes every few days due to the i915 driver (GPU hang)

#3 2025-09-29 11:57:48

Re: Kernel freezes every few days due to the i915 driver (GPU hang)

#4 2025-09-29 14:21:01

Re: Kernel freezes every few days due to the i915 driver (GPU hang)

#5 2025-09-30 10:17:31

Re: Kernel freezes every few days due to the i915 driver (GPU hang)

#6 2025-09-30 12:20:04

Re: Kernel freezes every few days due to the i915 driver (GPU hang)

#7 2025-10-03 18:29:52

Re: Kernel freezes every few days due to the i915 driver (GPU hang)

#8 2025-10-03 19:41:35

Re: Kernel freezes every few days due to the i915 driver (GPU hang)

#9 2025-10-06 16:24:11

Re: Kernel freezes every few days due to the i915 driver (GPU hang)

Board footer