You are not logged in.
Hello,
I'm experiencing a persistent issue with my hybrid graphics setup on Arch Linux that seems to be directly related to the power source. My system freezes and becomes unresponsive when using `prime-run` while the AC adapter is plugged in, but it works correctly on battery power.
**System Information:**
* **Laptop:** Acer Predator Helios 300
* **CPU:** Intel i7-10750H
* **GPU 1 (Integrated):** Intel CometLake-H GT2 [UHD Graphics]
* **GPU 2 (Dedicated):** NVIDIA GeForce RTX 3070 Mobile / Max-Q
* **Kernel:** `linux-lts`
* **Display Server:** Wayland
* **WM/Compositor:** Hyprland
* **NVIDIA Driver:** `nvidia-dkms` 575.64.05-1
**The Problem:**
The core issue is that `prime-run` only functions correctly when the laptop is running on battery.
1. **On Battery Power:** `prime-run` works as expected. I can offload applications to the NVIDIA GPU without any issues. HDMI output also works correctly if I boot the system on battery.
2. **When AC Adapter is Plugged In:** The system becomes unstable.
* If I boot with the charger connected, Hyprland is unresponsive, and I cannot open any applications (including terminals).
* If I plug in the charger while the system is running, any application started with `prime-run` freezes.
* `nvidia-smi` shows no running processes, but GPU memory usage appears to increase.
**Troubleshooting Steps Taken:**
I have already gone through the following steps:
1. Ensured `nvidia-prime` is installed to provide the `prime-run` command.
2. Investigated `udev` rules. A rule related to `slimbookbattery` was temporarily disabled, but this did not resolve the issue.
3. Switched from the standard `nvidia` package to `nvidia-dkms` to ensure the driver is correctly compiled for my kernel version. I have run `sudo dkms autoinstall` and am about to reboot.
4. Captured `journalctl` logs at the moment the issue occurs (when plugging in the AC adapter). The logs show a kernel panic originating from the NVIDIA driver.
**Relevant Log (Kernel Panic):**
Here is the call trace from `journalctl -k` when the freeze happens:
```
Jul 30 15:58:37 archlinux kernel: _nv013768rm+0x5dd/0x720 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux kernel: _nv013678rm+0xe2/0x880 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux kernel: _nv053604rm+0x594/0x770 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: _nv053136rm+0xd1/0x1f0 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: _nv052821rm+0xd3/0x1b0 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: _nv055004rm+0x3f5/0x500 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: _nv015696rm+0x424/0x680 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: _nv052978rm+0x69/0xd0 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: _nv013970rm+0x83/0xa0 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: _nv000792rm+0x7e9/0xe30 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: rm_ioctl+0x7f/0x400 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
Jul 30 15:58:37 archlinux.localdomain kernel: nvidia_unlocked_ioctl+0x52a/0x8e0 [nvidia c26937fa2e5546f9f2aa9dac61303df2fff932c1]
```
Given the kernel panic, this seems to be a driver-level issue, likely related to power management. I'm hoping someone might have encountered a similar problem or could offer some insight into what might be causing this behavior and how to resolve it.
Thank you in advance for your help.
Offline
Jul 31 07:31:55 archlinux kernel: wlan0: associate with 9a:3b:a8:17:73:74 (try 1/3)
Jul 31 07:31:55 archlinux kernel: wlan0: RX AssocResp from 9a:3b:a8:17:73:74 (capab=0x1431 status=0 aid=8)
Jul 31 07:31:55 archlinux kernel: wlan0: associated
Jul 31 07:31:55 archlinux kernel: wlan0: Limiting TX power to 20 (20 - 0) dBm as advertised by 9a:3b:a8:17:73:74
Jul 31 07:31:57 archlinux kernel: Initializing XFRM netlink socket
Jul 31 07:31:58 archlinux kernel: bridge: filtering via arp/ip/ip6tables is no longer available by default. Update your scripts to load br_netfilter if you need this.
Jul 31 07:31:59 archlinux kernel: Bluetooth: RFCOMM TTY layer initialized
Jul 31 07:31:59 archlinux kernel: Bluetooth: RFCOMM socket layer initialized
Jul 31 07:31:59 archlinux kernel: Bluetooth: RFCOMM ver 1.11
Jul 31 07:35:10 archlinux kernel: warning: `ThreadPoolForeg' uses wireless extensions which will stop working for Wi-Fi 7 hardware; use nl80211
Jul 31 07:39:13 archlinux unknown: ====== CHARGER PLUGGED IN ======
Jul 31 07:39:18 archlinux kernel: NVRM: GPU at PCI:0000:01:00: GPU-eb7cb673-1ed6-4bf5-6e5c-a6690d3a3439
Jul 31 07:39:18 archlinux kernel: NVRM: Xid (PCI:0000:01:00): 62, 3226d140 00016818 00000000 202a8b94 202aa94c 202a96ae 2025d264 202ac05a
Jul 31 07:39:18 archlinux kernel: ata1: SATA link down (SStatus 4 SControl 300)
Jul 31 07:39:18 archlinux kernel: r8169 0000:07:00.0 enp7s0: Link is Down
Jul 31 07:39:38 archlinux kernel: i915 0000:00:02.0: [drm] *ERROR* Atomic update failure on pipe A (start=140135 end=140136) time 156 us, min 1046, max 1079, scanline start 1042, end 1090
Offline
I was trying to video record what happens, but to my surprise, when OBS is recording, the error doesn’t happen at all.
It only crashes after I stop recording and try to start it again. Then I get this error:
So I guess this means that as long as the NVIDIA driver is actively doing something (like NVENC encoding), it doesn’t fail. But once it idles and I try to re-engage it, things fall apart.
This kind of confirms what I suspected earlier — something about power state transitions or the driver going idle on AC power is breaking everything.
Let me know if there’s a way to stop the GPU from sleeping or if this is something that can be patched.
Offline
https://wiki.archlinux.org/title/NVIDIA … ersistence
Main issue is probably the https://wiki.archlinux.org/title/CPU_frequency_scaling / turbo boost / performance vs. powersave setting and the combined CPU and GPU draw (profiles) exceeding the supply?
Offline
https://wiki.archlinux.org/title/NVIDIA … ersistence
Main issue is probably the https://wiki.archlinux.org/title/CPU_frequency_scaling / turbo boost / performance vs. powersave setting and the combined CPU and GPU draw (profiles) exceeding the supply?
Thanks for your reply, any recommendations/action to resolve this? I had Ubuntu on this machine previously that it worked like a charm so im sure its something that can be resolved
Offline
I had Ubuntu on this machine previously that it worked like a charm
Does it still?
Have you enabled the persistenced?
It's also possible that this is mainly caused by https://wiki.archlinux.org/title/NVIDIA … P_firmware
Offline
I had Ubuntu on this machine previously that it worked like a charm
Does it still?
Have you enabled the persistenced?
It's also possible that this is mainly caused by https://wiki.archlinux.org/title/NVIDIA … P_firmware
Yes I have tried that, also used NVreg_EnableGpuFirmware=0 buts still no luck
Offline
Ok i solved it, I removed all Nvidia drivers and installed an old one 535.261.03
Offline
That would most likely be "nvidia.NVreg_EnableGpuFirmware=0", can you please post the journal of a boot where you tried to disable the GSP?
Offline
new kernel 6.17 dropped support for 535, now im back with the same issue on 580
Offline
The kernels have never supported a particular OOT module and the 535xx drivers need patches for every major kernel update since quite a while - so you're basically waiting for that (and maybe use the LTS kernel in the meantime)
Also
That would most likely be "nvidia.NVreg_EnableGpuFirmware=0", can you please post the journal of a boot where you tried to disable the GSP?
Offline
Hi Seth,
Thanks for the suggestions.
I followed your advice and booted with the
nvidia.NVreg_EnableGpuFirmware=0 kernel parameter.
The result was the same as before. On battery power,
prime-run works fine (I tested with glmark2). As soon as I
plug in the AC adapter, it stops working. It seems
disabling the GSP didn't change the behavior.
Here is the journal from that boot session, as you
requested:
https://pastebin.com/stkWGhds
Offline
Oct 12 22:52:09 themodmin kernel: Command line: BOOT_IMAGE=/vmlinuz-linux root=UUID=… nvidia.NVreg_DynamicPowerManagement=0x02 nvidia.NVreg_PreserveVideoMemoryAllocations=1 nvidia.NVreg_EnableGpuFirmware=1
Offline
Sorry about that. I corrected my config, ran grub-mkconfig, and booted with nvidia.NVreg_EnableGpuFirmware=0.
The result is the same: prime-run fails on AC power. Disabling GSP had no effect.
Journal: https://pastebin.com/bNeJruhu
Offline
Oct 13 06:45:14 themodmin kernel: NVRM: loading NVIDIA UNIX x86_64 Kernel Module 580.95.05 Tue Sep 23 10:11:16 UTC 2025
Oct 13 06:45:15 themodmin systemd[1]: Started NVIDIA Persistence Daemon.
Oct 13 06:45:15 themodmin systemd[1]: Reached target Multi-User System.
Oct 13 06:45:15 themodmin kernel: [drm] Initialized nvidia-drm 0.0.0 for 0000:01:00.0 on minor 0
Oct 13 06:45:15 themodmin kernel: nvidia 0000:01:00.0: [drm] Cannot find any crtc or sizes
Oct 13 06:45:15 themodmin systemd-logind[2969]: New session '1' of user 'themodmin' with class 'user' and type 'tty'.
Oct 13 06:45:15 themodmin systemd-logind[2969]: New session '2' of user 'themodmin' with class 'manager' and type 'unspecified'.
Oct 13 06:45:15 themodmin kernel: NVRM: GPU at PCI:0000:01:00: GPU-eb7cb673-1ed6-4bf5-6e5c-a6690d3a3439
Oct 13 06:45:15 themodmin kernel: NVRM: Xid (PCI:0000:01:00): 62, 322d2367 00016998 00000000 202aab46 202ac8f4 202ab660 2025ec84 202ae002
Oct 13 06:45:15 themodmin kernel: NVRM: Xid (PCI:0000:01:00): 154, GPU recovery action changed from 0x0 (None) to 0x1 (GPU Reset Required)
Oct 13 06:45:15 themodmin uwsm[3228]: Selected compositor ID: hyprland.desktop
The GPU face-plants even before hyprland is started (but after your login)
Oct 13 06:45:13 themodmin systemd[1]: /etc/systemd/system/omarchy-seamless-login.service:13: Unknown key 'StartLimitIntervalSec' in section [Service], ignoring.
Oct 13 06:45:15 themodmin systemd[1]: Started Omarchy Seamless Auto-Login.
Oct 13 06:45:15 themodmin systemd[3371]: Started Omarchy Battery Monitor Timer.
Oct 13 06:45:15 themodmin systemd[3371]: Starting Omarchy Battery Monitor Check...
Oct 13 06:45:15 themodmin systemd[3371]: Finished Omarchy Battery Monitor Check.
can you skip the omarchy stuff?
(Battery monitor etc. makes me suspicious because of the problems precondition…)
Offline