You are not logged in.

#1 2025-09-05 16:22:26

Ksmnn
Member
Registered: 2025-09-05
Posts: 1

[NVIDIA] Dell G16 7630: kernel panic with nvidia-dkms

Hello,

I am experiencing a reproducible kernel panic during boot on a new Dell G16 7630 laptop when trying to use the nvidia-dkms driver. The system is a fresh Arch Linux install without a graphical environment yet, so the issue is at the kernel/module level, not a display manager configuration.

I have searched the forum and found several historical threads about NVIDIA ACPI issues, but none with the specific rm_acpi_notify page fault on this kernel version, and the suggested solutions (e.g., using acpi_osi= parameters) have not worked in my case.

System Information
Laptop: Dell G16 7630

BIOS: 1.23.0 (2025-03-04)

CPU: 13th Gen Intel i7-13650HX (Alderlake-RPLS integrated graphics)

dGPU: NVIDIA Device 28e0 (RTX 4060 Mobile)

Kernel: 6.12.45-1-lts (Also tested with linux 6.12.xx, same result)

Driver: nvidia-dkms 580.82.07-1

Boot Parameters: BOOT_IMAGE=/vmlinuz-linux-lts root=UUID=d317ba73-8512-444d-ab6e-ab00dcb62165 rw loglevel=3 quiet

Problem Description
The kernel panics shortly after the NVIDIA kernel modules (nvidia, nvidia_modeset, nvidia_uvm, nvidia_drm) are loaded. The panic is triggered by an ACPI interrupt, which leads to a page fault in the NVIDIA driver's rm_acpi_notify function. The system is completely unstable and reboots or hangs.

The open-source nouveau driver does not cause a panic and allows the system to boot, confirming a fundamental issue with the proprietary driver.

Key dmesg Output (Leading to Panic)
text
[ +0.000001] BUG: unable to handle page fault for address: fffffbeb5404a0c8
[ +0.000002] #PF: supervisor read access in kernel mode
[ +0.000001] #PF: error_code(0x0000) - not-present page
[ +0.000001] PGD 0 P4D 0
[ +0.000002] Oops: 0000 [#1] PREEMPT SMP PTI
[ +0.000003] CPU: 3 PID: 122 Comm: kworker/3:3 Not tainted 6.12.45-1-lts #1
[ +0.000002] Hardware name: Dell Inc. G16 7630/0P6FWP, BIOS 1.23.0 03/04/2025
[ +0.000001] Workqueue: kacpi_notify acpi_os_execute_deferred
[ +0.000002] RIP: 0010:kfree+0x6a/0x3a0
...
[ +0.000004] Call Trace:
[ +0.000001] <TASK>
[ +0.000001] ? _nv031312rm+0x55/0x380 [nvidia]
[ +0.000005] ? _nv056082rm+0x6e/0xd0 [nvidia]
[ +0.000003] ? _nv053813rm+0x1b3/0x220 [nvidia]
[ +0.000004] ? rm_acpi_notify+0x16e/0x220 [nvidia]
...
Steps Taken (Without Success)
Confirmed DKMS build: The nvidia module builds successfully against 6.12.45-1-lts headers.

Kernel Parameters: Tried numerous acpi_osi= variants ("Linux", "Windows", "Windows 2025"), acpi=off, pci=noacpi, nomodeset. Only acpi=off prevents the panic but breaks most hardware.

Kernel Types: Tested both linux-lts and linux kernels. Same result.

Module Blacklisting: Booting with modprobe.blacklist=nvidia allows a stable boot, proving the issue is 100% with the NVIDIA module load.

Analysis & Questions
The page fault in rm_acpi_notify suggests a severe incompatibility between the NVIDIA driver 580.82.07 and how ACPI interrupts are handled on this new platform with kernel 6.12.45.

Is this a known issue with the 6.12.y kernel series? Is there a tested DKMS patch or a downgrade path to a stable kernel version that works with this hardware?

Could this be a BIOS-related issue? My BIOS is reasonably recent. Are there specific BIOS settings (e.g., disabling Secure Boot, switching between Discrete/Hybrid graphics) that have proven effective for others on similar G-series laptops?

Given that this is a very new hardware stack, is the best course of action to wait for an updated nvidia-dkms package from Arch, or should I attempt to use a driver from a different source (e.g., the NVIDIA website directly)?

How can I best help debug this? I have full dmesg and journalctl logs available and can provide any other system information required.

Any guidance or pointers would be immensely helpful. Thank you for your time.

Offline

Board footer

Powered by FluxBB