Severe Display/Input Issues with RTX A500 Laptop GPU on Fedora 42 (Driver 570.153.02) - DRM Errors

Graphics / Linux Linux
1

Hello,

I am experiencing critical display and input issues on my Lenovo ThinkPad P14s Gen 4 laptop with an NVIDIA RTX A500 Laptop GPU running Fedora 42 and Nvidia driver 570.153.02 (from RPMFusion).

System Details:

  • Laptop Model: Lenovo ThinkPad P14s Gen 4 (21HF001XUS)

  • GPU: NVIDIA RTX A500 Laptop GPU (also has Intel Integrated Graphics - Optimus setup)

  • Operating System: Fedora 42 (Workstation Edition)

  • Kernel Version: 6.14.8-300.fc42.x86_64 (Problem also observed on other F42 kernels like 6.14.5, 6.14.6)

  • Nvidia Driver Version: 570.153.02 (installed via akmod-nvidia from RPMFusion)

  • Display Setup During Failure: Occurs on the internal laptop display, with and without an external Microsoft Thunderbolt Hub + monitors connected.

Problem Description:
The primary issue is a severe click/input misalignment. In both X11 and Wayland sessions, clicks in approximately the top 50% of application windows (observed in GNOME Terminal, Houdini, and other Qt applications) often register on UI elements below the actual cursor position or do not register at all. This makes applications very difficult to use.

Additionally:

  • Application windows (especially for Qt-based software like Houdini and qtdesigner) intermittently fail to open entirely. When this happens, no specific error appears in the launching terminal.

  • The system exhibits general graphics instability.

  • Display hotplugging (connecting/disconnecting the Thunderbolt hub and external monitors) also frequently leads to a degraded state where windows may not open or the click issues become more pronounced.

  • GNOME Shell itself logs errors related to window/actor allocation and monitor management (e.g., “Can’t update stage views actor … needs an allocation”, “meta_monitor_manager_get_logical_monitor_from_number: assertion … failed”) in both X11 and Wayland sessions.

Key Kernel Log Errors (from dmesg):
Even on a clean boot into an X11 session on the internal laptop display without any external peripherals, the following critical Nvidia DRM errors are consistently present:

  • nvidia 0000:03:00.0: [drm] No compatible format found

  • nvidia 0000:03:00.0: [drm] Cannot find any crtc or sizes

These errors appear early in the Nvidia DRM driver initialization.
Other persistent kernel errors include:

  • pcieport : Runtime PM usage count underflow!

  • (When not using nomodeset) i915 0000:00:02.0: [drm] ERROR Failed to get ACT after 3000 ms

Troubleshooting Steps Taken:

  • Confirmed problem across Wayland and X11 sessions.

  • Confirmed problem on internal laptop display only (no hub/external monitors).

  • Tested multiple Fedora 42 kernels (6.14.8, 6.14.6, 6.14.5) - issue persists.

  • Disabled all GNOME Shell extensions.

  • Reinstalled Nvidia drivers (akmod-nvidia, xorg-x11-drv-nvidia-cuda) via DNF.

  • Attempted dnf history undo of major system updates (failed due to unresolved dependencies).

  • Tested with nomodeset kernel parameter:

    • The i915 … Failed to get ACT error was suppressed.

    • However, the nvidia … [drm] No compatible format found and Cannot find any crtc or sizes errors persisted.

    • The click/input misalignment issue was still present in the (very sluggish) X11 session under nomodeset.

  • Verified no custom /etc/X11/xorg.conf or Nvidia-specific files in /etc/X11/xorg.conf.d/ are active.

  • BIOS/Firmware is reported as up-to-date by Lenovo Vantage / fwupdmgr.

  • nvidia-persistenced.service is inactive (which is typical for Optimus).

The issue seems to be a fundamental problem with the Nvidia driver’s ability to correctly initialize display outputs on this hardware/OS combination at the DRM level.

Attached is the nvidia-bug-report.log.gz generated from an X11 session exhibiting the problem.

Any assistance or insights would be greatly appreciated.

Thank you.

nvidia-bug-report.log.gz (370.6 KB)

2

Hello,

This is an update to my previous report regarding issues on a Lenovo ThinkPad P14s Gen 4 (Intel Iris Xe + NVIDIA RTX A500 Laptop GPU) running Fedora 42 (kernel 6.14.8-300.fc42.x86_64) with NVIDIA driver 570.153.02 from RPMFusion.

Summary of Progress:

  1. NVIDIA GPU Offload (prime-run) Now Functional:

  • The critical blocker preventing the NVIDIA GPU from being used via prime-run was the failure to load GSP firmware: Direct firmware load for nvidia/570.153.02/gsp_ga10x.bin failed with error -2.

  • This issue was resolved by:

    1. Installing the nvidia-gpu-firmware package (version 20250509-1.fc42) from the Fedora repositories.

    2. Ensuring a complete reinstallation of all xorg-x11-drv-nvidia-* packages, akmod-nvidia, and nvidia-settings from RPMFusion for driver version 570.153.02.

    3. The open-source nouveau driver remains blacklisted via kernel command line parameters (module_blacklist=nouveau rd.driver.blacklist=nouveau nouveau.modeset=0) as it was previously associated with kernel panics.

  • With these changes, prime-run glxinfo now correctly reports the NVIDIA RTX A500 GPU, and prime-run vkcube also successfully uses the NVIDIA GPU.

  1. UI Click/Input Misalignment & Window Visibility Issues Resolved by GNOME/Mutter Update:
  • The severe click/input misalignment and window visibility problems previously reported (affecting applications like Houdini and GNOME Terminal, especially with display scaling) were resolved by an update to GNOME’s Mutter compositor (mutter-48.3-2.fc42 from mutter-48.3-1.fc42) . This indicates the UI-specific issues were primarily related to the desktop environment’s handling of scaled windows rather than a direct NVIDIA rendering problem for that specific symptom.

Remaining NVIDIA-Related Issues:

  1. Persistent NVIDIA DRM Errors in dmesg:

  • Despite prime-run now functioning correctly for offload rendering, the following NVIDIA DRM errors continue to appear in dmesg immediately after [drm] Initialized nvidia-drm 0.0.0 for 0000:03:00.0 on minor X:

    • nvidia 0000:03:00.0: [drm] No compatible format found

    • nvidia 0000:03:00.0: [drm] Cannot find any crtc or sizes

  • While these don’t currently seem to break prime-run functionality, they suggest an incomplete or imperfect initialization of the display aspects of the NVIDIA DRM driver.

  1. nvidia-powerd Service Errors:

  • The nvidia-powerd service is logging errors at boot and potentially during operation:

    • SBIOS support not found for NVPCF GET_SUPPORTED function

    • No matching GPU found

    • Failed to initialize Dynamic Boost

    • Failed to detach GPU id 768 (or similar GPU ID)

  • These errors may impact dynamic boost functionality or other power management features for the NVIDIA GPU. System suspend is also currently failing (though this involves multiple components, the nvidia-powerd status might be relevant).

  1. PCI ROM Assignment Failure (Likely Minor):

  • The following message appears early in dmesg:

    • pci 0000:03:00.0: ROM [mem size 0x00080000 pref]: failed to assign

  • This seems to be non-critical as the driver loads and prime-run works, but it is a logged failure related to the NVIDIA GPU’s PCI initialization.

Other System Errors Resolved (Previously Mentioned as Potentially Related):

  • The Intel i915 DRM error Failed to get ACT after 3000 ms is no longer observed.

  • The pcieport : Runtime PM usage count underflow! error is no longer observed.

Request:
Could the NVIDIA team offer any insight into the persistent [drm] No compatible format found / Cannot find any crtc or sizes errors, even when prime-run is operational? Additionally, any guidance on the nvidia-powerd errors and their potential impact would be appreciated.

I will generate and attach a new nvidia-bug-report.log.gz from the current system state where prime-run is working but these DRM and nvidia-powerd messages are present.

Thank you for your support.

nvidia-bug-report.log.gz (466.1 KB)