Summary
I am seeing intermittent full-desktop flickering / corrupted rendering on Ubuntu
26.04 with Intel i915 graphics and GNOME Wayland.
When it happens, all active displays can flicker together. The issue is
intermittent and is more likely during window dragging or display reconfiguration.
## System
- Ubuntu 26.04
- GNOME Shell 50.1
- GDM Wayland session
- Intel Raptor Lake-P integrated graphics
- PCI ID: `8086:a7ab`
- Driver: `i915`
- Laptop: Xiaomi Redmi Book 14 2025
- Display topology:
- Internal eDP panel: 2240x1400 @ 60Hz, 133% scale
- External DP display: 3840x2160 @ 60Hz, 133% scale
- External HDMI display: 3840x2160 @ 60Hz, 133% scale
## Symptoms
The desktop intermittently flickers or displays rendering corruption across
multiple displays at the same time.
It can happen while rapidly dragging windows between monitors, but it is not
reliably reproducible. It may recover by itself after a few seconds.
## Relevant GNOME Shell logs
During a failure, GNOME Shell repeatedly logs:
```text
Failed to lock front buffer on /dev/dri/card1:
gbm_surface_lock_front_buffer failed
Failed to query buffer age, got error 3003
Page flip failed: drmModeAtomicCommit: Invalid argument
The errors can occur thousands of times during a short failure window.
Kernel logs
There are no i915 GPU hang, reset, or GPU timeout messages around the issue.
However, I have seen:
workqueue: i915_hotplug_work_func [i915] hogged CPU for >10000us
Workarounds already tested
I tested the following i915 boot parameters, individually and cumulatively:
i915.enable_psr=0
i915.enable_dc=0
i915.enable_fbc=0
i915.enable_dsb=0
The issue still occurs.
I also disabled the Ubuntu Tiling Assistant GNOME extension. The issue still
occurs, so it does not appear to be caused by that extension.
Changing one external monitor from USB-C/DP to native HDMI did not eliminate
the issue.
Kernel comparison
The issue occurred on Ubuntu kernel:
7.0.0-30-generic
I am currently testing upstream kernel:
7.2.0-070200-generic
with the same i915 parameters and display topology. So far it has not yet
reproduced on 7.2, but more testing time is needed before concluding it is
fixed.
Questions
1. Is this a known i915 / Mutter / GBM issue with multiple scaled 4K outputs
under GNOME Wayland?
2. Are there recommended logs or debug parameters to identify which atomic
display state is rejected?
3. Is there a known fix in a newer kernel, Mesa, Mutter, or GNOME Shell release?