Public bug reported:
# GNOME 50 / mutter: gnome-shell becomes unresponsive (both displays
freeze, audio continues) when an external display wired to the NVIDIA
dGPU comes up — hybrid Intel + NVIDIA, Wayland
Possibly the same class of bug as #4714 / #4732 / #4725 (secondary-GPU
scanout buffer handling in mutter 50, "onscreen/native: Use FBOs for
secondary GPU rendering", MR !4908). Reported as-is because the error
strings seen in those reports do **not** appear in this journal — here
the failure presents as a compositor main-loop hang rather than the
`Failed to ensure KMS FB ID` error loop.
## Affected versions
| Component | Version |
|---|---|
| Distro | Ubuntu 26.04.1 LTS (resolute) |
| GNOME Shell | 50.1 (`gnome-shell 50.1-0ubuntu1.2`) |
| Mutter | 50.1 (`libmutter-18-0 50.1-0ubuntu2.2`, `mutter-common-bin
50.1-0ubuntu2.2`) |
| Kernel | 7.0.0-31-generic |
| NVIDIA driver | 595.91.07 (`nvidia-driver-595-open`), `nvidia-drm modeset=1`,
`NVreg_DynamicPowerManagement=2` |
| Session | Wayland (only session type available on this system; no X11 session
shipped) |
## Hardware
- Laptop: ASUS ROG Strix SCAR 18 G835LW, BIOS G835LW.338, Intel Core Ultra 9
275HX
- Primary GPU: Intel iGPU `8086:7d67`, driver `i915`, `/dev/dri/card1` —
**only** connector `eDP-1` (internal 2560x1600@240 panel)
- Secondary GPU: NVIDIA RTX 5080 Laptop `10de:2c59`, driver `nvidia` (open),
`/dev/dri/card0` — connectors `DP-1`, `DP-2`, `HDMI-A-1`, `eDP-2` — **all
external outputs are wired to the dGPU**
- External display: MSI MAG 275QF (2560x1440), connected via a USB-C dock (VIA
Labs hub, DP Alt Mode) on a Thunderbolt 5 port
## Steps to reproduce
1. Boot into the Wayland session. Internal panel is on the iGPU (`card1-eDP-1
connected`); all dGPU connectors are `disconnected`; the NVIDIA GPU is in
runtime suspend (`power/runtime_status = suspended`, `power/control = auto`).
2. Connect the external monitor through the USB-C dock so that a DP link is
established on the NVIDIA dGPU output.
3. The dock re-enumerates, and gnome-shell starts using the dGPU:
```
2026-09-18 19:02:26 kernel: usb 5-2: USB disconnect, device number 72
2026-09-18 19:02:28 kernel: usb 4-2: USB disconnect, device number 28
2026-09-18 19:02:29 kernel: nvidia 0000:02:00.0: Enabling HDA controller
2026-09-18 19:02:32 kernel: ucsi_acpi USBC000:00: GET_CURRENT_CAM command
failed
2026-09-18 19:02:33 kernel: usb 4-1: New USB device found, idVendor=2109,
idProduct=0817 (dock, now on the other TB5 port)
2026-09-18 19:02:38 gnome-shell[926906]: Created gbm renderer for
'/dev/dri/card0' <-- dGPU used for a new output
2026-09-18 19:04:35 at-spi2-registryd[926957]: Disabling unresponsive app with
pid 926906 <-- gnome-shell hung
```
4. Both the internal panel and the external display show a frozen
picture. Audio playback (PipeWire / EasyEffects) keeps running, so the
rest of userspace is alive. No recovery without a hard power-off.
## Expected behaviour
The external display shows up as an additional monitor, as it did on
this same machine and the same package versions on 2026-09-14
(`~/.config/monitors.xml`: connector `DP-2`, product `MSI MAG 275QF`,
2560x1440 @ 143.993 Hz, `<colormode>bt2100</colormode>`).
## What was ruled out
- **No kernel-side GPU fault at the time of the freeze**: 0 occurrences of
`NVRM: Xid` or "GPU has fallen off the bus" inside the freeze window
(2026-09-18 19:00–19:05). The 4 Xid lines in that boot are all from an
unrelated game process (`Endfield.exe`) hours/days earlier, a separately
diagnosed issue.
- The dock's USB side is fully functional before and after (VIA Labs hub,
RTL8153 GbE, card reader all enumerate normally) — the display path is the only
thing that breaks.
- The compositor hang is not a crash: no coredump; gnome-shell simply stops
answering (at-spi reports it unresponsive) while the rest of the session keeps
running.
## Notes that may be relevant
- This machine's external display comes up reliably on the NVIDIA dGPU in some
attempts and not at all in others (separate, probably firmware-level issue:
ASUS UCSI reports `ucsi_acpi USBC000:00: GET_CURRENT_CAM command failed` /
`Swap rejected`, the Type-C partner reports `number_of_alternate_modes = 0`,
and a `2109:8817 USB Billboard Device` is enumerated). In the failing event the
DP link *did* come up (dock re-enumerated, gnome-shell opened card0) before the
freeze.
- The saved configuration for this external monitor has HDR enabled (`colormode
bt2100`, i.e. a 10-bit format). Given that #4732's description involves the
`AB30` 10-bit format being rejected, HDR may be a contributing factor. This has
not been isolated yet.
- `MUTTER_DEBUG_MULTI_GPU_FORCE_COPY_MODE` is present in the shipped library
with values `primary-gpu-cpu` / `primary-gpu-gpu` (the former is reported to
give black screens in #4714). Not tested here yet.
- Full journal extract, package versions and connector states: see the attached
evidence file `external-display-mutter-evidence-2026-09-18.txt`.
ProblemType: Bug
DistroRelease: Ubuntu 26.04
Package: mutter (not installed)
ProcVersionSignature: Ubuntu 7.0.0-31.31-generic 7.0.14
Uname: Linux 7.0.0-31-generic x86_64
ApportVersion: 2.34.1-0ubuntu0.1
Architecture: amd64
CasperMD5CheckResult: pass
CurrentDesktop: ubuntu:GNOME
Date: Fri Sep 18 19:20:03 2026
InstallationDate: Installed on 2026-09-05 (13 days ago)
InstallationMedia: Ubuntu 26.04 "Resolute Raccoon" - Release amd64 (20260423.1)
ProcEnviron:
LANG=zh_CN.UTF-8
PATH=(custom, no user)
SHELL=/bin/bash
TERM=xterm-256color
XDG_RUNTIME_DIR=<set>
SourcePackage: mutter
UpgradeStatus: No upgrade log present (probably fresh install)
** Affects: mutter (Ubuntu)
Importance: Undecided
Status: New
** Tags: amd64 apport-bug resolute wayland-session
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2167711
Title:
External display wired to the NVIDIA dGPU freezes the whole session
(hybrid Intel+NVIDIA, Wayland)
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/mutter/+bug/2167711/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
