Bug update: lock-screen failure still reproduces after lp2152207 PPA
packages

Summary:
The GNOME Wayland lock-screen/return-to-UI failure reproduced after installing 
the lp2152207 test packages from ppa:vanvugt/lp2152207. The graphical GNOME 
session was still present on tty2 while I was logged in on tty3 to collect 
diagnostics.

Package state at time of capture:
- libmutter-18-0, mutter-common, mutter-common-bin, gir1.2-mutter-18:
  50.1-0ubuntu2.1.lp2152207.1
- libegl-mesa0, libgbm1, libgl1-mesa-dri, libglx-mesa0, mesa-libgallium, 
mesa-vulkan-drivers:
  26.0.3-1ubuntu1.lp2152207.1
- gnome-shell: 50.1-0ubuntu1
- gdm3: 50.0-0ubuntu1
- nvidia-driver-595-open: 595.71.05-0ubuntu0.26.04.1
- nvidia-smi reported NVIDIA driver 595.71.05, CUDA 13.2

GPU state:
- System has two NVIDIA RTX A4500 GPUs.
- GPU 0 / bus 00000000:17:00.0: Display Active Off, 14 MiB / 20470 MiB used.
- GPU 1 / bus 00000000:65:00.0: Display Active On, 19388 MiB / 20470 MiB used, 
only 655 MiB free.
- On the display-attached GPU, /usr/bin/gnome-shell PID 918761 was using 17504 
MiB of GPU memory.
- Other GPU clients were much smaller by comparison, e.g. Chrome GPU process 
291 MiB, Thunderbird 221 MiB, Cursor 209 MiB, Nautilus 148 MiB.

High-signal log findings:
- In the active window immediately before capture, 2026-06-03 14:19-14:28, 
gnome-shell logged:
  - 578 x "Error 0x300d reported by eglSwapBuffers"
  - 578 x "meta_onscreen_native_swap_buffers_with_damage failed. Frame aborted."
  - 67 x "Failed to post KMS update: drmModeAtomicCommit: Invalid argument"
  - 159 x "Cursor update failed: drmModeAtomicCommit: Invalid argument"
  - 195 x "meta_monitor_manager_get_logical_monitor_from_number" assertion 
failures
  - 195 x "meta_workspace_get_work_area_for_monitor" assertion failures
- In the same 14:19-14:28 window, the kernel logged:
  - 1298 x "[nvidia-drm] Failed to allocate NVKMS memory for GEM object"
  - 2 x "workqueue: nv_drm_handle_hotplug_event [nvidia_drm] hogged CPU"
- Across June 3 in the captured boot journal, the same error family appears 
many more times:
  - 7321 x eglSwapBuffers error 0x300d
  - 7321 x meta_onscreen_native_swap_buffers_with_damage failed
  - 1419 x Failed to post KMS update / drmModeAtomicCommit invalid argument
  - 4805 x Cursor update failed / drmModeAtomicCommit invalid argument
  - 16617 x NVIDIA DRM failed to allocate NVKMS memory for GEM object

Interpretation:
This looks like the same lock-screen/Wayland compositor failure continuing 
after the lp2152207 test packages. At capture time, gnome-shell was still 
running but had consumed almost all VRAM on the display-attached RTX A4500, and 
GNOME Shell/NVIDIA DRM were repeatedly failing EGL/KMS/NVKMS operations. The 
capture therefore suggests a compositor/GPU memory leak or unreleased 
KMS/GBM/NVKMS resources leading to render/atomic commit failure, rather than a 
simple GNOME Shell process crash.


** Attachment added: "failed-after-lp2152207-20260603-142926.tar.gz"
   
https://bugs.launchpad.net/ubuntu/+source/mutter/+bug/2154316/+attachment/5975306/+files/failed-after-lp2152207-20260603-142926.tar.gz

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2154316

Title:
  GNOME Shell Wayland NVIDIA lock screen recovery fails with
  gbm_surface_lock_front_buffer and high gnome-shell GPU memory

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/mutter/+bug/2154316/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to