| Summary: | AMDGPU ring gfx_0.1.0 timeout / GPU hang when using KWin Wayland on Radeon iGPU (Ryzen 9950X3D) – complete freeze | ||
|---|---|---|---|
| Product: | [Plasma] kwin | Reporter: | mikael1022 <mikael1022bzh> |
| Component: | wayland-generic | Assignee: | KWin default assignee <kwin-bugs-null> |
| Status: | RESOLVED UPSTREAM | ||
| Severity: | grave | CC: | xaver.hugl |
| Priority: | NOR | ||
| Version First Reported In: | 6.5.3 | ||
| Target Milestone: | --- | ||
| Platform: | Fedora RPMs | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed/Implemented In: | ||
| Sentry Crash Report: | |||
| Attachments: | journalctl -b --priority=3 + journalctl --user -b | grep -i kwin | ||
Please report driver bugs to the GPU vendor, in this case at https://gitlab.freedesktop.org/mesa/mesa/-/issues for the reset happening in the first place (I assume, the log isn't clear about what triggered it), and https://gitlab.freedesktop.org/drm/amd/-/issues for failing to complete the GPU reset. |
Created attachment 187348 [details] journalctl -b --priority=3 + journalctl --user -b | grep -i kwin Hi, I am experiencing repeated GPU hangs when using KWin under Wayland with the AMD integrated GPU included in the AMD Ryzen 9 9950X3D (Granite Ridge / RDNA2 iGPU). The desktop freezes, screen outputs disappear, and the kernel logs show amdgpu ring timeouts and failed GPU resets. KWin reports “There are no outputs – creating placeholder screen”. The issue occurs on a fresh Fedora 43 installation, kernel 6.17.9, KDE Plasma 6.5.3, Mesa 24.x. Here are the exact kernel error messages: amdgpu 0000:76:00.0: amdgpu: ring gfx_0.1.0 timeout amdgpu 0000:76:00.0: amdgpu: Starting gfx_0.1.0 ring reset amdgpu 0000:76:00.0: amdgpu: Ring gfx_0.1.0 reset failed amdgpu 0000:76:00.0: amdgpu: GPU recovery failed: -62 amdgpu 0000:76:00.0: amdgpu: ASIC reset failed User logs show: plasmashell: “There are no outputs - creating placeholder screen” kwin_wayland: “XWAYLAND: mode -1x-1 is not available” The issue does not occur under Plasma X11. GPU coredump is sometimes created in /sys/class/drm/card0/device/devcoredump. Please tell me what additional logs you need.