Created attachment 177975 [details] Journal ctl SUMMARY While on a match on Halo Infinite, there's a chance that either Kwin or AMDGPU (Or mesa I think) freezes and crashes, leaving a no signal image with no form recovery. STEPS TO REPRODUCE 1. Have a AMD graphics card 2. Play Halo Infinite OBSERVED RESULT The screen freezes for a few seconds and then turns black, after a few seconds the monitor shows no signal. Sound still works. The AMDGPU tries to reset but fails to do so. EXPECTED RESULT The game continues with no problem SOFTWARE/OS VERSIONS Operating System: Fedora Linux 41 KDE Plasma Version: 6.2.5 KDE Frameworks Version: 6.10.0 Qt Version: 6.8.1 Kernel Version: 6.12.11-200.fc41.x86_64 (64-bit) Graphics Platform: Wayland ADDITIONAL INFORMATION Might be a mesa bug, but I'm unsure.
Relevant bits: fev 04 21:55:22 fedora kernel: gmc_v10_0_process_interrupt: 45 callbacks suppressed fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:4 pasid:32799) fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: in process HaloInfinite.ex pid 12774 thread vkd3d_queue pid 12936 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: in page starting at address 0x0000000000000000 from client 0x1b (UTCL2) fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00401431 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: Faulty UTCL2 client ID: SQC (data) (0xa) fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: MORE_FAULTS: 0x1 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: WALKER_ERROR: 0x0 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: PERMISSION_FAULTS: 0x3 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: MAPPING_ERROR: 0x0 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: RW: 0x0 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: [gfxhub] page fault (src_id:0 ring:24 vmid:4 pasid:32799) fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: in process HaloInfinite.ex pid 12774 thread vkd3d_queue pid 12936 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: in page starting at address 0x0000000000000000 from client 0x1b (UTCL2) fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: GCVM_L2_PROTECTION_FAULT_STATUS:0x00000000 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: Faulty UTCL2 client ID: CB/DB (0x0) fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: MORE_FAULTS: 0x0 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: WALKER_ERROR: 0x0 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: PERMISSION_FAULTS: 0x0 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: MAPPING_ERROR: 0x0 fev 04 21:55:22 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: RW: 0x0 fev 04 21:55:27 fedora kwin_wayland[2039]: kwin_scene_opengl: 0x3: GL_INVALID_OPERATION in glDrawBuffers(unsupported buffer GL_BACK_LEFT) fev 04 21:55:27 fedora kwin_wayland[2039]: kwin_scene_opengl: 0x3: GL_INVALID_OPERATION in glDrawBuffers(unsupported buffer GL_BACK_LEFT) fev 04 21:55:27 fedora kwin_wayland[2039]: kwin_scene_opengl: 0x3: GL_INVALID_OPERATION in glDrawBuffers(unsupported buffer GL_BACK_LEFT) fev 04 21:55:28 fedora kwin_wayland[2039]: kwin_scene_opengl: 0x3: GL_INVALID_OPERATION in glDrawBuffers(unsupported buffer GL_BACK_LEFT) fev 04 21:55:28 fedora kwin_wayland[2039]: kwin_scene_opengl: 0x3: GL_INVALID_OPERATION in glDrawBuffers(unsupported buffer GL_BACK_LEFT) fev 04 21:55:28 fedora kwin_wayland[2039]: kwin_scene_opengl: 0x3: GL_INVALID_OPERATION in glDrawBuffers(unsupported buffer GL_BACK_LEFT) fev 04 21:55:32 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: Dumping IP State fev 04 21:55:32 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: Dumping IP State Completed fev 04 21:55:32 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: ring gfx_0.0.0 timeout, signaled seq=12426568, emitted seq=12426570 fev 04 21:55:32 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: Process information: process HaloInfinite.ex pid 12774 thread vkd3d_queue pid 12935 fev 04 21:55:32 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: GPU reset begin! fev 04 21:55:32 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: MODE1 reset fev 04 21:55:32 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: GPU mode1 reset fev 04 21:55:32 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: GPU smu mode1 reset fev 04 21:55:44 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: GPU reset succeeded, trying to resume fev 04 21:55:44 fedora kernel: [drm] PCIE GART of 512M enabled (table at 0x0000008000300000). fev 04 21:55:44 fedora kernel: [drm] VRAM is lost due to GPU reset! fev 04 21:55:44 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: PSP is resuming... fev 04 21:55:52 fedora kernel: [drm:psp_v11_0_memory_training [amdgpu]] *ERROR* send training msg failed. fev 04 21:55:52 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: Failed to process memory training! fev 04 21:55:52 fedora kernel: [drm:amdgpu_device_fw_loading [amdgpu]] *ERROR* resume of IP block <psp> failed -62 fev 04 21:55:52 fedora kernel: amdgpu 0000:2d:00.0: amdgpu: GPU reset(2) failed fev 04 21:55:52 fedora kwin_wayland_wrapper[2167]: amdgpu: The CS has cancelled because the context is lost. This context is innocent. ... fev 04 21:55:57 fedora kwin_wayland[2039]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug --- It looks like a graphics reset occurs but the system doesn't recover properly.
Created attachment 178003 [details] Another journalctl log This one I waited 5 minutes for the gpu to possibly reset, but nothing. Weird boot messages.
From bug 494440 this ca be ignored: GL_INVALID_OPERATION in glDrawBuffers(unsupported buffer GL_BACK_LEFT) The crash looks to be entirely in the AMD gpu drivers
(In reply to TraceyC from comment #3) > From bug 494440 this ca be ignored: > GL_INVALID_OPERATION in glDrawBuffers(unsupported buffer GL_BACK_LEFT) > > The crash looks to be entirely in the AMD gpu drivers Then it is a Mesa bug. Wasn't sure which one crapped the bed as there are many components that could've done it.