I can confirm amdgpu.dcdebugmask=0x200 fixes it, no more stutters.
It would be very beneficial to include the patch making this behavior default (https://lore.kernel.org/amd-gfx/20250221160145.1730752-3-zaeem.moha...@amd.com/T/#u) in the relevant linux-stable trees, as this is a major issue for affected systems. I CC'd sta...@vger.kernel.org and ask for the patch to be included in linux-stable, even though it does not have a "Fixes:" tag, if the maintainers are okay with it. Thank you for the quick response! Tiago Feb 23, 2025, 17:26 by mario.limoncie...@amd.com: > Hello, > > This sounds like to me a problem with PSR-SU. There have been a variety of > these being reported the past few months. Enough so that the policy to > enable it by default is being rolled back for now. > > https://lore.kernel.org/amd-gfx/20250221160145.1730752-3-zaeem.moha...@amd.com/T/#u > > To confirm that's the root cause you can disable it with > amdgpu.dcdebugmask=0x200. > > Thanks, > > On 2/23/2025 10:08, m...@tteles.dev wrote: > >> Greetings, >> >> I wish to report a regression in amdgpu firmware introduced in commit >> c99eeb4d0e13f5831ae77f7ec521162594385d5f, the problem persists until git >> HEAD, and reverting to the previous commit fixes the issues with no further >> changes to the environment. >> >> The issue appears on a Lenovo IdeaPad Pro 5, with a Ryzen 8845HS processor, >> using the 780M iGPU (1002:1900 I believe). The screen on this laptop is >> 120Hz 2800x1800, supporting HDR. >> This was tested on Archlinux 6.13.3.arch1-1, with amdgpu drivers, on both >> Wayland (Gnome and Hyprland) and Xorg (i3wm). I'm using amd-pstate-epp >> scaling driver, tested with both performance and powersave governers, to the >> same effect. >> >> I will list the symptoms and attempt to guess at what the issue may be. >> >> - The screen stops updating if no mouse input is given while watching low >> resolution video (wherever it is, Brave and mpv, regardless of software or >> hw decoding). >> Low resolution matters here, I can play 4k video fine, 1080p with mild >> stutters, but lower becomes very bad. >> Here is an example using mpv with 360p sample video, recorded on my phone >> as to not disturb the environment its running in. >> (https://youtu.be/kYHqBjPxM2s You can tell it is very choppy, the original >> video https://www.youtube.com/watch?v=1HrXwe6s4W8 is not choppy) >> mpv did not report any dropped frames despite there obviously being a huge >> amount. >> >> - In some programs, text does not get displayed instantly from when typing >> it, having to type 5 or 6 characters for the previous to finally show. >> All these new characters show up at the same time. This was extremely >> noticeable with gnome's default "Console", it also happened in Brave >> Browser's search bar, and Signal, it does not seem to happen with Ghostty. >> >> Constantly waving my cursor fixes all these issues. Connecting a secondary >> display seems to fix the issue on all displays. >> >> From my limited knowledge this seems like a bug in damage detection for eDP >> + power management, where low enough power levels don't trigger redraws. >> >> I am confident the issue was introduced in that commit since I bisected >> linux-firmware commits affecting amdgpu until the issue was no longer found. >> >> >> Extra info: >> # dmesg | grep amd >> [ 0.414396] perf/amd_iommu: Detected AMD IOMMU #0 (2 banks, 4 >> counters/bank). >> [ 1.875492] [drm] amdgpu kernel modesetting enabled. >> [ 1.878563] amdgpu: Virtual CRAT table created for CPU >> [ 1.878573] amdgpu: Topology: Add CPU node >> [ 1.882607] amdgpu 0000:63:00.0: amdgpu: Fetched VBIOS from VFCT >> [ 1.882610] amdgpu: ATOM BIOS: 113-PHXGENERIC-001 >> [ 1.909536] amdgpu 0000:63:00.0: vgaarb: deactivate vga console >> [ 1.909542] amdgpu 0000:63:00.0: amdgpu: Trusted Memory Zone (TMZ) >> feature enabled >> [ 1.909612] amdgpu 0000:63:00.0: amdgpu: VRAM: 4096M 0x0000008000000000 - >> 0x00000080FFFFFFFF (4096M used) >> [ 1.909614] amdgpu 0000:63:00.0: amdgpu: GART: 512M 0x00007FFF00000000 - >> 0x00007FFF1FFFFFFF >> [ 1.909829] [drm] amdgpu: 4096M of VRAM memory ready >> [ 1.909832] [drm] amdgpu: 13932M of GTT memory ready. >> [ 1.935030] amdgpu 0000:63:00.0: amdgpu: reserve 0x4000000 from >> 0x80f8000000 for PSP TMR >> [ 2.482626] amdgpu 0000:63:00.0: amdgpu: RAS: optional ras ta ucode is >> not available >> [ 2.491068] amdgpu 0000:63:00.0: amdgpu: RAP: optional rap ta ucode is >> not available >> [ 2.491071] amdgpu 0000:63:00.0: amdgpu: SECUREDISPLAY: securedisplay ta >> ucode is not available >> [ 2.521960] amdgpu 0000:63:00.0: amdgpu: SMU is initialized successfully! >> [ 2.601611] kfd kfd: amdgpu: Allocated 3969056 bytes on gart >> [ 2.601628] kfd kfd: amdgpu: Total number of KFD nodes to be created: 1 >> [ 2.602048] amdgpu: Virtual CRAT table created for GPU >> [ 2.602191] amdgpu: Topology: Add dGPU node [0x1900:0x1002] >> [ 2.602193] kfd kfd: amdgpu: added device 1002:1900 >> [ 2.602206] amdgpu 0000:63:00.0: amdgpu: SE 1, SH per SE 2, CU per SH 6, >> active_cu_number 12 >> [ 2.602212] amdgpu 0000:63:00.0: amdgpu: ring gfx_0.0.0 uses VM inv eng 0 >> on hub 0 >> [ 2.602215] amdgpu 0000:63:00.0: amdgpu: ring comp_1.0.0 uses VM inv eng >> 1 on hub 0 >> [ 2.602216] amdgpu 0000:63:00.0: amdgpu: ring comp_1.1.0 uses VM inv eng >> 4 on hub 0 >> [ 2.602217] amdgpu 0000:63:00.0: amdgpu: ring comp_1.2.0 uses VM inv eng >> 6 on hub 0 >> [ 2.602219] amdgpu 0000:63:00.0: amdgpu: ring comp_1.3.0 uses VM inv eng >> 7 on hub 0 >> [ 2.602220] amdgpu 0000:63:00.0: amdgpu: ring comp_1.0.1 uses VM inv eng >> 8 on hub 0 >> [ 2.602221] amdgpu 0000:63:00.0: amdgpu: ring comp_1.1.1 uses VM inv eng >> 9 on hub 0 >> [ 2.602222] amdgpu 0000:63:00.0: amdgpu: ring comp_1.2.1 uses VM inv eng >> 10 on hub 0 >> [ 2.602223] amdgpu 0000:63:00.0: amdgpu: ring comp_1.3.1 uses VM inv eng >> 11 on hub 0 >> [ 2.602224] amdgpu 0000:63:00.0: amdgpu: ring sdma0 uses VM inv eng 12 on >> hub 0 >> [ 2.602225] amdgpu 0000:63:00.0: amdgpu: ring vcn_unified_0 uses VM inv >> eng 0 on hub 8 >> [ 2.602226] amdgpu 0000:63:00.0: amdgpu: ring jpeg_dec uses VM inv eng 1 >> on hub 8 >> [ 2.602227] amdgpu 0000:63:00.0: amdgpu: ring mes_kiq_3.1.0 uses VM inv >> eng 13 on hub 0 >> [ 2.609481] amdgpu 0000:63:00.0: amdgpu: Runtime PM not available >> [ 2.610787] [drm] Initialized amdgpu 3.60.0 for 0000:63:00.0 on minor 1 >> [ 2.617618] fbcon: amdgpudrmfb (fb0) is primary device >> [ 3.788332] amdgpu 0000:63:00.0: [drm] fb0: amdgpudrmfb frame buffer >> device >> [ 5.120419] kvm_amd: TSC scaling supported >> [ 5.120426] kvm_amd: Nested Virtualization enabled >> [ 5.120428] kvm_amd: Nested Paging enabled >> [ 5.120430] kvm_amd: LBR virtualization supported >> [ 5.120443] kvm_amd: Virtual GIF supported >> [ 5.120444] kvm_amd: Virtual NMI enabled >> [ 5.170720] snd_hda_intel 0000:63:00.1: bound 0000:63:00.0 (ops >> amdgpu_dm_audio_component_bind_ops [amdgpu]) >> [ 5.191443] amd_atl: AMD Address Translation Library initialized >> >> >> Thank you for your time, >> Tiago Teles >>
