Black screen on launch with patches of colored noise after kernel DRM update

[TheColorman]

Hyprland version

v0.52.1, but issue only appears after Linux kernel commit 6e223d5d (bisected). The commit is currently present in kernel versions 6.12.58 and 6.12.59.
Other DEs tested that still work after this kernel commit:

• KDE Plasma 6.5.2
• Gnome 48.4
• Sway 1.11

Regression

Yes, but unsure whether it's a regression in the Linux kernel or that the change in the kernel revealed a preexisting bug in Hyprland.

Describe the bug

Launching Hyprland after the above kernel commit opens a black screen instead of the expected wallpaper, as well as two distinct rectangles of color noise/TV static. Keybinds and applications still seemingly work - I can open a terminal and type commands despite there being no visual feedback (only tested with reboot).

After finding the problematic kernel commit, I tried launching every Hyprland release back to v0.49.0, but v0.50.0 and v0.49.0 seemed to crash on launch for unrelated reasons.

The fact that other Wayland DEs seem to work makes me less suspicious of the Linux kernel, but please let me know if this isn't an issue with Hyprland, but instead with the Linux kernel (though I'm not sure what to do if that is indeed the case).

Reproduction steps

1. Compile and use commit 6e223d5d or later of the Linux kernel.
2. Use any release of Hyprland after v0.50.0.
3. Launch Hyprland either with uwsm start default or directly with Hyprland from tty.
4. See glitched output.

System info and config

Logs saved after booting, launching Hyprland, then pressing SUPER+M to exit Hyprland:

• dmesg.log
• hyprland.log (debug:disable_logs = false)
• hyprland.conf.txt (hyprctl systeminfo -c output)
• monitor_recording.mp4

I run NixOS with 2 identical Asus monitors (the specifics are in hyprland.log). Both monitors are connected directly to my "03:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Navi 48 [Radeon RX 9070/9070 XT/9070 GRE] (rev c0)" GPU using DisplayPort, with cables that came with the monitors (Cable model E118077-Z).

I have included a minimal NixOS configuration that I can boot into to reproduce this issue for reference. The logs above were collected from my normal OS configuration, and not this minimal reproduction.

Reproduction NixOS configuration

{
  inputs = {
    nixpkgs.url = "github:nixos/nixpkgs/1a41aa8"; # <- Last good nixpkgs commit before kernel bump in 9ffa607f which introduced my issue.
    hyprland.url = "github:hyprwm/Hyprland/v0.52.1"; # <- Not following above nixpkgs
  };

  outputs = {
    nixpkgs,
    hyprland,
    ...
  }: {
    # Saber is my system name
    nixosConfigurations.saber = nixpkgs.lib.nixosSystem {
      modules = [
        ({
          pkgs,
          modulesPath,
          lib,
          config,
          ...
        }: {
          # === KERNEL OVERRIDE === #
          boot.kernelPackages =
            pkgs.linuxPackagesFor
            (pkgs.linux_6_12.override {
              argsOverride = {
                src = pkgs.fetchFromGitHub {
                  owner = "torvalds";
                  repo = "linux";
                  # 6e223d5dc81d2e7553c5303b4c642700618648c2 is the first bad commit
                  # commit 6e223d5dc81d2e7553c5303b4c642700618648c2 (HEAD)
                  # Author: Mario Limonciello <[email protected]>
                  # Date:   Wed Aug 27 12:17:19 2025 -0500
                  #
                  #     drm/amd/display: Add fallback path for YCBCR422
                  #
                  #     [ Upstream commit db291ed1732e02e79dca431838713bbf602bda1c ]
                  #
                  #     [Why]
                  #     DP validation may fail with multiple displays and higher color depths.
                  #     The sink may support others though.
                  #
                  #     [How]
                  #     When DP bandwidth validation fails, progressively fallback through:
                  #     - YUV422 8bpc (bandwidth efficient)
                  #     - YUV422 6bpc (reduced color depth)
                  #     - YUV420 (last resort)
                  #
                  #     This resolves cases where displays would show no image due to insufficient
                  #     DP link bandwidth for the requested RGB mode.
                  #
                  #     Suggested-by: Mauri Carvalho <[email protected]>
                  #     Reviewed-by: Wayne Lin <[email protected]>
                  #     Signed-off-by: Mario Limonciello <[email protected]>
                  #     Signed-off-by: Ray Wu <[email protected]>
                  #     Tested-by: Daniel Wheeler <[email protected]>
                  #     Signed-off-by: Alex Deucher <[email protected]>
                  #     Signed-off-by: Sasha Levin <[email protected]>
                  #
                  #  drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 45 ++++++++++++++++++++++++++++++++++++---------
                  #  drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h |  1 +
                  #  2 files changed, 37 insertions(+), 9 deletions(-)

                  # rev = "c5f45678"; # Last good
                  # hash = "sha256-DbJAwDXvOhKh1129yjFGWOV9maiImgAjEuNHwrR3hgA=";
                  # rev = "d5dc97879a97b328a89ec092271faa3db9f2bff3"; # 6.12.59, BAD
                  # hash = "sha256-xyFNSaT+gebkCu/eUNTrH9uqpBLzm4uBB4NSgYpiczs=";
                  rev = "6e223d5dc81d2e7553c5303b4c642700618648c2";
                  hash = "sha256-igqteOacLPUZ70ftPp3G9ls4+3HlvCOiT6HiJA0J6ag=";
                };
                dontStrip = true;
                version = "6.12.57";
                modDirVersion = "6.12.57";
              };
            });

          # === USE UPSTREAM HYPRLAND === #
          nixpkgs.overlays = [
            (_final: _prev: {
              inherit (hyprland.packages.x86_64-linux) hyprland xdg-desktop-portal-hyprland;
            })
          ];
          programs.hyprland.enable = true; # Launch with `Hyprland`

          # === OTHER SYSTEM STATE === #
          system.stateVersion = "24.05"; # This is my current stateVersion

          i18n = {
            defaultLocale = "en_DK.UTF-8";
            extraLocaleSettings = {
              LC_ADDRESS = "da_DK.UTF-8";
              LC_IDENTIFICATION = "da_DK.UTF-8";
              LC_MEASUREMENT = "da_DK.UTF-8";
              LC_MONETARY = "da_DK.UTF-8";
              LC_NAME = "da_DK.UTF-8";
              LC_NUMERIC = "en_US.UTF-8";
              LC_PAPER = "da_DK.UTF-8";
              LC_TELEPHONE = "da_DK.UTF-8";
              LC_TIME = "da_DK.UTF-8";
            };
          };

          users.users."color" = {
            isNormalUser = true;
            password = "password";
            description = "color";
            extraGroups = ["wheel"];
          };

          time.timeZone = "Europe/Copenhagen";

          boot.loader = {
            grub = {
              enable = true;
              device = "nodev";
              configurationLimit = 40;
              efiSupport = true;
            };
            efi.canTouchEfiVariables = true;
          };

          # === HARDWARE-CONFIGURATION.NIX - DEPENDS ON HARDWARE! === #

          imports = [
            (modulesPath + "/installer/scan/not-detected.nix")
          ];

          boot.initrd.availableKernelModules = ["nvme" "xhci_pci" "ahci" "usbhid"];
          boot.initrd.kernelModules = [];
          boot.kernelModules = ["kvm-amd"];
          boot.extraModulePackages = [];

          fileSystems."/" = {
            device = "/dev/disk/by-uuid/deb9b618-c83e-4600-9802-94be15bb8417";
            fsType = "ext4";
          };

          fileSystems."/bin" = {
            device = "/usr/bin";
            fsType = "none";
            options = ["bind"];
          };

          fileSystems."/boot" = {
            device = "/dev/disk/by-uuid/30E7-B53D";
            fsType = "vfat";
            options = ["fmask=0077" "dmask=0077"];
          };

          swapDevices = [];

          # Enables DHCP on each ethernet and wireless interface. In case of scripted networking
          # (the default) this is the recommended approach. When using systemd-networkd it's
          # still possible to use this option, but it's recommended to use it in conjunction
          # with explicit per-interface declarations with `networking.interfaces.<interface>.useDHCP`.
          networking.useDHCP = lib.mkDefault true;

          nixpkgs.hostPlatform = lib.mkDefault "x86_64-linux";
          hardware.cpu.amd.updateMicrocode = lib.mkDefault config.hardware.enableRedistributableFirmware;
        })
      ];
    };
  };
}

[vaxerski]

but issue only appears after Linux kernel commit

blame kernel then. Kernel has a strict no-breaking-userspace policy, I'm pretty sure they will not let this slide into a full release

[TheColorman]

But the issue is present in 6.12.58, a full LTS kernel release, so something's gone wrong somewhere. I was surprised an issue was released to the kernel, hence why I thought it might have been Hyprland/Aquamarine accidentally misusing something that just happened to work before this.

I'll trust your judgement though, I dont know nearly enough about this topic to draw any conclusion 😁

[vaxerski]

just call linus he'll fix it

[UjinT34]

Are you sure that your monitor can handle 10bpc at max res and rate without chroma subsampling? Other compositors probably limit it to 8bpc or set lower refresh rate.

[TheColorman]

@UjinT34 I can't believe that's not the first thing I tried, but switching my from 260hz to 240hz in my Hyprland config solved the issue, so thank you so much! When I set my Hyprland config to output 10bpc with 240hz, my monitors work, but report 8bpc, which I assume is the kernel commit working as it should.

I feel stupid for not considering my bandwidth limits for a kernel commit that literally changes the behavior of limited bandwidth connections...