AMDGPU PSR/DMCUB Display Freezes (flip_done timed out)

Summary

The AMDGPU driver's Panel Self-Refresh (PSR) and Display Micro-Controller Unit (DMCUB) can cause display freezes on Framework Laptop 13 AMD models and Framework Desktop. The DMCUB firmware hangs, producing flip_done timed out errors in the kernel log and making the display unresponsive. The issue is tracked upstream as drm/amd#4141 and has recurred across multiple kernel versions from 6.12 through 6.18¹²³.

Symptoms

• Display becomes completely unresponsive — mouse cursor may still move briefly, then everything freezes¹⁴
• DMCUB error spam in dmesg¹:

  amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data

• Pageflip / CRTC timeout in dmesg¹:

  amdgpu 0000:c1:00.0: [drm] *ERROR* [CRTC:80:crtc-0] flip_done timed out

• Compositor warnings — kwin_wayland reports Pageflip timed out! This is a bug in the amdgpu kernel driver¹
• System may become unbearably laggy before fully freezing, or recover after a long delay⁴
• Sound may continue playing even when the display is frozen⁵
• Night light / color temperature features (GNOME Night Light, KDE Night Color) appear to trigger the issue¹⁶

Affected Models

• Framework Laptop 13 (AMD Ryzen 7040 Series) — most commonly reported¹⁴⁷
• Framework Laptop 13 (AMD Ryzen AI 300 Series)⁸
• Framework Desktop (AMD Ryzen AI Max 300 Series)⁹

The issue is not Framework-specific — it affects other AMD laptops using the same GPU family (e.g., ThinkPad Z13 with Ryzen 7 PRO 6850U)³.

Root Cause

The root cause is a DMCUB (Display Micro-Controller Unit Bootloader) firmware hang during PSR (Panel Self-Refresh) state transitions. PSR is a power-saving feature that stops sending display data when the screen content is static, instead having the panel self-refresh from its own memory.

When the DMCUB firmware hangs, it can no longer process page flip requests, causing the flip_done timed out errors and freezing the display pipeline²³. Several factors increase the likelihood of triggering the hang:

• Night light / color temperature changes — these force PSR exit/re-entry cycles that expose the firmware bug¹⁶
• X11 compositors — PSR compatibility with Xorg is particularly problematic; Framework's Mario Limonciello (an AMD employee) confirmed this is "a bug caused by how Xorg vs Wayland work" and submitted a patch to detect X11 and disable PSR automatically⁶¹⁰
• Kernel regressions — the bug has been fixed and regressed multiple times. A regression in kernel versions 6.12.29, 6.14.7, and 6.15rc7 caused the issue to reappear⁷. Users on kernel 6.18.x have again started reporting crashes¹

A kernel patch series proposing page_flip_timeout() support was submitted in February 2026 but is still under review¹¹.

Diagnosis

1. Check kernel logs for DMCUB errors:

   sudo dmesg | grep -E '(DMCUB error|flip_done timed out)'

2. Check journal for crash timeline:

   journalctl -b -1 | grep -E '(amdgpu|DMCUB|flip_done)' | tail -30

3. Verify kernel version — regressions are known in 6.12.29+, 6.14.7+, 6.15rc7, and 6.18.x⁷:

   uname -r

Workarounds & Fixes

Solution	Effectiveness	Notes
amdgpu.dcdebugmask=0x10 kernel parameter468	Most effective	Disables PSR entirely. Add to GRUB_CMDLINE_LINUX_DEFAULT in /etc/default/grub, then run sudo grub2-mkconfig -o /boot/grub2/grub.cfg (BIOS) or sudo grub2-mkconfig -o /boot/efi/EFI/fedora/grub.cfg (UEFI)
amdgpu.dcdebugmask=0x610 kernel parameter8	Effective	More targeted PSR/Replay disable. Tested by Framework's Mario Limonciello
Disable Night Light / Night Color16	May prevent triggers	Reduces PSR state transitions. Works for some users but not all
Switch to Wayland610	Reduces frequency	Avoids X11-specific PSR triggers. Gives better battery life than the dcdebugmask workaround
GPU recovery without reboot4	Temporary fix	sudo cat /sys/kernel/debug/dri/0/amdgpu_gpu_recover (or dri/1/ depending on GPU). Requires SSH access if display is frozen
amdgpu.runpm=0 kernel parameter3	Effective	Disables GPU runtime power management. Significantly higher idle power consumption

Resolution

There is no permanent fix yet. AMD is still investigating the root cause of the DMCUB firmware hangs³. A kernel patch to implement page_flip_timeout() support was proposed in February 2026 as a potential mitigation, but it remains under review on the amd-gfx mailing list¹¹. Separately, a patch to detect X11 and automatically disable PSR was submitted by a Framework engineer¹⁰.

Users should apply the amdgpu.dcdebugmask=0x10 workaround and monitor upstream development on drm/amd#4141³.

Related

• FW13 AMD UI freeze — Framework Community (original tracking thread)
• AI 300 Display locking up with Fedora 42 Xfce — Framework Community
• AMD GPU MES Timeouts Causing System Hangs on Framework Laptop 13 (AMD AI 300 Series) — Framework Community
• drm/amd#4141 — Upstream bug tracker
• drm/amd#4238 — Kernel regression tracker
• ArchWiki — AMDGPU Troubleshooting

Footnotes

1. AMDGPU crashes with 6.18.13-100.fc42.x86_64 — Framework Community ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰

2. AMDGPU Pageflip screen unresponsive — Arch Linux Forums ↩ ↩²

3. [amdgpu][drm] ERROR flip_done timed out — Ubuntu Bug #2047154 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

4. FW13 AMD UI freeze — Framework Community ↩ ↩² ↩³ ↩⁴ ↩⁵

5. System freeze (amdgpu - flip_done timed out) after switch to kernel 6.12 — Manjaro Forum ↩

6. AI 300 Display locking up incessantly with Fedora 42 Xfce — Framework Community ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

7. FW13 AMD UI freeze — regression report — Framework Community (drm/amd#4238) ↩ ↩² ↩³

8. AMD GPU MES Timeouts Causing System Hangs on Framework Laptop 13 (AMD AI 300 Series) — Framework Community ↩ ↩² ↩³

9. AMD Drivers Frequently Hanging and Crashing — Framework Community ↩

10. [PATCH] drm/amd/display: Fix Xorg desktop unresponsive on Replay panel — Framework Community (Mario Limonciello) ↩ ↩² ↩³

11. [PATCH v2 2/2] drm/amdgpu: implement page_flip_timeout() support — amd-gfx Mailing List ↩ ↩²