Full system freeze on Framework Laptop 13 AMD Ryzen 7040 / DMCUB + xhci_hcd errors

Summary

My Framework Laptop 13 (AMD Ryzen 7040 Series) completely froze while I was opening a file in Emacs. The keyboard and mouse became unresponsive, the screen was frozen, and the only recovery was a cold restart (hold power button). After inspecting the logs from the previous boot, I found a cluster of AMDGPU DMCUB errors immediately before the freeze, along with a long history of xhci_hcd USB-controller power-state errors.

System Information

Item Value
Model Framework Laptop 13 (AMD Ryzen 7040 Series)
CPU AMD Ryzen 5 7640U w/ Radeon 760M Graphics
BIOS / System Firmware 03.18 (0.0.3.18)
OS Fedora Workstation 44
Kernel 7.0.12-201.fc44.x86_64
Desktop GNOME (Wayland session)
fwupd status No firmware updates available on stable LVFS channel

Problem Description

  • What I was doing: Opening a file in Emacs (emacs / emacsclient).
  • What happened: The entire graphical session froze. No keyboard input, no mouse cursor movement, no response to Ctrl+Alt+F3 or SysRq.
  • Recovery: Cold restart by holding the power button.
  • Frequency: This appears to be part of a longer pattern — last -x shows many historical crash events over the past several months, not just this single incident.

Timeline from the previous boot

The system had been running normally since the morning. Roughly one minute before the hard freeze, the kernel began logging AMDGPU DMCUB errors:

Jun 20 18:51:20 [framework-13.fritz.box] kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Jun 20 18:51:20 [framework-13.fritz.box] kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Jun 20 18:51:20 [framework-13.fritz.box] kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data

After these messages, the journal ends abruptly. There is no clean systemd shutdown sequence, no kernel panic message, and no OOM killer entry — the system simply stopped logging, consistent with a hard GPU/display-controller lockup.

Relevant Log Excerpts

1. AMDGPU DMCUB errors (immediately before freeze)

Jun 20 18:51:20 [framework-13.fritz.box] kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Jun 20 18:51:20 [framework-13.fritz.box] kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Jun 20 18:51:20 [framework-13.fritz.box] kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data

2. xhci_hcd USB power-state errors (chronic, throughout the day)

These errors appeared continuously during the previous boot:

Jun 20 18:35:52 [framework-13.fritz.box] kernel: xhci_hcd 0000:c1:00.3: Refused to change power state from D0 to D3hot
Jun 20 18:36:04 [framework-13.fritz.box] kernel: xhci_hcd 0000:c1:00.3: Refused to change power state from D0 to D3hot
Jun 20 18:36:18 [framework-13.fritz.box] kernel: xhci_hcd 0000:c1:00.3: Refused to change power state from D0 to D3hot
Jun 20 18:36:32 [framework-13.fritz.box] kernel: xhci_hcd 0000:c1:00.3: Refused to change power state from D0 to D3hot
... (repeats hundreds of times) ...
Jun 20 18:51:28 [framework-13.fritz.box] kernel: xhci_hcd 0000:c1:00.3: Refused to change power state from D0 to D3hot

3. Boot / uptime confirmation

$ uptime && who -b
 18:55:23 up 3 min,  1 user,  load average: 0,43, 0,44, 0,19
         system boot  2026-06-20 18:52

This confirms the cold restart occurred at 18:52, shortly after the DMCUB errors at 18:51:20.

4. Current firmware version from fwupdmgr

├─System Firmware:
│     Current version:    0.0.3.18
│     Vendor:             Framework (DMI:INSYDE Corp.)
│     Update State:       Success

fwupdmgr get-updates reports:

Devices with the latest available firmware version:
 • System Firmware
No updates available

The LVFS metadata was refreshed within the last 24 hours, so 03.18 is currently the newest stable BIOS available to this machine.

Additional Context

Historical crashes

The output of last -x shows a long history of crash entries for this laptop over many months, suggesting this is a chronic stability issue rather than a one-time incident.

Emacs crashes (possibly related, possibly separate)

coredumpctl shows several recent Emacs (emacs-30.2-pgtk) segmentation faults:

Sat 2026-06-06 19:31:47 CEST    6622 1000 1000 SIGSEGV present  /usr/bin/emacs-30.2-pgtk        37.6M
Sat 2026-06-06 19:32:39 CEST    7271 1000 1000 SIGSEGV present  /usr/bin/emacs-30.2-pgtk        40.7M
Sat 2026-06-06 19:33:35 CEST    8599 1000 1000 SIGSEGV present  /usr/bin/emacs-30.2-pgtk        38.9M

These are application-level crashes and did not freeze the whole system. I mention them only because the most recent full-system freeze also happened while using Emacs; the DMCUB error, however, points to a GPU/display-firmware issue rather than an Emacs bug.

Ruled out

  • Out of memory: No OOM messages in the journal; free -h shows 26 GiB available after reboot.
  • Disk full: Root filesystem is 84% full (151 GiB free), not a critical state.
  • Kernel panic: No panic, hung_task, or BUG messages in the previous boot log.

Steps Already Taken

  1. Cold-restarted the laptop.
  2. Checked fwupdmgr get-updates — no BIOS update available on stable LVFS.
  3. Reviewed previous-boot journal with journalctl -b -1 and found the DMCUB + xhci_hcd errors above.

Request for Help

Has anyone else with the Framework 13 AMD Ryzen 7040 series seen this exact DMCUB + xhci_hcd pattern on BIOS 03.18? I would appreciate guidance on:

  1. Whether a newer BIOS (e.g., 03.19+) exists or is expected on LVFS soon.
  2. Whether enabling the LVFS testing channel is worth the risk for this model.
  3. Which kernel boot parameters are currently recommended for Ryzen 7040 DMCUB stability. I am considering starting with amdgpu.dcdebugmask=0x10 to disable PSR-SU.
  4. Any other diagnostic steps I should take before opening a support ticket.

Thanks in advance.

Do you also have any amdgpu flip_done timed out errors in the journal? Either way (especially if yes) try the amdgpu.dcdebugmask=0x10 kernel command line paramter workaround mentioned in that thread.

I’ve been experiencing freezes today as well (AMD 7040, 96gb ram, 1tb ssd), Fedora 43 latest kernel, BIOS 3.18. Whenever I sudo dnf upgrade, I always manually set the kernel to the latest. These freezes happened after the latest restart on the newest kernel.

Screen resolution was 166%. Just tried switching to 200% to see if this helps with the freezing. Thought it was Chrome related, but it happened on firefox-based browsers as well. Temps are ok, not plugged in on battery power when it happens.

I know this doesn’t necessarily contribute to an answer but I hope it provides good points of reference as we troubleshoot. I’ll update if I find anything else of note.

Update: I upgraded to the newest Fedora version with a higher screen resolution, ran sudo dnf upgrade a few times, and haven’t had a full system freeze since. Perhaps it was a bad kernel version? I can’t figure out what caused it.

The only error logs I was running into before this, around the times of the freezes, were wireplumber errors.

I had this crash two days ago–the system entirely locked up while I was editing a file. It’s never happened before and it hasn’t happened since, but yes, I see the “flip_done timed out” error.

jul 05 17:42:36 ... kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
jul 05 17:42:36 ... kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
jul 05 17:42:36 ... kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
jul 05 17:42:46 ... kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* [CRTC:369:crtc-0] flip_done timed out
jul 05 17:43:09 ... kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* flip_done timed out
jul 05 17:43:09 ... kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* [CRTC:369:crtc-0] commit wait timed out
jul 05 17:43:19 ... kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* flip_done timed out
jul 05 17:43:19 ... kernel: amdgpu 0000:c1:00.0: [drm] *ERROR* [CONNECTOR:383:eDP-1] commit wait timed out