Fedora Account System
Red Hat Associate
Red Hat Customer
1. Please describe the problem: After upgrading to kernel-7.2.5-200.fc44.x86_64, my second monitor goes completely blank on boot despite being detected correctly by the desktop (KDE Plasma / KWin Wayland shows it in Settings > Display, and DRM enumerates all its connectors). The kernel log shows the amdgpu driver failing to load the DMCUB (Display Micro-Controller) firmware via the PSP, which cascades into EDID/HDMI-infoframe failures on every connector and eventually a pageflip timeout on the affected CRTC. Hardware: Bosgame EffiZen Series, AMD Ryzen 7 6800H with Radeon Graphics, AMD Radeon 680M (Rembrandt / YELLOW_CARP, PCI ID 1002:1681, subsystem 1002:0124, rev c8), driver "amdgpu". Two monitors; the second one is the one that goes blank. Desktop: KDE Plasma 6.7.5 (plasma-workspace-6.7.5-1.fc44, kwin-6.7.5-1.fc44), Wayland session (kwin_wayland). mesa-dri-drivers-26.1.8-1.fc44.x86_64. 2. What is the Version-Release number of the kernel: kernel-7.2.5-200.fc44.x86_64 (built Fri Sep 11 15:11:05 UTC 2026) 3. Did it work previously in Fedora? If so, what kernel version did the issue *first* appear? Yes. kernel-7.2.4-200.fc44.x86_64 (built Mon Sep 7 19:10:19 UTC 2026) works correctly with both monitors, using the exact same linux-firmware-20260910-1 / amd-gpu-firmware-20260910-1 packages (installed in the same dnf transaction as kernel 7.2.5, transaction #69, 2026-09-13). The issue first appeared with kernel-7.2.5-200.fc44.x86_64. Since both kernels were tested against the identical firmware version, this looks like a regression in the amdgpu driver code shipped in 7.2.5, not solely a firmware-blob issue. 4. Can you reproduce this issue? If so, please provide the steps to reproduce the issue below: 100% reproducible on every 7.2.5 boot, 0% on every 7.2.4 boot. Steps: 1. Have a dual-monitor setup on an AMD Rembrandt/YELLOW_CARP APU (Radeon 680M or similar). 2. Boot kernel-7.2.5-200.fc44.x86_64. 3. Observe the second monitor. Actual results: Second monitor stays blank/no signal. It is still detected/listed in system display settings and enumerated by DRM (kernel sees all connectors), but never displays an image. journalctl -k for that boot is flooded with hundreds of amdgpu DMCUB/DMUB errors, and KWin logs a pageflip timeout. Expected results: Both monitors display correctly, as they do on kernel 7.2.4. Verified across 6 separate boots into kernel 7.2.5 (140-503 "DMCUB error" log lines each, 3-28 "EDID fail" occurrences, 1 "DMCUB ucode load fail" each); 0 occurrences across multiple boots into kernel 7.2.4 with the identical linux-firmware-20260910-1 package installed (including a boot on 7.2.4 taken *after* the firmware update, confirming it's not firmware alone). 5. Does this problem occur with the latest Rawhide kernel? Not tested. 6. Are you running any modules that not shipped with directly Fedora's kernel? No. 7. Please attach the kernel logs. Key excerpt from an affected boot (kernel 7.2.5, 2026-09-13 22:27): amdgpu 0000:64:00.0: [drm] Loading DMUB firmware via PSP: version=0x0400004A amdgpu 0000:64:00.0: [drm] *ERROR* Error getting DMUB FW meta info: 4 amdgpu 0000:64:00.0: failed to load ucode DMCUB(0x3F) amdgpu 0000:64:00.0: psp gfx command LOAD_IP_FW(0x6) failed and response status is (0xFFFF0008) amdgpu 0000:64:00.0: [drm] Wait for DMUB auto-load failed: 3 amdgpu 0000:64:00.0: [drm] DMUB hardware initialized: version=0x0400004A amdgpu 0000:64:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data [... repeats hundreds of times per boot ...] amdgpu 0000:64:00.0: [drm] *ERROR* EDID CEA parser failed amdgpu 0000:64:00.0: [drm] HDMI-A-1: PSR support 0, DC PSR ver -1, sink PSR ver 0 DPCD caps 0x0 su_y_granularity 0 amdgpu 0000:64:00.0: [drm] *ERROR* EDID CEA parser failed amdgpu 0000:64:00.0: [drm] HDMI-A-2: PSR support 0, DC PSR ver -1, sink PSR ver 0 DPCD caps 0x0 su_y_granularity 0 amdgpu 0000:64:00.0: [drm] DP-1: PSR support 0, DC PSR ver -1, sink PSR ver 0 DPCD caps 0x0 su_y_granularity 0 amdgpu 0000:64:00.0: [drm] *ERROR* dpia_query_hpd_status: for link(4) dpia(0) failed with status(0), current_hpd_status(0) new_hpd_status(0) amdgpu 0000:64:00.0: [drm] *ERROR* Failed to setup vendor infoframe on connector HDMI-A-1: -22 ... amdgpu 0000:64:00.0: [drm] *ERROR* [CRTC:376:crtc-0] hw_done or flip_done timed out kwin_wayland[3488]: Pageflip timed out! This is a bug in the amdgpu kernel driver kwin_wayland[3488]: Please report this at https://gitlab.freedesktop.org/drm/amd/-/issues kwin_wayland[3488]: With the output of 'sudo dmesg' and 'journalctl --user-unit plasma-kwin_wayland --boot 0' Full journalctl -k output for the affected boot will be attached separately. This looks related to (but distinct from) a report on the CachyOS forum describing the same DMCUB-load failure on Radeon 680M/Rembrandt after the linux-firmware-amdgpu 20260910-1 update, attributed there to the firmware bump alone: https://discuss.cachyos.org/t/regression-linux-firmware-amdgpu-20260910-1-dmcub-fails-to-load-on-amd-radeon-680m-rembrandt-causing-slow-boot-and-visual-glitches/35623 However, on this machine the same new firmware works fine under kernel 7.2.4 and only breaks under kernel 7.2.5 - a detail not covered in that thread - so the two kernels' amdgpu driver code likely differ in how they negotiate with the updated firmware. Workaround: boot the previous kernel, kernel-7.2.4-200.fc44.x86_64, via the GRUB/systemd-boot menu until a fix lands. Reproducible: Always
I'm seeing the exact same symptoms (secondary monitor has been working but went black when I updated from 7.2.4-200 to 7.2.5-200) Also using Fedora 44 KDE Plasma (latest shipping versions) Primary monitor is working, Secondary monitor shows up in the system settings, but stays blank. I see the same log entries as above (e.g., "Pageflip timed out! This is a bug in the amdgpu kernel driver") My hardware is AMD 7735 HS processor using the onboard Radeon M680 graphics. It's configured with 32GB of DDR5 RAM and two HDMI video monitors running at 1920x1080 If I use GRUB to boot back to the previous 7.2.4-200 kernel, things are fine again (both monitors work as normal) This is just a "me too" report/comment in case it helps with anything.
Confirming this regression on a Minisforum UM690 (AMD Ryzen 9 6900HX / Radeon 680M - Rembrandt / DCN 3.1). Setup: - OS: Fedora 43 - Displays: Dual Philips 275S (1x DP, 1x HDMI) Behavior: - On kernel 7.2.5-100.fc43: HDMI monitor gets no signal ("No video input"), remains black, and blocks login until physically disconnected. - Journal log shows the same DMCUB cascaded failure and EDID CEA parsing error: `amdgpu 0000:e4:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data` `amdgpu 0000:e4:00.0: [drm] REG_WAIT timeout 1us * 100 tries - dcn31_program_compbuf_size line:142` `amdgpu 0000:e4:00.0: [drm] *ERROR* EDID CEA parser failed` Workaround: - Booting back into kernel 7.1.13-100.fc43 completely resolves the issue, restoring both DP and HDMI outputs immediately.
I don't know how we get the attention of someone at Redhat/Fedora so they can update the kernel, but this issue has been being worked on here for the last month: https://gitlab.freedesktop.org/drm/amd/-/work_items/5616 With the resulting patch from AMD being available here: https://lore.kernel.org/amd-gfx/20260908113338.2433445-59-chen-yu.chen@amd.com/
I have just published a bug on the fedora message board: https://discussion.fedoraproject.org/t/kernel-7-2-5-amdgpu-regression-on-rembrandt-radeon-680m-patch-available-in-upstream-lore/202259 Let's see if this helps.
This morning, Fedora 44 KDE Plasma updated to Linux Kernel 7.2.6-200.fc44.x86_64 (64-bit) and the issue is gone. Specifically, my secondary monitor is properly found during boot, and it works works fine. Thanks!
Just upgraded and it's resolved for me as well. Having said that, I see in this thread that it apparently was due to the amdgpu firmware: https://bodhi.fedoraproject.org/updates/FEDORA-2026-1649616d86, which was apparently resolved in this linux-firmware package released 6 days ago: https://bodhi.fedoraproject.org/updates/FEDORA-2026-91cb8c4497 However, upgrading that package doesn't rebuild the initramfs, so it was only after upgrading the kernel, which triggered rebuilding the initramfs, that this actually resolved for us. Maybe initramfs rebuilds should be triggered on packages like linux-firmware?
I am seeing a very similar regression on a Beelink SER6 with an AMD Radeon 680M (Rembrandt), running Fedora 44. Kernel 7.1.13: HDMI output to my TV works normally. Kernel 7.2.x (including 7.2.7): when using Gamescope as the compositor for Steam, the TV receives no HDMI signal. Returning to kernel 7.1.13 immediately restores HDMI output. I have already updated both linux-firmware and amd-gpu-firmware. The issue is specific to the Gamescope/Steam setup: if I start a normal GNOME session using Mutter, the HDMI output works even with the newer kernel. The relevant setup is: Hardware: Beelink SER6 GPU: AMD Radeon 680M OS: Fedora 44 Affected kernels: 7.2.x, tested through 7.2.7 Working kernel: 7.1.13 Display: HDMI-connected TV Affected compositor: Gamescope Working compositor: Mutter This makes me wonder whether there is a regression in the AMDGPU/DRM/KMS path used by Gamescope, in addition to the DMCUB/HDMI issues already reported here.
That sounds like it might be a separate regression, I'd recommend searching https://gitlab.freedesktop.org/drm/amd/-/issues and opening an issue there if you can't find anything. Make sure you include logs, package versions, and anything else you can think of.