Following is generated by Claude Code CLI. Beep boop! Confirming this on a third hardware configuration, with a different card vendor and board from the reports so far.
HARDWARE GPU: Gigabyte Radeon RX 9070 XT, Navi 48, 1002:7550, subsystem 1458:2424, rev c0 VBIOS: 113-R907XTGOL-F51 DCN: 4.0.1, Display Core v3.2.369 SMU: firmware 104.76.0 (0x00684c00), if version 0x33, driver if version 0x2e CPU: AMD Ryzen 5 5600X (no iGPU, so this is the only GPU in the system) Motherboard: Gigabyte B550 GAMING X V2, BIOS F13 Displays: 3 connected (1x DisplayPort 2560x1440, 2x HDMI 1920x1080) SOFTWARE OS: Linux Mint 22.3 Cinnamon (noble base), X11 Kernel: 7.0.0-42-generic Secure Boot: disabled xserver-xorg-video-amdgpu: 23.0.0-1ubuntu0.24.04.1 Mesa/libdrm: 26.2.6~kisak1~n / 2.4.134-3~kisak~n (kisak PPA, not the Ubuntu archive) SYMPTOMS With 0ubuntu3.4, normal boot stops on a black screen with the Mint splash logo. Booting with nomodeset reaches the desktop, but amdgpu fails to probe (error -22). This failed on all 4 boots with amdgpu active on 0ubuntu3.4, all on 7.0.0-42, with 66 to 220 DMCUB errors per boot. Unlike #2173040, those boots' journals have no Xorg core dump or lightdm restart limit message. The failure may show up differently depending on the configuration. FIRMWARE HEADERS Decompressed and read from the common firmware header of each file: package file links to header ip ucode_version 3.3 dcn_4_0_1_dmcub regular file 4.0 0x0A000800 3.3 dcn_3_2_0_dmcub regular file 3.2 0x07002A00 3.4 dcn_3_2_0_dmcub regular file 3.2 0x07002F00 3.4 dcn_4_0_1_dmcub dcn_3_2_0_dmcub 3.2 0x07002F00 So on 3.4 the driver for a DCN 4.0.1 GPU receives a blob whose own header says IP 3.2. The installed 3.3 file here has the same sha256 as in #2173040 (ecd8d2b8...). KERNEL LOG WITH 0ubuntu3.4 [10.144827] amdgpu 0000:08:00.0: [drm] Loading DMUB firmware via PSP: version=0x07002F00 [10.352950] amdgpu 0000:08:00.0: failed to load ucode DMCUB(0x3F) [10.352956] amdgpu 0000:08:00.0: psp gfx command LOAD_IP_FW(0x6) failed and response status is (0x11) [10.392213] amdgpu 0000:08:00.0: [drm] Display Core v3.2.369 initialized on DCN 4.0.1 [10.494706] amdgpu 0000:08:00.0: [drm] Wait for DMUB auto-load failed: 3 [10.756906] amdgpu 0000:08:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data [19.942720] amdgpu 0000:08:00.0: [drm] *ERROR* EDID CEA parser failed [26.502576] [drm:dcn20_wait_for_blank_complete [amdgpu]] *ERROR* DC: failed to blank crtc! [27.015594] amdgpu 0000:08:00.0: [drm] REG_WAIT timeout 1us * 10 tries - optc3_lock line:128 [27.604897] amdgpu 0000:08:00.0: [drm] REG_WAIT timeout 1us * 100000 tries - optc1_wait_for_state line:844 KERNEL LOG WITH 0ubuntu3.3 (same kernel, Mesa and hardware) amdgpu 0000:08:00.0: [drm] Loading DMUB firmware via PSP: version=0x0A000800 amdgpu 0000:08:00.0: [drm] Display Core v3.2.369 initialized on DCN 4.0.1 amdgpu 0000:08:00.0: [drm] DMUB hardware initialized: version=0x0A000800 No DMCUB errors. radeonsi hardware acceleration works (gfx1201). WORKAROUND Downgraded the whole package to 0ubuntu3.3 and ran apt-mark hold. The 3.3 .deb is no longer in the archive pockets, but it is still available from Launchpad. NOTES FOR THE FIX - Scope is wider than RX 9070 XT: #2173041 has the same errors on 1002:7590 (Navi 44, RX 9050 / 9060 XT). - On this system the initramfs contains neither amdgpu nor its firmware, so both load from the root filesystem. The bad file is picked up without any initramfs involvement here. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2173041 Title: black screen after upgrading linux-firmware-amd-graphics to 0ubuntu3.4 To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux-firmware-amd-graphics/+bug/2173041/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
