** Attachment added: "daily live 2026-04-18 normal boot with builtin display via eDP and external display via USB-C-Radeon - both are working!" https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2144530/+attachment/5962794/+files/gpu-goodboot23-u2604.zip -- You received this bug notification because you are subscribed to linux in Ubuntu. Matching subscriptions: Bgg, Bmail, Nb https://bugs.launchpad.net/bugs/2144530 Title: Regression on Acer Nitro AN17-41: amdgpu DC/DMUB fails to light up internal eDP and Radeon-routed USB-C/DP outputs Status in linux package in Ubuntu: Invalid Status in linux-firmware package in Ubuntu: Invalid Status in linux-firmware-amd-graphics package in Ubuntu: Fix Released Status in linux source package in Resolute: Invalid Status in linux-firmware source package in Resolute: Invalid Status in linux-firmware-amd-graphics source package in Resolute: Fix Released Bug description: [ Impact ] here’s a regression reported by community in 1e2c1534: https://gitlab.freedesktop.org/drm/amd/-/work_items/5143#note_3411517 The last known good version is 43c969c1, and there’s a revert happening in upstream: https://gitlab.freedesktop.org/drm/amd/-/work_items/5143#note_3418637 [ Fix ] Cherry-pick the following revert commit: - d3a8c25 (amdgpu: Revert Yellow Carp DMUB fw to 0x4000045) Firmware in Noble and Questing currently doesn't include the bad version. Only Resolute needs this. [ Test ] Install the firmware package including this revert and boot. The display should light up normally. [ Where the problem could occur] Whatever was introduce after commit 43c969c1 would be reverted. Also, there's an ongoing SRU in Q and N to update the very same file in LP:#2147396. If it happens that they need firmware after commit 43c969c1 to fix the issue, then one of those 2 issues will likely not get fixed. ================= ORIGINAL DESCRIPTION ================= Problem summary =============== On an Acer Nitro AN17-41 laptop, Ubuntu 26.04 development branch with kernel 6.19.0-9-generic fails to display video when the internal panel and/or external USB-C display output are owned by the AMD iGPU. This appears to be a regression or behavior change in the newer Linux graphics stack, because the same machine and same monitor path work under: - Windows 11 - Ubuntu 24.04.4 live ISO The machine has: - Acer Nitro AN17-41 - BIOS V1.25 - AMD Rembrandt / Radeon 680M iGPU (1002:1681) - NVIDIA GeForce RTX 4050 Mobile / Max-Q dGPU (10de:28e1) I collected logs over SSH from another device in the failing cases because the machine was still alive even when both displays were black. Observed behavior on Ubuntu 26.04 / kernel 6.19 ================================================ When AMD owns the display outputs, the system boots but the displays stay black. This affects: - the internal eDP panel - an external monitor connected through the USB-C / DP output routed to Radeon The logs repeatedly show amdgpu DC / DMUB failures such as: - "Failed to enable ASSR" - "No reply for DMUB command: status=3" - "DMCUB error - collecting diagnostic data" - "Error queueing DMUB command: status=2" - "enabling link ... failed: 15" In the failing cases, DRM can still detect AMD-owned outputs as connected and enabled, but they remain black. Environment =========== OS: - Ubuntu 26.04 (Resolute Raccoon development branch) Kernel: - 6.19.0-9-generic Secure Boot: - enabled Hardware: - Acer Nitro AN17-41 - BIOS V1.25 - AMD Radeon 680M iGPU - NVIDIA RTX 4050 Mobile dGPU Regression evidence =================== I now have a working Linux control case on the same laptop and same physical USB-C monitor path. Tested with: - Ubuntu 24.04.4 LTS live ISO Results on Ubuntu 24.04.4: 1. Auto / hybrid mode: - internal panel works - external monitor connected to the Radeon-routed USB-C/DP output also works 2. NVIDIA GPU Only mode: - internal panel works - the same external monitor on the Radeon-routed USB-C/DP output also works in my test In both Ubuntu 24.04.4 test cases, the displays light up normally and I do not see the repeated amdgpu DC/DMUB failure pattern that appears on Ubuntu 26.04 / kernel 6.19. This strongly suggests: - the hardware is capable of driving both outputs under Linux - the Radeon-routed USB-C output is not fundamentally broken on this laptop - the issue is more likely a regression or missing quirk in the newer graphics/kernel stack used in Ubuntu 26.04 Additional control evidence =========================== The same USB-C monitor connected to the Radeon-routed USB-C output also works in Windows 11 when switching to Extend mode with Win+P. This further suggests that the physical hardware path is functional, and that the failing configuration is specific to the newer Linux graphics stack. What works ========== 1. NVIDIA GPU Only mode in firmware / BIOS on Ubuntu 26.04: - the internal laptop panel works 2. Windows 11: - the USB-C monitor connected to the Radeon-routed USB-C output works in Extend mode 3. Ubuntu 24.04.4 live ISO: - Auto / hybrid mode works - internal panel works - Radeon-routed USB-C/DP external monitor works - NVIDIA GPU Only mode also worked in my test, including the same USB-C monitor What does not work ================== On Ubuntu 26.04 / kernel 6.19: 1. Hybrid / normal mode: - if the internal panel is owned by AMD, the panel stays black 2. AMD-only test: - booting with module_blacklist=nvidia,nvidia_drm,nvidia_modeset,nvidia_uvm - internal eDP stays black - external USB-C/DP monitor routed to Radeon also stays black This strongly suggests the issue is not caused by nvidia-drm handoff or by both DRM stacks loading together. How to reproduce ================ Case A: Hybrid / normal mode on Ubuntu 26.04 1. Boot Ubuntu 26.04 on Acer Nitro AN17-41. 2. Let the machine use its normal hybrid graphics routing. 3. If the internal panel is owned by AMD, the panel remains black. Case B: AMD-only test on Ubuntu 26.04 1. Boot with: module_blacklist=nvidia,nvidia_drm,nvidia_modeset,nvidia_uvm 2. Connect an external monitor to the USB-C / DP output routed to Radeon before power-on. 3. Boot Ubuntu. 4. Result: both the internal panel and the external USB-C display remain black, although the system is alive and reachable over SSH. Case C: Verbose DRM logging on Ubuntu 26.04 1. Boot with: drm.debug=0x1e log_buf_len=4M module_blacklist=nvidia,nvidia_drm,nvidia_modeset,nvidia_uvm 2. Result: same black screens, with detailed DRM / amdgpu logging showing DMUB / ASSR / link enable failures. Expected result =============== When AMD owns the internal eDP panel and/or Radeon-routed USB-C/DP output, the displays should light up normally and the system should reach a visible graphical session. Actual result ============= The system boots, but the displays stay black. Logs indicate failures in amdgpu DC / DMUB while trying to bring up AMD-owned display links. Important log patterns ====================== Examples consistently seen in the affected Ubuntu 26.04 boots: - "Skipping amdgpu DM backlight registration" - "Failed to get EDID from ACPI: -19" (seen in verbose DRM logging) - "boot timing validation failed ..." - "Failed to enable ASSR" - "No reply for DMUB command: status=3" - "DMCUB error - collecting diagnostic data" - "Error queueing DMUB command: status=2" - "enabling link 0 failed: 15" - "enabling link 2 failed: 15" Why I think this is a kernel/amdgpu issue ========================================= The issue reproduces when NVIDIA modules are blacklisted, so it does not appear to depend on nvidia-drm being active. The internal panel and the Radeon-routed external USB-C/DP output can both be detected as connected/enabled by DRM, but they still remain black. The same hardware and same USB-C path work: - on Windows 11 - on Ubuntu 24.04.4 live ISO This makes it look like an amdgpu DC / DMUB / link bring-up failure in the newer Ubuntu 26.04 graphics stack, rather than: - a simple EDID-not-found case - a generic Optimus/MUX problem - a pure backlight-only issue - a userspace / compositor-only issue - a fundamental hardware limitation of the laptop Tests already performed ======================= On Ubuntu 26.04: 1. Normal hybrid boot - black internal display when AMD owns it 2. NVIDIA GPU Only mode - internal panel works 3. AMD-only boot with: module_blacklist=nvidia,nvidia_drm,nvidia_modeset,nvidia_uvm - both internal eDP and Radeon-routed USB-C/DP stayed black 4. Verbose DRM boot with: drm.debug=0x1e log_buf_len=4M - same failure, more detailed DMUB / ASSR / link logs 5. Test with: amdgpu.sg_display=0 - no improvement 6. Test with: amdgpu.abmlevel=0 - no improvement On Ubuntu 24.04.4 live ISO: 7. Auto / hybrid mode - internal panel works - Radeon-routed USB-C/DP external monitor works 8. NVIDIA GPU Only mode - internal panel works - the same Radeon-routed USB-C/DP external monitor also worked in my test Attachments =========== I am attaching log bundles from these scenarios: - gpu-badboot4-nvidiablacklist.zip - gpu-badboot5-nvidiablacklistamdusbc.zip - gpu-badboot6-nvidiablacklistamdusbcdbg.zip - gpu-goodboot7-nvidiaonly-w-usbc-radeon.zip - gpu-badboot8-sgzero.zip - gpu-badboot9-abmzero.zip - gpu-goodboot10-u24044.zip - gpu-goodboot11-u24045.zip These include: - journalctl / dmesg output - DRM connector status and enabled state - lspci / module state - backlight info - EDID decode output - verbose DRM logs where applicable I can reproduce this reliably and I can collect additional logs over SSH because the machine stays alive even when both displays are black. ProblemType: Bug DistroRelease: Ubuntu 26.04 Package: linux-image-6.19.0-9-generic 6.19.0-9.9 ProcVersionSignature: Ubuntu 6.19.0-9.9-generic 6.19.5 Uname: Linux 6.19.0-9-generic x86_64 ApportVersion: 2.33.1-0ubuntu3 Architecture: amd64 AudioDevicesInUse: USER PID ACCESS COMMAND /dev/snd/controlC2: denixx 6079 F.... wireplumber /dev/snd/controlC0: denixx 6079 F.... wireplumber /dev/snd/controlC1: denixx 6079 F.... wireplumber /dev/snd/seq: denixx 6059 F.... pipewire CasperMD5CheckResult: pass CurrentDesktop: ubuntu:GNOME Date: Mon Mar 16 08:26:28 2026 InstallationDate: Installed on 2026-03-15 (1 days ago) InstallationMedia: Ubuntu 26.04 LTS "Resolute Raccoon" - Daily amd64 (20260313) IwDevWlp4s0Link: Not connected. MachineType: Acer Nitro AN17-41 ProcEnviron: LANG=uk_UA.UTF-8 PATH=(custom, no user) SHELL=/bin/bash TERM=xterm-256color ProcFB: 0 amdgpudrmfb 1 nvidia-drmdrmfb ProcKernelCmdLine: BOOT_IMAGE=/vmlinuz-6.19.0-9-generic root=/dev/mapper/ubuntu--vg-ubuntu--lv ro quiet splash crashkernel=2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M SourcePackage: linux UpgradeStatus: No upgrade log present (probably fresh install) dmi.bios.date: 09/18/2024 dmi.bios.release: 1.25 dmi.bios.vendor: INSYDE Corp. dmi.bios.version: V1.25 dmi.board.asset.tag: Base Board Asset Tag dmi.board.name: Spacia_PEH dmi.board.vendor: RB dmi.board.version: V1.25 dmi.chassis.type: 10 dmi.chassis.vendor: Acer dmi.chassis.version: Chassis Version dmi.ec.firmware.release: 1.11 dmi.modalias: dmi:bvnINSYDECorp.:bvrV1.25:bd09/18/2024:br1.25:efr1.11:svnAcer:pnNitroAN17-41:pvrV1.25:rvnRB:rnSpacia_PEH:rvrV1.25:cvnAcer:ct10:cvrChassisVersion:sku0000000000000000: dmi.product.family: Acer Nitro 17 dmi.product.name: Nitro AN17-41 dmi.product.sku: 0000000000000000 dmi.product.version: V1.25 dmi.sys.vendor: Acer --- ProblemType: Bug ApportVersion: 2.33.1-0ubuntu3 Architecture: amd64 AudioDevicesInUse: USER PID ACCESS COMMAND /dev/snd/controlC2: ubuntu 4446 F.... wireplumber /dev/snd/controlC0: ubuntu 4446 F.... wireplumber /dev/snd/controlC1: ubuntu 4446 F.... wireplumber /dev/snd/seq: ubuntu 4430 F.... pipewire CasperMD5CheckResult: pass CasperVersion: 26.04.0 CloudArchitecture: x86_64 CloudID: nocloud CloudName: unknown CloudPlatform: nocloud CloudSubPlatform: seed-dir (/var/lib/cloud/seed/nocloud) CurrentDesktop: ubuntu:GNOME DistroRelease: Ubuntu 26.04 IwDevWlp4s0Link: Not connected. LiveMediaBuild: Ubuntu 26.04 LTS "Resolute Raccoon" - Daily amd64 (20260316) MachineType: Acer Nitro AN17-41 NonfreeKernelModules: zfs Package: linux (not installed) ProcEnviron: LANG=C.UTF-8 PATH=(custom, no user) SHELL=/bin/bash TERM=xterm-256color XDG_RUNTIME_DIR=<set> ProcFB: 0 amdgpudrmfb 1 nouveaudrmfb ProcKernelCmdLine: BOOT_IMAGE=/casper/vmlinuz --- quiet splash ProcVersionSignature: Ubuntu 6.19.0-9.9-generic 6.19.5 Tags: resolute wayland-session Uname: Linux 6.19.0-9-generic x86_64 UpgradeStatus: No upgrade log present (probably fresh install) UserGroups: adm cdrom dip lpadmin plugdev sambashare sudo users _MarkForUpload: True dmi.bios.date: 09/18/2024 dmi.bios.release: 1.25 dmi.bios.vendor: INSYDE Corp. dmi.bios.version: V1.25 dmi.board.asset.tag: Base Board Asset Tag dmi.board.name: Spacia_PEH dmi.board.vendor: RB dmi.board.version: V1.25 dmi.chassis.type: 10 dmi.chassis.vendor: Acer dmi.chassis.version: Chassis Version dmi.ec.firmware.release: 1.11 dmi.modalias: dmi:bvnINSYDECorp.:bvrV1.25:bd09/18/2024:br1.25:efr1.11:svnAcer:pnNitroAN17-41:pvrV1.25:rvnRB:rnSpacia_PEH:rvrV1.25:cvnAcer:ct10:cvrChassisVersion:sku0000000000000000: dmi.product.family: Acer Nitro 17 dmi.product.name: Nitro AN17-41 dmi.product.sku: 0000000000000000 dmi.product.version: V1.25 dmi.sys.vendor: Acer --- ProblemType: Bug ApportVersion: 2.33.1-0ubuntu4 Architecture: amd64 AudioDevicesInUse: USER PID ACCESS COMMAND /dev/snd/controlC2: ubuntu 4386 F.... wireplumber /dev/snd/controlC0: ubuntu 4386 F.... wireplumber /dev/snd/controlC1: ubuntu 4386 F.... wireplumber /dev/snd/seq: ubuntu 4370 F.... pipewire CasperMD5CheckMismatches: ./boot/grub/i386-pc/eltorito.img CasperMD5CheckResult: fail CasperVersion: 26.04.1 CloudArchitecture: x86_64 CloudID: nocloud CloudName: unknown CloudPlatform: nocloud CloudSubPlatform: seed-dir (/var/lib/cloud/seed/nocloud) CurrentDesktop: ubuntu:GNOME DistroRelease: Ubuntu 26.04 IwDevWlp4s0Link: Not connected. LiveMediaBuild: Ubuntu 26.04 "Resolute Raccoon" - Daily amd64 (20260322) MachineType: Acer Nitro AN17-41 NonfreeKernelModules: zfs Package: linux (not installed) ProcEnviron: LANG=C.UTF-8 PATH=(custom, no user) SHELL=/bin/bash TERM=xterm-256color XDG_RUNTIME_DIR=<set> ProcFB: 0 amdgpudrmfb 1 nouveaudrmfb ProcKernelCmdLine: BOOT_IMAGE=/casper/vmlinuz --- quiet splash drm.debug=0x1e log_buf_len=4M ProcVersionSignature: Ubuntu 7.0.0-7.7-generic 7.0.0-rc3 Tags: resolute wayland-session Uname: Linux 7.0.0-7-generic x86_64 UpgradeStatus: No upgrade log present (probably fresh install) UserGroups: adm cdrom dip lpadmin plugdev sambashare sudo users _MarkForUpload: True dmi.bios.date: 09/18/2024 dmi.bios.release: 1.25 dmi.bios.vendor: INSYDE Corp. dmi.bios.version: V1.25 dmi.board.asset.tag: Base Board Asset Tag dmi.board.name: Spacia_PEH dmi.board.vendor: RB dmi.board.version: V1.25 dmi.chassis.type: 10 dmi.chassis.vendor: Acer dmi.chassis.version: Chassis Version dmi.ec.firmware.release: 1.11 dmi.modalias: dmi:bvnINSYDECorp.:bvrV1.25:bd09/18/2024:br1.25:efr1.11:svnAcer:pnNitroAN17-41:pvrV1.25:rvnRB:rnSpacia_PEH:rvrV1.25:cvnAcer:ct10:cvrChassisVersion:sku0000000000000000:pfaAcerNitro17: dmi.product.family: Acer Nitro 17 dmi.product.name: Nitro AN17-41 dmi.product.sku: 0000000000000000 dmi.product.version: V1.25 dmi.sys.vendor: Acer To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2144530/+subscriptions
Комментариев нет:
Отправить комментарий