Hello Proxmox Community,
I am facing a critical issue where all my VMs configured with OVMF (UEFI) BIOS fail to boot, while identical VMs configured with SeaBIOS (Legacy BIOS) work perfectly. This issue directly impacts my ability to use PCI Passthrough for my GPU.
System Specifications:
Any VM I create using OVMF (UEFI) BIOS (including VM 106, my main Gaming BATOCERA VM, and a diagnostic VM 107) consistently fails to boot.
Questions to the Community:
I am facing a critical issue where all my VMs configured with OVMF (UEFI) BIOS fail to boot, while identical VMs configured with SeaBIOS (Legacy BIOS) work perfectly. This issue directly impacts my ability to use PCI Passthrough for my GPU.
System Specifications:
- Proxmox VE Version: PVE 8.x (kernel 6.8.12-9-pve)
- CPU: Intel Core i3-6100 (with Intel HD Graphics 530 iGPU)
- RAM: 24 GB
- Storage: 512GB NVMe (Proxmox OS on ZFS), 3TB HDD, 14TB HDD
- GPU: NVIDIA GeForce GTX 1080 Ti (for passthrough attempts)
Any VM I create using OVMF (UEFI) BIOS (including VM 106, my main Gaming BATOCERA VM, and a diagnostic VM 107) consistently fails to boot.
- Symptoms:
- VM status in Proxmox GUI shows "running".
- No video output on the attached physical GPU monitor (for VM 106 with GTX 1080 Ti pass-through attempts).
- No output or errors in the NoVNC console (when vga: virtio is used), typically showing:
failed to load boot 0001 UEFI QEMU DVDROM prom pci root 0x0 ...<br>No bootable option or device found<br>
(Similar errors also occurred when trying to boot from an imported .qcow2 disk or a passed-through physical USB stick). - journalctl -u qemu-server@VMID.service shows -- No entries -- or very early, unhelpful logs.
- VM Configuration (Example - VM 107 "TEST-BOOT" with CorePlus-16.0.iso):
bios: ovmf - boot: order=ide2;net0
- cores: 1
- cpu: kvm64
- efidisk0: local-zfs:vm-106-disk-0,efitype=4m,size=1M <-- (VM 107 uses its own efidisk)
- ide2: local:iso/CorePlus-16.0.iso,media=cdrom,size=273M
- br>machine: q35
- memory: 512
- vga: virtio
(Note: VM 106 uses cpu: host and has hostpci0 configured, but showed the same boot errors even when hostpci0 was removed).
- To isolate the problem, I created a new minimal VM (ID 108, "TEST-LEGACY-BOOT") with SeaBIOS.
- VM Configuration (VM 108):
bios: seabios <-- Key difference<br>boot: order=ide2;net0<br>cores: 1<br>cpu: kvm64<br>ide2: local:iso/CorePlus-16.0.iso,media=cdrom,size=273M<br>machine: q35<br>memory: 512<br>vga: virtio<br> - Result: VM 108 boots successfully from CorePlus-16.0.iso and displays the Tiny Core Linux desktop in the NoVNC console.
Questions to the Community:
- Is this a known bug or incompatibility with OVMF in Proxmox VE 8.x (or the underlying QEMU/OVMF versions) on certain Intel 6th Gen / 200 Series PCH hardware?
- Are there specific OVMF settings, qm set parameters, or kernel arguments that can help debug or fix OVMF boot issues?
- Are there ways to "reset" or reinstall the OVMF firmware within Proxmox without reinstalling the entire host?
- Given that OVMF is failing, are there any obscure workarounds to achieve dGPU passthrough with SeaBIOS (though this is generally not supported)?