Windows11 bootloops during install

[Nimrodium]

windows 11 boot loops after restarting while being installed with the latest installer (25H2), I don't have any diagnostic data because I'm unaware how to find that. I was using the mvisor-git AUR package, but switched to manual compilation (I had to apply the patch to spic_agent.cc) to see if that would fix it. but still the same behavior.

windows will boot for 4 seconds and then freeze, then restart.

when i attempted to run the half installed qcow2 image in qemu, it immediately fails and fallsback to the UEFI Shell. manually launching efi/boot/bootx64.efi from the shell gives the same result.

booting the installer in qemu suprisingly failed to boot, and fell back to UEFI Shell quicker than mvisor did in the partially installed medium, but I tried with virt-machine-manager and it was able to boot the installer, i was able to fully install windows 11 in there. attempting to boot this install in mvisor after installing in qemu (i didnt add a TPM module to qemu) it fails to boot as usual.

[78]

Hello,

Thanks for reporting this issue. It sounds like a difficult boot loop problem!

Could you please check for any error logs? Alternatively, could you try enabling debug logging in your mvisor's YAML configuration file and see if any error messages appear when you attempt to boot Windows 11?

This information would be very helpful for diagnosing the boot failure.

[devtinoimao]

i have a similar issue, where it bootloops due to a raw disk image reading error, it just boots for 10 seconds then restarts, the only way to fix this problem was to use a legacy bios and not uefi bios, or switch to windows 10, which is what i have done, but then when i install the vgpu drivers it just hangs and reboots and wouldn't install the vgpu drivers (so i have 2 issues here the raw disk image issue and the vgpu driver issue)

[devtinoimao]

There is a windows for docker image cuz i used a firmware that is used for windows for docker, i've tried others too like regular ovmf from qemu
it just hangs at a certain point during boot and reports hard disk
[38463.868] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4```
./start.sh
[0.000] gui/vnc/server.cc:37 VncServer() debug: VNC server started port=5900
[4563.564] devices/usb/xhci_host.cc:1329 CompleteTransfer() warn: transfer failed status=6
[13586.532] core/machine.cc:160 Reset() debug: Resettings vCPUs
[13586.774] core/vcpu.cc:795 LoadStateFrom() warn: KVM_CAP_TSC_CONTROL was not supported
[13586.910] core/vcpu.cc:795 LoadStateFrom() warn: KVM_CAP_TSC_CONTROL was not supported
[19853.014] devices/usb/usb_device.cc:263 CopyStringsDescriptor() warn: invalid string index=0xee length=18
[19853.083] devices/usb/xhci_host.cc:1329 CompleteTransfer() warn: transfer failed status=6
[33437.390] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[33438.518] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[33440.228] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[33442.073] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[33443.417] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[33444.704] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38446.702] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38450.071] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38454.551] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38456.930] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38458.581] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38460.593] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38465.839] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38468.310] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[38470.339] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[61261.386] core/machine.cc:160 Reset() debug: Resettings vCPUs

[61261.515] core/vcpu.cc:795 LoadStateFrom() warn: KVM_CAP_TSC_CONTROL was not supported
[65258.290] devices/usb/xhci_host.cc:1329 CompleteTransfer() warn: transfer failed status=6
[139079.441] core/machine.cc:160 Reset() debug: Resettings vCPUs
[139079.639] core/vcpu.cc:795 LoadStateFrom() warn: KVM_CAP_TSC_CONTROL was not supported
[139079.709] core/vcpu.cc:795 LoadStateFrom() warn: KVM_CAP_TSC_CONTROL was not supported
[144536.338] devices/usb/usb_device.cc:263 CopyStringsDescriptor() warn: invalid string index=0xee length=18
[144536.349] devices/usb/xhci_host.cc:1329 CompleteTransfer() warn: transfer failed status=6
[157163.700] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[157165.187] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[157166.474] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[157167.685] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[157169.221] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
^AA^A[171496.712] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[171498.889] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[171501.000] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[171503.495] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4
[171505.679] images/raw.cc:100 HandleIoRequest() error: unhandled io request 4


`

[78]

i have a similar issue, where it bootloops due to a raw disk image reading error, it just boots for 10 seconds then restarts, the only way to fix this problem was to use a legacy bios and not uefi bios, or switch to windows 10, which is what i have done, but then when i install the vgpu drivers it just hangs and reboots and wouldn't install the vgpu drivers (so i have 2 issues here the raw disk image issue and the vgpu driver issue)

I haven't tried a Windows 11 25H2 installation yet.
The QCOW2 image format is fully supported and better than the raw image format. You can convert the disk image from raw to QCOW2 with qemu-img.

[devtinoimao]

i had no luck even with qcow2, windows 11 23h2
so i will stick with legacy boot for now
i will also try disabling secure boot in the vm, because perhaps that is whats making the vgpu driver not work

[78]

@devtinoimao You mentioned vGPU. Could you provide your YAML configuration file?

[devtinoimao]

i successfully managed to install windows 11 25h2 under legacy bios, do i need uefi for vgpu?

name: Default configuration
base: q35.yaml

machine:
  memory: 8G
  vcpu: 2
  bios: ../share/bios-256k.bin
  # Set vcpu thread priority value [-20, 19]
  # A higher value means a lower priority
  priority: 2
  # Turn on BIOS output and performance measurement
  debug: No
  # Turn on hypervisor to lower CPU usage (Hyper-V is used for Windows)
  hypervisor: Yes

objects:
  - class: qxl
  - class: spice-agent
  - class: usb-tablet

  - class: virtio-network
    mac: 00:50:00:11:22:33
    map: tcp:0.0.0.0:8022-:22
    # If you have configured the tap interface, you can uncomment the following lines
    # backend: tap
    # ifname: tap0

  - class: ata-disk
    image: /home/abokaborda/Desktop/win11.qcow2
    snapshot: No
  - class: ata-cdrom
    image: /home/abokaborda/Downloads/virtio-win-0.1.285.iso
    snapshot: No
    
  # - class: floppy
  #   image: /data/images/floppy.img

  # - class: virtio-block
  #   image: /data/empty.qcow2
  #   snapshot: No

  # - class: virtio-fs
  #   path: /tmp/fuse
  #   disk_name: mvisor-fs   #   inode_count: 200

  #- class: vfio-pci
  #  sysfs: /sys/bus/mdev/devices/cf8e0a65-e94b-4ac0-bfa7-ed2c6f8de82f
  #  debug: Yes

  - class: virtio-vgpu
    memory: 2G
    staging: No
    blob: No
    node: /dev/dri/renderD129

  # - class: virtio-cuda
  #   memory: 1G
  #   gpu_memory: 8G
  #   visible_gpus: GPU-0

here is my yaml

I have a video regarding the issue but i cant upload it now, upon installing the inf for the vgpu kernelmode it hangs then reboots, even in windows 10

[78]

@nooodles2023 Take a look!

[Nimrodium]

Hello,

Thanks for reporting this issue. It sounds like a difficult boot loop problem!

Could you please check for any error logs? Alternatively, could you try enabling debug logging in your mvisor's YAML configuration file and see if any error messages appear when you attempt to boot Windows 11?

This information would be very helpful for diagnosing the boot failure.

What do I need to add to my yaml config to enable debug logging? I could try legacy bios, i was under the impression windows 11 didn't support legacy bios.

edit: i just looked at the posted yaml config again and noticed the debug: No line in machine, im guessing that's it ? when I get home today I'll add this line and attempt to boot the installed image, then I'll try installing with legacy bios

[Nimrodium]

when running with Debug: Yes it spits out a lot of debug diagnostics but i think the actual error is from HandleMmio()

[66401.052] core/device_manager.cc:502 HandleMmio() error: unhandled mmio read base: 0x00000000ffc00fff size: 1 data: 0000000000000000 errno=4, Interrupted system call

full log https://gist.github.com/Nimrodium/fdeb0b3fbfebdbc139a54d20829f1ff3

i can confirm i can fully install when using the provided bios_256k.bin firmware
when trying to install vgpu drivers it also does hang then i tried ^C'ing the batch file and it BSOD'd and then rebooted.

there are no special messages in the console when it BSODS, just

[823990.276] core/device_manager.cc:485 HandleMmio() debug: pio count=0  mmio count=441
[824991.463] core/device_manager.cc:485 HandleMmio() debug: pio count=0  mmio count=201
[826003.995] core/device_manager.cc:485 HandleMmio() debug: pio count=42  mmio count=442
[827018.385] core/device_manager.cc:485 HandleMmio() debug: pio count=6  mmio count=1198
[827403.692] core/memory_manager.cc:317 Unmap() debug: unmap region PCI BAR MMIO gpa=0xfebf6000 size=0x1000 type=Device
[827403.783] core/memory_manager.cc:317 Unmap() debug: unmap region PCI BAR MMIO gpa=0xfebec000 size=0x4000 type=Device
[827403.934] core/memory_manager.cc:302 Map() debug: map region PCI BAR MMIO gpa=0xfebf6000 size=0x1000 type=Device
[827403.981] core/memory_manager.cc:302 Map() debug: map region PCI BAR MMIO gpa=0xfebec000 size=0x4000 type=Device

which it had been spamming throughout running. so i don't think its anything

[nooodles2023]

The driver I have built it using WDK is only support for win10, it may cause crash on other windows os.
You can try to resize the system memory to 16G and tested it if the crash was called by 2G dedicated and continuous physical memory.

I'm planning to rebuild mvisor-win-vgpu-driver by using IVSHMEM to deliver virgl commands instead of windows device driver, that could be easy to test any opengl cases and get rid of WDK.

[Nimrodium]

with system memory 16G and virtio-vgpu memory at 1G it doesnt BSOD during install, it just says

Adding driver package: vgpu.inf
Driver package added successfully. (Already exists in the system)
The operation completed successfully.
...
no errors. however on reboot, initially it said please wait before the lock screen, but it meant nothing, no blur on the lock screen, and device manager says basic display adapter

[nooodles2023]

with system memory 16G and virtio-vgpu memory at 1G it doesnt BSOD during install, it just says

Adding driver package: vgpu.inf Driver package added successfully. (Already exists in the system) The operation completed successfully. ... no errors. however on reboot, initially it said please wait before the lock screen, but it meant nothing, no blur on the lock screen, and device manager says basic display adapter

Yes, it's a system device, not a dispaly adapter. It only helps to deliver virgl commands from guest vm to host

[Nimrodium]

Ah I see, yeah I see it under system devices, but it of course doesn't seem to work still, since it's just a system device, does it still need the Red Hat Virtio driver? I installed it and it still doesn't work, I imagine there isn't much I can do from here to get it to work on win11, I'll just wait for the rebuild with IVSHMEM unless there's anything else I should try