kvm-guest-drivers-windows: Libvirt - SeekEnd(Os { code: 29, kind: NotSeekable, message: "Illegal seek" })

When used with libvirt, the new rust daemon crashes with this message:

virtiofsd[10356]: Use of deprecated option format '-o': Please specify options without it (e.g., '--cache auto' instead of '-o cache=auto')
virtiofsd[10356]: Waiting for vhost-user socket connection...
virtiofsd[10356]: Client connected, servicing requests
virtiofsd[10356]: Waiting for daemon failed: HandleRequest(ReqHandlerError(Custom { kind: Other, error: SeekEnd(Os { code: 29, kind: NotSeekable, message: "Illegal seek" }) }))

The config file used:

<filesystem type="mount" accessmode="passthrough">
  <driver type="virtiofs" queue="1024"/>
  <binary path="/home/launch_virtiofs">
    <thread_pool size="1"/>
  </binary>
  <source dir="/home/Windows"/>
  <target dir="Exchange"/>
  <address type="pci" domain="0x0000" bus="0x0a" slot="0x00" function="0x0"/>
</filesystem>

The launch_virtiofs script:

#!/bin/bash
/usr/lib/virtiofsd --syslog --log-level debug "$@"

Host operating system: EndeavourOS, Kernel: 6.2.12-arch1-1, Guest operating system: Windows 11, ISO installer virtio-win-0.1.229.iso was used in the guest OS, virtiofsd 1.6 QEMU 8.0.0

After raising the issue on gitlab of virtiofsd, the devs there said I should raise it here

About this issue

  • Original URL
  • State: closed
  • Created a year ago
  • Comments: 27 (4 by maintainers)

Most upvoted comments

Because @gnif is not willing to even communicate normally I opened an issue at qemus gitlab: https://gitlab.com/qemu-project/qemu/-/issues/1877

Hopefully someone there is willing to help normal users (and dont just brush them rudely off) who even lack the skills to debug this problem between looking-glass and virtiofs.

This is an UPSTREAM issue with SOMEONE ELSE’S CODE! You expect me to dig into a bug triggered upstream. Yup, so rude of me to refer you to the correct people to report this issue.

This is an UPSTREAM issue with SOMEONE ELSE’S CODE! You expect me to dig into a bug triggered upstream. Yup, so rude of me to refer you to the correct people to report this issue.

Hm, more yelling in Capslock and new excuses. Really nice, now i really regret sponsoring you in the past

Yelling in capslock… lol, capitalization can be used for emphasis too you know.

I don’t care what you think, I have made it clear that this is an upstream issue with QEMU, and simply referred you to report it to them. Clearly you have zero idea of how software development works and have little idea that debugging this without knowing the code base in QEMU (aka, blind), is MANY hours of work, The maintainer of the code that is crashing vritiofs needs to investigate why there is crash when combined with the QEMU (not Looking Glass) IVSHMEM device.

new excuses

What? no new excuses, this is not a bug in LG, or caused by LG, it’s a bug in QEMU that this particular configuration happens to trigger. Correlation does not equal causation.

now i really regret sponsoring you in the past

So because you sponsored me you expect me to be at your beckoning whim to debug code written by someone else in a project I do not maintain? 🤦

Edit: Umm, yelling in caps? bit of the pot calling the kettle black don’t you think. 2023-09-11-184751_915x221_scrot https://github.com/gnif/LookingGlass/issues/1089#issuecomment-1713259976

@viktor-prutyanov will look at this next week