[Virtio-fs] [virtiofsd] Issue opened: virtiofsd with writeback sometimes ends up with stale handles

[Virtio-fs] [virtiofsd] Issue opened: virtiofsd with writeback sometimes ends up with stale handles

[virtiofs-bot]

I am no specialist in virtiofsd, but here's the case:

I'm running virtiofsd with `--writeback` mode. On guest, I am using a Debian system with it's root partition actually being on a virtiofsd device.

When running `update-initramfs -u` I'm getting this error:

```sync: error syncing '/boot/initrd.img-5.17.0-1-amd64': Stale file handle```

As a side note, the /boot folder is not on a different filesystem, using the same virtiofsd device.
When removing `--writeback` from virtiofsd - it all starts to work.

Full virtiofsd command line parameters:

```/usr/bin/virtiofsd --writeback --socket-path=/<snip>/vfsd.sock --shared-dir /<snip>/fs --announce-submounts --inode-file-handles=fallback --seccomp=log --sandbox=namespace --socket-group kvm-instances```

Full qemu command line parameters (this one is huge though):

```/usr/bin/qemu-system-x86_64 -vnc none -enable-kvm -name <snip> -smp cores=4,threads=1,sockets=1,maxcpus=4 -cpu host -device virtio-balloon,deflate-on-oom=on,free-page-reporting=on,guest-stats-polling-interval=10,id=balloon0 -object memory-backend-memfd,id=ram0,size=8G,share=on -machine memory-backend=ram0 -m 8G -chardev socket,id=char0,path=/<snip>/vfsd.sock -device vhost-user-fs-pci,chardev=char0,tag=rootfs,cache-size=8G -kernel /<snip>/fs/vmlinuz -initrd /<snip>/fs/initrd.img -append root=rootfs rootfstype=virtiofs rw quiet -netdev type=bridge,id=virt0,br=virt0 -device virtio-net-pci,netdev=virt0,host_mtu=9000,mac=52:54:01:00:00:01 -monitor unix:/<snip>/monitor.sock,server=on,wait=off```

The qemu is the one built from https://gitlab.com/virtio-fs/qemu the branch virtio-fs-dev
The virtiofsd was built from the git source code, main branch.
---
https://gitlab.com/virtio-fs/virtiofsd/-/issues/47