Re: [Virtio-fs] xfstest results for virtio-fs on aarch64

["Dr. David Alan Gilbert" <dgilbert@redhat.com>]

* qi.fuli@fujitsu.com (qi.fuli@fujitsu.com) wrote:
> Hello,
> 
> We have run the generic tests of xfstest for virtio-fs[1] on aarch64[2],
> here we selected some tests that did not run or failed to run,
> and we categorized them, basing on the reasons in our understanding.

Hi,
  Apologies for taking a while to respond.
What I've done is to repeat these and create gitlab issues for each
categroy that you defined.  Now we can go and look at these
inidividually.

>   * Category 1: generic/003, generic/192
>     Error: access time error
>     Reason: file_accessed() not run

https://gitlab.com/virtio-fs/qemu/issues/8

>   * Category 2: generic/089, generic/478, generic/484, generic/504
>     Error: lock error

https://gitlab.com/virtio-fs/qemu/issues/9

>   * Category 3: generic/426, generic/467, generic/477
>     Error: open_by_handle error

https://gitlab.com/virtio-fs/qemu/issues/10

>   * Category 4: generic/551
>     Error: kvm panic

This for me just caused an OOM; when I increased the guest size to 32G
it was fine; can you retest?  I'm not sure how much RAM xfstests
expects.

>   * Category 5: generic/011, generic/013
>     Error: cannot remove file
>     Reason: NFS backend

https://gitlab.com/virtio-fs/qemu/issues/11

>   * Category 6: generic/035
>     Error: nlink is 1, should be 0

https://gitlab.com/virtio-fs/qemu/issues/12

>   * Category 7: generic/125, generic/193, generic/314
>     Error: open/chown/mkdir permission error

There were OK for me after I'd created the users/groups that
the xfstests manual asked for (fsgqa, and 123456-fsgqa) - can
you confirm the error you saw.

>   * Category 8: generic/469
>     Error: fallocate keep_size is needed

https://gitlab.com/virtio-fs/qemu/issues/13

>     Reason: NFS4.0 backend
>   * Category 9: generic/323
>     Error: system hang
>     Reason: fd is close before AIO finished

https://gitlab.com/virtio-fs/qemu/issues/14

I reproduced all these on x86, so they're not aarch
specific.

> We would like to know if virtio-fs does not support these tests in
> the specification or they are bugs that need to be fixed.
> It would be very appreciated if anyone could give some comments.

Now I've got them reproduced easily and also on x86, we'll
dig into them and figure out what's going on.

Dave

> [1] qemu: https://gitlab.com/virtio-fs/qemu/tree/virtio-fs-dev
>      start qemu script:
>      $VIRTIOFSD -o vhost_user_socket=/tmp/vhostqemu1 -o 
> source=/root/virtio-fs/test1/ -o cache=always -o xattr -o flock -d &
>      $VIRTIOFSD -o vhost_user_socket=/tmp/vhostqemu2 -o 
> source=/root/virtio-fs/test2/ -o cache=always -o xattr -o flock -d &
>      $QEMU -M virt,accel=kvm,gic_version=3 \
>          -cpu host \
>          -smp 8 \
>          -m 8192\
>          -nographic \
>          -serial mon:stdio \
>          -netdev tap,id=net0 -device 
> virtio-net-pci,netdev=net0,id=net0,mac=XX:XX:XX:XX:XX:XX \
>          -object 
> memory-backend-file,id=mem,size=8G,mem-path=/dev/shm,share=on \
>          -numa node,memdev=mem \
>          -drive 
> file=/root/virtio-fs/AAVMF/AAVMF_CODE.fd,if=pflash,format=raw,unit=0,readonly=on 
> \
>          -drive file=$VARS,if=pflash,format=raw,unit=1 \
>          -chardev socket,id=char1,path=/tmp/vhostqemu1 \
>          -device 
> vhost-user-fs-pci,queue-size=1024,chardev=char1,tag=myfs1,cache-size=0 \
>          -chardev socket,id=char2,path=/tmp/vhostqemu2 \
>          -device 
> vhost-user-fs-pci,queue-size=1024,chardev=char2,tag=myfs2,cache-size=0 \
>          -drive if=virtio,file=/var/lib/libvirt/images/guest.img
> 
> [2] host kernel: 4.18.0-80.4.2.el8_0.aarch64
>      guest kernel: 5.4-rc1
>      Arch: Arm64
>      backend: NFS 4.0
> 
> Thanks,
> QI Fuli
> 
> _______________________________________________
> Virtio-fs mailing list
> Virtio-fs@redhat.com
> https://www.redhat.com/mailman/listinfo/virtio-fs
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK