[Qemu-devel] virtfs error reporting?

[Qemu-devel] virtfs error reporting?

[Rob Landley]

So on the host side I'm trying to do this:

$ qemu -cpu pentium3 -nographic -no-reboot -kernel bzImage \
  -hda hda.sqf -append 'root=/dev/hda rw init=/sbin/init.sh panic=1 \
  PATH=/bin:/sbin console=ttyS0 HOST=i686 ' -net nic,model=e1000 \
  -net user -virtfs local,path=../root-filesystem-i686,security_model=mapped,mount_tag=root

And on the guest side I'm trying to do this:

  mount -t 9p -o ro,trans=virtio,version=9p2000.L root /mnt

And it's saying:

  mount: mounting root on /mnt failed: No such device

No idea what I'm doing wrong, "no such device" is not an informative
error message.

There's no obvious way to get the Linux client to tell me what tags
are available.  "What can I mount?  Do you see anything?"  I've
looked under /proc and /sys for "virtio"...  It's entirely possible
I'm missing something in my kernel .config but who knows what?

I note that qemu was sitting down and SULKING unless I specified the
security_model (I'm mounting read only, I don't care, but there
doesn't seem to be any way to DO a read-only export).  I'm guessing
based on the fact it stopped complaining that it is in fact exporting
something (and thus not upset I gave it a relative path, although giving
it an absolute path made no difference...)

Any clues?

Rob

Re: [Qemu-devel] virtfs error reporting?

[M. Mohan Kumar]

[-- Attachment #1: Type: text/plain, Size: 1489 bytes --]

On Monday 25 April 2011 1:01:38 am Rob Landley wrote:
> So on the host side I'm trying to do this:
> 
> $ qemu -cpu pentium3 -nographic -no-reboot -kernel bzImage \
>   -hda hda.sqf -append 'root=/dev/hda rw init=/sbin/init.sh panic=1 \
>   PATH=/bin:/sbin console=ttyS0 HOST=i686 ' -net nic,model=e1000 \
>   -net user -virtfs
> local,path=../root-filesystem-i686,security_model=mapped,mount_tag=root
> 
> And on the guest side I'm trying to do this:
> 
>   mount -t 9p -o ro,trans=virtio,version=9p2000.L root /mnt
> 
> And it's saying:
> 
>   mount: mounting root on /mnt failed: No such device
> 
> No idea what I'm doing wrong, "no such device" is not an informative
> error message.
> 
> There's no obvious way to get the Linux client to tell me what tags
> are available.  "What can I mount?  Do you see anything?"  I've
> looked under /proc and /sys for "virtio"...  It's entirely possible
> I'm missing something in my kernel .config but who knows what?

You can get available tags from 
/sys/devices/virtio-pci/virtio*/mount_tag


> 
> I note that qemu was sitting down and SULKING unless I specified the
> security_model (I'm mounting read only, I don't care, but there
> doesn't seem to be any way to DO a read-only export).  I'm guessing
> based on the fact it stopped complaining that it is in fact exporting
> something (and thus not upset I gave it a relative path, although giving
> it an absolute path made no difference...)
> 
> Any clues?
> 
> Rob

----
M. Mohan Kumar

[-- Attachment #2: Type: text/html, Size: 8106 bytes --]

Re: [Qemu-devel] virtfs error reporting?

[Eric Van Hensbergen]

On Sun, Apr 24, 2011 at 2:31 PM, Rob Landley <rlandley@parallels.com> wrote:
> So on the host side I'm trying to do this:
>
> $ qemu -cpu pentium3 -nographic -no-reboot -kernel bzImage \
>  -hda hda.sqf -append 'root=/dev/hda rw init=/sbin/init.sh panic=1 \
>  PATH=/bin:/sbin console=ttyS0 HOST=i686 ' -net nic,model=e1000 \
>  -net user -virtfs local,path=../root-filesystem-i686,security_model=mapped,mount_tag=root
>
> And on the guest side I'm trying to do this:
>
>  mount -t 9p -o ro,trans=virtio,version=9p2000.L root /mnt
>
> And it's saying:
>
>  mount: mounting root on /mnt failed: No such device
>

Not that I expect this to be the problem (or if it is, I'm not sure of
a good reason for it) -- but are you sure this isn't a different case
of the same problem you had with a TCP server?

In other words, put the root /mnt before the -o:

        mount -t 9p root /mnt -o ro,trans=virtio,version=9p2000.L

        -eric