Re: [libvirt] [Qemu-devel] live snapshot wiki updated
On 07/19/11 16:24, Eric Blake wrote:
[adding the libvir-list]
On 07/19/2011 08:09 AM, Jes Sorensen wrote:
> Urgh, libvirt parsing image files is really unfortunate, it really
> doesn't give me warm fuzzy feelings :( libvirt really should not know
> about internals of image formats.
But even if you add new features to qemu to avoid needing this in the
future, it doesn't change the past - libvirt will always have to know
how to parse image files understood by older qemu, and so as long as
libvirt already knows how to do that parsing, we might as well take
advantage of it.
What has been done here in the past is plain wrong. Continuing to do it
isn't the right thing to do here.
Besides, I feel that having a well-documented file format, so that
independent applications can both parse the same file with the same
semantics by obeying the file format specification, is a good design goal.
We all know that documentation is rarely uptodate, new features may not
get added and libvirt will never be able to keep up. The driver for a
file format belongs in QEMU and nowhere else.
>> It would be nice if libvirt had a way to pass fds for every
disk and
>> backing file up front; then, SELinux can work around the lack of NFS
>> per-file labelling by blocking open() in qemu. In fact, this has
>> already been proposed:
>
> A cleaner solution seems to have libvirt provide a call-back allowing
> QEMU to call out and have libvirt open a file descriptor instead. This
> way libvirt can validate it and open it for QEMU and pass it back.
Yes, that could probably be made to work with libvirt.
I am a little frustrated this approach wasn't taken up front instead of
the evil hack of having libvirt attempt to parse image files.
> If we cannot do something like this, I would prefer to have
backing
> files on NFS should simply not be supported when running in an selinux
> setup.
As nice as that sentiment is, it will never fly, because it would be a
regression in current behavior. The whole reason that the virt_use_nfs
SELinux bool exists is that some people are willing to make the partial
security tradeoff. Besides, the use of sVirt via SELinux is more than
just open() protection - while the current virt_use_nfs bool makes NFS
less secure than otherwise possible, it still gives some nice guarantees
to the rest of the qemu process such as passthrough accesses to local
pci devices.
Well leaving things at status quo is not making it worse, it just leaves
an evil in place.
Just because it is currently not as secure to mix NFS shared storage
with backing files doesn't stop some people from wanting to do it [in
fact, that's my current development setup - I use qcow2 images on NFS
shared storage, keep SELinux enabled, and enable the virt_use_nfs bool].
This discussion is about adding enhancements that make SELinux even
more powerful when using NFS shared storage, by adding fd passing
(whether libvirt parses in advance, or whether qemu raises an event and
requires feedback from libvirt), and not about crippling the existing
capability to use the virt_use_nfs selinux bool.
I do not believe we should try and add extra interfaces to support
something which is inherently broken. This really boils down to whether
we should support fd passing for snapshots in the first place. If it is
to support the broken setup of libvirt parsing image files, then I am
totally against it, if we work on a proper solution that involves this
in some way, then we can discuss it.
Cheers,
Jes