On Thu, Jul 23, 2020 at 09:47:33AM -0400, Vivek Goyal wrote:
> On Thu, Jul 23, 2020 at 01:28:50PM +0100, Stefan Hajnoczi wrote:
> > On Wed, Jul 22, 2020 at 06:58:20PM +0100, Dr. David Alan Gilbert wrote:
> > > * Stefan Hajnoczi (stefanha@redhat.com) wrote:
> > > > virtiofsd cannot run in an unprivileged container because CAP_SYS_ADMIN
> > > > is required to create namespaces.
> > > > 
> > > > Introduce a weaker sandbox that is sufficient in container environments
> > > > because the container runtime already sets up namespaces. Use chroot to
> > > > restrict path traversal to the shared directory.
> > > > 
> > > > virtiofsd loses the following:
> > > > 
> > > > 1. Mount namespace. The process chroots to the shared directory but
> > > >    leaves the mounts in place. Seccomp rejects mount(2)/umount(2)
> > > >    syscalls.
> > > 
> > > OK, I'm guessing the behaviour of what happens if the host adds another
> > > mount afterwards might be different?
> > 
> > Running inside a container with -o chroot:
> >  - The container has its own mount namespace and it is therefore not
> >    affected, modulo shared subtrees (see mount(8)).
> 
> How does virtiofsd inside container gets the directory it wants to
> export to guest? I am assuming that its probably a volume mount
> inside container. If yes, volume mounts can set propagation
> properties say "slave" and that means mounts done later will
> propagate inside container in that volume (and hopefully be
> visible inside guest once submount patches are upstream).

Yes, the shared directory is a volume mount and propagation works as
expected:

  $ docker run -v path/to/shared-dir:/shared-dir:slave \
	       -v /tmp/socket-dir:/var/run \
	       virtiofsd

When I mount things in path/to/shared-dir on the host they appear in
/shared-dir inside the guest. When I omit the "slave" option (the
default is "private") then the mount does not propagate.

The following Dockerfile was used:

  $ cp ~/src/qemu/virtiofsd . # copy the executable
  $ cat >Dockerfile
  FROM fedora:32
  COPY --chown=0:0 virtiofsd /usr/bin/virtiofsd
  RUN dnf update -qy && dnf install -qy libseccomp liburing zlib glib2 nettle gnutls libcap-ng
  VOLUME ["/shared-dir", "/var/run"]
  CMD ["/usr/bin/virtiofsd", "--socket-path=/var/run/virtiofsd.sock", "-o", "source=/shared-dir", "-o", "cache=auto", "-o", "no_posix_lock", "-o", "xattr", "-o", "chroot"]

Stefan