[REVIEW][PATCH 1/2] userns: Better restrictions on when proc and sysfs can be mounted

[Eric W. Biederman]

Gao feng <gaofeng at cn.fujitsu.com> writes:

> Ok,I agree with you that we should make container security by default.
>
> What's your idea that introduces option MS_NOT_A_LOCK just like Andy's
> advisement?

It might be doable but it is unnecessary.

> In libvirt, host creates dev and devpts directories for container,then
> mount devpts, tmpfs on them and create device nodes inside these dirs
> for container. and then in container, these filesystems are moved to
> container's /dev/ /dev/pts directory. We really have no need to lock
> these mounts. they are just created for container.

If the global root creates the namespace and performs all of the mounts
it is unnecessary.

Now I believe you can create those directories for the most part as
non-root in libvirt and gain some interesting applications.

That said if you don't want locked mounts you just just be able to
create a temporary mount namespace as the global root, and do your prep
work.

Then create your unprivileged mount namespace and bind mount the
directories where you want them, and then pivot_root away the bits you
don't want.

There is already more mechanism than I like to deal with the mount
namespace I would really rather not invent/debug/support any more.

Eric