On Fri, 28 Sep 2018, Jann Horn wrote:
quoted
so with this hard-coded logic, you are saying this case is
'safe' in a sidechannel context.
Which hints at the deeper issue that containers are a userland
abstraction. Protection of containers needs to be defined by userland
policy.
Or just compare mount namespaces additionally/instead. I think that
containers will always use those, because AFAIK nobody uses chroot()
for containers, given that the kernel makes absolutely no security
guarantees about chroot().
We can't define this in the kernel. It has no concept of containers.
People utilize some combination of namespaces and cgroups and call them
containers, but we can't make assumptions from the kernel on what any of
this means from a security point of view, and hard-code kernel policy
based on those assumptions.
This is violating the principal of separating mechanism and policy, and
also imposing semantics across the kernel/user boundary. The latter
creates an ABI which we can then never break.
--
James Morris
[off-list ref]